diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,18253 +1,66927 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.0,
-  "global_step": 3038,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 9555,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.0,
-      "learning_rate": 2.173913043478261e-07,
-      "loss": 3.3606,
+      "epoch": 0.00010465724751439037,
+      "grad_norm": 62.41193288917299,
+      "learning_rate": 6.968641114982578e-08,
+      "loss": 2.0951,
       "step": 1
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 4.347826086956522e-07,
-      "loss": 3.1084,
+      "epoch": 0.00020931449502878074,
+      "grad_norm": 62.41976397338786,
+      "learning_rate": 1.3937282229965157e-07,
+      "loss": 1.9349,
       "step": 2
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 6.521739130434783e-07,
-      "loss": 2.9241,
+      "epoch": 0.0003139717425431711,
+      "grad_norm": 86.35511116310265,
+      "learning_rate": 2.090592334494774e-07,
+      "loss": 2.2482,
       "step": 3
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 8.695652173913044e-07,
-      "loss": 2.7249,
+      "epoch": 0.0004186289900575615,
+      "grad_norm": 66.12787272745858,
+      "learning_rate": 2.7874564459930313e-07,
+      "loss": 1.9908,
       "step": 4
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 1.0869565217391306e-06,
-      "loss": 3.0549,
+      "epoch": 0.0005232862375719519,
+      "grad_norm": 60.71641111176538,
+      "learning_rate": 3.4843205574912896e-07,
+      "loss": 1.9762,
       "step": 5
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 1.3043478260869566e-06,
-      "loss": 3.4679,
+      "epoch": 0.0006279434850863422,
+      "grad_norm": 56.78291196095233,
+      "learning_rate": 4.181184668989548e-07,
+      "loss": 1.949,
       "step": 6
     },
     {
-      "epoch": 0.0,
-      "learning_rate": 1.521739130434783e-06,
-      "loss": 2.6214,
+      "epoch": 0.0007326007326007326,
+      "grad_norm": 38.93880117457696,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 1.9651,
       "step": 7
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.7391304347826088e-06,
-      "loss": 3.1068,
+      "epoch": 0.000837257980115123,
+      "grad_norm": 52.09622789077548,
+      "learning_rate": 5.574912891986063e-07,
+      "loss": 1.8705,
       "step": 8
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 1.956521739130435e-06,
-      "loss": 2.6991,
+      "epoch": 0.0009419152276295133,
+      "grad_norm": 49.92480355404063,
+      "learning_rate": 6.271777003484321e-07,
+      "loss": 1.8799,
       "step": 9
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.173913043478261e-06,
-      "loss": 2.1599,
+      "epoch": 0.0010465724751439038,
+      "grad_norm": 32.20342595275959,
+      "learning_rate": 6.968641114982579e-07,
+      "loss": 1.6128,
       "step": 10
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.391304347826087e-06,
-      "loss": 2.3008,
+      "epoch": 0.001151229722658294,
+      "grad_norm": 53.634110936725754,
+      "learning_rate": 7.665505226480836e-07,
+      "loss": 1.7474,
       "step": 11
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.6086956521739132e-06,
-      "loss": 2.0458,
+      "epoch": 0.0012558869701726845,
+      "grad_norm": 36.607148713405984,
+      "learning_rate": 8.362369337979096e-07,
+      "loss": 1.7438,
       "step": 12
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 2.8260869565217393e-06,
-      "loss": 2.1521,
+      "epoch": 0.0013605442176870747,
+      "grad_norm": 3.9957477431354564,
+      "learning_rate": 9.059233449477353e-07,
+      "loss": 0.7564,
       "step": 13
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.043478260869566e-06,
-      "loss": 1.8656,
+      "epoch": 0.0014652014652014652,
+      "grad_norm": 20.58298531785394,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 1.5063,
       "step": 14
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.2608695652173914e-06,
-      "loss": 1.9967,
+      "epoch": 0.0015698587127158557,
+      "grad_norm": 15.444768444817306,
+      "learning_rate": 1.045296167247387e-06,
+      "loss": 1.4041,
       "step": 15
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.4782608695652175e-06,
-      "loss": 1.6188,
+      "epoch": 0.001674515960230246,
+      "grad_norm": 12.107828560978803,
+      "learning_rate": 1.1149825783972125e-06,
+      "loss": 1.3931,
       "step": 16
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.6956521739130436e-06,
-      "loss": 1.6884,
+      "epoch": 0.0017791732077446364,
+      "grad_norm": 10.814057953845309,
+      "learning_rate": 1.1846689895470384e-06,
+      "loss": 1.3134,
       "step": 17
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 3.91304347826087e-06,
-      "loss": 1.652,
+      "epoch": 0.0018838304552590266,
+      "grad_norm": 7.249315729621174,
+      "learning_rate": 1.2543554006968642e-06,
+      "loss": 1.3668,
       "step": 18
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.130434782608696e-06,
-      "loss": 1.6944,
+      "epoch": 0.001988487702773417,
+      "grad_norm": 7.144445780657985,
+      "learning_rate": 1.32404181184669e-06,
+      "loss": 1.2966,
       "step": 19
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.347826086956522e-06,
-      "loss": 1.7657,
+      "epoch": 0.0020931449502878076,
+      "grad_norm": 8.323575437600967,
+      "learning_rate": 1.3937282229965158e-06,
+      "loss": 1.2833,
       "step": 20
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.565217391304348e-06,
-      "loss": 1.4748,
+      "epoch": 0.002197802197802198,
+      "grad_norm": 6.584038254790013,
+      "learning_rate": 1.4634146341463414e-06,
+      "loss": 1.3209,
       "step": 21
     },
     {
-      "epoch": 0.01,
-      "learning_rate": 4.782608695652174e-06,
-      "loss": 1.5941,
+      "epoch": 0.002302459445316588,
+      "grad_norm": 7.650578965105668,
+      "learning_rate": 1.5331010452961673e-06,
+      "loss": 1.185,
       "step": 22
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5e-06,
-      "loss": 1.5832,
+      "epoch": 0.0024071166928309787,
+      "grad_norm": 6.43808617250343,
+      "learning_rate": 1.602787456445993e-06,
+      "loss": 1.1619,
       "step": 23
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.2173913043478265e-06,
-      "loss": 1.5005,
+      "epoch": 0.002511773940345369,
+      "grad_norm": 6.450238348514814,
+      "learning_rate": 1.6724738675958191e-06,
+      "loss": 1.1685,
       "step": 24
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.4347826086956525e-06,
-      "loss": 1.6937,
+      "epoch": 0.0026164311878597592,
+      "grad_norm": 6.596585528409829,
+      "learning_rate": 1.742160278745645e-06,
+      "loss": 1.2266,
       "step": 25
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.652173913043479e-06,
-      "loss": 1.747,
+      "epoch": 0.0027210884353741495,
+      "grad_norm": 6.701054827351807,
+      "learning_rate": 1.8118466898954705e-06,
+      "loss": 1.2315,
       "step": 26
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 5.8695652173913055e-06,
-      "loss": 1.7165,
+      "epoch": 0.00282574568288854,
+      "grad_norm": 6.515865205471776,
+      "learning_rate": 1.8815331010452964e-06,
+      "loss": 1.2951,
       "step": 27
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.086956521739132e-06,
-      "loss": 1.572,
+      "epoch": 0.0029304029304029304,
+      "grad_norm": 5.495232722232645,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.1448,
       "step": 28
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.304347826086958e-06,
-      "loss": 1.3604,
+      "epoch": 0.0030350601779173206,
+      "grad_norm": 5.290875061084644,
+      "learning_rate": 2.020905923344948e-06,
+      "loss": 1.1407,
       "step": 29
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.521739130434783e-06,
-      "loss": 1.3899,
+      "epoch": 0.0031397174254317113,
+      "grad_norm": 5.3025777909139045,
+      "learning_rate": 2.090592334494774e-06,
+      "loss": 1.127,
       "step": 30
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.739130434782609e-06,
-      "loss": 1.3702,
+      "epoch": 0.0032443746729461016,
+      "grad_norm": 5.245351898269604,
+      "learning_rate": 2.1602787456445995e-06,
+      "loss": 1.1103,
       "step": 31
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 6.956521739130435e-06,
-      "loss": 1.5494,
+      "epoch": 0.003349031920460492,
+      "grad_norm": 5.300255475950051,
+      "learning_rate": 2.229965156794425e-06,
+      "loss": 1.1917,
       "step": 32
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.173913043478261e-06,
-      "loss": 1.474,
+      "epoch": 0.003453689167974882,
+      "grad_norm": 4.572317456714689,
+      "learning_rate": 2.299651567944251e-06,
+      "loss": 1.0874,
       "step": 33
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.391304347826087e-06,
-      "loss": 1.4282,
+      "epoch": 0.0035583464154892728,
+      "grad_norm": 4.107821899876536,
+      "learning_rate": 2.3693379790940767e-06,
+      "loss": 1.0825,
       "step": 34
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.608695652173914e-06,
-      "loss": 1.4724,
+      "epoch": 0.003663003663003663,
+      "grad_norm": 4.21060285666979,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 1.2219,
       "step": 35
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 7.82608695652174e-06,
-      "loss": 1.45,
+      "epoch": 0.0037676609105180532,
+      "grad_norm": 4.328566900505888,
+      "learning_rate": 2.5087108013937284e-06,
+      "loss": 1.0502,
       "step": 36
     },
     {
-      "epoch": 0.02,
-      "learning_rate": 8.043478260869566e-06,
-      "loss": 1.3026,
+      "epoch": 0.003872318158032444,
+      "grad_norm": 4.291665457341644,
+      "learning_rate": 2.578397212543554e-06,
+      "loss": 1.1426,
       "step": 37
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 8.260869565217392e-06,
-      "loss": 1.5396,
+      "epoch": 0.003976975405546834,
+      "grad_norm": 3.8850933278156563,
+      "learning_rate": 2.64808362369338e-06,
+      "loss": 1.0759,
       "step": 38
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 8.478260869565218e-06,
-      "loss": 1.4549,
+      "epoch": 0.004081632653061225,
+      "grad_norm": 4.413531892474245,
+      "learning_rate": 2.7177700348432056e-06,
+      "loss": 1.1848,
       "step": 39
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 8.695652173913044e-06,
-      "loss": 1.5142,
+      "epoch": 0.004186289900575615,
+      "grad_norm": 4.844856614218897,
+      "learning_rate": 2.7874564459930316e-06,
+      "loss": 1.0984,
       "step": 40
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 8.91304347826087e-06,
-      "loss": 1.5001,
+      "epoch": 0.004290947148090005,
+      "grad_norm": 4.748816166519876,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.0605,
       "step": 41
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 9.130434782608697e-06,
-      "loss": 1.4477,
+      "epoch": 0.004395604395604396,
+      "grad_norm": 4.777444061979255,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 1.2714,
       "step": 42
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 9.347826086956523e-06,
-      "loss": 1.2861,
+      "epoch": 0.004500261643118786,
+      "grad_norm": 4.522554555361124,
+      "learning_rate": 2.996515679442509e-06,
+      "loss": 1.0385,
       "step": 43
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 9.565217391304349e-06,
-      "loss": 1.3809,
+      "epoch": 0.004604918890633176,
+      "grad_norm": 4.075973193199012,
+      "learning_rate": 3.0662020905923345e-06,
+      "loss": 1.0521,
       "step": 44
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 9.782608695652175e-06,
-      "loss": 1.6367,
+      "epoch": 0.004709576138147566,
+      "grad_norm": 4.3231014366459934,
+      "learning_rate": 3.13588850174216e-06,
+      "loss": 1.0546,
       "step": 45
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1e-05,
-      "loss": 1.4074,
+      "epoch": 0.0048142333856619575,
+      "grad_norm": 5.680111667722333,
+      "learning_rate": 3.205574912891986e-06,
+      "loss": 1.0781,
       "step": 46
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.0217391304347829e-05,
-      "loss": 1.4236,
+      "epoch": 0.004918890633176348,
+      "grad_norm": 4.1612873974328854,
+      "learning_rate": 3.2752613240418118e-06,
+      "loss": 1.027,
       "step": 47
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.0434782608695653e-05,
-      "loss": 1.6102,
+      "epoch": 0.005023547880690738,
+      "grad_norm": 4.992758604601489,
+      "learning_rate": 3.3449477351916382e-06,
+      "loss": 1.1227,
       "step": 48
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.0652173913043479e-05,
-      "loss": 1.4612,
+      "epoch": 0.005128205128205128,
+      "grad_norm": 4.226754919676679,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 1.1298,
       "step": 49
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.0869565217391305e-05,
-      "loss": 1.3934,
+      "epoch": 0.0052328623757195184,
+      "grad_norm": 5.514671205059567,
+      "learning_rate": 3.48432055749129e-06,
+      "loss": 1.1269,
       "step": 50
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.1086956521739131e-05,
-      "loss": 1.4294,
+      "epoch": 0.005337519623233909,
+      "grad_norm": 4.351259557452271,
+      "learning_rate": 3.5540069686411155e-06,
+      "loss": 1.0444,
       "step": 51
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.1304347826086957e-05,
-      "loss": 1.5092,
+      "epoch": 0.005442176870748299,
+      "grad_norm": 5.399366279585442,
+      "learning_rate": 3.623693379790941e-06,
+      "loss": 1.1222,
       "step": 52
     },
     {
-      "epoch": 0.03,
-      "learning_rate": 1.1521739130434783e-05,
-      "loss": 1.2208,
+      "epoch": 0.00554683411826269,
+      "grad_norm": 4.028877965786529,
+      "learning_rate": 3.693379790940767e-06,
+      "loss": 1.1525,
       "step": 53
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.1739130434782611e-05,
-      "loss": 1.2975,
+      "epoch": 0.00565149136577708,
+      "grad_norm": 4.683154305540764,
+      "learning_rate": 3.7630662020905927e-06,
+      "loss": 0.9988,
       "step": 54
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.1956521739130435e-05,
-      "loss": 1.4067,
+      "epoch": 0.0057561486132914706,
+      "grad_norm": 3.937666148096688,
+      "learning_rate": 3.832752613240418e-06,
+      "loss": 1.0655,
       "step": 55
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.2173913043478263e-05,
-      "loss": 1.3488,
+      "epoch": 0.005860805860805861,
+      "grad_norm": 4.766869445012149,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.7926,
       "step": 56
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.2391304347826088e-05,
-      "loss": 1.3825,
+      "epoch": 0.005965463108320251,
+      "grad_norm": 4.319598323365691,
+      "learning_rate": 3.97212543554007e-06,
+      "loss": 1.0314,
       "step": 57
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.2608695652173915e-05,
-      "loss": 1.3548,
+      "epoch": 0.006070120355834641,
+      "grad_norm": 4.576774250115059,
+      "learning_rate": 4.041811846689896e-06,
+      "loss": 1.1618,
       "step": 58
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.282608695652174e-05,
-      "loss": 1.4098,
+      "epoch": 0.0061747776033490315,
+      "grad_norm": 5.167500660632476,
+      "learning_rate": 4.111498257839722e-06,
+      "loss": 1.0677,
       "step": 59
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.3043478260869566e-05,
-      "loss": 1.5195,
+      "epoch": 0.006279434850863423,
+      "grad_norm": 4.38060785735574,
+      "learning_rate": 4.181184668989548e-06,
+      "loss": 1.0157,
       "step": 60
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.3260869565217392e-05,
-      "loss": 1.4409,
+      "epoch": 0.006384092098377813,
+      "grad_norm": 4.893113671421494,
+      "learning_rate": 4.250871080139373e-06,
+      "loss": 1.1794,
       "step": 61
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.3478260869565218e-05,
-      "loss": 1.5297,
+      "epoch": 0.006488749345892203,
+      "grad_norm": 4.849041572456957,
+      "learning_rate": 4.320557491289199e-06,
+      "loss": 1.046,
       "step": 62
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.3695652173913046e-05,
-      "loss": 1.5921,
+      "epoch": 0.006593406593406593,
+      "grad_norm": 3.991277540603915,
+      "learning_rate": 4.390243902439025e-06,
+      "loss": 1.0056,
       "step": 63
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.391304347826087e-05,
-      "loss": 1.2743,
+      "epoch": 0.006698063840920984,
+      "grad_norm": 4.542337282469629,
+      "learning_rate": 4.45993031358885e-06,
+      "loss": 1.0251,
       "step": 64
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.4130434782608698e-05,
-      "loss": 1.3369,
+      "epoch": 0.006802721088435374,
+      "grad_norm": 4.28521583781609,
+      "learning_rate": 4.529616724738676e-06,
+      "loss": 1.0708,
       "step": 65
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.4347826086956522e-05,
-      "loss": 1.3681,
+      "epoch": 0.006907378335949764,
+      "grad_norm": 4.125637623705243,
+      "learning_rate": 4.599303135888502e-06,
+      "loss": 1.0472,
       "step": 66
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.456521739130435e-05,
-      "loss": 1.4715,
+      "epoch": 0.007012035583464155,
+      "grad_norm": 5.268122050018008,
+      "learning_rate": 4.668989547038328e-06,
+      "loss": 0.99,
       "step": 67
     },
     {
-      "epoch": 0.04,
-      "learning_rate": 1.4782608695652174e-05,
-      "loss": 1.4458,
+      "epoch": 0.0071166928309785455,
+      "grad_norm": 5.0648320868276375,
+      "learning_rate": 4.738675958188153e-06,
+      "loss": 1.0457,
       "step": 68
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.5000000000000002e-05,
-      "loss": 1.3767,
+      "epoch": 0.007221350078492936,
+      "grad_norm": 3.948886008013687,
+      "learning_rate": 4.8083623693379794e-06,
+      "loss": 1.1263,
       "step": 69
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.5217391304347828e-05,
-      "loss": 1.4051,
+      "epoch": 0.007326007326007326,
+      "grad_norm": 4.195666188025583,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 1.0227,
       "step": 70
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.5434782608695654e-05,
-      "loss": 1.5299,
+      "epoch": 0.007430664573521716,
+      "grad_norm": 4.413992487953222,
+      "learning_rate": 4.947735191637631e-06,
+      "loss": 1.0551,
       "step": 71
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.565217391304348e-05,
-      "loss": 1.4234,
+      "epoch": 0.0075353218210361065,
+      "grad_norm": 5.788427063687076,
+      "learning_rate": 5.017421602787457e-06,
+      "loss": 1.094,
       "step": 72
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.5869565217391306e-05,
-      "loss": 1.4651,
+      "epoch": 0.007639979068550497,
+      "grad_norm": 4.474163687473138,
+      "learning_rate": 5.087108013937283e-06,
+      "loss": 1.0967,
       "step": 73
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.6086956521739132e-05,
-      "loss": 1.6008,
+      "epoch": 0.007744636316064888,
+      "grad_norm": 4.512506707897861,
+      "learning_rate": 5.156794425087108e-06,
+      "loss": 1.1004,
       "step": 74
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.630434782608696e-05,
-      "loss": 1.4812,
+      "epoch": 0.007849293563579277,
+      "grad_norm": 4.228651613816608,
+      "learning_rate": 5.226480836236935e-06,
+      "loss": 1.0283,
       "step": 75
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.6521739130434785e-05,
-      "loss": 1.3406,
+      "epoch": 0.007953950811093667,
+      "grad_norm": 4.3737582716641175,
+      "learning_rate": 5.29616724738676e-06,
+      "loss": 0.9979,
       "step": 76
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.673913043478261e-05,
-      "loss": 1.4521,
+      "epoch": 0.00805860805860806,
+      "grad_norm": 3.9566453802387094,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 1.0923,
       "step": 77
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.6956521739130437e-05,
-      "loss": 1.5115,
+      "epoch": 0.00816326530612245,
+      "grad_norm": 4.071859232406034,
+      "learning_rate": 5.435540069686411e-06,
+      "loss": 1.0203,
       "step": 78
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.7173913043478263e-05,
-      "loss": 1.3607,
+      "epoch": 0.00826792255363684,
+      "grad_norm": 4.0563972426988295,
+      "learning_rate": 5.505226480836237e-06,
+      "loss": 1.127,
       "step": 79
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.739130434782609e-05,
-      "loss": 1.5419,
+      "epoch": 0.00837257980115123,
+      "grad_norm": 4.657288632168375,
+      "learning_rate": 5.574912891986063e-06,
+      "loss": 1.1196,
       "step": 80
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.7608695652173915e-05,
-      "loss": 1.4382,
+      "epoch": 0.00847723704866562,
+      "grad_norm": 4.090819887573403,
+      "learning_rate": 5.644599303135889e-06,
+      "loss": 1.0944,
       "step": 81
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.782608695652174e-05,
-      "loss": 1.4131,
+      "epoch": 0.00858189429618001,
+      "grad_norm": 4.482297147920703,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.0254,
       "step": 82
     },
     {
-      "epoch": 0.05,
-      "learning_rate": 1.8043478260869567e-05,
-      "loss": 1.2361,
+      "epoch": 0.008686551543694401,
+      "grad_norm": 4.152492838084387,
+      "learning_rate": 5.7839721254355405e-06,
+      "loss": 1.0602,
       "step": 83
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.8260869565217393e-05,
-      "loss": 1.3589,
+      "epoch": 0.008791208791208791,
+      "grad_norm": 4.327953715994116,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.978,
       "step": 84
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.847826086956522e-05,
-      "loss": 1.511,
+      "epoch": 0.008895866038723181,
+      "grad_norm": 3.0360168594436097,
+      "learning_rate": 5.923344947735193e-06,
+      "loss": 0.7671,
       "step": 85
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.8695652173913045e-05,
-      "loss": 1.4226,
+      "epoch": 0.009000523286237572,
+      "grad_norm": 4.211239355773984,
+      "learning_rate": 5.993031358885018e-06,
+      "loss": 1.0706,
       "step": 86
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.891304347826087e-05,
-      "loss": 1.4287,
+      "epoch": 0.009105180533751962,
+      "grad_norm": 4.660606741232546,
+      "learning_rate": 6.062717770034844e-06,
+      "loss": 1.0697,
       "step": 87
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9130434782608697e-05,
-      "loss": 1.4446,
+      "epoch": 0.009209837781266352,
+      "grad_norm": 3.96480287896892,
+      "learning_rate": 6.132404181184669e-06,
+      "loss": 1.0302,
       "step": 88
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9347826086956523e-05,
-      "loss": 1.3839,
+      "epoch": 0.009314495028780742,
+      "grad_norm": 4.126939118609246,
+      "learning_rate": 6.202090592334495e-06,
+      "loss": 1.1167,
       "step": 89
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.956521739130435e-05,
-      "loss": 1.4631,
+      "epoch": 0.009419152276295133,
+      "grad_norm": 4.87024424753776,
+      "learning_rate": 6.27177700348432e-06,
+      "loss": 0.993,
       "step": 90
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9782608695652176e-05,
-      "loss": 1.4006,
+      "epoch": 0.009523809523809525,
+      "grad_norm": 3.9263349606700695,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 1.0546,
       "step": 91
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 2e-05,
-      "loss": 1.4534,
+      "epoch": 0.009628466771323915,
+      "grad_norm": 3.985792246955264,
+      "learning_rate": 6.411149825783972e-06,
+      "loss": 0.9981,
       "step": 92
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.999999431403446e-05,
-      "loss": 1.5352,
+      "epoch": 0.009733124018838305,
+      "grad_norm": 3.924993510446252,
+      "learning_rate": 6.480836236933798e-06,
+      "loss": 0.9752,
       "step": 93
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9999977256144297e-05,
-      "loss": 1.4683,
+      "epoch": 0.009837781266352695,
+      "grad_norm": 4.292200608183251,
+      "learning_rate": 6.5505226480836235e-06,
+      "loss": 1.0762,
       "step": 94
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9999948826348907e-05,
-      "loss": 1.5199,
+      "epoch": 0.009942438513867086,
+      "grad_norm": 4.485480700913085,
+      "learning_rate": 6.62020905923345e-06,
+      "loss": 1.104,
       "step": 95
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9999909024680632e-05,
-      "loss": 1.398,
+      "epoch": 0.010047095761381476,
+      "grad_norm": 4.5984608130045235,
+      "learning_rate": 6.6898954703832765e-06,
+      "loss": 1.051,
       "step": 96
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9999857851184724e-05,
-      "loss": 1.5483,
+      "epoch": 0.010151753008895866,
+      "grad_norm": 4.449884370761991,
+      "learning_rate": 6.759581881533102e-06,
+      "loss": 0.9836,
       "step": 97
     },
     {
-      "epoch": 0.06,
-      "learning_rate": 1.9999795305919377e-05,
-      "loss": 1.4268,
+      "epoch": 0.010256410256410256,
+      "grad_norm": 4.1157967239658575,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 1.0645,
       "step": 98
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9999721388955722e-05,
-      "loss": 1.2935,
+      "epoch": 0.010361067503924647,
+      "grad_norm": 3.864292249565954,
+      "learning_rate": 6.898954703832753e-06,
+      "loss": 1.0302,
       "step": 99
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9999636100377818e-05,
-      "loss": 1.5527,
+      "epoch": 0.010465724751439037,
+      "grad_norm": 5.087581226627934,
+      "learning_rate": 6.96864111498258e-06,
+      "loss": 0.9327,
       "step": 100
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9999539440282648e-05,
-      "loss": 1.2354,
+      "epoch": 0.010570381998953427,
+      "grad_norm": 4.4876214527409655,
+      "learning_rate": 7.038327526132405e-06,
+      "loss": 1.0256,
       "step": 101
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.999943140878014e-05,
-      "loss": 1.3866,
+      "epoch": 0.010675039246467817,
+      "grad_norm": 4.768444458649774,
+      "learning_rate": 7.108013937282231e-06,
+      "loss": 0.9977,
       "step": 102
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.999931200599314e-05,
-      "loss": 1.4202,
+      "epoch": 0.010779696493982208,
+      "grad_norm": 4.8508061117938075,
+      "learning_rate": 7.177700348432056e-06,
+      "loss": 1.0055,
       "step": 103
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9999181232057437e-05,
-      "loss": 1.3781,
+      "epoch": 0.010884353741496598,
+      "grad_norm": 4.971725250363107,
+      "learning_rate": 7.247386759581882e-06,
+      "loss": 0.9967,
       "step": 104
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9999039087121746e-05,
-      "loss": 1.1009,
+      "epoch": 0.01098901098901099,
+      "grad_norm": 4.397737680675271,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 1.0356,
       "step": 105
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9998885571347712e-05,
-      "loss": 1.3709,
+      "epoch": 0.01109366823652538,
+      "grad_norm": 3.973868161061544,
+      "learning_rate": 7.386759581881534e-06,
+      "loss": 1.0277,
       "step": 106
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9998720684909913e-05,
-      "loss": 1.4178,
+      "epoch": 0.01119832548403977,
+      "grad_norm": 4.846756354395717,
+      "learning_rate": 7.4564459930313594e-06,
+      "loss": 1.113,
       "step": 107
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.999854442799585e-05,
-      "loss": 1.2833,
+      "epoch": 0.01130298273155416,
+      "grad_norm": 4.364102429082418,
+      "learning_rate": 7.5261324041811855e-06,
+      "loss": 1.0448,
       "step": 108
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9998356800805973e-05,
-      "loss": 1.3406,
+      "epoch": 0.01140763997906855,
+      "grad_norm": 4.772968488932661,
+      "learning_rate": 7.595818815331011e-06,
+      "loss": 0.9974,
       "step": 109
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9998157803553638e-05,
-      "loss": 1.4118,
+      "epoch": 0.011512297226582941,
+      "grad_norm": 4.338651147105108,
+      "learning_rate": 7.665505226480837e-06,
+      "loss": 1.0383,
       "step": 110
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9997947436465157e-05,
-      "loss": 1.4097,
+      "epoch": 0.011616954474097331,
+      "grad_norm": 4.665177086985384,
+      "learning_rate": 7.735191637630662e-06,
+      "loss": 0.9858,
       "step": 111
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.9997725699779747e-05,
-      "loss": 1.3415,
+      "epoch": 0.011721611721611722,
+      "grad_norm": 4.226627817628828,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 1.1107,
       "step": 112
     },
     {
-      "epoch": 0.07,
-      "learning_rate": 1.999749259374957e-05,
-      "loss": 1.4323,
+      "epoch": 0.011826268969126112,
+      "grad_norm": 4.365500077240299,
+      "learning_rate": 7.874564459930314e-06,
+      "loss": 0.9337,
       "step": 113
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.999724811863971e-05,
-      "loss": 1.4016,
+      "epoch": 0.011930926216640502,
+      "grad_norm": 4.57360703127662,
+      "learning_rate": 7.94425087108014e-06,
+      "loss": 1.0088,
       "step": 114
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.999699227472818e-05,
-      "loss": 1.4401,
+      "epoch": 0.012035583464154892,
+      "grad_norm": 4.672192335977975,
+      "learning_rate": 8.013937282229966e-06,
+      "loss": 1.0041,
       "step": 115
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9996725062305934e-05,
-      "loss": 1.4507,
+      "epoch": 0.012140240711669283,
+      "grad_norm": 4.678168288512447,
+      "learning_rate": 8.083623693379791e-06,
+      "loss": 1.0472,
       "step": 116
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.999644648167684e-05,
-      "loss": 1.387,
+      "epoch": 0.012244897959183673,
+      "grad_norm": 4.804225775870655,
+      "learning_rate": 8.153310104529616e-06,
+      "loss": 1.0888,
       "step": 117
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9996156533157692e-05,
-      "loss": 1.2825,
+      "epoch": 0.012349555206698063,
+      "grad_norm": 4.403636503472787,
+      "learning_rate": 8.222996515679443e-06,
+      "loss": 1.13,
       "step": 118
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.999585521707822e-05,
-      "loss": 1.2307,
+      "epoch": 0.012454212454212455,
+      "grad_norm": 4.208797982507968,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 1.0292,
       "step": 119
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.999554253378108e-05,
-      "loss": 1.301,
+      "epoch": 0.012558869701726845,
+      "grad_norm": 5.510093081064813,
+      "learning_rate": 8.362369337979095e-06,
+      "loss": 1.0361,
       "step": 120
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9995218483621858e-05,
-      "loss": 1.302,
+      "epoch": 0.012663526949241236,
+      "grad_norm": 5.492522216479893,
+      "learning_rate": 8.43205574912892e-06,
+      "loss": 1.0391,
       "step": 121
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9994883066969055e-05,
-      "loss": 1.342,
+      "epoch": 0.012768184196755626,
+      "grad_norm": 4.227779135360482,
+      "learning_rate": 8.501742160278746e-06,
+      "loss": 1.1114,
       "step": 122
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9994536284204106e-05,
-      "loss": 1.5534,
+      "epoch": 0.012872841444270016,
+      "grad_norm": 3.8028192975551014,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.0596,
       "step": 123
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9994178135721372e-05,
-      "loss": 1.2546,
+      "epoch": 0.012977498691784406,
+      "grad_norm": 4.700182372142366,
+      "learning_rate": 8.641114982578398e-06,
+      "loss": 1.1663,
       "step": 124
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.999380862192813e-05,
-      "loss": 1.3045,
+      "epoch": 0.013082155939298797,
+      "grad_norm": 4.393426240660029,
+      "learning_rate": 8.710801393728223e-06,
+      "loss": 0.9717,
       "step": 125
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.99934277432446e-05,
-      "loss": 1.3729,
+      "epoch": 0.013186813186813187,
+      "grad_norm": 5.209722101990777,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 1.0548,
       "step": 126
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9993035500103904e-05,
-      "loss": 1.4059,
+      "epoch": 0.013291470434327577,
+      "grad_norm": 4.44296200002556,
+      "learning_rate": 8.850174216027875e-06,
+      "loss": 1.0316,
       "step": 127
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.9992631892952108e-05,
-      "loss": 1.3092,
+      "epoch": 0.013396127681841967,
+      "grad_norm": 3.2269766969577525,
+      "learning_rate": 8.9198606271777e-06,
+      "loss": 0.7394,
       "step": 128
     },
     {
-      "epoch": 0.08,
-      "learning_rate": 1.999221692224818e-05,
-      "loss": 1.2541,
+      "epoch": 0.013500784929356358,
+      "grad_norm": 4.000784251757387,
+      "learning_rate": 8.989547038327527e-06,
+      "loss": 0.9277,
       "step": 129
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9991790588464034e-05,
-      "loss": 1.5616,
+      "epoch": 0.013605442176870748,
+      "grad_norm": 3.76241263382153,
+      "learning_rate": 9.059233449477352e-06,
+      "loss": 1.0673,
       "step": 130
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9991352892084483e-05,
-      "loss": 1.3117,
+      "epoch": 0.013710099424385138,
+      "grad_norm": 4.24231924403224,
+      "learning_rate": 9.12891986062718e-06,
+      "loss": 0.9685,
       "step": 131
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.999090383360728e-05,
-      "loss": 1.4864,
+      "epoch": 0.013814756671899528,
+      "grad_norm": 2.3758209921949343,
+      "learning_rate": 9.198606271777004e-06,
+      "loss": 0.7123,
       "step": 132
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9990443413543086e-05,
-      "loss": 1.1912,
+      "epoch": 0.01391941391941392,
+      "grad_norm": 3.3277360996370966,
+      "learning_rate": 9.268292682926831e-06,
+      "loss": 1.0094,
       "step": 133
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.998997163241549e-05,
-      "loss": 1.3293,
+      "epoch": 0.01402407116692831,
+      "grad_norm": 3.9700536041385353,
+      "learning_rate": 9.337979094076656e-06,
+      "loss": 1.029,
       "step": 134
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9989488490761e-05,
-      "loss": 1.3906,
+      "epoch": 0.0141287284144427,
+      "grad_norm": 3.8999757749740973,
+      "learning_rate": 9.407665505226482e-06,
+      "loss": 1.099,
       "step": 135
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9988993989129035e-05,
-      "loss": 1.2405,
+      "epoch": 0.014233385661957091,
+      "grad_norm": 4.007407594413442,
+      "learning_rate": 9.477351916376307e-06,
+      "loss": 0.9941,
       "step": 136
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9988488128081943e-05,
-      "loss": 1.4372,
+      "epoch": 0.014338042909471481,
+      "grad_norm": 4.372638054940945,
+      "learning_rate": 9.547038327526134e-06,
+      "loss": 0.998,
       "step": 137
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.998797090819499e-05,
-      "loss": 1.2302,
+      "epoch": 0.014442700156985872,
+      "grad_norm": 3.781559177153234,
+      "learning_rate": 9.616724738675959e-06,
+      "loss": 1.0021,
       "step": 138
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9987442330056352e-05,
-      "loss": 1.3168,
+      "epoch": 0.014547357404500262,
+      "grad_norm": 3.8918297025335358,
+      "learning_rate": 9.686411149825786e-06,
+      "loss": 1.1862,
       "step": 139
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9986902394267117e-05,
-      "loss": 1.3076,
+      "epoch": 0.014652014652014652,
+      "grad_norm": 4.87861639008213,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 1.0808,
       "step": 140
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9986351101441308e-05,
-      "loss": 1.1699,
+      "epoch": 0.014756671899529042,
+      "grad_norm": 4.2959641485004365,
+      "learning_rate": 9.825783972125436e-06,
+      "loss": 1.0558,
       "step": 141
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9985788452205847e-05,
-      "loss": 1.3322,
+      "epoch": 0.014861329147043432,
+      "grad_norm": 3.6891019296746816,
+      "learning_rate": 9.895470383275261e-06,
+      "loss": 0.9425,
       "step": 142
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9985214447200566e-05,
-      "loss": 1.3678,
+      "epoch": 0.014965986394557823,
+      "grad_norm": 2.421295406110499,
+      "learning_rate": 9.965156794425088e-06,
+      "loss": 0.7397,
       "step": 143
     },
     {
-      "epoch": 0.09,
-      "learning_rate": 1.9984629087078234e-05,
-      "loss": 1.2653,
+      "epoch": 0.015070643642072213,
+      "grad_norm": 4.319425469746967,
+      "learning_rate": 1.0034843205574913e-05,
+      "loss": 1.0007,
       "step": 144
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.998403237250451e-05,
-      "loss": 1.3188,
+      "epoch": 0.015175300889586603,
+      "grad_norm": 6.872837574225154,
+      "learning_rate": 1.0104529616724739e-05,
+      "loss": 1.0009,
       "step": 145
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9983424304157976e-05,
-      "loss": 1.4451,
+      "epoch": 0.015279958137100993,
+      "grad_norm": 4.1481133816472875,
+      "learning_rate": 1.0174216027874565e-05,
+      "loss": 1.1726,
       "step": 146
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.998280488273012e-05,
-      "loss": 1.3479,
+      "epoch": 0.015384615384615385,
+      "grad_norm": 4.440751447995416,
+      "learning_rate": 1.024390243902439e-05,
+      "loss": 1.1265,
       "step": 147
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9982174108925346e-05,
-      "loss": 1.4162,
+      "epoch": 0.015489272632129776,
+      "grad_norm": 4.171124786786287,
+      "learning_rate": 1.0313588850174216e-05,
+      "loss": 1.0745,
       "step": 148
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.998153198346097e-05,
-      "loss": 1.4135,
+      "epoch": 0.015593929879644166,
+      "grad_norm": 4.2679960358102385,
+      "learning_rate": 1.0383275261324041e-05,
+      "loss": 1.0301,
       "step": 149
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9980878507067207e-05,
-      "loss": 1.3905,
+      "epoch": 0.015698587127158554,
+      "grad_norm": 4.083629703686367,
+      "learning_rate": 1.045296167247387e-05,
+      "loss": 1.0725,
       "step": 150
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9980213680487186e-05,
-      "loss": 1.4447,
+      "epoch": 0.015803244374672946,
+      "grad_norm": 4.756910822842423,
+      "learning_rate": 1.0522648083623695e-05,
+      "loss": 1.1538,
       "step": 151
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9979537504476945e-05,
-      "loss": 1.4285,
+      "epoch": 0.015907901622187335,
+      "grad_norm": 4.175459761454693,
+      "learning_rate": 1.059233449477352e-05,
+      "loss": 1.0311,
       "step": 152
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.997884997980543e-05,
-      "loss": 1.4105,
+      "epoch": 0.016012558869701727,
+      "grad_norm": 4.660499784764975,
+      "learning_rate": 1.0662020905923345e-05,
+      "loss": 1.0348,
       "step": 153
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9978151107254477e-05,
-      "loss": 1.2898,
+      "epoch": 0.01611721611721612,
+      "grad_norm": 5.477934194364109,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.947,
       "step": 154
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9977440887618855e-05,
-      "loss": 1.3787,
+      "epoch": 0.016221873364730507,
+      "grad_norm": 3.8257154240108475,
+      "learning_rate": 1.0801393728222997e-05,
+      "loss": 1.0325,
       "step": 155
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9976719321706213e-05,
-      "loss": 1.419,
+      "epoch": 0.0163265306122449,
+      "grad_norm": 4.473013121990363,
+      "learning_rate": 1.0871080139372822e-05,
+      "loss": 1.0694,
       "step": 156
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.997598641033711e-05,
-      "loss": 1.4784,
+      "epoch": 0.016431187859759288,
+      "grad_norm": 3.7487977727407107,
+      "learning_rate": 1.0940766550522648e-05,
+      "loss": 0.9867,
       "step": 157
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.997524215434501e-05,
-      "loss": 1.4202,
+      "epoch": 0.01653584510727368,
+      "grad_norm": 3.6501719775560844,
+      "learning_rate": 1.1010452961672475e-05,
+      "loss": 1.0029,
       "step": 158
     },
     {
-      "epoch": 0.1,
-      "learning_rate": 1.9974486554576274e-05,
-      "loss": 1.2834,
+      "epoch": 0.01664050235478807,
+      "grad_norm": 4.465954207259012,
+      "learning_rate": 1.1080139372822301e-05,
+      "loss": 1.1428,
       "step": 159
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9973719611890164e-05,
-      "loss": 1.4078,
+      "epoch": 0.01674515960230246,
+      "grad_norm": 3.7784565534446597,
+      "learning_rate": 1.1149825783972127e-05,
+      "loss": 1.0845,
       "step": 160
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9972941327158848e-05,
-      "loss": 1.2659,
+      "epoch": 0.01684981684981685,
+      "grad_norm": 4.3826834909022825,
+      "learning_rate": 1.1219512195121953e-05,
+      "loss": 1.029,
       "step": 161
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.997215170126738e-05,
-      "loss": 1.3769,
+      "epoch": 0.01695447409733124,
+      "grad_norm": 5.076983929836477,
+      "learning_rate": 1.1289198606271779e-05,
+      "loss": 0.9905,
       "step": 162
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9971350735113718e-05,
-      "loss": 1.3037,
+      "epoch": 0.01705913134484563,
+      "grad_norm": 3.87202080289898,
+      "learning_rate": 1.1358885017421604e-05,
+      "loss": 0.9471,
       "step": 163
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9970538429608716e-05,
-      "loss": 1.3615,
+      "epoch": 0.01716378859236002,
+      "grad_norm": 3.947159057053555,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.0517,
       "step": 164
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.996971478567612e-05,
-      "loss": 1.4036,
+      "epoch": 0.01726844583987441,
+      "grad_norm": 3.5664649583899886,
+      "learning_rate": 1.1498257839721256e-05,
+      "loss": 1.0465,
       "step": 165
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.996887980425258e-05,
-      "loss": 1.2713,
+      "epoch": 0.017373103087388802,
+      "grad_norm": 3.889725720772939,
+      "learning_rate": 1.1567944250871081e-05,
+      "loss": 1.0653,
       "step": 166
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.996803348628762e-05,
-      "loss": 1.3764,
+      "epoch": 0.01747776033490319,
+      "grad_norm": 3.9170851836947143,
+      "learning_rate": 1.1637630662020906e-05,
+      "loss": 1.034,
       "step": 167
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9967175832743677e-05,
-      "loss": 1.4328,
+      "epoch": 0.017582417582417582,
+      "grad_norm": 3.72251997612241,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 1.0297,
       "step": 168
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9966306844596063e-05,
-      "loss": 1.4352,
+      "epoch": 0.017687074829931974,
+      "grad_norm": 3.576705734526832,
+      "learning_rate": 1.177700348432056e-05,
+      "loss": 1.0623,
       "step": 169
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9965426522832985e-05,
-      "loss": 1.2599,
+      "epoch": 0.017791732077446363,
+      "grad_norm": 3.263183539989399,
+      "learning_rate": 1.1846689895470385e-05,
+      "loss": 1.0284,
       "step": 170
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9964534868455544e-05,
-      "loss": 1.3638,
+      "epoch": 0.017896389324960755,
+      "grad_norm": 4.86530329059982,
+      "learning_rate": 1.191637630662021e-05,
+      "loss": 1.1147,
       "step": 171
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9963631882477716e-05,
-      "loss": 1.3307,
+      "epoch": 0.018001046572475143,
+      "grad_norm": 3.74319247595939,
+      "learning_rate": 1.1986062717770036e-05,
+      "loss": 1.0216,
       "step": 172
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9962717565926376e-05,
-      "loss": 1.36,
+      "epoch": 0.018105703819989535,
+      "grad_norm": 3.4535523228322274,
+      "learning_rate": 1.2055749128919862e-05,
+      "loss": 1.1057,
       "step": 173
     },
     {
-      "epoch": 0.11,
-      "learning_rate": 1.9961791919841275e-05,
-      "loss": 1.3391,
+      "epoch": 0.018210361067503924,
+      "grad_norm": 3.5433866410709784,
+      "learning_rate": 1.2125435540069688e-05,
+      "loss": 1.1274,
       "step": 174
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9960854945275055e-05,
-      "loss": 1.4955,
+      "epoch": 0.018315018315018316,
+      "grad_norm": 3.8964338737423567,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 1.1093,
       "step": 175
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.995990664329323e-05,
-      "loss": 1.4003,
+      "epoch": 0.018419675562532704,
+      "grad_norm": 3.7316530734886197,
+      "learning_rate": 1.2264808362369338e-05,
+      "loss": 1.0501,
       "step": 176
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9958947014974216e-05,
-      "loss": 1.2154,
+      "epoch": 0.018524332810047096,
+      "grad_norm": 3.8491712042277872,
+      "learning_rate": 1.2334494773519165e-05,
+      "loss": 1.1212,
       "step": 177
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9957976061409277e-05,
-      "loss": 1.3286,
+      "epoch": 0.018628990057561485,
+      "grad_norm": 3.4777164601747863,
+      "learning_rate": 1.240418118466899e-05,
+      "loss": 1.0654,
       "step": 178
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9956993783702594e-05,
-      "loss": 1.4478,
+      "epoch": 0.018733647305075877,
+      "grad_norm": 2.261327289656553,
+      "learning_rate": 1.2473867595818815e-05,
+      "loss": 0.7282,
       "step": 179
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9956000182971192e-05,
-      "loss": 1.2906,
+      "epoch": 0.018838304552590265,
+      "grad_norm": 3.552001161873253,
+      "learning_rate": 1.254355400696864e-05,
+      "loss": 0.9128,
       "step": 180
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9954995260344993e-05,
-      "loss": 1.1524,
+      "epoch": 0.018942961800104657,
+      "grad_norm": 5.287461211801188,
+      "learning_rate": 1.2613240418118469e-05,
+      "loss": 1.051,
       "step": 181
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.995397901696679e-05,
-      "loss": 1.3923,
+      "epoch": 0.01904761904761905,
+      "grad_norm": 4.4864817842055436,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 1.0427,
       "step": 182
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.995295145399224e-05,
-      "loss": 1.3211,
+      "epoch": 0.019152276295133438,
+      "grad_norm": 4.588279167310623,
+      "learning_rate": 1.275261324041812e-05,
+      "loss": 1.0934,
       "step": 183
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9951912572589892e-05,
-      "loss": 1.3806,
+      "epoch": 0.01925693354264783,
+      "grad_norm": 5.248743013483407,
+      "learning_rate": 1.2822299651567945e-05,
+      "loss": 1.0224,
       "step": 184
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9950862373941145e-05,
-      "loss": 1.3799,
+      "epoch": 0.01936159079016222,
+      "grad_norm": 3.599591701450194,
+      "learning_rate": 1.2891986062717772e-05,
+      "loss": 0.942,
       "step": 185
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.994980085924028e-05,
-      "loss": 1.3846,
+      "epoch": 0.01946624803767661,
+      "grad_norm": 4.22948244161219,
+      "learning_rate": 1.2961672473867597e-05,
+      "loss": 1.096,
       "step": 186
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.994872802969445e-05,
-      "loss": 1.4632,
+      "epoch": 0.019570905285191,
+      "grad_norm": 3.883052580674137,
+      "learning_rate": 1.3031358885017422e-05,
+      "loss": 1.0867,
       "step": 187
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.994764388652366e-05,
-      "loss": 1.3105,
+      "epoch": 0.01967556253270539,
+      "grad_norm": 3.9641257982365024,
+      "learning_rate": 1.3101045296167247e-05,
+      "loss": 1.0495,
       "step": 188
     },
     {
-      "epoch": 0.12,
-      "learning_rate": 1.9946548430960796e-05,
-      "loss": 1.3987,
+      "epoch": 0.01978021978021978,
+      "grad_norm": 3.98653960238745,
+      "learning_rate": 1.3170731707317076e-05,
+      "loss": 1.1492,
       "step": 189
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9945441664251605e-05,
-      "loss": 1.4065,
+      "epoch": 0.01988487702773417,
+      "grad_norm": 2.3679885080466554,
+      "learning_rate": 1.32404181184669e-05,
+      "loss": 0.7717,
       "step": 190
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9944323587654685e-05,
-      "loss": 1.4344,
+      "epoch": 0.01998953427524856,
+      "grad_norm": 4.239294493465942,
+      "learning_rate": 1.3310104529616726e-05,
+      "loss": 1.1124,
       "step": 191
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9943194202441514e-05,
-      "loss": 1.3795,
+      "epoch": 0.020094191522762952,
+      "grad_norm": 4.028830074043196,
+      "learning_rate": 1.3379790940766553e-05,
+      "loss": 1.1247,
       "step": 192
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9942053509896417e-05,
-      "loss": 1.552,
+      "epoch": 0.02019884877027734,
+      "grad_norm": 3.522476393811863,
+      "learning_rate": 1.3449477351916378e-05,
+      "loss": 1.0408,
       "step": 193
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9940901511316585e-05,
-      "loss": 1.3405,
+      "epoch": 0.020303506017791732,
+      "grad_norm": 3.420820328724481,
+      "learning_rate": 1.3519163763066203e-05,
+      "loss": 1.0528,
       "step": 194
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.993973820801206e-05,
-      "loss": 1.3728,
+      "epoch": 0.02040816326530612,
+      "grad_norm": 3.855766622211791,
+      "learning_rate": 1.3588850174216028e-05,
+      "loss": 1.0484,
       "step": 195
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.993856360130574e-05,
-      "loss": 1.4416,
+      "epoch": 0.020512820512820513,
+      "grad_norm": 4.09125354171716,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 1.1015,
       "step": 196
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9937377692533382e-05,
-      "loss": 1.3926,
+      "epoch": 0.020617477760334905,
+      "grad_norm": 3.3241462228706653,
+      "learning_rate": 1.372822299651568e-05,
+      "loss": 1.1006,
       "step": 197
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.99361804830436e-05,
-      "loss": 1.2651,
+      "epoch": 0.020722135007849293,
+      "grad_norm": 4.516135214785916,
+      "learning_rate": 1.3797909407665506e-05,
+      "loss": 1.0546,
       "step": 198
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9934971974197843e-05,
-      "loss": 1.2976,
+      "epoch": 0.020826792255363685,
+      "grad_norm": 3.3260375654673564,
+      "learning_rate": 1.3867595818815331e-05,
+      "loss": 1.1214,
       "step": 199
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.993375216737042e-05,
-      "loss": 1.4121,
+      "epoch": 0.020931449502878074,
+      "grad_norm": 4.0907170025291935,
+      "learning_rate": 1.393728222996516e-05,
+      "loss": 1.0947,
       "step": 200
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9932521063948494e-05,
-      "loss": 1.276,
+      "epoch": 0.021036106750392466,
+      "grad_norm": 3.9206762011268337,
+      "learning_rate": 1.4006968641114985e-05,
+      "loss": 1.0005,
       "step": 201
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9931278665332057e-05,
-      "loss": 1.3322,
+      "epoch": 0.021140763997906854,
+      "grad_norm": 3.74058600438095,
+      "learning_rate": 1.407665505226481e-05,
+      "loss": 1.0507,
       "step": 202
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9930024972933966e-05,
-      "loss": 1.3596,
+      "epoch": 0.021245421245421246,
+      "grad_norm": 4.046674364114655,
+      "learning_rate": 1.4146341463414635e-05,
+      "loss": 1.0381,
       "step": 203
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.9928759988179908e-05,
-      "loss": 1.4512,
+      "epoch": 0.021350078492935635,
+      "grad_norm": 4.070409637629066,
+      "learning_rate": 1.4216027874564462e-05,
+      "loss": 1.0869,
       "step": 204
     },
     {
-      "epoch": 0.13,
-      "learning_rate": 1.992748371250841e-05,
-      "loss": 1.4053,
+      "epoch": 0.021454735740450027,
+      "grad_norm": 2.477212456664358,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.7274,
       "step": 205
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.992619614737085e-05,
-      "loss": 1.2646,
+      "epoch": 0.021559392987964415,
+      "grad_norm": 4.809989174382412,
+      "learning_rate": 1.4355400696864112e-05,
+      "loss": 0.9624,
       "step": 206
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9924897294231434e-05,
-      "loss": 1.5216,
+      "epoch": 0.021664050235478807,
+      "grad_norm": 4.855067955207067,
+      "learning_rate": 1.4425087108013938e-05,
+      "loss": 1.056,
       "step": 207
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9923587154567214e-05,
-      "loss": 1.4538,
+      "epoch": 0.021768707482993196,
+      "grad_norm": 4.666013151279942,
+      "learning_rate": 1.4494773519163764e-05,
+      "loss": 0.9354,
       "step": 208
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9922265729868065e-05,
-      "loss": 1.4267,
+      "epoch": 0.021873364730507588,
+      "grad_norm": 3.940057891989908,
+      "learning_rate": 1.456445993031359e-05,
+      "loss": 1.014,
       "step": 209
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.992093302163671e-05,
-      "loss": 1.5343,
+      "epoch": 0.02197802197802198,
+      "grad_norm": 5.015568593900297,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 1.0634,
       "step": 210
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9919589031388688e-05,
-      "loss": 1.3869,
+      "epoch": 0.022082679225536368,
+      "grad_norm": 3.5260271531233927,
+      "learning_rate": 1.470383275261324e-05,
+      "loss": 0.9887,
       "step": 211
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.991823376065238e-05,
-      "loss": 1.3431,
+      "epoch": 0.02218733647305076,
+      "grad_norm": 4.142631353724366,
+      "learning_rate": 1.4773519163763069e-05,
+      "loss": 1.0409,
       "step": 212
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.991686721096899e-05,
-      "loss": 1.2332,
+      "epoch": 0.02229199372056515,
+      "grad_norm": 4.18461357230588,
+      "learning_rate": 1.4843205574912894e-05,
+      "loss": 1.0713,
       "step": 213
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.991548938389255e-05,
-      "loss": 1.4032,
+      "epoch": 0.02239665096807954,
+      "grad_norm": 4.157498562874197,
+      "learning_rate": 1.4912891986062719e-05,
+      "loss": 1.0933,
       "step": 214
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.991410028098991e-05,
-      "loss": 1.3987,
+      "epoch": 0.02250130821559393,
+      "grad_norm": 4.027272133207346,
+      "learning_rate": 1.4982578397212544e-05,
+      "loss": 0.9743,
       "step": 215
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9912699903840754e-05,
-      "loss": 1.5462,
+      "epoch": 0.02260596546310832,
+      "grad_norm": 4.379374705129572,
+      "learning_rate": 1.5052264808362371e-05,
+      "loss": 1.0026,
       "step": 216
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.991128825403758e-05,
-      "loss": 1.3819,
+      "epoch": 0.02271062271062271,
+      "grad_norm": 3.239907491980648,
+      "learning_rate": 1.5121951219512196e-05,
+      "loss": 1.0678,
       "step": 217
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9909865333185704e-05,
-      "loss": 1.3906,
+      "epoch": 0.0228152799581371,
+      "grad_norm": 4.0127649077633665,
+      "learning_rate": 1.5191637630662021e-05,
+      "loss": 1.1173,
       "step": 218
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9908431142903263e-05,
-      "loss": 1.4463,
+      "epoch": 0.02291993720565149,
+      "grad_norm": 4.264440457331995,
+      "learning_rate": 1.5261324041811848e-05,
+      "loss": 0.9743,
       "step": 219
     },
     {
-      "epoch": 0.14,
-      "learning_rate": 1.9906985684821208e-05,
-      "loss": 1.3995,
+      "epoch": 0.023024594453165882,
+      "grad_norm": 3.997508098429076,
+      "learning_rate": 1.5331010452961673e-05,
+      "loss": 1.0646,
       "step": 220
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9905528960583305e-05,
-      "loss": 1.1931,
+      "epoch": 0.02312925170068027,
+      "grad_norm": 3.47902625924093,
+      "learning_rate": 1.54006968641115e-05,
+      "loss": 0.9376,
       "step": 221
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.990406097184613e-05,
-      "loss": 1.328,
+      "epoch": 0.023233908948194663,
+      "grad_norm": 4.175847455405455,
+      "learning_rate": 1.5470383275261324e-05,
+      "loss": 1.16,
       "step": 222
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9902581720279074e-05,
-      "loss": 1.3747,
+      "epoch": 0.02333856619570905,
+      "grad_norm": 3.8697528880278775,
+      "learning_rate": 1.554006968641115e-05,
+      "loss": 1.1035,
       "step": 223
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9901091207564326e-05,
-      "loss": 1.356,
+      "epoch": 0.023443223443223443,
+      "grad_norm": 3.7062219666594522,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 1.0437,
       "step": 224
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9899589435396888e-05,
-      "loss": 1.5297,
+      "epoch": 0.023547880690737835,
+      "grad_norm": 3.604348929857401,
+      "learning_rate": 1.5679442508710803e-05,
+      "loss": 1.1209,
       "step": 225
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9898076405484565e-05,
-      "loss": 1.2664,
+      "epoch": 0.023652537938252224,
+      "grad_norm": 4.264573745725593,
+      "learning_rate": 1.5749128919860628e-05,
+      "loss": 1.0217,
       "step": 226
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9896552119547964e-05,
-      "loss": 1.3391,
+      "epoch": 0.023757195185766616,
+      "grad_norm": 5.445543416503218,
+      "learning_rate": 1.5818815331010456e-05,
+      "loss": 1.0786,
       "step": 227
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9895016579320495e-05,
-      "loss": 1.5459,
+      "epoch": 0.023861852433281004,
+      "grad_norm": 5.26347191325024,
+      "learning_rate": 1.588850174216028e-05,
+      "loss": 1.1497,
       "step": 228
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.989346978654836e-05,
-      "loss": 1.3014,
+      "epoch": 0.023966509680795396,
+      "grad_norm": 3.5473340256833548,
+      "learning_rate": 1.5958188153310107e-05,
+      "loss": 0.9933,
       "step": 229
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9891911742990567e-05,
-      "loss": 1.4365,
+      "epoch": 0.024071166928309785,
+      "grad_norm": 3.654247972110555,
+      "learning_rate": 1.6027874564459932e-05,
+      "loss": 1.0457,
       "step": 230
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9890342450418907e-05,
-      "loss": 1.5066,
+      "epoch": 0.024175824175824177,
+      "grad_norm": 3.6750069709677047,
+      "learning_rate": 1.6097560975609757e-05,
+      "loss": 1.1164,
       "step": 231
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9888761910617967e-05,
-      "loss": 1.3875,
+      "epoch": 0.024280481423338565,
+      "grad_norm": 3.542624844100214,
+      "learning_rate": 1.6167247386759582e-05,
+      "loss": 1.1167,
       "step": 232
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9887170125385135e-05,
-      "loss": 1.3043,
+      "epoch": 0.024385138670852957,
+      "grad_norm": 3.9872277812389996,
+      "learning_rate": 1.6236933797909408e-05,
+      "loss": 1.1433,
       "step": 233
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9885567096530566e-05,
-      "loss": 1.3391,
+      "epoch": 0.024489795918367346,
+      "grad_norm": 3.7156048056910382,
+      "learning_rate": 1.6306620209059233e-05,
+      "loss": 1.0351,
       "step": 234
     },
     {
-      "epoch": 0.15,
-      "learning_rate": 1.9883952825877225e-05,
-      "loss": 1.51,
+      "epoch": 0.024594453165881738,
+      "grad_norm": 3.7816100969884348,
+      "learning_rate": 1.637630662020906e-05,
+      "loss": 1.0736,
       "step": 235
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.988232731526084e-05,
-      "loss": 1.4932,
+      "epoch": 0.024699110413396126,
+      "grad_norm": 2.5076547499457362,
+      "learning_rate": 1.6445993031358887e-05,
+      "loss": 0.7651,
       "step": 236
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9880690566529933e-05,
-      "loss": 1.3678,
+      "epoch": 0.024803767660910518,
+      "grad_norm": 3.3947366530434184,
+      "learning_rate": 1.6515679442508712e-05,
+      "loss": 1.1833,
       "step": 237
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.987904258154581e-05,
-      "loss": 1.4132,
+      "epoch": 0.02490842490842491,
+      "grad_norm": 3.9432345380153713,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 1.0596,
       "step": 238
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.987738336218254e-05,
-      "loss": 1.4784,
+      "epoch": 0.0250130821559393,
+      "grad_norm": 4.098078794977877,
+      "learning_rate": 1.6655052264808366e-05,
+      "loss": 1.1229,
       "step": 239
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9875712910326982e-05,
-      "loss": 1.2517,
+      "epoch": 0.02511773940345369,
+      "grad_norm": 3.9973498865337533,
+      "learning_rate": 1.672473867595819e-05,
+      "loss": 0.9674,
       "step": 240
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.987403122787876e-05,
-      "loss": 1.3915,
+      "epoch": 0.02522239665096808,
+      "grad_norm": 3.5983933762145037,
+      "learning_rate": 1.6794425087108016e-05,
+      "loss": 1.2461,
       "step": 241
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9872338316750264e-05,
-      "loss": 1.3817,
+      "epoch": 0.02532705389848247,
+      "grad_norm": 3.8197581407276977,
+      "learning_rate": 1.686411149825784e-05,
+      "loss": 1.033,
       "step": 242
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9870634178866676e-05,
-      "loss": 1.3482,
+      "epoch": 0.02543171114599686,
+      "grad_norm": 2.3297919147992445,
+      "learning_rate": 1.6933797909407666e-05,
+      "loss": 0.7915,
       "step": 243
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.986891881616592e-05,
-      "loss": 1.4669,
+      "epoch": 0.02553636839351125,
+      "grad_norm": 3.8568402426046853,
+      "learning_rate": 1.700348432055749e-05,
+      "loss": 1.1112,
       "step": 244
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9867192230598692e-05,
-      "loss": 1.521,
+      "epoch": 0.02564102564102564,
+      "grad_norm": 3.3775792696962434,
+      "learning_rate": 1.7073170731707317e-05,
+      "loss": 1.0562,
       "step": 245
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.986545442412846e-05,
-      "loss": 1.3481,
+      "epoch": 0.025745682888540032,
+      "grad_norm": 3.8185082938369206,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 1.2041,
       "step": 246
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9863705398731444e-05,
-      "loss": 1.2345,
+      "epoch": 0.02585034013605442,
+      "grad_norm": 3.946632969495481,
+      "learning_rate": 1.721254355400697e-05,
+      "loss": 1.1593,
       "step": 247
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.986194515639662e-05,
-      "loss": 1.39,
+      "epoch": 0.025954997383568813,
+      "grad_norm": 3.4354194263721243,
+      "learning_rate": 1.7282229965156796e-05,
+      "loss": 1.0919,
       "step": 248
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.986017369912573e-05,
-      "loss": 1.4151,
+      "epoch": 0.0260596546310832,
+      "grad_norm": 4.090271178228001,
+      "learning_rate": 1.735191637630662e-05,
+      "loss": 1.1536,
       "step": 249
     },
     {
-      "epoch": 0.16,
-      "learning_rate": 1.9858391028933258e-05,
-      "loss": 1.3328,
+      "epoch": 0.026164311878597593,
+      "grad_norm": 3.7495551186516574,
+      "learning_rate": 1.7421602787456446e-05,
+      "loss": 1.024,
       "step": 250
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9856597147846447e-05,
-      "loss": 1.3425,
+      "epoch": 0.02626896912611198,
+      "grad_norm": 3.682112510859972,
+      "learning_rate": 1.7491289198606275e-05,
+      "loss": 1.1665,
       "step": 251
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.985479205790528e-05,
-      "loss": 1.485,
+      "epoch": 0.026373626373626374,
+      "grad_norm": 3.3296117604440783,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 1.0428,
       "step": 252
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.98529757611625e-05,
-      "loss": 1.2616,
+      "epoch": 0.026478283621140766,
+      "grad_norm": 3.200300909223925,
+      "learning_rate": 1.7630662020905925e-05,
+      "loss": 1.0666,
       "step": 253
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9851148259683586e-05,
-      "loss": 1.1877,
+      "epoch": 0.026582940868655154,
+      "grad_norm": 3.747386619716652,
+      "learning_rate": 1.770034843205575e-05,
+      "loss": 1.154,
       "step": 254
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.984930955554676e-05,
-      "loss": 1.489,
+      "epoch": 0.026687598116169546,
+      "grad_norm": 3.9792918121852483,
+      "learning_rate": 1.7770034843205575e-05,
+      "loss": 0.9988,
       "step": 255
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.984745965084298e-05,
-      "loss": 1.4196,
+      "epoch": 0.026792255363683935,
+      "grad_norm": 3.1282408579177283,
+      "learning_rate": 1.78397212543554e-05,
+      "loss": 0.9897,
       "step": 256
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9845598547675945e-05,
-      "loss": 1.3344,
+      "epoch": 0.026896912611198327,
+      "grad_norm": 2.056362322374514,
+      "learning_rate": 1.7909407665505226e-05,
+      "loss": 0.7055,
       "step": 257
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9843726248162096e-05,
-      "loss": 1.3713,
+      "epoch": 0.027001569858712715,
+      "grad_norm": 2.5522966724650624,
+      "learning_rate": 1.7979094076655054e-05,
+      "loss": 0.8675,
       "step": 258
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.984184275443059e-05,
-      "loss": 1.2307,
+      "epoch": 0.027106227106227107,
+      "grad_norm": 3.5782559255097457,
+      "learning_rate": 1.804878048780488e-05,
+      "loss": 1.0991,
       "step": 259
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.983994806862333e-05,
-      "loss": 1.4117,
+      "epoch": 0.027210884353741496,
+      "grad_norm": 3.928265536667273,
+      "learning_rate": 1.8118466898954705e-05,
+      "loss": 1.05,
       "step": 260
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9838042192894935e-05,
-      "loss": 1.4123,
+      "epoch": 0.027315541601255888,
+      "grad_norm": 3.551876974379943,
+      "learning_rate": 1.818815331010453e-05,
+      "loss": 1.0506,
       "step": 261
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9836125129412756e-05,
-      "loss": 1.1802,
+      "epoch": 0.027420198848770276,
+      "grad_norm": 2.60968997475218,
+      "learning_rate": 1.825783972125436e-05,
+      "loss": 0.7637,
       "step": 262
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9834196880356863e-05,
-      "loss": 1.4824,
+      "epoch": 0.027524856096284668,
+      "grad_norm": 3.2023640410720504,
+      "learning_rate": 1.8327526132404184e-05,
+      "loss": 1.1014,
       "step": 263
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.983225744792005e-05,
-      "loss": 1.3801,
+      "epoch": 0.027629513343799057,
+      "grad_norm": 4.254369956741867,
+      "learning_rate": 1.839721254355401e-05,
+      "loss": 1.1119,
       "step": 264
     },
     {
-      "epoch": 0.17,
-      "learning_rate": 1.9830306834307828e-05,
-      "loss": 1.4142,
+      "epoch": 0.02773417059131345,
+      "grad_norm": 4.456237791959698,
+      "learning_rate": 1.8466898954703834e-05,
+      "loss": 1.0861,
       "step": 265
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9828345041738413e-05,
-      "loss": 1.2993,
+      "epoch": 0.02783882783882784,
+      "grad_norm": 3.326487692184649,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 1.1481,
       "step": 266
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9826372072442747e-05,
-      "loss": 1.24,
+      "epoch": 0.02794348508634223,
+      "grad_norm": 3.4486894326387985,
+      "learning_rate": 1.8606271777003488e-05,
+      "loss": 0.9798,
       "step": 267
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9824387928664478e-05,
-      "loss": 1.3416,
+      "epoch": 0.02804814233385662,
+      "grad_norm": 3.5440191404796875,
+      "learning_rate": 1.8675958188153313e-05,
+      "loss": 1.1801,
       "step": 268
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9822392612659964e-05,
-      "loss": 1.4188,
+      "epoch": 0.02815279958137101,
+      "grad_norm": 3.978193923401927,
+      "learning_rate": 1.8745644599303138e-05,
+      "loss": 1.0222,
       "step": 269
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9820386126698256e-05,
-      "loss": 1.401,
+      "epoch": 0.0282574568288854,
+      "grad_norm": 2.5739712495009224,
+      "learning_rate": 1.8815331010452963e-05,
+      "loss": 0.7676,
       "step": 270
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.981836847306112e-05,
-      "loss": 1.262,
+      "epoch": 0.02836211407639979,
+      "grad_norm": 3.7240630754036523,
+      "learning_rate": 1.888501742160279e-05,
+      "loss": 1.0653,
       "step": 271
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.981633965404302e-05,
-      "loss": 1.1779,
+      "epoch": 0.028466771323914182,
+      "grad_norm": 3.1834334045524337,
+      "learning_rate": 1.8954703832752614e-05,
+      "loss": 1.0548,
       "step": 272
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9814299671951113e-05,
-      "loss": 1.3676,
+      "epoch": 0.02857142857142857,
+      "grad_norm": 4.050102203011576,
+      "learning_rate": 1.902439024390244e-05,
+      "loss": 1.0478,
       "step": 273
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9812248529105254e-05,
-      "loss": 1.3851,
+      "epoch": 0.028676085818942963,
+      "grad_norm": 3.1797343765169783,
+      "learning_rate": 1.9094076655052267e-05,
+      "loss": 1.0334,
       "step": 274
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9810186227837986e-05,
-      "loss": 1.4388,
+      "epoch": 0.02878074306645735,
+      "grad_norm": 3.6752536373704454,
+      "learning_rate": 1.9163763066202093e-05,
+      "loss": 1.1476,
       "step": 275
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9808112770494546e-05,
-      "loss": 1.1772,
+      "epoch": 0.028885400313971743,
+      "grad_norm": 4.244905455321221,
+      "learning_rate": 1.9233449477351918e-05,
+      "loss": 1.0884,
       "step": 276
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9806028159432854e-05,
-      "loss": 1.2482,
+      "epoch": 0.02899005756148613,
+      "grad_norm": 2.548313029921991,
+      "learning_rate": 1.9303135888501743e-05,
+      "loss": 0.7592,
       "step": 277
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9803932397023512e-05,
-      "loss": 1.3222,
+      "epoch": 0.029094714809000524,
+      "grad_norm": 4.015070153207213,
+      "learning_rate": 1.937282229965157e-05,
+      "loss": 1.1131,
       "step": 278
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9801825485649816e-05,
-      "loss": 1.3487,
+      "epoch": 0.029199372056514912,
+      "grad_norm": 3.5031056198015382,
+      "learning_rate": 1.9442508710801397e-05,
+      "loss": 0.9966,
       "step": 279
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9799707427707723e-05,
-      "loss": 1.1701,
+      "epoch": 0.029304029304029304,
+      "grad_norm": 3.7962075539310467,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 1.1491,
       "step": 280
     },
     {
-      "epoch": 0.18,
-      "learning_rate": 1.9797578225605875e-05,
-      "loss": 1.3803,
+      "epoch": 0.029408686551543696,
+      "grad_norm": 3.575956042142137,
+      "learning_rate": 1.9581881533101047e-05,
+      "loss": 1.0795,
       "step": 281
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.979543788176559e-05,
-      "loss": 1.2782,
+      "epoch": 0.029513343799058084,
+      "grad_norm": 3.5922001990846497,
+      "learning_rate": 1.9651567944250872e-05,
+      "loss": 1.0633,
       "step": 282
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.979328639862085e-05,
-      "loss": 1.4364,
+      "epoch": 0.029618001046572476,
+      "grad_norm": 3.176407389065215,
+      "learning_rate": 1.9721254355400697e-05,
+      "loss": 1.0095,
       "step": 283
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9791123778618303e-05,
-      "loss": 1.3674,
+      "epoch": 0.029722658294086865,
+      "grad_norm": 3.9198920216298063,
+      "learning_rate": 1.9790940766550523e-05,
+      "loss": 1.1247,
       "step": 284
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.978895002421727e-05,
-      "loss": 1.3909,
+      "epoch": 0.029827315541601257,
+      "grad_norm": 3.7058792697424425,
+      "learning_rate": 1.9860627177700348e-05,
+      "loss": 1.0748,
       "step": 285
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.978676513788973e-05,
-      "loss": 1.3825,
+      "epoch": 0.029931972789115645,
+      "grad_norm": 3.1240553549271075,
+      "learning_rate": 1.9930313588850176e-05,
+      "loss": 1.0794,
       "step": 286
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.978456912212032e-05,
-      "loss": 1.2967,
+      "epoch": 0.030036630036630037,
+      "grad_norm": 4.161059734877122,
+      "learning_rate": 2e-05,
+      "loss": 1.2648,
       "step": 287
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9782361979406334e-05,
-      "loss": 1.2333,
+      "epoch": 0.030141287284144426,
+      "grad_norm": 3.292538569117115,
+      "learning_rate": 1.9999999425489863e-05,
+      "loss": 1.132,
       "step": 288
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9780143712257715e-05,
-      "loss": 1.3207,
+      "epoch": 0.030245944531658818,
+      "grad_norm": 3.3817728904500988,
+      "learning_rate": 1.9999997701959524e-05,
+      "loss": 1.119,
       "step": 289
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9777914323197063e-05,
-      "loss": 1.322,
+      "epoch": 0.030350601779173206,
+      "grad_norm": 3.698433893111136,
+      "learning_rate": 1.999999482940917e-05,
+      "loss": 1.1232,
       "step": 290
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.977567381475963e-05,
-      "loss": 1.4313,
+      "epoch": 0.0304552590266876,
+      "grad_norm": 3.867287084222632,
+      "learning_rate": 1.9999990807839145e-05,
+      "loss": 1.0713,
       "step": 291
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.97734221894933e-05,
-      "loss": 1.4255,
+      "epoch": 0.030559916274201987,
+      "grad_norm": 3.4207322827343276,
+      "learning_rate": 1.99999856372499e-05,
+      "loss": 1.131,
       "step": 292
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9771159449958612e-05,
-      "loss": 1.3717,
+      "epoch": 0.03066457352171638,
+      "grad_norm": 3.489327239181255,
+      "learning_rate": 1.9999979317642035e-05,
+      "loss": 1.0825,
       "step": 293
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9768885598728734e-05,
-      "loss": 1.2229,
+      "epoch": 0.03076923076923077,
+      "grad_norm": 4.019683220751094,
+      "learning_rate": 1.9999971849016274e-05,
+      "loss": 1.0214,
       "step": 294
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.976660063838947e-05,
-      "loss": 1.4424,
+      "epoch": 0.03087388801674516,
+      "grad_norm": 3.58104376114611,
+      "learning_rate": 1.9999963231373474e-05,
+      "loss": 1.0346,
       "step": 295
     },
     {
-      "epoch": 0.19,
-      "learning_rate": 1.9764304571539266e-05,
-      "loss": 1.3519,
+      "epoch": 0.03097854526425955,
+      "grad_norm": 4.29557092301096,
+      "learning_rate": 1.9999953464714628e-05,
+      "loss": 1.1057,
       "step": 296
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.9761997400789194e-05,
-      "loss": 1.3743,
+      "epoch": 0.03108320251177394,
+      "grad_norm": 4.031376664345659,
+      "learning_rate": 1.9999942549040857e-05,
+      "loss": 1.199,
       "step": 297
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.9759679128762946e-05,
-      "loss": 1.3588,
+      "epoch": 0.031187859759288332,
+      "grad_norm": 3.2778277667278837,
+      "learning_rate": 1.9999930484353412e-05,
+      "loss": 1.1605,
       "step": 298
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.9757349758096855e-05,
-      "loss": 1.3905,
+      "epoch": 0.031292517006802724,
+      "grad_norm": 3.57216344766695,
+      "learning_rate": 1.9999917270653686e-05,
+      "loss": 1.2141,
       "step": 299
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.975500929143986e-05,
-      "loss": 1.4055,
+      "epoch": 0.03139717425431711,
+      "grad_norm": 3.32290052080157,
+      "learning_rate": 1.9999902907943195e-05,
+      "loss": 1.0385,
       "step": 300
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.9752657731453523e-05,
-      "loss": 1.3259,
+      "epoch": 0.0315018315018315,
+      "grad_norm": 3.3614253151991718,
+      "learning_rate": 1.9999887396223584e-05,
+      "loss": 0.8076,
       "step": 301
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.9750295080812025e-05,
-      "loss": 1.342,
+      "epoch": 0.03160648874934589,
+      "grad_norm": 3.999164186143186,
+      "learning_rate": 1.999987073549664e-05,
+      "loss": 1.1335,
       "step": 302
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.974792134220215e-05,
-      "loss": 1.4263,
+      "epoch": 0.031711145996860285,
+      "grad_norm": 3.8434118240060937,
+      "learning_rate": 1.9999852925764277e-05,
+      "loss": 1.0446,
       "step": 303
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.9745536518323304e-05,
-      "loss": 1.393,
+      "epoch": 0.03181580324437467,
+      "grad_norm": 4.069391175340414,
+      "learning_rate": 1.9999833967028542e-05,
+      "loss": 1.0591,
       "step": 304
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.9743140611887487e-05,
-      "loss": 1.3359,
+      "epoch": 0.03192046049188906,
+      "grad_norm": 3.8316602155773998,
+      "learning_rate": 1.999981385929161e-05,
+      "loss": 1.1707,
       "step": 305
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.974073362561931e-05,
-      "loss": 1.4202,
+      "epoch": 0.032025117739403454,
+      "grad_norm": 4.737562782836604,
+      "learning_rate": 1.99997926025558e-05,
+      "loss": 1.0614,
       "step": 306
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.9738315562255983e-05,
-      "loss": 1.3547,
+      "epoch": 0.032129774986917846,
+      "grad_norm": 3.8265410957341515,
+      "learning_rate": 1.9999770196823544e-05,
+      "loss": 1.1765,
       "step": 307
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.9735886424547306e-05,
-      "loss": 1.4709,
+      "epoch": 0.03223443223443224,
+      "grad_norm": 3.265062349948494,
+      "learning_rate": 1.999974664209742e-05,
+      "loss": 1.0048,
       "step": 308
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.973344621525568e-05,
-      "loss": 1.2986,
+      "epoch": 0.03233908948194662,
+      "grad_norm": 3.642030345306258,
+      "learning_rate": 1.9999721938380133e-05,
+      "loss": 1.1304,
       "step": 309
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.97309949371561e-05,
-      "loss": 1.2082,
+      "epoch": 0.032443746729461015,
+      "grad_norm": 3.4292052143727405,
+      "learning_rate": 1.999969608567453e-05,
+      "loss": 1.1584,
       "step": 310
     },
     {
-      "epoch": 0.2,
-      "learning_rate": 1.9728532593036132e-05,
-      "loss": 1.3314,
+      "epoch": 0.03254840397697541,
+      "grad_norm": 3.7985067604810565,
+      "learning_rate": 1.999966908398357e-05,
+      "loss": 1.0927,
       "step": 311
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.972605918569594e-05,
-      "loss": 1.387,
+      "epoch": 0.0326530612244898,
+      "grad_norm": 4.314018944438203,
+      "learning_rate": 1.999964093331036e-05,
+      "loss": 1.0461,
       "step": 312
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.972357471794827e-05,
-      "loss": 1.1319,
+      "epoch": 0.032757718472004184,
+      "grad_norm": 3.083704093004486,
+      "learning_rate": 1.9999611633658142e-05,
+      "loss": 1.0489,
       "step": 313
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.972107919261844e-05,
-      "loss": 1.2984,
+      "epoch": 0.032862375719518576,
+      "grad_norm": 3.261286750192,
+      "learning_rate": 1.999958118503027e-05,
+      "loss": 1.0078,
       "step": 314
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.9718572612544345e-05,
-      "loss": 1.4288,
+      "epoch": 0.03296703296703297,
+      "grad_norm": 3.4697246675242375,
+      "learning_rate": 1.9999549587430252e-05,
+      "loss": 1.1564,
       "step": 315
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.9716054980576446e-05,
-      "loss": 1.213,
+      "epoch": 0.03307169021454736,
+      "grad_norm": 3.5137061405420478,
+      "learning_rate": 1.9999516840861714e-05,
+      "loss": 0.9702,
       "step": 316
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.971352629957778e-05,
-      "loss": 1.3471,
+      "epoch": 0.033176347462061745,
+      "grad_norm": 3.7031729797465465,
+      "learning_rate": 1.9999482945328422e-05,
+      "loss": 1.1463,
       "step": 317
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.9710986572423947e-05,
-      "loss": 1.2475,
+      "epoch": 0.03328100470957614,
+      "grad_norm": 3.526121821006833,
+      "learning_rate": 1.9999447900834266e-05,
+      "loss": 0.9697,
       "step": 318
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.9708435802003105e-05,
-      "loss": 1.4886,
+      "epoch": 0.03338566195709053,
+      "grad_norm": 3.582049192765141,
+      "learning_rate": 1.9999411707383273e-05,
+      "loss": 1.1307,
       "step": 319
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.9705873991215973e-05,
-      "loss": 1.3517,
+      "epoch": 0.03349031920460492,
+      "grad_norm": 3.71675942867392,
+      "learning_rate": 1.9999374364979608e-05,
+      "loss": 1.1452,
       "step": 320
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.970330114297583e-05,
-      "loss": 1.4266,
+      "epoch": 0.033594976452119306,
+      "grad_norm": 3.5193147394190074,
+      "learning_rate": 1.9999335873627555e-05,
+      "loss": 1.1801,
       "step": 321
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.970071726020849e-05,
-      "loss": 1.3975,
+      "epoch": 0.0336996336996337,
+      "grad_norm": 3.5695487111516617,
+      "learning_rate": 1.999929623333154e-05,
+      "loss": 1.0796,
       "step": 322
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.969812234585233e-05,
-      "loss": 1.2978,
+      "epoch": 0.03380429094714809,
+      "grad_norm": 3.32206675867571,
+      "learning_rate": 1.999925544409612e-05,
+      "loss": 1.0898,
       "step": 323
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.9695516402858277e-05,
-      "loss": 1.2647,
+      "epoch": 0.03390894819466248,
+      "grad_norm": 2.9565893128467153,
+      "learning_rate": 1.9999213505925975e-05,
+      "loss": 1.1213,
       "step": 324
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.9692899434189786e-05,
-      "loss": 1.3324,
+      "epoch": 0.034013605442176874,
+      "grad_norm": 3.595043850818446,
+      "learning_rate": 1.9999170418825928e-05,
+      "loss": 1.1474,
       "step": 325
     },
     {
-      "epoch": 0.21,
-      "learning_rate": 1.969027144282285e-05,
-      "loss": 1.475,
+      "epoch": 0.03411826268969126,
+      "grad_norm": 3.003534456854358,
+      "learning_rate": 1.9999126182800932e-05,
+      "loss": 1.0788,
       "step": 326
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9687632431746014e-05,
-      "loss": 1.2854,
+      "epoch": 0.03422291993720565,
+      "grad_norm": 3.9622027216915914,
+      "learning_rate": 1.9999080797856063e-05,
+      "loss": 1.1529,
       "step": 327
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9684982403960332e-05,
-      "loss": 1.3711,
+      "epoch": 0.03432757718472004,
+      "grad_norm": 3.7438205661201955,
+      "learning_rate": 1.9999034263996543e-05,
+      "loss": 1.1211,
       "step": 328
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.968232136247941e-05,
-      "loss": 1.3683,
+      "epoch": 0.034432234432234435,
+      "grad_norm": 3.1550464025240004,
+      "learning_rate": 1.9998986581227718e-05,
+      "loss": 1.1448,
       "step": 329
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9679649310329353e-05,
-      "loss": 1.3178,
+      "epoch": 0.03453689167974882,
+      "grad_norm": 3.0209980125708435,
+      "learning_rate": 1.999893774955506e-05,
+      "loss": 1.053,
       "step": 330
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9676966250548807e-05,
-      "loss": 1.3448,
+      "epoch": 0.03464154892726321,
+      "grad_norm": 3.214470861959865,
+      "learning_rate": 1.999888776898419e-05,
+      "loss": 1.0786,
       "step": 331
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.967427218618893e-05,
-      "loss": 1.3387,
+      "epoch": 0.034746206174777604,
+      "grad_norm": 3.747250332756373,
+      "learning_rate": 1.9998836639520843e-05,
+      "loss": 1.1941,
       "step": 332
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.967156712031339e-05,
-      "loss": 1.3689,
+      "epoch": 0.034850863422291996,
+      "grad_norm": 3.406359341493149,
+      "learning_rate": 1.9998784361170893e-05,
+      "loss": 1.1671,
       "step": 333
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.966885105599837e-05,
-      "loss": 1.3894,
+      "epoch": 0.03495552066980638,
+      "grad_norm": 2.8938267258829944,
+      "learning_rate": 1.9998730933940355e-05,
+      "loss": 1.0199,
       "step": 334
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9666123996332564e-05,
-      "loss": 1.3845,
+      "epoch": 0.03506017791732077,
+      "grad_norm": 3.1924364336563045,
+      "learning_rate": 1.999867635783536e-05,
+      "loss": 1.133,
       "step": 335
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.966338594441716e-05,
-      "loss": 1.3974,
+      "epoch": 0.035164835164835165,
+      "grad_norm": 3.0001944625011707,
+      "learning_rate": 1.9998620632862184e-05,
+      "loss": 1.066,
       "step": 336
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9660636903365854e-05,
-      "loss": 1.4133,
+      "epoch": 0.03526949241234956,
+      "grad_norm": 3.3739671593672447,
+      "learning_rate": 1.9998563759027228e-05,
+      "loss": 0.993,
       "step": 337
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9657876876304835e-05,
-      "loss": 1.2042,
+      "epoch": 0.03537414965986395,
+      "grad_norm": 3.0009291832792693,
+      "learning_rate": 1.9998505736337022e-05,
+      "loss": 1.0397,
       "step": 338
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.965510586637279e-05,
-      "loss": 1.3591,
+      "epoch": 0.035478806907378334,
+      "grad_norm": 3.57452005557107,
+      "learning_rate": 1.9998446564798243e-05,
+      "loss": 1.1239,
       "step": 339
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.965232387672089e-05,
-      "loss": 1.4954,
+      "epoch": 0.035583464154892726,
+      "grad_norm": 3.4334676273713898,
+      "learning_rate": 1.9998386244417686e-05,
+      "loss": 1.059,
       "step": 340
     },
     {
-      "epoch": 0.22,
-      "learning_rate": 1.9649530910512796e-05,
-      "loss": 1.3161,
+      "epoch": 0.03568812140240712,
+      "grad_norm": 3.4932921996468846,
+      "learning_rate": 1.9998324775202277e-05,
+      "loss": 1.0886,
       "step": 341
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9646726970924645e-05,
-      "loss": 1.374,
+      "epoch": 0.03579277864992151,
+      "grad_norm": 3.293255380782077,
+      "learning_rate": 1.9998262157159087e-05,
+      "loss": 0.8386,
       "step": 342
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9643912061145065e-05,
-      "loss": 1.4118,
+      "epoch": 0.035897435897435895,
+      "grad_norm": 3.7924349792602188,
+      "learning_rate": 1.99981983902953e-05,
+      "loss": 1.1097,
       "step": 343
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9641086184375148e-05,
-      "loss": 1.3582,
+      "epoch": 0.03600209314495029,
+      "grad_norm": 3.4574803613996297,
+      "learning_rate": 1.9998133474618254e-05,
+      "loss": 1.1045,
       "step": 344
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9638249343828463e-05,
-      "loss": 1.2793,
+      "epoch": 0.03610675039246468,
+      "grad_norm": 3.2469937969404272,
+      "learning_rate": 1.9998067410135403e-05,
+      "loss": 1.1738,
       "step": 345
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9635401542731043e-05,
-      "loss": 1.4125,
+      "epoch": 0.03621140763997907,
+      "grad_norm": 1.9876001648747574,
+      "learning_rate": 1.999800019685434e-05,
+      "loss": 0.7593,
       "step": 346
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.963254278432139e-05,
-      "loss": 1.3889,
+      "epoch": 0.036316064887493456,
+      "grad_norm": 3.247944294880278,
+      "learning_rate": 1.9997931834782783e-05,
+      "loss": 1.0247,
       "step": 347
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9629673071850463e-05,
-      "loss": 1.2159,
+      "epoch": 0.03642072213500785,
+      "grad_norm": 2.30114299659168,
+      "learning_rate": 1.999786232392859e-05,
+      "loss": 0.7979,
       "step": 348
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.962679240858168e-05,
-      "loss": 1.2363,
+      "epoch": 0.03652537938252224,
+      "grad_norm": 3.0984066897700964,
+      "learning_rate": 1.999779166429975e-05,
+      "loss": 1.0613,
       "step": 349
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9623900797790914e-05,
-      "loss": 1.2471,
+      "epoch": 0.03663003663003663,
+      "grad_norm": 4.166217793012304,
+      "learning_rate": 1.999771985590438e-05,
+      "loss": 1.1616,
       "step": 350
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9620998242766482e-05,
-      "loss": 1.3256,
+      "epoch": 0.036734693877551024,
+      "grad_norm": 3.8538711686650697,
+      "learning_rate": 1.999764689875073e-05,
+      "loss": 1.034,
       "step": 351
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9618084746809147e-05,
-      "loss": 1.2357,
+      "epoch": 0.03683935112506541,
+      "grad_norm": 3.2661014681676273,
+      "learning_rate": 1.9997572792847186e-05,
+      "loss": 1.0767,
       "step": 352
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9615160313232122e-05,
-      "loss": 1.4569,
+      "epoch": 0.0369440083725798,
+      "grad_norm": 3.1154642175823195,
+      "learning_rate": 1.999749753820226e-05,
+      "loss": 1.1105,
       "step": 353
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.961222494536105e-05,
-      "loss": 1.4797,
+      "epoch": 0.03704866562009419,
+      "grad_norm": 2.72464228837991,
+      "learning_rate": 1.99974211348246e-05,
+      "loss": 1.0197,
       "step": 354
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.960927864653401e-05,
-      "loss": 1.2745,
+      "epoch": 0.037153322867608585,
+      "grad_norm": 4.151004358135479,
+      "learning_rate": 1.9997343582722984e-05,
+      "loss": 1.0986,
       "step": 355
     },
     {
-      "epoch": 0.23,
-      "learning_rate": 1.9606321420101514e-05,
-      "loss": 1.3715,
+      "epoch": 0.03725798011512297,
+      "grad_norm": 3.2194916123198096,
+      "learning_rate": 1.9997264881906324e-05,
+      "loss": 0.9507,
       "step": 356
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.96033532694265e-05,
-      "loss": 1.2943,
+      "epoch": 0.03736263736263736,
+      "grad_norm": 3.094166715267168,
+      "learning_rate": 1.9997185032383663e-05,
+      "loss": 1.0726,
       "step": 357
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.960037419788433e-05,
-      "loss": 1.338,
+      "epoch": 0.037467294610151754,
+      "grad_norm": 2.858902354576879,
+      "learning_rate": 1.9997104034164176e-05,
+      "loss": 0.9607,
       "step": 358
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9597384208862776e-05,
-      "loss": 1.246,
+      "epoch": 0.037571951857666146,
+      "grad_norm": 3.59796942440944,
+      "learning_rate": 1.9997021887257166e-05,
+      "loss": 1.1792,
       "step": 359
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9594383305762045e-05,
-      "loss": 1.3237,
+      "epoch": 0.03767660910518053,
+      "grad_norm": 3.4966139920154546,
+      "learning_rate": 1.9996938591672076e-05,
+      "loss": 1.1609,
       "step": 360
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9591371491994735e-05,
-      "loss": 1.3279,
+      "epoch": 0.03778126635269492,
+      "grad_norm": 3.2932872126047164,
+      "learning_rate": 1.9996854147418477e-05,
+      "loss": 1.1843,
       "step": 361
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.958834877098586e-05,
-      "loss": 1.3185,
+      "epoch": 0.037885923600209315,
+      "grad_norm": 3.1771915837350897,
+      "learning_rate": 1.999676855450607e-05,
+      "loss": 1.1945,
       "step": 362
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9585315146172843e-05,
-      "loss": 1.3534,
+      "epoch": 0.03799058084772371,
+      "grad_norm": 3.0394535417384465,
+      "learning_rate": 1.999668181294469e-05,
+      "loss": 1.1143,
       "step": 363
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9582270621005494e-05,
-      "loss": 1.1566,
+      "epoch": 0.0380952380952381,
+      "grad_norm": 2.8537817079806067,
+      "learning_rate": 1.9996593922744308e-05,
+      "loss": 0.9112,
       "step": 364
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9579215198946028e-05,
-      "loss": 1.2134,
+      "epoch": 0.038199895342752484,
+      "grad_norm": 3.322133711285726,
+      "learning_rate": 1.9996504883915017e-05,
+      "loss": 1.1074,
       "step": 365
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9576148883469054e-05,
-      "loss": 1.3722,
+      "epoch": 0.038304552590266876,
+      "grad_norm": 3.0632114432772037,
+      "learning_rate": 1.999641469646705e-05,
+      "loss": 1.0545,
       "step": 366
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9573071678061565e-05,
-      "loss": 1.3042,
+      "epoch": 0.03840920983778127,
+      "grad_norm": 3.7007048473022235,
+      "learning_rate": 1.999632336041077e-05,
+      "loss": 1.174,
       "step": 367
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.956998358622293e-05,
-      "loss": 1.4297,
+      "epoch": 0.03851386708529566,
+      "grad_norm": 3.2674017560088946,
+      "learning_rate": 1.999623087575667e-05,
+      "loss": 1.0591,
       "step": 368
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9566884611464916e-05,
-      "loss": 1.3346,
+      "epoch": 0.038618524332810045,
+      "grad_norm": 3.522836020354182,
+      "learning_rate": 1.9996137242515376e-05,
+      "loss": 1.0767,
       "step": 369
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9563774757311646e-05,
-      "loss": 1.2136,
+      "epoch": 0.03872318158032444,
+      "grad_norm": 3.827590637468791,
+      "learning_rate": 1.9996042460697654e-05,
+      "loss": 1.046,
       "step": 370
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.956065402729963e-05,
-      "loss": 1.4095,
+      "epoch": 0.03882783882783883,
+      "grad_norm": 3.301858757724581,
+      "learning_rate": 1.9995946530314384e-05,
+      "loss": 1.0542,
       "step": 371
     },
     {
-      "epoch": 0.24,
-      "learning_rate": 1.9557522424977745e-05,
-      "loss": 1.2655,
+      "epoch": 0.03893249607535322,
+      "grad_norm": 3.119223795376493,
+      "learning_rate": 1.9995849451376593e-05,
+      "loss": 1.1795,
       "step": 372
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.9554379953907217e-05,
-      "loss": 1.3623,
+      "epoch": 0.039037153322867606,
+      "grad_norm": 3.623763948471002,
+      "learning_rate": 1.999575122389544e-05,
+      "loss": 1.0188,
       "step": 373
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.955122661766165e-05,
-      "loss": 1.3615,
+      "epoch": 0.039141810570382,
+      "grad_norm": 2.947104901430164,
+      "learning_rate": 1.9995651847882208e-05,
+      "loss": 1.0813,
       "step": 374
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.954806241982699e-05,
-      "loss": 1.4546,
+      "epoch": 0.03924646781789639,
+      "grad_norm": 3.284552361791963,
+      "learning_rate": 1.9995551323348314e-05,
+      "loss": 1.1506,
       "step": 375
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.954488736400155e-05,
-      "loss": 1.5023,
+      "epoch": 0.03935112506541078,
+      "grad_norm": 3.4207057905425566,
+      "learning_rate": 1.999544965030531e-05,
+      "loss": 1.0482,
       "step": 376
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.9541701453795976e-05,
-      "loss": 1.3962,
+      "epoch": 0.03945578231292517,
+      "grad_norm": 3.9718583768537226,
+      "learning_rate": 1.999534682876488e-05,
+      "loss": 1.107,
       "step": 377
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.9538504692833263e-05,
-      "loss": 1.313,
+      "epoch": 0.03956043956043956,
+      "grad_norm": 3.4233946505392274,
+      "learning_rate": 1.9995242858738834e-05,
+      "loss": 1.055,
       "step": 378
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.9535297084748747e-05,
-      "loss": 1.2892,
+      "epoch": 0.03966509680795395,
+      "grad_norm": 3.4939287719910905,
+      "learning_rate": 1.999513774023912e-05,
+      "loss": 1.1309,
       "step": 379
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.9532078633190097e-05,
-      "loss": 1.3584,
+      "epoch": 0.03976975405546834,
+      "grad_norm": 2.846888454584744,
+      "learning_rate": 1.9995031473277822e-05,
+      "loss": 1.0401,
       "step": 380
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.9528849341817316e-05,
-      "loss": 1.4247,
+      "epoch": 0.039874411302982735,
+      "grad_norm": 3.4884066840679826,
+      "learning_rate": 1.9994924057867142e-05,
+      "loss": 1.1144,
       "step": 381
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.9525609214302727e-05,
-      "loss": 1.4444,
+      "epoch": 0.03997906855049712,
+      "grad_norm": 3.2953541736196095,
+      "learning_rate": 1.999481549401943e-05,
+      "loss": 1.0356,
       "step": 382
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.9522358254330987e-05,
-      "loss": 1.3144,
+      "epoch": 0.04008372579801151,
+      "grad_norm": 3.8515521747365353,
+      "learning_rate": 1.999470578174715e-05,
+      "loss": 1.0333,
       "step": 383
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.951909646559906e-05,
-      "loss": 1.4022,
+      "epoch": 0.040188383045525904,
+      "grad_norm": 3.862495676492481,
+      "learning_rate": 1.999459492106292e-05,
+      "loss": 1.0685,
       "step": 384
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.951582385181623e-05,
-      "loss": 1.3433,
+      "epoch": 0.040293040293040296,
+      "grad_norm": 3.8607869501733596,
+      "learning_rate": 1.999448291197947e-05,
+      "loss": 1.0893,
       "step": 385
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.9512540416704095e-05,
-      "loss": 1.4016,
+      "epoch": 0.04039769754055468,
+      "grad_norm": 3.108316063988891,
+      "learning_rate": 1.999436975450967e-05,
+      "loss": 1.129,
       "step": 386
     },
     {
-      "epoch": 0.25,
-      "learning_rate": 1.950924616399655e-05,
-      "loss": 1.3799,
+      "epoch": 0.04050235478806907,
+      "grad_norm": 3.1460718799511724,
+      "learning_rate": 1.9994255448666528e-05,
+      "loss": 1.0922,
       "step": 387
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9505941097439805e-05,
-      "loss": 1.4395,
+      "epoch": 0.040607012035583465,
+      "grad_norm": 4.815195078193566,
+      "learning_rate": 1.9994139994463174e-05,
+      "loss": 1.0287,
       "step": 388
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9502625220792345e-05,
-      "loss": 1.303,
+      "epoch": 0.04071166928309786,
+      "grad_norm": 3.2133489103871464,
+      "learning_rate": 1.9994023391912873e-05,
+      "loss": 0.9636,
       "step": 389
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9499298537824972e-05,
-      "loss": 1.3902,
+      "epoch": 0.04081632653061224,
+      "grad_norm": 3.0323586032384466,
+      "learning_rate": 1.9993905641029024e-05,
+      "loss": 1.1492,
       "step": 390
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.949596105232077e-05,
-      "loss": 1.3047,
+      "epoch": 0.040920983778126634,
+      "grad_norm": 3.408180618813306,
+      "learning_rate": 1.999378674182516e-05,
+      "loss": 1.1212,
       "step": 391
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9492612768075094e-05,
-      "loss": 1.1657,
+      "epoch": 0.041025641025641026,
+      "grad_norm": 2.832485184741635,
+      "learning_rate": 1.9993666694314938e-05,
+      "loss": 1.1055,
       "step": 392
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9489253688895594e-05,
-      "loss": 1.4182,
+      "epoch": 0.04113029827315542,
+      "grad_norm": 3.3779814563246826,
+      "learning_rate": 1.999354549851215e-05,
+      "loss": 1.0783,
       "step": 393
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9485883818602194e-05,
-      "loss": 1.205,
+      "epoch": 0.04123495552066981,
+      "grad_norm": 3.15338999466506,
+      "learning_rate": 1.9993423154430732e-05,
+      "loss": 1.1143,
       "step": 394
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.948250316102709e-05,
-      "loss": 1.3438,
+      "epoch": 0.041339612768184195,
+      "grad_norm": 3.651309843882222,
+      "learning_rate": 1.9993299662084734e-05,
+      "loss": 1.1113,
       "step": 395
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9479111720014737e-05,
-      "loss": 1.3099,
+      "epoch": 0.04144427001569859,
+      "grad_norm": 2.5508921394839152,
+      "learning_rate": 1.9993175021488343e-05,
+      "loss": 0.8452,
       "step": 396
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.947570949942186e-05,
-      "loss": 1.346,
+      "epoch": 0.04154892726321298,
+      "grad_norm": 3.3357843841110077,
+      "learning_rate": 1.9993049232655882e-05,
+      "loss": 1.1295,
       "step": 397
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9472296503117437e-05,
-      "loss": 1.4133,
+      "epoch": 0.04165358451072737,
+      "grad_norm": 3.9695559727733483,
+      "learning_rate": 1.999292229560181e-05,
+      "loss": 0.9869,
       "step": 398
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9468872734982705e-05,
-      "loss": 1.5194,
+      "epoch": 0.041758241758241756,
+      "grad_norm": 3.7523305234520024,
+      "learning_rate": 1.9992794210340707e-05,
+      "loss": 1.1228,
       "step": 399
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9465438198911158e-05,
-      "loss": 1.3365,
+      "epoch": 0.04186289900575615,
+      "grad_norm": 3.5321821977005503,
+      "learning_rate": 1.999266497688729e-05,
+      "loss": 1.0244,
       "step": 400
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.946199289880852e-05,
-      "loss": 1.3004,
+      "epoch": 0.04196755625327054,
+      "grad_norm": 3.433702518298409,
+      "learning_rate": 1.9992534595256414e-05,
+      "loss": 1.1635,
       "step": 401
     },
     {
-      "epoch": 0.26,
-      "learning_rate": 1.9458536838592764e-05,
-      "loss": 1.4811,
+      "epoch": 0.04207221350078493,
+      "grad_norm": 3.3240435185890482,
+      "learning_rate": 1.9992403065463053e-05,
+      "loss": 1.1308,
       "step": 402
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9455070022194094e-05,
-      "loss": 1.3259,
+      "epoch": 0.04217687074829932,
+      "grad_norm": 2.418715245595869,
+      "learning_rate": 1.9992270387522327e-05,
+      "loss": 0.7399,
       "step": 403
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9451592453554956e-05,
-      "loss": 1.2946,
+      "epoch": 0.04228152799581371,
+      "grad_norm": 3.4217501748246497,
+      "learning_rate": 1.999213656144947e-05,
+      "loss": 1.0257,
       "step": 404
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.944810413663001e-05,
-      "loss": 1.2141,
+      "epoch": 0.0423861852433281,
+      "grad_norm": 3.5574880915366642,
+      "learning_rate": 1.9992001587259872e-05,
+      "loss": 1.1526,
       "step": 405
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9444605075386156e-05,
-      "loss": 1.4021,
+      "epoch": 0.04249084249084249,
+      "grad_norm": 3.439254247799819,
+      "learning_rate": 1.9991865464969035e-05,
+      "loss": 1.0449,
       "step": 406
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.944109527380249e-05,
-      "loss": 1.455,
+      "epoch": 0.042595499738356885,
+      "grad_norm": 3.5204318338726273,
+      "learning_rate": 1.99917281945926e-05,
+      "loss": 1.1471,
       "step": 407
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9437574735870346e-05,
-      "loss": 1.4715,
+      "epoch": 0.04270015698587127,
+      "grad_norm": 3.858873349887819,
+      "learning_rate": 1.9991589776146335e-05,
+      "loss": 1.0934,
       "step": 408
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9434043465593247e-05,
-      "loss": 1.4539,
+      "epoch": 0.04280481423338566,
+      "grad_norm": 3.362234619886235,
+      "learning_rate": 1.9991450209646153e-05,
+      "loss": 1.1658,
       "step": 409
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9430501466986932e-05,
-      "loss": 1.4801,
+      "epoch": 0.042909471480900054,
+      "grad_norm": 3.220742237697856,
+      "learning_rate": 1.999130949510809e-05,
+      "loss": 1.1169,
       "step": 410
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9426948744079338e-05,
-      "loss": 1.2148,
+      "epoch": 0.043014128728414446,
+      "grad_norm": 2.9342623132519536,
+      "learning_rate": 1.999116763254831e-05,
+      "loss": 0.9709,
       "step": 411
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.94233853009106e-05,
-      "loss": 1.2596,
+      "epoch": 0.04311878597592883,
+      "grad_norm": 3.0789423142360572,
+      "learning_rate": 1.9991024621983115e-05,
+      "loss": 1.0549,
       "step": 412
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9419811141533032e-05,
-      "loss": 1.2145,
+      "epoch": 0.04322344322344322,
+      "grad_norm": 3.5769673216250717,
+      "learning_rate": 1.9990880463428938e-05,
+      "loss": 1.0587,
       "step": 413
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9416226270011152e-05,
-      "loss": 1.2662,
+      "epoch": 0.043328100470957615,
+      "grad_norm": 3.387771571360907,
+      "learning_rate": 1.9990735156902337e-05,
+      "loss": 1.1727,
       "step": 414
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9412630690421648e-05,
-      "loss": 1.302,
+      "epoch": 0.043432757718472007,
+      "grad_norm": 3.8004457138302237,
+      "learning_rate": 1.9990588702420017e-05,
+      "loss": 1.0929,
       "step": 415
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.940902440685339e-05,
-      "loss": 1.2591,
+      "epoch": 0.04353741496598639,
+      "grad_norm": 2.9911775987883606,
+      "learning_rate": 1.9990441099998802e-05,
+      "loss": 1.0605,
       "step": 416
     },
     {
-      "epoch": 0.27,
-      "learning_rate": 1.9405407423407414e-05,
-      "loss": 1.4317,
+      "epoch": 0.043642072213500784,
+      "grad_norm": 3.3225073464157875,
+      "learning_rate": 1.999029234965565e-05,
+      "loss": 1.101,
       "step": 417
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.9401779744196934e-05,
-      "loss": 1.3283,
+      "epoch": 0.043746729461015176,
+      "grad_norm": 3.142201219435067,
+      "learning_rate": 1.9990142451407658e-05,
+      "loss": 1.1804,
       "step": 418
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.9398141373347317e-05,
-      "loss": 1.2165,
+      "epoch": 0.04385138670852957,
+      "grad_norm": 3.2497520075753905,
+      "learning_rate": 1.9989991405272043e-05,
+      "loss": 1.09,
       "step": 419
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.93944923149961e-05,
-      "loss": 1.3613,
+      "epoch": 0.04395604395604396,
+      "grad_norm": 2.8504351959056646,
+      "learning_rate": 1.9989839211266164e-05,
+      "loss": 1.089,
       "step": 420
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.9390832573292964e-05,
-      "loss": 1.3744,
+      "epoch": 0.044060701203558345,
+      "grad_norm": 3.425903015712905,
+      "learning_rate": 1.998968586940751e-05,
+      "loss": 1.1222,
       "step": 421
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.938716215239974e-05,
-      "loss": 1.1967,
+      "epoch": 0.044165358451072737,
+      "grad_norm": 3.12857526376575,
+      "learning_rate": 1.9989531379713697e-05,
+      "loss": 1.1248,
       "step": 422
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.9383481056490402e-05,
-      "loss": 1.2715,
+      "epoch": 0.04427001569858713,
+      "grad_norm": 2.753045621554163,
+      "learning_rate": 1.998937574220248e-05,
+      "loss": 1.076,
       "step": 423
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.9379789289751073e-05,
-      "loss": 1.3597,
+      "epoch": 0.04437467294610152,
+      "grad_norm": 3.9313106712974,
+      "learning_rate": 1.998921895689174e-05,
+      "loss": 0.9975,
       "step": 424
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.9376086856380007e-05,
-      "loss": 1.4245,
+      "epoch": 0.044479330193615906,
+      "grad_norm": 3.1879760675576465,
+      "learning_rate": 1.9989061023799486e-05,
+      "loss": 1.0489,
       "step": 425
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.937237376058758e-05,
-      "loss": 1.3464,
+      "epoch": 0.0445839874411303,
+      "grad_norm": 3.5604479042648447,
+      "learning_rate": 1.9988901942943874e-05,
+      "loss": 1.0179,
       "step": 426
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.93686500065963e-05,
-      "loss": 1.4296,
+      "epoch": 0.04468864468864469,
+      "grad_norm": 3.4144953084922376,
+      "learning_rate": 1.998874171434318e-05,
+      "loss": 0.9328,
       "step": 427
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.9364915598640796e-05,
-      "loss": 1.2627,
+      "epoch": 0.04479330193615908,
+      "grad_norm": 2.955908397360233,
+      "learning_rate": 1.998858033801581e-05,
+      "loss": 1.1155,
       "step": 428
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.936117054096781e-05,
-      "loss": 1.2727,
+      "epoch": 0.044897959183673466,
+      "grad_norm": 3.6880350580445254,
+      "learning_rate": 1.9988417813980315e-05,
+      "loss": 1.1311,
       "step": 429
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.9357414837836192e-05,
-      "loss": 1.2721,
+      "epoch": 0.04500261643118786,
+      "grad_norm": 3.3607014595989178,
+      "learning_rate": 1.9988254142255362e-05,
+      "loss": 1.1893,
       "step": 430
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.935364849351691e-05,
-      "loss": 1.3247,
+      "epoch": 0.04510727367870225,
+      "grad_norm": 2.8727076333580395,
+      "learning_rate": 1.998808932285976e-05,
+      "loss": 0.9684,
       "step": 431
     },
     {
-      "epoch": 0.28,
-      "learning_rate": 1.934987151229302e-05,
-      "loss": 1.5809,
+      "epoch": 0.04521193092621664,
+      "grad_norm": 3.439241913580116,
+      "learning_rate": 1.9987923355812448e-05,
+      "loss": 1.0141,
       "step": 432
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.934608389845968e-05,
-      "loss": 1.3678,
+      "epoch": 0.04531658817373103,
+      "grad_norm": 2.9285693240861783,
+      "learning_rate": 1.9987756241132494e-05,
+      "loss": 1.0593,
       "step": 433
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.9342285656324135e-05,
-      "loss": 1.4506,
+      "epoch": 0.04542124542124542,
+      "grad_norm": 3.670926244435246,
+      "learning_rate": 1.99875879788391e-05,
+      "loss": 1.0491,
       "step": 434
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.9338476790205726e-05,
-      "loss": 1.2935,
+      "epoch": 0.04552590266875981,
+      "grad_norm": 3.006014723154654,
+      "learning_rate": 1.99874185689516e-05,
+      "loss": 1.1722,
       "step": 435
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.933465730443586e-05,
-      "loss": 1.1895,
+      "epoch": 0.0456305599162742,
+      "grad_norm": 3.7892151003300305,
+      "learning_rate": 1.9987248011489462e-05,
+      "loss": 1.0608,
       "step": 436
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.9330827203358044e-05,
-      "loss": 1.248,
+      "epoch": 0.045735217163788595,
+      "grad_norm": 3.2332272253735765,
+      "learning_rate": 1.998707630647228e-05,
+      "loss": 1.1638,
       "step": 437
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.932698649132783e-05,
-      "loss": 1.3905,
+      "epoch": 0.04583987441130298,
+      "grad_norm": 3.480839997662857,
+      "learning_rate": 1.998690345391978e-05,
+      "loss": 1.0832,
       "step": 438
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.932313517271285e-05,
-      "loss": 1.3918,
+      "epoch": 0.04594453165881737,
+      "grad_norm": 3.8054324863562385,
+      "learning_rate": 1.9986729453851833e-05,
+      "loss": 1.0328,
       "step": 439
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.9319273251892805e-05,
-      "loss": 1.3228,
+      "epoch": 0.046049188906331764,
+      "grad_norm": 3.8488459066381497,
+      "learning_rate": 1.998655430628843e-05,
+      "loss": 0.9417,
       "step": 440
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.931540073325944e-05,
-      "loss": 1.4809,
+      "epoch": 0.046153846153846156,
+      "grad_norm": 3.1253018165861786,
+      "learning_rate": 1.9986378011249684e-05,
+      "loss": 1.1043,
       "step": 441
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.9311517621216555e-05,
-      "loss": 1.4098,
+      "epoch": 0.04625850340136054,
+      "grad_norm": 3.5195863235640408,
+      "learning_rate": 1.9986200568755863e-05,
+      "loss": 1.1416,
       "step": 442
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.930762392018e-05,
-      "loss": 1.2479,
+      "epoch": 0.04636316064887493,
+      "grad_norm": 3.38872021874277,
+      "learning_rate": 1.9986021978827353e-05,
+      "loss": 1.1378,
       "step": 443
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.9303719634577668e-05,
-      "loss": 1.4779,
+      "epoch": 0.046467817896389325,
+      "grad_norm": 3.2198779898823475,
+      "learning_rate": 1.9985842241484678e-05,
+      "loss": 1.1032,
       "step": 444
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.929980476884948e-05,
-      "loss": 1.2986,
+      "epoch": 0.04657247514390372,
+      "grad_norm": 3.037988731718675,
+      "learning_rate": 1.998566135674848e-05,
+      "loss": 1.0834,
       "step": 445
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.92958793274474e-05,
-      "loss": 1.3807,
+      "epoch": 0.0466771323914181,
+      "grad_norm": 3.113016605036802,
+      "learning_rate": 1.998547932463955e-05,
+      "loss": 1.0245,
       "step": 446
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.9291943314835406e-05,
-      "loss": 1.535,
+      "epoch": 0.046781789638932494,
+      "grad_norm": 2.981225578291425,
+      "learning_rate": 1.9985296145178803e-05,
+      "loss": 1.1454,
       "step": 447
     },
     {
-      "epoch": 0.29,
-      "learning_rate": 1.928799673548951e-05,
-      "loss": 1.1961,
+      "epoch": 0.046886446886446886,
+      "grad_norm": 3.2756449737549818,
+      "learning_rate": 1.998511181838729e-05,
+      "loss": 1.0624,
       "step": 448
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.9284039593897734e-05,
-      "loss": 1.327,
+      "epoch": 0.04699110413396128,
+      "grad_norm": 3.2146017721872417,
+      "learning_rate": 1.9984926344286184e-05,
+      "loss": 1.121,
       "step": 449
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.928007189456011e-05,
-      "loss": 1.5256,
+      "epoch": 0.04709576138147567,
+      "grad_norm": 3.6515983191470642,
+      "learning_rate": 1.99847397228968e-05,
+      "loss": 1.0705,
       "step": 450
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.9276093641988684e-05,
-      "loss": 1.5381,
+      "epoch": 0.047200418628990055,
+      "grad_norm": 2.7925140240267536,
+      "learning_rate": 1.998455195424058e-05,
+      "loss": 1.0412,
       "step": 451
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.9272104840707487e-05,
-      "loss": 1.323,
+      "epoch": 0.04730507587650445,
+      "grad_norm": 3.2058786484672184,
+      "learning_rate": 1.99843630383391e-05,
+      "loss": 1.0438,
       "step": 452
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.9268105495252568e-05,
-      "loss": 1.3281,
+      "epoch": 0.04740973312401884,
+      "grad_norm": 3.206199705921322,
+      "learning_rate": 1.9984172975214068e-05,
+      "loss": 1.1241,
       "step": 453
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.9264095610171944e-05,
-      "loss": 1.3403,
+      "epoch": 0.04751439037153323,
+      "grad_norm": 2.96917837023507,
+      "learning_rate": 1.998398176488732e-05,
+      "loss": 1.0307,
       "step": 454
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.926007519002564e-05,
-      "loss": 1.3414,
+      "epoch": 0.047619047619047616,
+      "grad_norm": 2.2142798706812514,
+      "learning_rate": 1.9983789407380828e-05,
+      "loss": 0.7792,
       "step": 455
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.925604423938564e-05,
-      "loss": 1.2793,
+      "epoch": 0.04772370486656201,
+      "grad_norm": 2.1393293337536754,
+      "learning_rate": 1.9983595902716693e-05,
+      "loss": 0.8014,
       "step": 456
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.9252002762835917e-05,
-      "loss": 1.3011,
+      "epoch": 0.0478283621140764,
+      "grad_norm": 3.088242788316876,
+      "learning_rate": 1.998340125091715e-05,
+      "loss": 1.0111,
       "step": 457
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.924795076497241e-05,
-      "loss": 1.2783,
+      "epoch": 0.04793301936159079,
+      "grad_norm": 3.001125565502091,
+      "learning_rate": 1.9983205452004566e-05,
+      "loss": 0.994,
       "step": 458
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.924388825040303e-05,
-      "loss": 1.2557,
+      "epoch": 0.04803767660910518,
+      "grad_norm": 3.783192428283837,
+      "learning_rate": 1.9983008506001437e-05,
+      "loss": 1.1141,
       "step": 459
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.9239815223747632e-05,
-      "loss": 1.3768,
+      "epoch": 0.04814233385661957,
+      "grad_norm": 3.004204145609777,
+      "learning_rate": 1.9982810412930393e-05,
+      "loss": 1.1247,
       "step": 460
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.9235731689638035e-05,
-      "loss": 1.2036,
+      "epoch": 0.04824699110413396,
+      "grad_norm": 3.1685378042245147,
+      "learning_rate": 1.9982611172814197e-05,
+      "loss": 1.1299,
       "step": 461
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.9231637652718004e-05,
-      "loss": 1.2012,
+      "epoch": 0.04835164835164835,
+      "grad_norm": 3.1587611211309907,
+      "learning_rate": 1.9982410785675735e-05,
+      "loss": 1.112,
       "step": 462
     },
     {
-      "epoch": 0.3,
-      "learning_rate": 1.9227533117643256e-05,
-      "loss": 1.4446,
+      "epoch": 0.048456305599162745,
+      "grad_norm": 3.458436913532201,
+      "learning_rate": 1.9982209251538043e-05,
+      "loss": 1.0351,
       "step": 463
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.922341808908144e-05,
-      "loss": 1.3174,
+      "epoch": 0.04856096284667713,
+      "grad_norm": 2.969870331031836,
+      "learning_rate": 1.998200657042427e-05,
+      "loss": 1.1774,
       "step": 464
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.921929257171213e-05,
-      "loss": 1.3249,
+      "epoch": 0.04866562009419152,
+      "grad_norm": 3.763085180450192,
+      "learning_rate": 1.9981802742357704e-05,
+      "loss": 1.0587,
       "step": 465
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.9215156570226844e-05,
-      "loss": 1.255,
+      "epoch": 0.048770277341705914,
+      "grad_norm": 3.33040533060541,
+      "learning_rate": 1.998159776736177e-05,
+      "loss": 1.0296,
       "step": 466
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.921101008932901e-05,
-      "loss": 1.2195,
+      "epoch": 0.048874934589220306,
+      "grad_norm": 3.290492374884369,
+      "learning_rate": 1.998139164546002e-05,
+      "loss": 1.2095,
       "step": 467
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.920685313373398e-05,
-      "loss": 1.3089,
+      "epoch": 0.04897959183673469,
+      "grad_norm": 2.857263563106438,
+      "learning_rate": 1.998118437667613e-05,
+      "loss": 0.9923,
       "step": 468
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.920268570816901e-05,
-      "loss": 1.2939,
+      "epoch": 0.04908424908424908,
+      "grad_norm": 3.053876121052991,
+      "learning_rate": 1.9980975961033925e-05,
+      "loss": 0.8369,
       "step": 469
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.9198507817373274e-05,
-      "loss": 1.4276,
+      "epoch": 0.049188906331763475,
+      "grad_norm": 3.177534083313426,
+      "learning_rate": 1.998076639855735e-05,
+      "loss": 1.0123,
       "step": 470
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.919431946609784e-05,
-      "loss": 1.3933,
+      "epoch": 0.04929356357927787,
+      "grad_norm": 4.140885802398993,
+      "learning_rate": 1.998055568927048e-05,
+      "loss": 0.9744,
       "step": 471
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.9190120659105663e-05,
-      "loss": 1.3057,
+      "epoch": 0.04939822082679225,
+      "grad_norm": 2.9041664415140396,
+      "learning_rate": 1.9980343833197528e-05,
+      "loss": 1.0679,
       "step": 472
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.9185911401171615e-05,
-      "loss": 1.3188,
+      "epoch": 0.049502878074306644,
+      "grad_norm": 2.942481635418397,
+      "learning_rate": 1.998013083036284e-05,
+      "loss": 1.099,
       "step": 473
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.918169169708242e-05,
-      "loss": 1.2174,
+      "epoch": 0.049607535321821036,
+      "grad_norm": 2.9546879658340486,
+      "learning_rate": 1.9979916680790885e-05,
+      "loss": 1.0112,
       "step": 474
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.91774615516367e-05,
-      "loss": 1.4041,
+      "epoch": 0.04971219256933543,
+      "grad_norm": 3.0965296597879113,
+      "learning_rate": 1.997970138450627e-05,
+      "loss": 1.0086,
       "step": 475
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.9173220969644947e-05,
-      "loss": 1.1018,
+      "epoch": 0.04981684981684982,
+      "grad_norm": 3.855575845463165,
+      "learning_rate": 1.997948494153374e-05,
+      "loss": 1.1754,
       "step": 476
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.916896995592953e-05,
-      "loss": 1.3083,
+      "epoch": 0.049921507064364205,
+      "grad_norm": 3.993215917358464,
+      "learning_rate": 1.997926735189816e-05,
+      "loss": 1.0746,
       "step": 477
     },
     {
-      "epoch": 0.31,
-      "learning_rate": 1.916470851532466e-05,
-      "loss": 1.1981,
+      "epoch": 0.0500261643118786,
+      "grad_norm": 3.0492290153570094,
+      "learning_rate": 1.9979048615624526e-05,
+      "loss": 1.1768,
       "step": 478
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.9160436652676425e-05,
-      "loss": 1.4576,
+      "epoch": 0.05013082155939299,
+      "grad_norm": 3.6838636144975068,
+      "learning_rate": 1.997882873273798e-05,
+      "loss": 1.0034,
       "step": 479
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.9156154372842757e-05,
-      "loss": 1.1978,
+      "epoch": 0.05023547880690738,
+      "grad_norm": 3.4794216494384926,
+      "learning_rate": 1.9978607703263783e-05,
+      "loss": 1.0218,
       "step": 480
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.9151861680693432e-05,
-      "loss": 1.193,
+      "epoch": 0.050340136054421766,
+      "grad_norm": 3.241477320849195,
+      "learning_rate": 1.9978385527227334e-05,
+      "loss": 1.0337,
       "step": 481
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.914755858111008e-05,
-      "loss": 1.2737,
+      "epoch": 0.05044479330193616,
+      "grad_norm": 3.6191060732119857,
+      "learning_rate": 1.9978162204654155e-05,
+      "loss": 1.1394,
       "step": 482
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.914324507898614e-05,
-      "loss": 1.3568,
+      "epoch": 0.05054945054945055,
+      "grad_norm": 2.8341172658181253,
+      "learning_rate": 1.9977937735569915e-05,
+      "loss": 1.0184,
       "step": 483
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.9138921179226908e-05,
-      "loss": 1.4216,
+      "epoch": 0.05065410779696494,
+      "grad_norm": 3.16341632485204,
+      "learning_rate": 1.99777121200004e-05,
+      "loss": 1.0631,
       "step": 484
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.9134586886749492e-05,
-      "loss": 1.2753,
+      "epoch": 0.05075876504447933,
+      "grad_norm": 3.015431914842525,
+      "learning_rate": 1.9977485357971535e-05,
+      "loss": 1.2249,
       "step": 485
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.9130242206482817e-05,
-      "loss": 1.3887,
+      "epoch": 0.05086342229199372,
+      "grad_norm": 15.10309744378473,
+      "learning_rate": 1.997725744950938e-05,
+      "loss": 1.0752,
       "step": 486
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.912588714336762e-05,
-      "loss": 1.3849,
+      "epoch": 0.05096807953950811,
+      "grad_norm": 3.0329623555885776,
+      "learning_rate": 1.9977028394640113e-05,
+      "loss": 1.1592,
       "step": 487
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.912152170235646e-05,
-      "loss": 1.3427,
+      "epoch": 0.0510727367870225,
+      "grad_norm": 2.993215052911289,
+      "learning_rate": 1.997679819339006e-05,
+      "loss": 1.1015,
       "step": 488
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.9117145888413678e-05,
-      "loss": 1.3543,
+      "epoch": 0.05117739403453689,
+      "grad_norm": 3.028558942288711,
+      "learning_rate": 1.997656684578567e-05,
+      "loss": 1.0943,
       "step": 489
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.911275970651542e-05,
-      "loss": 1.3034,
+      "epoch": 0.05128205128205128,
+      "grad_norm": 2.6575824106168677,
+      "learning_rate": 1.9976334351853522e-05,
+      "loss": 1.1031,
       "step": 490
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.910836316164962e-05,
-      "loss": 1.4744,
+      "epoch": 0.05138670852956567,
+      "grad_norm": 3.262048948295798,
+      "learning_rate": 1.997610071162033e-05,
+      "loss": 1.0627,
       "step": 491
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.9103956258816008e-05,
-      "loss": 1.3314,
+      "epoch": 0.051491365777080064,
+      "grad_norm": 2.942164825678646,
+      "learning_rate": 1.997586592511295e-05,
+      "loss": 1.1196,
       "step": 492
     },
     {
-      "epoch": 0.32,
-      "learning_rate": 1.909953900302607e-05,
-      "loss": 1.4215,
+      "epoch": 0.051596023024594456,
+      "grad_norm": 3.143793552468755,
+      "learning_rate": 1.997562999235835e-05,
+      "loss": 1.1187,
       "step": 493
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.909511139930309e-05,
-      "loss": 1.2962,
+      "epoch": 0.05170068027210884,
+      "grad_norm": 3.6555595068602846,
+      "learning_rate": 1.997539291338364e-05,
+      "loss": 0.9745,
       "step": 494
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9090673452682108e-05,
-      "loss": 1.3132,
+      "epoch": 0.05180533751962323,
+      "grad_norm": 3.3509907787982494,
+      "learning_rate": 1.9975154688216064e-05,
+      "loss": 0.9672,
       "step": 495
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9086225168209917e-05,
-      "loss": 1.2975,
+      "epoch": 0.051909994767137625,
+      "grad_norm": 3.446789550285205,
+      "learning_rate": 1.997491531688299e-05,
+      "loss": 1.0936,
       "step": 496
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9081766550945084e-05,
-      "loss": 1.3207,
+      "epoch": 0.05201465201465202,
+      "grad_norm": 3.215446069445594,
+      "learning_rate": 1.9974674799411927e-05,
+      "loss": 1.1857,
       "step": 497
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.907729760595791e-05,
-      "loss": 1.3353,
+      "epoch": 0.0521193092621664,
+      "grad_norm": 3.694397906430922,
+      "learning_rate": 1.9974433135830505e-05,
+      "loss": 1.0549,
       "step": 498
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9072818338330455e-05,
-      "loss": 1.2535,
+      "epoch": 0.052223966509680794,
+      "grad_norm": 3.5507510319349462,
+      "learning_rate": 1.9974190326166498e-05,
+      "loss": 1.0954,
       "step": 499
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9068328753156514e-05,
-      "loss": 1.0754,
+      "epoch": 0.052328623757195186,
+      "grad_norm": 2.911759573200471,
+      "learning_rate": 1.9973946370447804e-05,
+      "loss": 1.1315,
       "step": 500
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9063828855541605e-05,
-      "loss": 1.3487,
+      "epoch": 0.05243328100470958,
+      "grad_norm": 3.830707804003234,
+      "learning_rate": 1.9973701268702454e-05,
+      "loss": 1.1961,
       "step": 501
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9059318650602986e-05,
-      "loss": 1.4404,
+      "epoch": 0.05253793825222396,
+      "grad_norm": 3.5874003170282953,
+      "learning_rate": 1.9973455020958602e-05,
+      "loss": 1.1831,
       "step": 502
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.905479814346962e-05,
-      "loss": 1.2397,
+      "epoch": 0.052642595499738355,
+      "grad_norm": 5.067523741996903,
+      "learning_rate": 1.9973207627244556e-05,
+      "loss": 1.06,
       "step": 503
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9050267339282214e-05,
-      "loss": 1.378,
+      "epoch": 0.05274725274725275,
+      "grad_norm": 3.126368066369804,
+      "learning_rate": 1.9972959087588734e-05,
+      "loss": 0.7553,
       "step": 504
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9045726243193155e-05,
-      "loss": 1.2465,
+      "epoch": 0.05285190999476714,
+      "grad_norm": 3.7926866203618923,
+      "learning_rate": 1.9972709402019696e-05,
+      "loss": 1.1616,
       "step": 505
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.904117486036655e-05,
-      "loss": 1.3214,
+      "epoch": 0.05295656724228153,
+      "grad_norm": 3.098260850078986,
+      "learning_rate": 1.9972458570566134e-05,
+      "loss": 1.0706,
       "step": 506
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9036613195978198e-05,
-      "loss": 1.3161,
+      "epoch": 0.053061224489795916,
+      "grad_norm": 3.121342346390446,
+      "learning_rate": 1.9972206593256863e-05,
+      "loss": 1.0746,
       "step": 507
     },
     {
-      "epoch": 0.33,
-      "learning_rate": 1.9032041255215596e-05,
-      "loss": 1.2936,
+      "epoch": 0.05316588173731031,
+      "grad_norm": 2.3263594607795484,
+      "learning_rate": 1.997195347012084e-05,
+      "loss": 0.8619,
       "step": 508
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.9027459043277923e-05,
-      "loss": 1.436,
+      "epoch": 0.0532705389848247,
+      "grad_norm": 4.313521927576008,
+      "learning_rate": 1.997169920118714e-05,
+      "loss": 0.9864,
       "step": 509
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.902286656537603e-05,
-      "loss": 1.2758,
+      "epoch": 0.05337519623233909,
+      "grad_norm": 2.8736994783195966,
+      "learning_rate": 1.9971443786485e-05,
+      "loss": 1.1794,
       "step": 510
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.9018263826732467e-05,
-      "loss": 1.3259,
+      "epoch": 0.05347985347985348,
+      "grad_norm": 2.67925787298318,
+      "learning_rate": 1.9971187226043746e-05,
+      "loss": 0.92,
       "step": 511
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.9013650832581424e-05,
-      "loss": 1.3638,
+      "epoch": 0.05358451072736787,
+      "grad_norm": 2.7647374456372544,
+      "learning_rate": 1.997092951989287e-05,
+      "loss": 1.1168,
       "step": 512
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.9009027588168773e-05,
-      "loss": 1.357,
+      "epoch": 0.05368916797488226,
+      "grad_norm": 3.207257190593513,
+      "learning_rate": 1.9970670668061977e-05,
+      "loss": 1.182,
       "step": 513
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.900439409875203e-05,
-      "loss": 1.3922,
+      "epoch": 0.05379382522239665,
+      "grad_norm": 2.9763787397307384,
+      "learning_rate": 1.997041067058081e-05,
+      "loss": 1.0417,
       "step": 514
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.899975036960037e-05,
-      "loss": 1.2587,
+      "epoch": 0.05389848246991104,
+      "grad_norm": 3.3432997678318634,
+      "learning_rate": 1.997014952747925e-05,
+      "loss": 1.051,
       "step": 515
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.8995096405994614e-05,
-      "loss": 1.4294,
+      "epoch": 0.05400313971742543,
+      "grad_norm": 2.789610544658323,
+      "learning_rate": 1.996988723878729e-05,
+      "loss": 1.1173,
       "step": 516
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.899043221322721e-05,
-      "loss": 1.244,
+      "epoch": 0.05410779696493982,
+      "grad_norm": 3.7956635314776,
+      "learning_rate": 1.9969623804535084e-05,
+      "loss": 1.0319,
       "step": 517
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.898575779660225e-05,
-      "loss": 1.1213,
+      "epoch": 0.054212454212454214,
+      "grad_norm": 3.0432435518135597,
+      "learning_rate": 1.9969359224752884e-05,
+      "loss": 1.0832,
       "step": 518
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.8981073161435447e-05,
-      "loss": 1.4039,
+      "epoch": 0.054317111459968606,
+      "grad_norm": 3.060370744438756,
+      "learning_rate": 1.9969093499471106e-05,
+      "loss": 1.0148,
       "step": 519
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.897637831305414e-05,
-      "loss": 1.3606,
+      "epoch": 0.05442176870748299,
+      "grad_norm": 3.0599503263592265,
+      "learning_rate": 1.996882662872027e-05,
+      "loss": 0.9405,
       "step": 520
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.8971673256797275e-05,
-      "loss": 1.4622,
+      "epoch": 0.05452642595499738,
+      "grad_norm": 2.8659069744524346,
+      "learning_rate": 1.996855861253105e-05,
+      "loss": 1.0298,
       "step": 521
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.8966957998015403e-05,
-      "loss": 1.3161,
+      "epoch": 0.054631083202511775,
+      "grad_norm": 3.698825550046783,
+      "learning_rate": 1.9968289450934235e-05,
+      "loss": 1.163,
       "step": 522
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.896223254207069e-05,
-      "loss": 1.3393,
+      "epoch": 0.05473574045002617,
+      "grad_norm": 3.0714368145204616,
+      "learning_rate": 1.9968019143960755e-05,
+      "loss": 1.0566,
       "step": 523
     },
     {
-      "epoch": 0.34,
-      "learning_rate": 1.89574968943369e-05,
-      "loss": 1.3398,
+      "epoch": 0.05484039769754055,
+      "grad_norm": 2.9908123101769912,
+      "learning_rate": 1.9967747691641667e-05,
+      "loss": 1.1947,
       "step": 524
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.895275106019937e-05,
-      "loss": 1.462,
+      "epoch": 0.054945054945054944,
+      "grad_norm": 3.455634138607162,
+      "learning_rate": 1.996747509400816e-05,
+      "loss": 1.1672,
       "step": 525
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.8947995045055023e-05,
-      "loss": 1.3711,
+      "epoch": 0.055049712192569336,
+      "grad_norm": 2.910243752165646,
+      "learning_rate": 1.9967201351091562e-05,
+      "loss": 1.1195,
       "step": 526
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.894322885431238e-05,
-      "loss": 1.4027,
+      "epoch": 0.05515436944008373,
+      "grad_norm": 2.779316157067133,
+      "learning_rate": 1.9966926462923324e-05,
+      "loss": 0.8195,
       "step": 527
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.8938452493391515e-05,
-      "loss": 1.4405,
+      "epoch": 0.05525902668759811,
+      "grad_norm": 3.1584634955840154,
+      "learning_rate": 1.996665042953503e-05,
+      "loss": 1.1175,
       "step": 528
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.8933665967724072e-05,
-      "loss": 1.3696,
+      "epoch": 0.055363683935112505,
+      "grad_norm": 2.6746711631574938,
+      "learning_rate": 1.9966373250958395e-05,
+      "loss": 0.9986,
       "step": 529
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.892886928275325e-05,
-      "loss": 1.3695,
+      "epoch": 0.0554683411826269,
+      "grad_norm": 2.956338657336674,
+      "learning_rate": 1.9966094927225272e-05,
+      "loss": 1.0119,
       "step": 530
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.892406244393382e-05,
-      "loss": 1.2519,
+      "epoch": 0.05557299843014129,
+      "grad_norm": 3.287242958068802,
+      "learning_rate": 1.9965815458367633e-05,
+      "loss": 1.0788,
       "step": 531
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.891924545673207e-05,
-      "loss": 1.438,
+      "epoch": 0.05567765567765568,
+      "grad_norm": 2.8717406338940283,
+      "learning_rate": 1.99655348444176e-05,
+      "loss": 1.1331,
       "step": 532
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.891441832662586e-05,
-      "loss": 1.2753,
+      "epoch": 0.055782312925170066,
+      "grad_norm": 3.1316748048585477,
+      "learning_rate": 1.996525308540741e-05,
+      "loss": 1.0685,
       "step": 533
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.8909581059104557e-05,
-      "loss": 1.3365,
+      "epoch": 0.05588697017268446,
+      "grad_norm": 3.1662562257885165,
+      "learning_rate": 1.996497018136944e-05,
+      "loss": 1.1702,
       "step": 534
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.8904733659669072e-05,
-      "loss": 1.3209,
+      "epoch": 0.05599162742019885,
+      "grad_norm": 2.9159713047319022,
+      "learning_rate": 1.9964686132336193e-05,
+      "loss": 1.1428,
       "step": 535
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.8899876133831835e-05,
-      "loss": 1.1781,
+      "epoch": 0.05609628466771324,
+      "grad_norm": 2.9197107556733863,
+      "learning_rate": 1.9964400938340312e-05,
+      "loss": 1.2099,
       "step": 536
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.8895008487116793e-05,
-      "loss": 1.3314,
+      "epoch": 0.05620094191522763,
+      "grad_norm": 3.103396125023537,
+      "learning_rate": 1.996411459941456e-05,
+      "loss": 1.0532,
       "step": 537
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.8890130725059396e-05,
-      "loss": 1.2742,
+      "epoch": 0.05630559916274202,
+      "grad_norm": 2.9755305851979985,
+      "learning_rate": 1.9963827115591843e-05,
+      "loss": 1.1459,
       "step": 538
     },
     {
-      "epoch": 0.35,
-      "learning_rate": 1.8885242853206606e-05,
-      "loss": 1.3207,
+      "epoch": 0.05641025641025641,
+      "grad_norm": 3.1958262641616835,
+      "learning_rate": 1.996353848690519e-05,
+      "loss": 1.1823,
       "step": 539
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.888034487711687e-05,
-      "loss": 1.2713,
+      "epoch": 0.0565149136577708,
+      "grad_norm": 3.044198764425719,
+      "learning_rate": 1.9963248713387767e-05,
+      "loss": 0.9377,
       "step": 540
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.8875436802360142e-05,
-      "loss": 1.4202,
+      "epoch": 0.05661957090528519,
+      "grad_norm": 2.7930578822333825,
+      "learning_rate": 1.9962957795072874e-05,
+      "loss": 0.9952,
       "step": 541
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.8870518634517843e-05,
-      "loss": 1.3311,
+      "epoch": 0.05672422815279958,
+      "grad_norm": 2.9967717314841598,
+      "learning_rate": 1.9962665731993925e-05,
+      "loss": 1.1515,
       "step": 542
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.8865590379182885e-05,
-      "loss": 1.3785,
+      "epoch": 0.05682888540031397,
+      "grad_norm": 3.0834478594381025,
+      "learning_rate": 1.9962372524184493e-05,
+      "loss": 1.1046,
       "step": 543
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.8860652041959643e-05,
-      "loss": 1.3968,
+      "epoch": 0.056933542647828364,
+      "grad_norm": 3.0669052108361456,
+      "learning_rate": 1.996207817167826e-05,
+      "loss": 1.1256,
       "step": 544
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.8855703628463957e-05,
-      "loss": 1.3965,
+      "epoch": 0.05703819989534275,
+      "grad_norm": 2.9144654203085496,
+      "learning_rate": 1.996178267450905e-05,
+      "loss": 1.0783,
       "step": 545
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.8850745144323137e-05,
-      "loss": 1.2262,
+      "epoch": 0.05714285714285714,
+      "grad_norm": 2.9822664939930537,
+      "learning_rate": 1.9961486032710813e-05,
+      "loss": 1.1593,
       "step": 546
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.884577659517593e-05,
-      "loss": 1.358,
+      "epoch": 0.05724751439037153,
+      "grad_norm": 3.1486295364529227,
+      "learning_rate": 1.9961188246317644e-05,
+      "loss": 1.0789,
       "step": 547
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.884079798667254e-05,
-      "loss": 1.286,
+      "epoch": 0.057352171637885925,
+      "grad_norm": 2.7926732727072054,
+      "learning_rate": 1.9960889315363747e-05,
+      "loss": 1.1259,
       "step": 548
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.88358093244746e-05,
-      "loss": 1.3762,
+      "epoch": 0.05745682888540032,
+      "grad_norm": 3.057281352342737,
+      "learning_rate": 1.9960589239883474e-05,
+      "loss": 1.1181,
       "step": 549
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.883081061425519e-05,
-      "loss": 1.3745,
+      "epoch": 0.0575614861329147,
+      "grad_norm": 3.254756100588196,
+      "learning_rate": 1.996028801991131e-05,
+      "loss": 1.047,
       "step": 550
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.882580186169881e-05,
-      "loss": 1.2543,
+      "epoch": 0.057666143380429094,
+      "grad_norm": 2.770630465752798,
+      "learning_rate": 1.9959985655481855e-05,
+      "loss": 1.053,
       "step": 551
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.882078307250137e-05,
-      "loss": 1.3128,
+      "epoch": 0.057770800627943486,
+      "grad_norm": 2.816610825508362,
+      "learning_rate": 1.9959682146629862e-05,
+      "loss": 1.1418,
       "step": 552
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.8815754252370205e-05,
-      "loss": 1.4083,
+      "epoch": 0.05787545787545788,
+      "grad_norm": 2.0688323215063646,
+      "learning_rate": 1.9959377493390198e-05,
+      "loss": 0.8396,
       "step": 553
     },
     {
-      "epoch": 0.36,
-      "learning_rate": 1.881071540702406e-05,
-      "loss": 1.4121,
+      "epoch": 0.05798011512297226,
+      "grad_norm": 2.1085687907562662,
+      "learning_rate": 1.995907169579787e-05,
+      "loss": 0.793,
       "step": 554
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.8805666542193073e-05,
-      "loss": 1.2899,
+      "epoch": 0.058084772370486655,
+      "grad_norm": 3.06254545731126,
+      "learning_rate": 1.9958764753888015e-05,
+      "loss": 1.1242,
       "step": 555
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.8800607663618774e-05,
-      "loss": 1.4122,
+      "epoch": 0.05818942961800105,
+      "grad_norm": 2.75776805013571,
+      "learning_rate": 1.99584566676959e-05,
+      "loss": 0.9819,
       "step": 556
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.879553877705409e-05,
-      "loss": 1.2897,
+      "epoch": 0.05829408686551544,
+      "grad_norm": 3.275437113764389,
+      "learning_rate": 1.995814743725693e-05,
+      "loss": 1.1104,
       "step": 557
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.8790459888263323e-05,
-      "loss": 1.3716,
+      "epoch": 0.058398744113029824,
+      "grad_norm": 3.203542046928147,
+      "learning_rate": 1.995783706260663e-05,
+      "loss": 1.0711,
       "step": 558
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.8785371003022145e-05,
-      "loss": 1.2673,
+      "epoch": 0.058503401360544216,
+      "grad_norm": 3.3467386858572934,
+      "learning_rate": 1.9957525543780663e-05,
+      "loss": 1.1925,
       "step": 559
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.8780272127117606e-05,
-      "loss": 1.2832,
+      "epoch": 0.05860805860805861,
+      "grad_norm": 2.855041752035653,
+      "learning_rate": 1.9957212880814826e-05,
+      "loss": 0.8485,
       "step": 560
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.8775163266348116e-05,
-      "loss": 1.3121,
+      "epoch": 0.058712715855573,
+      "grad_norm": 3.4035834251768855,
+      "learning_rate": 1.9956899073745046e-05,
+      "loss": 0.9867,
       "step": 561
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.877004442652343e-05,
-      "loss": 1.2356,
+      "epoch": 0.05881737310308739,
+      "grad_norm": 2.891412935735807,
+      "learning_rate": 1.9956584122607373e-05,
+      "loss": 1.1266,
       "step": 562
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.8764915613464657e-05,
-      "loss": 1.2424,
+      "epoch": 0.05892203035060178,
+      "grad_norm": 2.741554961840439,
+      "learning_rate": 1.9956268027438006e-05,
+      "loss": 1.1316,
       "step": 563
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.8759776833004256e-05,
-      "loss": 1.4182,
+      "epoch": 0.05902668759811617,
+      "grad_norm": 2.629765377645471,
+      "learning_rate": 1.9955950788273255e-05,
+      "loss": 0.9629,
       "step": 564
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.8754628090986e-05,
-      "loss": 1.141,
+      "epoch": 0.05913134484563056,
+      "grad_norm": 3.303903521460224,
+      "learning_rate": 1.9955632405149577e-05,
+      "loss": 1.1504,
       "step": 565
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.8749469393265016e-05,
-      "loss": 1.3906,
+      "epoch": 0.05923600209314495,
+      "grad_norm": 2.4920423274313244,
+      "learning_rate": 1.9955312878103554e-05,
+      "loss": 1.1054,
       "step": 566
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.874430074570773e-05,
-      "loss": 1.2956,
+      "epoch": 0.05934065934065934,
+      "grad_norm": 2.4458587609353613,
+      "learning_rate": 1.9954992207171898e-05,
+      "loss": 0.8266,
       "step": 567
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.87391221541919e-05,
-      "loss": 1.3609,
+      "epoch": 0.05944531658817373,
+      "grad_norm": 3.0320812785910274,
+      "learning_rate": 1.995467039239146e-05,
+      "loss": 1.0863,
       "step": 568
     },
     {
-      "epoch": 0.37,
-      "learning_rate": 1.873393362460658e-05,
-      "loss": 1.3083,
+      "epoch": 0.05954997383568812,
+      "grad_norm": 3.1180464676532025,
+      "learning_rate": 1.995434743379921e-05,
+      "loss": 1.0776,
       "step": 569
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.8728735162852125e-05,
-      "loss": 1.4665,
+      "epoch": 0.059654631083202514,
+      "grad_norm": 2.0897952402688795,
+      "learning_rate": 1.995402333143226e-05,
+      "loss": 0.8537,
       "step": 570
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.87235267748402e-05,
-      "loss": 1.3554,
+      "epoch": 0.0597592883307169,
+      "grad_norm": 3.3450036899077102,
+      "learning_rate": 1.9953698085327856e-05,
+      "loss": 1.0668,
       "step": 571
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.8718308466493745e-05,
-      "loss": 1.3665,
+      "epoch": 0.05986394557823129,
+      "grad_norm": 3.719131888220057,
+      "learning_rate": 1.995337169552336e-05,
+      "loss": 1.1326,
       "step": 572
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.871308024374698e-05,
-      "loss": 1.4273,
+      "epoch": 0.05996860282574568,
+      "grad_norm": 3.887030150352439,
+      "learning_rate": 1.9953044162056275e-05,
+      "loss": 1.0363,
       "step": 573
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.8707842112545406e-05,
-      "loss": 1.344,
+      "epoch": 0.060073260073260075,
+      "grad_norm": 3.301925941503779,
+      "learning_rate": 1.9952715484964242e-05,
+      "loss": 1.1164,
       "step": 574
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.870259407884579e-05,
-      "loss": 1.4037,
+      "epoch": 0.06017791732077447,
+      "grad_norm": 3.126271535901025,
+      "learning_rate": 1.9952385664285024e-05,
+      "loss": 0.9704,
       "step": 575
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.8697336148616162e-05,
-      "loss": 1.3336,
+      "epoch": 0.06028257456828885,
+      "grad_norm": 3.0038883887884067,
+      "learning_rate": 1.9952054700056518e-05,
+      "loss": 1.0455,
       "step": 576
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.86920683278358e-05,
-      "loss": 1.3447,
+      "epoch": 0.060387231815803244,
+      "grad_norm": 3.0353412809857003,
+      "learning_rate": 1.995172259231675e-05,
+      "loss": 1.126,
       "step": 577
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.868679062249524e-05,
-      "loss": 1.3287,
+      "epoch": 0.060491889063317636,
+      "grad_norm": 3.041115370144133,
+      "learning_rate": 1.9951389341103885e-05,
+      "loss": 1.1008,
       "step": 578
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.8681503038596248e-05,
-      "loss": 1.2865,
+      "epoch": 0.06059654631083203,
+      "grad_norm": 3.577476624319239,
+      "learning_rate": 1.995105494645621e-05,
+      "loss": 1.157,
       "step": 579
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.8676205582151822e-05,
-      "loss": 1.1761,
+      "epoch": 0.06070120355834641,
+      "grad_norm": 3.1702638194747808,
+      "learning_rate": 1.9950719408412145e-05,
+      "loss": 1.0744,
       "step": 580
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.8670898259186202e-05,
-      "loss": 1.2776,
+      "epoch": 0.060805860805860805,
+      "grad_norm": 2.6999043263976747,
+      "learning_rate": 1.9950382727010254e-05,
+      "loss": 1.0654,
       "step": 581
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.8665581075734836e-05,
-      "loss": 1.2189,
+      "epoch": 0.0609105180533752,
+      "grad_norm": 2.5781649995482216,
+      "learning_rate": 1.9950044902289214e-05,
+      "loss": 0.7968,
       "step": 582
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.866025403784439e-05,
-      "loss": 1.3302,
+      "epoch": 0.06101517530088959,
+      "grad_norm": 2.8874420445573032,
+      "learning_rate": 1.994970593428784e-05,
+      "loss": 1.1324,
       "step": 583
     },
     {
-      "epoch": 0.38,
-      "learning_rate": 1.865491715157273e-05,
-      "loss": 1.3992,
+      "epoch": 0.061119832548403974,
+      "grad_norm": 3.5373274248424806,
+      "learning_rate": 1.994936582304509e-05,
+      "loss": 1.0826,
       "step": 584
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.864957042298893e-05,
-      "loss": 1.3607,
+      "epoch": 0.061224489795918366,
+      "grad_norm": 3.2429438002106203,
+      "learning_rate": 1.994902456860003e-05,
+      "loss": 1.0991,
       "step": 585
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.864421385817325e-05,
-      "loss": 1.3804,
+      "epoch": 0.06132914704343276,
+      "grad_norm": 2.83907357900198,
+      "learning_rate": 1.9948682170991884e-05,
+      "loss": 1.0165,
       "step": 586
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.8638847463217145e-05,
-      "loss": 1.2228,
+      "epoch": 0.06143380429094715,
+      "grad_norm": 2.756315965811872,
+      "learning_rate": 1.9948338630259988e-05,
+      "loss": 1.1076,
       "step": 587
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.8633471244223232e-05,
-      "loss": 1.2283,
+      "epoch": 0.06153846153846154,
+      "grad_norm": 3.0042678927369333,
+      "learning_rate": 1.9947993946443814e-05,
+      "loss": 1.1311,
       "step": 588
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.8628085207305325e-05,
-      "loss": 1.3609,
+      "epoch": 0.06164311878597593,
+      "grad_norm": 2.6664016639707975,
+      "learning_rate": 1.9947648119582973e-05,
+      "loss": 1.0378,
       "step": 589
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.8622689358588373e-05,
-      "loss": 1.3056,
+      "epoch": 0.06174777603349032,
+      "grad_norm": 3.1499853449871784,
+      "learning_rate": 1.9947301149717194e-05,
+      "loss": 1.0647,
       "step": 590
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.8617283704208507e-05,
-      "loss": 1.3889,
+      "epoch": 0.06185243328100471,
+      "grad_norm": 3.2629382002059293,
+      "learning_rate": 1.9946953036886346e-05,
+      "loss": 1.0228,
       "step": 591
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.8611868250312994e-05,
-      "loss": 1.389,
+      "epoch": 0.0619570905285191,
+      "grad_norm": 3.129418827715554,
+      "learning_rate": 1.994660378113043e-05,
+      "loss": 1.1271,
       "step": 592
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.860644300306026e-05,
-      "loss": 1.3262,
+      "epoch": 0.06206174777603349,
+      "grad_norm": 2.9554401950010414,
+      "learning_rate": 1.994625338248958e-05,
+      "loss": 1.1953,
       "step": 593
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.8601007968619847e-05,
-      "loss": 1.3581,
+      "epoch": 0.06216640502354788,
+      "grad_norm": 3.2598690152125314,
+      "learning_rate": 1.994590184100405e-05,
+      "loss": 1.1584,
       "step": 594
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.8595563153172448e-05,
-      "loss": 1.3574,
+      "epoch": 0.06227106227106227,
+      "grad_norm": 2.437511527082557,
+      "learning_rate": 1.9945549156714236e-05,
+      "loss": 0.8374,
       "step": 595
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.8590108562909862e-05,
-      "loss": 1.1755,
+      "epoch": 0.062375719518576664,
+      "grad_norm": 2.6096105594208985,
+      "learning_rate": 1.994519532966066e-05,
+      "loss": 1.1617,
       "step": 596
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.858464420403502e-05,
-      "loss": 1.4054,
+      "epoch": 0.06248037676609105,
+      "grad_norm": 3.421662115903477,
+      "learning_rate": 1.9944840359883983e-05,
+      "loss": 1.1564,
       "step": 597
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.857917008276195e-05,
-      "loss": 1.3015,
+      "epoch": 0.06258503401360545,
+      "grad_norm": 1.9096028175870379,
+      "learning_rate": 1.9944484247424986e-05,
+      "loss": 0.7336,
       "step": 598
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.857368620531578e-05,
-      "loss": 1.3561,
+      "epoch": 0.06268969126111983,
+      "grad_norm": 2.8077084469651816,
+      "learning_rate": 1.9944126992324592e-05,
+      "loss": 1.1911,
       "step": 599
     },
     {
-      "epoch": 0.39,
-      "learning_rate": 1.8568192577932746e-05,
-      "loss": 1.4332,
+      "epoch": 0.06279434850863422,
+      "grad_norm": 2.8556439666255695,
+      "learning_rate": 1.9943768594623844e-05,
+      "loss": 1.0634,
       "step": 600
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.8562689206860155e-05,
-      "loss": 1.3722,
+      "epoch": 0.06289900575614861,
+      "grad_norm": 3.117743097234504,
+      "learning_rate": 1.994340905436393e-05,
+      "loss": 1.1191,
       "step": 601
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.8557176098356406e-05,
-      "loss": 1.2927,
+      "epoch": 0.063003663003663,
+      "grad_norm": 3.1601015532368866,
+      "learning_rate": 1.9943048371586158e-05,
+      "loss": 0.9725,
       "step": 602
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.8551653258690972e-05,
-      "loss": 1.4575,
+      "epoch": 0.0631083202511774,
+      "grad_norm": 2.208608976213274,
+      "learning_rate": 1.994268654633197e-05,
+      "loss": 0.7586,
       "step": 603
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.854612069414438e-05,
-      "loss": 1.4379,
+      "epoch": 0.06321297749869179,
+      "grad_norm": 2.1158633727476026,
+      "learning_rate": 1.994232357864294e-05,
+      "loss": 0.7692,
       "step": 604
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.8540578411008233e-05,
-      "loss": 1.3694,
+      "epoch": 0.06331763474620618,
+      "grad_norm": 2.9413326628625205,
+      "learning_rate": 1.9941959468560782e-05,
+      "loss": 1.1525,
       "step": 605
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.8535026415585168e-05,
-      "loss": 1.2722,
+      "epoch": 0.06342229199372057,
+      "grad_norm": 2.8204049381108742,
+      "learning_rate": 1.9941594216127323e-05,
+      "loss": 0.872,
       "step": 606
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.8529464714188888e-05,
-      "loss": 1.192,
+      "epoch": 0.06352694924123496,
+      "grad_norm": 3.190529808813044,
+      "learning_rate": 1.994122782138453e-05,
+      "loss": 1.1846,
       "step": 607
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.852389331314411e-05,
-      "loss": 1.2234,
+      "epoch": 0.06363160648874934,
+      "grad_norm": 3.08211718908032,
+      "learning_rate": 1.9940860284374515e-05,
+      "loss": 0.9465,
       "step": 608
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.85183122187866e-05,
-      "loss": 1.3328,
+      "epoch": 0.06373626373626373,
+      "grad_norm": 3.0616588635006123,
+      "learning_rate": 1.99404916051395e-05,
+      "loss": 0.9674,
       "step": 609
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.8512721437463136e-05,
-      "loss": 1.2733,
+      "epoch": 0.06384092098377812,
+      "grad_norm": 2.9537487271769116,
+      "learning_rate": 1.9940121783721846e-05,
+      "loss": 1.1373,
       "step": 610
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.8507120975531517e-05,
-      "loss": 1.3759,
+      "epoch": 0.06394557823129252,
+      "grad_norm": 2.9668339388811407,
+      "learning_rate": 1.9939750820164044e-05,
+      "loss": 1.0418,
       "step": 611
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.850151083936055e-05,
-      "loss": 1.2952,
+      "epoch": 0.06405023547880691,
+      "grad_norm": 2.8848768351239658,
+      "learning_rate": 1.9939378714508728e-05,
+      "loss": 0.9834,
       "step": 612
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.849589103533004e-05,
-      "loss": 1.3329,
+      "epoch": 0.0641548927263213,
+      "grad_norm": 2.759886281009913,
+      "learning_rate": 1.9939005466798648e-05,
+      "loss": 1.0321,
       "step": 613
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.84902615698308e-05,
-      "loss": 1.4101,
+      "epoch": 0.06425954997383569,
+      "grad_norm": 3.0129757609858245,
+      "learning_rate": 1.9938631077076692e-05,
+      "loss": 1.1952,
       "step": 614
     },
     {
-      "epoch": 0.4,
-      "learning_rate": 1.8484622449264608e-05,
-      "loss": 1.3062,
+      "epoch": 0.06436420722135008,
+      "grad_norm": 2.573835926627507,
+      "learning_rate": 1.9938255545385875e-05,
+      "loss": 0.9685,
       "step": 615
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.8478973680044237e-05,
-      "loss": 1.3987,
+      "epoch": 0.06446886446886448,
+      "grad_norm": 3.453838047274854,
+      "learning_rate": 1.993787887176935e-05,
+      "loss": 1.129,
       "step": 616
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.8473315268593432e-05,
-      "loss": 1.3151,
+      "epoch": 0.06457352171637885,
+      "grad_norm": 3.1074759117744137,
+      "learning_rate": 1.9937501056270397e-05,
+      "loss": 1.0101,
       "step": 617
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.846764722134689e-05,
-      "loss": 1.3956,
+      "epoch": 0.06467817896389325,
+      "grad_norm": 2.755027947394353,
+      "learning_rate": 1.9937122098932428e-05,
+      "loss": 0.9722,
       "step": 618
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.8461969544750286e-05,
-      "loss": 1.392,
+      "epoch": 0.06478283621140764,
+      "grad_norm": 2.9013696059112224,
+      "learning_rate": 1.9936741999798985e-05,
+      "loss": 1.1728,
       "step": 619
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.845628224526023e-05,
-      "loss": 1.5057,
+      "epoch": 0.06488749345892203,
+      "grad_norm": 3.1607523426322586,
+      "learning_rate": 1.993636075891374e-05,
+      "loss": 0.895,
       "step": 620
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.8450585329344277e-05,
-      "loss": 1.2634,
+      "epoch": 0.06499215070643642,
+      "grad_norm": 3.1382276779001597,
+      "learning_rate": 1.9935978376320503e-05,
+      "loss": 1.1031,
       "step": 621
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.8444878803480925e-05,
-      "loss": 1.3566,
+      "epoch": 0.06509680795395081,
+      "grad_norm": 2.52595770004853,
+      "learning_rate": 1.9935594852063208e-05,
+      "loss": 0.938,
       "step": 622
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.8439162674159595e-05,
-      "loss": 1.3247,
+      "epoch": 0.0652014652014652,
+      "grad_norm": 2.650381943162927,
+      "learning_rate": 1.993521018618592e-05,
+      "loss": 1.0193,
       "step": 623
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.8433436947880626e-05,
-      "loss": 1.3977,
+      "epoch": 0.0653061224489796,
+      "grad_norm": 2.651852123933312,
+      "learning_rate": 1.9934824378732846e-05,
+      "loss": 1.0853,
       "step": 624
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.842770163115528e-05,
-      "loss": 1.3899,
+      "epoch": 0.06541077969649398,
+      "grad_norm": 3.306163119634711,
+      "learning_rate": 1.9934437429748306e-05,
+      "loss": 1.154,
       "step": 625
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.842195673050572e-05,
-      "loss": 1.308,
+      "epoch": 0.06551543694400837,
+      "grad_norm": 2.784487003175816,
+      "learning_rate": 1.993404933927677e-05,
+      "loss": 1.0449,
       "step": 626
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.8416202252465e-05,
-      "loss": 1.4328,
+      "epoch": 0.06562009419152276,
+      "grad_norm": 2.759785634839251,
+      "learning_rate": 1.9933660107362825e-05,
+      "loss": 1.0753,
       "step": 627
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.841043820357708e-05,
-      "loss": 1.3289,
+      "epoch": 0.06572475143903715,
+      "grad_norm": 2.731464057889386,
+      "learning_rate": 1.9933269734051196e-05,
+      "loss": 0.9269,
       "step": 628
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.8404664590396796e-05,
-      "loss": 1.3466,
+      "epoch": 0.06582940868655154,
+      "grad_norm": 2.837489067099457,
+      "learning_rate": 1.9932878219386738e-05,
+      "loss": 1.0685,
       "step": 629
     },
     {
-      "epoch": 0.41,
-      "learning_rate": 1.8398881419489854e-05,
-      "loss": 1.3224,
+      "epoch": 0.06593406593406594,
+      "grad_norm": 3.017220693367057,
+      "learning_rate": 1.9932485563414436e-05,
+      "loss": 1.1035,
       "step": 630
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.8393088697432847e-05,
-      "loss": 1.3337,
+      "epoch": 0.06603872318158033,
+      "grad_norm": 3.3205197744653194,
+      "learning_rate": 1.9932091766179408e-05,
+      "loss": 1.1205,
       "step": 631
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.838728643081321e-05,
-      "loss": 1.3094,
+      "epoch": 0.06614338042909472,
+      "grad_norm": 2.891684092310922,
+      "learning_rate": 1.99316968277269e-05,
+      "loss": 1.1477,
       "step": 632
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.8381474626229245e-05,
-      "loss": 1.3409,
+      "epoch": 0.06624803767660911,
+      "grad_norm": 2.9884409256785447,
+      "learning_rate": 1.9931300748102294e-05,
+      "loss": 1.1761,
       "step": 633
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.8375653290290093e-05,
-      "loss": 1.4036,
+      "epoch": 0.06635269492412349,
+      "grad_norm": 2.536315479596216,
+      "learning_rate": 1.99309035273511e-05,
+      "loss": 1.0411,
       "step": 634
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.836982242961574e-05,
-      "loss": 1.266,
+      "epoch": 0.06645735217163788,
+      "grad_norm": 2.9476335588680365,
+      "learning_rate": 1.9930505165518958e-05,
+      "loss": 1.1562,
       "step": 635
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.8363982050836996e-05,
-      "loss": 1.2164,
+      "epoch": 0.06656200941915227,
+      "grad_norm": 3.0500527548211704,
+      "learning_rate": 1.993010566265164e-05,
+      "loss": 1.1814,
       "step": 636
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.835813216059551e-05,
-      "loss": 1.3615,
+      "epoch": 0.06666666666666667,
+      "grad_norm": 2.6707063897105905,
+      "learning_rate": 1.9929705018795055e-05,
+      "loss": 1.0591,
       "step": 637
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.8352272765543723e-05,
-      "loss": 1.3961,
+      "epoch": 0.06677132391418106,
+      "grad_norm": 2.618480485695442,
+      "learning_rate": 1.9929303233995227e-05,
+      "loss": 1.0465,
       "step": 638
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.8346403872344907e-05,
-      "loss": 1.3247,
+      "epoch": 0.06687598116169545,
+      "grad_norm": 2.66882534809794,
+      "learning_rate": 1.9928900308298334e-05,
+      "loss": 1.1071,
       "step": 639
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.8340525487673128e-05,
-      "loss": 1.4376,
+      "epoch": 0.06698063840920984,
+      "grad_norm": 2.9292541219026496,
+      "learning_rate": 1.9928496241750667e-05,
+      "loss": 1.1123,
       "step": 640
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.8334637618213238e-05,
-      "loss": 1.3172,
+      "epoch": 0.06708529565672423,
+      "grad_norm": 2.7357462801221546,
+      "learning_rate": 1.992809103439865e-05,
+      "loss": 1.132,
       "step": 641
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.8328740270660887e-05,
-      "loss": 1.3366,
+      "epoch": 0.06718995290423861,
+      "grad_norm": 3.003236810512024,
+      "learning_rate": 1.9927684686288854e-05,
+      "loss": 1.1636,
       "step": 642
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.83228334517225e-05,
-      "loss": 1.2715,
+      "epoch": 0.067294610151753,
+      "grad_norm": 3.313716591599714,
+      "learning_rate": 1.992727719746796e-05,
+      "loss": 1.0848,
       "step": 643
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.831691716811526e-05,
-      "loss": 1.387,
+      "epoch": 0.0673992673992674,
+      "grad_norm": 3.0324704855965163,
+      "learning_rate": 1.9926868567982788e-05,
+      "loss": 1.177,
       "step": 644
     },
     {
-      "epoch": 0.42,
-      "learning_rate": 1.8310991426567137e-05,
-      "loss": 1.2501,
+      "epoch": 0.06750392464678179,
+      "grad_norm": 3.172131349022228,
+      "learning_rate": 1.9926458797880294e-05,
+      "loss": 0.8632,
       "step": 645
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.8305056233816833e-05,
-      "loss": 1.1845,
+      "epoch": 0.06760858189429618,
+      "grad_norm": 3.0648739329108077,
+      "learning_rate": 1.9926047887207565e-05,
+      "loss": 1.0914,
       "step": 646
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.8299111596613815e-05,
-      "loss": 1.4116,
+      "epoch": 0.06771323914181057,
+      "grad_norm": 2.4156718664411696,
+      "learning_rate": 1.992563583601181e-05,
+      "loss": 0.947,
       "step": 647
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.8293157521718284e-05,
-      "loss": 1.2341,
+      "epoch": 0.06781789638932496,
+      "grad_norm": 2.7382356031716646,
+      "learning_rate": 1.9925222644340372e-05,
+      "loss": 1.0838,
       "step": 648
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.8287194015901167e-05,
-      "loss": 1.2673,
+      "epoch": 0.06792255363683936,
+      "grad_norm": 2.9469770945822047,
+      "learning_rate": 1.9924808312240737e-05,
+      "loss": 1.1064,
       "step": 649
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.8281221085944128e-05,
-      "loss": 1.3124,
+      "epoch": 0.06802721088435375,
+      "grad_norm": 3.1261406037905513,
+      "learning_rate": 1.9924392839760505e-05,
+      "loss": 1.1091,
       "step": 650
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.827523873863954e-05,
-      "loss": 1.3819,
+      "epoch": 0.06813186813186813,
+      "grad_norm": 3.1877565587473575,
+      "learning_rate": 1.9923976226947417e-05,
+      "loss": 1.1858,
       "step": 651
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.8269246980790482e-05,
-      "loss": 1.2699,
+      "epoch": 0.06823652537938252,
+      "grad_norm": 2.8216280798542885,
+      "learning_rate": 1.992355847384934e-05,
+      "loss": 1.0689,
       "step": 652
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.8263245819210745e-05,
-      "loss": 1.4479,
+      "epoch": 0.06834118262689691,
+      "grad_norm": 2.9795946286363875,
+      "learning_rate": 1.9923139580514284e-05,
+      "loss": 1.0729,
       "step": 653
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.825723526072481e-05,
-      "loss": 1.2992,
+      "epoch": 0.0684458398744113,
+      "grad_norm": 2.5206528985759324,
+      "learning_rate": 1.9922719546990367e-05,
+      "loss": 0.8577,
       "step": 654
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.8251215312167833e-05,
-      "loss": 1.2057,
+      "epoch": 0.0685504971219257,
+      "grad_norm": 2.9454318451318455,
+      "learning_rate": 1.992229837332586e-05,
+      "loss": 0.8568,
       "step": 655
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.8245185980385673e-05,
-      "loss": 1.2626,
+      "epoch": 0.06865515436944009,
+      "grad_norm": 2.774719223512651,
+      "learning_rate": 1.9921876059569158e-05,
+      "loss": 0.9953,
       "step": 656
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.823914727223483e-05,
-      "loss": 1.2323,
+      "epoch": 0.06875981161695448,
+      "grad_norm": 3.19787444851232,
+      "learning_rate": 1.9921452605768783e-05,
+      "loss": 1.0818,
       "step": 657
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.823309919458249e-05,
-      "loss": 1.2677,
+      "epoch": 0.06886446886446887,
+      "grad_norm": 3.4049791357976105,
+      "learning_rate": 1.9921028011973388e-05,
+      "loss": 1.1536,
       "step": 658
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.8227041754306484e-05,
-      "loss": 1.3755,
+      "epoch": 0.06896912611198326,
+      "grad_norm": 2.557786514794364,
+      "learning_rate": 1.9920602278231765e-05,
+      "loss": 0.986,
       "step": 659
     },
     {
-      "epoch": 0.43,
-      "learning_rate": 1.822097495829529e-05,
-      "loss": 1.1518,
+      "epoch": 0.06907378335949764,
+      "grad_norm": 3.1241150572729808,
+      "learning_rate": 1.9920175404592827e-05,
+      "loss": 0.9868,
       "step": 660
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.821489881344803e-05,
-      "loss": 1.3471,
+      "epoch": 0.06917844060701203,
+      "grad_norm": 3.2992236992022788,
+      "learning_rate": 1.9919747391105626e-05,
+      "loss": 1.0701,
       "step": 661
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.8208813326674444e-05,
-      "loss": 1.2778,
+      "epoch": 0.06928309785452642,
+      "grad_norm": 3.4812568274144655,
+      "learning_rate": 1.991931823781934e-05,
+      "loss": 0.9236,
       "step": 662
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.8202718504894917e-05,
-      "loss": 1.3122,
+      "epoch": 0.06938775510204082,
+      "grad_norm": 2.7811990200114276,
+      "learning_rate": 1.991888794478328e-05,
+      "loss": 1.0129,
       "step": 663
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.8196614355040436e-05,
-      "loss": 1.4677,
+      "epoch": 0.06949241234955521,
+      "grad_norm": 3.0685242353855267,
+      "learning_rate": 1.9918456512046886e-05,
+      "loss": 1.1438,
       "step": 664
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.819050088405259e-05,
-      "loss": 1.284,
+      "epoch": 0.0695970695970696,
+      "grad_norm": 2.6809973552138824,
+      "learning_rate": 1.9918023939659735e-05,
+      "loss": 1.0722,
       "step": 665
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.818437809888359e-05,
-      "loss": 1.4112,
+      "epoch": 0.06970172684458399,
+      "grad_norm": 2.5126108916830545,
+      "learning_rate": 1.9917590227671523e-05,
+      "loss": 0.8222,
       "step": 666
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.817824600649621e-05,
-      "loss": 1.3281,
+      "epoch": 0.06980638409209838,
+      "grad_norm": 3.1637408972668135,
+      "learning_rate": 1.9917155376132092e-05,
+      "loss": 1.0751,
       "step": 667
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.8172104613863838e-05,
-      "loss": 1.329,
+      "epoch": 0.06991104133961276,
+      "grad_norm": 2.7641342208213695,
+      "learning_rate": 1.9916719385091404e-05,
+      "loss": 0.9868,
       "step": 668
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.816595392797041e-05,
-      "loss": 1.1344,
+      "epoch": 0.07001569858712715,
+      "grad_norm": 3.369758095168238,
+      "learning_rate": 1.9916282254599556e-05,
+      "loss": 1.073,
       "step": 669
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.8159793955810453e-05,
-      "loss": 1.4005,
+      "epoch": 0.07012035583464155,
+      "grad_norm": 2.9489363996307576,
+      "learning_rate": 1.9915843984706773e-05,
+      "loss": 1.0683,
       "step": 670
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.815362470438904e-05,
-      "loss": 1.3454,
+      "epoch": 0.07022501308215594,
+      "grad_norm": 3.103782003060308,
+      "learning_rate": 1.9915404575463414e-05,
+      "loss": 1.0495,
       "step": 671
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.8147446180721802e-05,
-      "loss": 1.3562,
+      "epoch": 0.07032967032967033,
+      "grad_norm": 3.442901028729499,
+      "learning_rate": 1.991496402691997e-05,
+      "loss": 1.0754,
       "step": 672
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.8141258391834914e-05,
-      "loss": 1.2067,
+      "epoch": 0.07043432757718472,
+      "grad_norm": 2.861095313005916,
+      "learning_rate": 1.991452233912706e-05,
+      "loss": 1.0522,
       "step": 673
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.8135061344765088e-05,
-      "loss": 1.2327,
+      "epoch": 0.07053898482469911,
+      "grad_norm": 2.9472040327867144,
+      "learning_rate": 1.9914079512135434e-05,
+      "loss": 1.1586,
       "step": 674
     },
     {
-      "epoch": 0.44,
-      "learning_rate": 1.812885504655956e-05,
-      "loss": 1.1144,
+      "epoch": 0.0706436420722135,
+      "grad_norm": 3.1812368240216657,
+      "learning_rate": 1.9913635545995972e-05,
+      "loss": 1.1145,
       "step": 675
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8122639504276096e-05,
-      "loss": 1.2406,
+      "epoch": 0.0707482993197279,
+      "grad_norm": 2.9277804198588666,
+      "learning_rate": 1.9913190440759693e-05,
+      "loss": 1.0576,
       "step": 676
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.811641472498296e-05,
-      "loss": 1.3028,
+      "epoch": 0.07085295656724228,
+      "grad_norm": 3.271562835349517,
+      "learning_rate": 1.9912744196477732e-05,
+      "loss": 0.9771,
       "step": 677
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8110180715758934e-05,
-      "loss": 1.3957,
+      "epoch": 0.07095761381475667,
+      "grad_norm": 3.0718494789605604,
+      "learning_rate": 1.9912296813201372e-05,
+      "loss": 1.0435,
       "step": 678
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8103937483693288e-05,
-      "loss": 1.3159,
+      "epoch": 0.07106227106227106,
+      "grad_norm": 2.1086283397641576,
+      "learning_rate": 1.991184829098201e-05,
+      "loss": 0.8067,
       "step": 679
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.809768503588578e-05,
-      "loss": 1.4038,
+      "epoch": 0.07116692830978545,
+      "grad_norm": 2.0279982104266643,
+      "learning_rate": 1.9911398629871187e-05,
+      "loss": 0.6897,
       "step": 680
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8091423379446658e-05,
-      "loss": 1.3685,
+      "epoch": 0.07127158555729984,
+      "grad_norm": 2.7850815226838184,
+      "learning_rate": 1.991094782992057e-05,
+      "loss": 1.0727,
       "step": 681
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8085152521496627e-05,
-      "loss": 1.3142,
+      "epoch": 0.07137624280481424,
+      "grad_norm": 2.7710574348613313,
+      "learning_rate": 1.9910495891181956e-05,
+      "loss": 1.1439,
       "step": 682
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8078872469166864e-05,
-      "loss": 1.3661,
+      "epoch": 0.07148090005232863,
+      "grad_norm": 2.970435833694052,
+      "learning_rate": 1.9910042813707272e-05,
+      "loss": 1.1369,
       "step": 683
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8072583229599004e-05,
-      "loss": 1.3521,
+      "epoch": 0.07158555729984302,
+      "grad_norm": 2.618973989205177,
+      "learning_rate": 1.990958859754858e-05,
+      "loss": 1.0571,
       "step": 684
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8066284809945128e-05,
-      "loss": 1.4548,
+      "epoch": 0.0716902145473574,
+      "grad_norm": 3.5278999615831323,
+      "learning_rate": 1.9909133242758066e-05,
+      "loss": 1.177,
       "step": 685
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8059977217367753e-05,
-      "loss": 1.2938,
+      "epoch": 0.07179487179487179,
+      "grad_norm": 3.266029063508325,
+      "learning_rate": 1.990867674938806e-05,
+      "loss": 0.9545,
       "step": 686
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8053660459039833e-05,
-      "loss": 1.2422,
+      "epoch": 0.07189952904238618,
+      "grad_norm": 3.0659580296418856,
+      "learning_rate": 1.9908219117491004e-05,
+      "loss": 1.0684,
       "step": 687
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8047334542144738e-05,
-      "loss": 1.3967,
+      "epoch": 0.07200418628990057,
+      "grad_norm": 3.164931997476543,
+      "learning_rate": 1.9907760347119485e-05,
+      "loss": 1.0845,
       "step": 688
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8040999473876258e-05,
-      "loss": 1.428,
+      "epoch": 0.07210884353741497,
+      "grad_norm": 2.8716658890938174,
+      "learning_rate": 1.990730043832622e-05,
+      "loss": 1.0957,
       "step": 689
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8034655261438588e-05,
-      "loss": 1.2242,
+      "epoch": 0.07221350078492936,
+      "grad_norm": 2.545281120610984,
+      "learning_rate": 1.990683939116405e-05,
+      "loss": 1.0906,
       "step": 690
     },
     {
-      "epoch": 0.45,
-      "learning_rate": 1.8028301912046325e-05,
-      "loss": 1.3014,
+      "epoch": 0.07231815803244375,
+      "grad_norm": 3.317532471187599,
+      "learning_rate": 1.9906377205685953e-05,
+      "loss": 1.1177,
       "step": 691
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.8021939432924456e-05,
-      "loss": 1.3546,
+      "epoch": 0.07242281527995814,
+      "grad_norm": 3.2165985399160575,
+      "learning_rate": 1.9905913881945027e-05,
+      "loss": 1.0378,
       "step": 692
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.8015567831308344e-05,
-      "loss": 1.3246,
+      "epoch": 0.07252747252747253,
+      "grad_norm": 2.8761721393997193,
+      "learning_rate": 1.990544941999452e-05,
+      "loss": 1.1571,
       "step": 693
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.800918711444373e-05,
-      "loss": 1.3759,
+      "epoch": 0.07263212977498691,
+      "grad_norm": 3.4562167899000076,
+      "learning_rate": 1.9904983819887788e-05,
+      "loss": 0.9759,
       "step": 694
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.8002797289586726e-05,
-      "loss": 1.3185,
+      "epoch": 0.0727367870225013,
+      "grad_norm": 4.0163161813819155,
+      "learning_rate": 1.9904517081678342e-05,
+      "loss": 1.0135,
       "step": 695
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.7996398364003797e-05,
-      "loss": 1.3874,
+      "epoch": 0.0728414442700157,
+      "grad_norm": 2.999273333333365,
+      "learning_rate": 1.99040492054198e-05,
+      "loss": 1.1353,
       "step": 696
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.7989990344971747e-05,
-      "loss": 1.277,
+      "epoch": 0.07294610151753009,
+      "grad_norm": 2.947720739435947,
+      "learning_rate": 1.990358019116593e-05,
+      "loss": 1.0866,
       "step": 697
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.7983573239777746e-05,
-      "loss": 1.3898,
+      "epoch": 0.07305075876504448,
+      "grad_norm": 2.559467975109432,
+      "learning_rate": 1.9903110038970617e-05,
+      "loss": 1.1007,
       "step": 698
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.797714705571927e-05,
-      "loss": 1.3666,
+      "epoch": 0.07315541601255887,
+      "grad_norm": 2.6168641864922084,
+      "learning_rate": 1.9902638748887886e-05,
+      "loss": 1.0792,
       "step": 699
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.7970711800104138e-05,
-      "loss": 1.3797,
+      "epoch": 0.07326007326007326,
+      "grad_norm": 3.3899484327660367,
+      "learning_rate": 1.9902166320971888e-05,
+      "loss": 0.9802,
       "step": 700
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.7964267480250476e-05,
-      "loss": 1.3042,
+      "epoch": 0.07336473050758766,
+      "grad_norm": 3.5929670401239706,
+      "learning_rate": 1.9901692755276905e-05,
+      "loss": 1.14,
       "step": 701
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.7957814103486716e-05,
-      "loss": 1.2159,
+      "epoch": 0.07346938775510205,
+      "grad_norm": 3.5869697848385083,
+      "learning_rate": 1.9901218051857354e-05,
+      "loss": 1.0971,
       "step": 702
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.7951351677151604e-05,
-      "loss": 1.3311,
+      "epoch": 0.07357404500261643,
+      "grad_norm": 2.74561495745866,
+      "learning_rate": 1.9900742210767773e-05,
+      "loss": 1.1434,
       "step": 703
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.7944880208594156e-05,
-      "loss": 1.3292,
+      "epoch": 0.07367870225013082,
+      "grad_norm": 3.1605580372961306,
+      "learning_rate": 1.9900265232062844e-05,
+      "loss": 1.0261,
       "step": 704
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.7938399705173687e-05,
-      "loss": 1.443,
+      "epoch": 0.07378335949764521,
+      "grad_norm": 3.0884788086378463,
+      "learning_rate": 1.9899787115797373e-05,
+      "loss": 1.0976,
       "step": 705
     },
     {
-      "epoch": 0.46,
-      "learning_rate": 1.7931910174259777e-05,
-      "loss": 1.148,
+      "epoch": 0.0738880167451596,
+      "grad_norm": 2.904397093628698,
+      "learning_rate": 1.9899307862026288e-05,
+      "loss": 1.0172,
       "step": 706
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.792541162323228e-05,
-      "loss": 1.3056,
+      "epoch": 0.073992673992674,
+      "grad_norm": 2.550496674073018,
+      "learning_rate": 1.9898827470804663e-05,
+      "loss": 1.1443,
       "step": 707
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.7918904059481303e-05,
-      "loss": 1.1624,
+      "epoch": 0.07409733124018839,
+      "grad_norm": 3.0699639273230828,
+      "learning_rate": 1.9898345942187695e-05,
+      "loss": 1.1312,
       "step": 708
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.79123874904072e-05,
-      "loss": 1.2849,
+      "epoch": 0.07420198848770278,
+      "grad_norm": 3.151696553595531,
+      "learning_rate": 1.9897863276230712e-05,
+      "loss": 0.9484,
       "step": 709
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.7905861923420568e-05,
-      "loss": 1.1711,
+      "epoch": 0.07430664573521717,
+      "grad_norm": 2.936271363452074,
+      "learning_rate": 1.9897379472989177e-05,
+      "loss": 1.0499,
       "step": 710
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.7899327365942242e-05,
-      "loss": 1.3396,
+      "epoch": 0.07441130298273155,
+      "grad_norm": 3.427173808865134,
+      "learning_rate": 1.989689453251867e-05,
+      "loss": 1.0594,
       "step": 711
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.7892783825403272e-05,
-      "loss": 1.3184,
+      "epoch": 0.07451596023024594,
+      "grad_norm": 2.7577070019256165,
+      "learning_rate": 1.9896408454874924e-05,
+      "loss": 1.0995,
       "step": 712
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.788623130924493e-05,
-      "loss": 1.3957,
+      "epoch": 0.07462061747776033,
+      "grad_norm": 2.737358494436629,
+      "learning_rate": 1.989592124011378e-05,
+      "loss": 0.9775,
       "step": 713
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.7879669824918685e-05,
-      "loss": 1.3386,
+      "epoch": 0.07472527472527472,
+      "grad_norm": 2.637102118855309,
+      "learning_rate": 1.9895432888291228e-05,
+      "loss": 1.1897,
       "step": 714
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.7873099379886217e-05,
-      "loss": 1.2501,
+      "epoch": 0.07482993197278912,
+      "grad_norm": 2.8640305527444525,
+      "learning_rate": 1.9894943399463373e-05,
+      "loss": 1.068,
       "step": 715
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.7866519981619396e-05,
-      "loss": 1.4069,
+      "epoch": 0.07493458922030351,
+      "grad_norm": 2.506044517062919,
+      "learning_rate": 1.9894452773686463e-05,
+      "loss": 1.1048,
       "step": 716
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.7859931637600256e-05,
-      "loss": 1.4704,
+      "epoch": 0.0750392464678179,
+      "grad_norm": 2.731113694615565,
+      "learning_rate": 1.9893961011016873e-05,
+      "loss": 1.1107,
       "step": 717
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.7853334355321025e-05,
-      "loss": 1.2778,
+      "epoch": 0.07514390371533229,
+      "grad_norm": 2.892316241004994,
+      "learning_rate": 1.9893468111511106e-05,
+      "loss": 1.0969,
       "step": 718
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.784672814228409e-05,
-      "loss": 1.3288,
+      "epoch": 0.07524856096284668,
+      "grad_norm": 2.55870522176311,
+      "learning_rate": 1.98929740752258e-05,
+      "loss": 1.0443,
       "step": 719
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.784011300600198e-05,
-      "loss": 1.4501,
+      "epoch": 0.07535321821036106,
+      "grad_norm": 2.9295733001233044,
+      "learning_rate": 1.989247890221771e-05,
+      "loss": 1.0639,
       "step": 720
     },
     {
-      "epoch": 0.47,
-      "learning_rate": 1.783348895399739e-05,
-      "loss": 1.3109,
+      "epoch": 0.07545787545787545,
+      "grad_norm": 3.0665452355998073,
+      "learning_rate": 1.9891982592543748e-05,
+      "loss": 1.1135,
       "step": 721
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7826855993803148e-05,
-      "loss": 1.2988,
+      "epoch": 0.07556253270538985,
+      "grad_norm": 2.8998275546803223,
+      "learning_rate": 1.9891485146260926e-05,
+      "loss": 0.9916,
       "step": 722
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7820214132962205e-05,
-      "loss": 1.4882,
+      "epoch": 0.07566718995290424,
+      "grad_norm": 2.5710268709337294,
+      "learning_rate": 1.989098656342641e-05,
+      "loss": 1.0469,
       "step": 723
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7813563379027644e-05,
-      "loss": 1.5883,
+      "epoch": 0.07577184720041863,
+      "grad_norm": 3.0608910480478535,
+      "learning_rate": 1.989048684409749e-05,
+      "loss": 1.1399,
       "step": 724
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7806903739562652e-05,
-      "loss": 1.2206,
+      "epoch": 0.07587650444793302,
+      "grad_norm": 2.8003344138495887,
+      "learning_rate": 1.988998598833158e-05,
+      "loss": 1.0059,
       "step": 725
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7800235222140525e-05,
-      "loss": 1.6747,
+      "epoch": 0.07598116169544741,
+      "grad_norm": 2.9559350963453785,
+      "learning_rate": 1.988948399618623e-05,
+      "loss": 1.1011,
       "step": 726
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7793557834344662e-05,
-      "loss": 1.2666,
+      "epoch": 0.0760858189429618,
+      "grad_norm": 2.6079665939299583,
+      "learning_rate": 1.988898086771912e-05,
+      "loss": 0.9801,
       "step": 727
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7786871583768536e-05,
-      "loss": 1.2639,
+      "epoch": 0.0761904761904762,
+      "grad_norm": 2.6799211618975143,
+      "learning_rate": 1.9888476602988063e-05,
+      "loss": 1.0884,
       "step": 728
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7780176478015706e-05,
-      "loss": 1.2828,
+      "epoch": 0.07629513343799058,
+      "grad_norm": 2.6807547704920665,
+      "learning_rate": 1.9887971202050996e-05,
+      "loss": 1.0336,
       "step": 729
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7773472524699803e-05,
-      "loss": 1.2552,
+      "epoch": 0.07639979068550497,
+      "grad_norm": 2.8884167531514233,
+      "learning_rate": 1.9887464664965997e-05,
+      "loss": 1.108,
       "step": 730
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7766759731444512e-05,
-      "loss": 1.2183,
+      "epoch": 0.07650444793301936,
+      "grad_norm": 3.1096081530713118,
+      "learning_rate": 1.9886956991791258e-05,
+      "loss": 1.0963,
       "step": 731
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.776003810588358e-05,
-      "loss": 1.3489,
+      "epoch": 0.07660910518053375,
+      "grad_norm": 2.5776344588421156,
+      "learning_rate": 1.9886448182585122e-05,
+      "loss": 1.071,
       "step": 732
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7753307655660792e-05,
-      "loss": 1.3426,
+      "epoch": 0.07671376242804814,
+      "grad_norm": 2.7119013146953814,
+      "learning_rate": 1.9885938237406048e-05,
+      "loss": 1.0899,
       "step": 733
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7746568388429966e-05,
-      "loss": 1.3664,
+      "epoch": 0.07681841967556254,
+      "grad_norm": 2.5283853043631153,
+      "learning_rate": 1.9885427156312627e-05,
+      "loss": 1.1398,
       "step": 734
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7739820311854957e-05,
-      "loss": 1.2674,
+      "epoch": 0.07692307692307693,
+      "grad_norm": 2.563527440025152,
+      "learning_rate": 1.988491493936359e-05,
+      "loss": 1.0551,
       "step": 735
     },
     {
-      "epoch": 0.48,
-      "learning_rate": 1.7733063433609623e-05,
-      "loss": 1.4536,
+      "epoch": 0.07702773417059132,
+      "grad_norm": 2.4468041205415805,
+      "learning_rate": 1.9884401586617785e-05,
+      "loss": 1.1602,
       "step": 736
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.7726297761377847e-05,
-      "loss": 1.2702,
+      "epoch": 0.0771323914181057,
+      "grad_norm": 3.40716823965318,
+      "learning_rate": 1.98838870981342e-05,
+      "loss": 1.0084,
       "step": 737
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.7719523302853498e-05,
-      "loss": 1.3485,
+      "epoch": 0.07723704866562009,
+      "grad_norm": 2.9705969422751113,
+      "learning_rate": 1.988337147397195e-05,
+      "loss": 1.2072,
       "step": 738
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.771274006574045e-05,
-      "loss": 1.4078,
+      "epoch": 0.07734170591313448,
+      "grad_norm": 2.4958733209485864,
+      "learning_rate": 1.9882854714190283e-05,
+      "loss": 0.8871,
       "step": 739
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.7705948057752547e-05,
-      "loss": 1.1793,
+      "epoch": 0.07744636316064887,
+      "grad_norm": 2.890941680142748,
+      "learning_rate": 1.9882336818848576e-05,
+      "loss": 0.9567,
       "step": 740
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.769914728661362e-05,
-      "loss": 1.3122,
+      "epoch": 0.07755102040816327,
+      "grad_norm": 3.002499996703523,
+      "learning_rate": 1.9881817788006335e-05,
+      "loss": 1.1373,
       "step": 741
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.7692337760057454e-05,
-      "loss": 1.3285,
+      "epoch": 0.07765567765567766,
+      "grad_norm": 2.697898190678546,
+      "learning_rate": 1.9881297621723194e-05,
+      "loss": 0.9728,
       "step": 742
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.76855194858278e-05,
-      "loss": 1.4051,
+      "epoch": 0.07776033490319205,
+      "grad_norm": 3.317056448747156,
+      "learning_rate": 1.9880776320058926e-05,
+      "loss": 1.0526,
       "step": 743
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.767869247167835e-05,
-      "loss": 1.3203,
+      "epoch": 0.07786499215070644,
+      "grad_norm": 3.070014562054591,
+      "learning_rate": 1.988025388307343e-05,
+      "loss": 1.1524,
       "step": 744
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.7671856725372737e-05,
-      "loss": 1.3042,
+      "epoch": 0.07796964939822083,
+      "grad_norm": 2.942240381438674,
+      "learning_rate": 1.9879730310826733e-05,
+      "loss": 1.1355,
       "step": 745
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.7665012254684522e-05,
-      "loss": 1.2165,
+      "epoch": 0.07807430664573521,
+      "grad_norm": 2.4967345724379144,
+      "learning_rate": 1.9879205603379e-05,
+      "loss": 1.0319,
       "step": 746
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.76581590673972e-05,
-      "loss": 1.3246,
+      "epoch": 0.0781789638932496,
+      "grad_norm": 2.5917380077833445,
+      "learning_rate": 1.9878679760790508e-05,
+      "loss": 0.8069,
       "step": 747
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.7651297171304157e-05,
-      "loss": 1.3853,
+      "epoch": 0.078283621140764,
+      "grad_norm": 1.830900388771415,
+      "learning_rate": 1.9878152783121686e-05,
+      "loss": 0.7969,
       "step": 748
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.7644426574208698e-05,
-      "loss": 1.3546,
+      "epoch": 0.07838827838827839,
+      "grad_norm": 3.1335346676454017,
+      "learning_rate": 1.9877624670433086e-05,
+      "loss": 1.1493,
       "step": 749
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.763754728392402e-05,
-      "loss": 1.3478,
+      "epoch": 0.07849293563579278,
+      "grad_norm": 2.3535864496436356,
+      "learning_rate": 1.987709542278539e-05,
+      "loss": 0.8744,
       "step": 750
     },
     {
-      "epoch": 0.49,
-      "learning_rate": 1.7630659308273204e-05,
-      "loss": 1.2798,
+      "epoch": 0.07859759288330717,
+      "grad_norm": 2.9044705756712044,
+      "learning_rate": 1.9876565040239404e-05,
+      "loss": 0.9138,
       "step": 751
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7623762655089208e-05,
-      "loss": 1.2209,
+      "epoch": 0.07870225013082156,
+      "grad_norm": 3.197340259960328,
+      "learning_rate": 1.9876033522856072e-05,
+      "loss": 1.0667,
       "step": 752
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.761685733221486e-05,
-      "loss": 1.3514,
+      "epoch": 0.07880690737833596,
+      "grad_norm": 2.672736829036044,
+      "learning_rate": 1.987550087069647e-05,
+      "loss": 1.0583,
       "step": 753
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7609943347502846e-05,
-      "loss": 1.3803,
+      "epoch": 0.07891156462585033,
+      "grad_norm": 2.9518811281187585,
+      "learning_rate": 1.9874967083821796e-05,
+      "loss": 1.0376,
       "step": 754
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7603020708815695e-05,
-      "loss": 1.3253,
+      "epoch": 0.07901622187336473,
+      "grad_norm": 3.0369965552675917,
+      "learning_rate": 1.9874432162293388e-05,
+      "loss": 1.0123,
       "step": 755
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7596089424025794e-05,
-      "loss": 1.2287,
+      "epoch": 0.07912087912087912,
+      "grad_norm": 2.527882866802051,
+      "learning_rate": 1.9873896106172705e-05,
+      "loss": 1.0585,
       "step": 756
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7589149501015345e-05,
-      "loss": 1.5018,
+      "epoch": 0.07922553636839351,
+      "grad_norm": 2.619647602913425,
+      "learning_rate": 1.9873358915521345e-05,
+      "loss": 0.9611,
       "step": 757
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.758220094767638e-05,
-      "loss": 1.1544,
+      "epoch": 0.0793301936159079,
+      "grad_norm": 3.167480526301533,
+      "learning_rate": 1.9872820590401028e-05,
+      "loss": 1.0156,
       "step": 758
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.757524377191075e-05,
-      "loss": 1.2132,
+      "epoch": 0.0794348508634223,
+      "grad_norm": 2.573985063306484,
+      "learning_rate": 1.9872281130873608e-05,
+      "loss": 1.0724,
       "step": 759
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7568277981630107e-05,
-      "loss": 1.3695,
+      "epoch": 0.07953950811093669,
+      "grad_norm": 2.946085000736679,
+      "learning_rate": 1.9871740537001078e-05,
+      "loss": 1.1423,
       "step": 760
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7561303584755903e-05,
-      "loss": 1.2031,
+      "epoch": 0.07964416535845108,
+      "grad_norm": 2.569944570278952,
+      "learning_rate": 1.9871198808845546e-05,
+      "loss": 1.0442,
       "step": 761
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7554320589219368e-05,
-      "loss": 1.285,
+      "epoch": 0.07974882260596547,
+      "grad_norm": 2.1772171170370136,
+      "learning_rate": 1.987065594646926e-05,
+      "loss": 1.0529,
       "step": 762
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.754732900296151e-05,
-      "loss": 1.2214,
+      "epoch": 0.07985347985347985,
+      "grad_norm": 2.845399072310291,
+      "learning_rate": 1.9870111949934597e-05,
+      "loss": 1.0669,
       "step": 763
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.754032883393313e-05,
-      "loss": 1.3069,
+      "epoch": 0.07995813710099424,
+      "grad_norm": 2.987591086180299,
+      "learning_rate": 1.9869566819304064e-05,
+      "loss": 1.1109,
       "step": 764
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7533320090094763e-05,
-      "loss": 1.3448,
+      "epoch": 0.08006279434850863,
+      "grad_norm": 2.9260073141957794,
+      "learning_rate": 1.986902055464029e-05,
+      "loss": 1.1047,
       "step": 765
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7526302779416703e-05,
-      "loss": 1.287,
+      "epoch": 0.08016745159602302,
+      "grad_norm": 2.978042723105865,
+      "learning_rate": 1.9868473156006055e-05,
+      "loss": 1.1091,
       "step": 766
     },
     {
-      "epoch": 0.5,
-      "learning_rate": 1.7519276909878988e-05,
-      "loss": 1.3828,
+      "epoch": 0.08027210884353742,
+      "grad_norm": 3.0411024163342995,
+      "learning_rate": 1.9867924623464244e-05,
+      "loss": 1.072,
       "step": 767
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.751224248947139e-05,
-      "loss": 1.4209,
+      "epoch": 0.08037676609105181,
+      "grad_norm": 2.804164851030498,
+      "learning_rate": 1.986737495707789e-05,
+      "loss": 1.1031,
       "step": 768
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.7505199526193398e-05,
-      "loss": 1.2545,
+      "epoch": 0.0804814233385662,
+      "grad_norm": 2.531079552784771,
+      "learning_rate": 1.986682415691015e-05,
+      "loss": 1.0565,
       "step": 769
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.749814802805423e-05,
-      "loss": 1.3509,
+      "epoch": 0.08058608058608059,
+      "grad_norm": 3.015392851420843,
+      "learning_rate": 1.9866272223024314e-05,
+      "loss": 1.0387,
       "step": 770
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.7491088003072798e-05,
-      "loss": 1.2408,
+      "epoch": 0.08069073783359498,
+      "grad_norm": 3.0741672862303258,
+      "learning_rate": 1.9865719155483794e-05,
+      "loss": 1.0325,
       "step": 771
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.748401945927771e-05,
-      "loss": 1.3241,
+      "epoch": 0.08079539508110936,
+      "grad_norm": 2.6992994649398785,
+      "learning_rate": 1.9865164954352146e-05,
+      "loss": 1.0413,
       "step": 772
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.7476942404707267e-05,
-      "loss": 1.2471,
+      "epoch": 0.08090005232862375,
+      "grad_norm": 2.652137381945092,
+      "learning_rate": 1.986460961969305e-05,
+      "loss": 1.0665,
       "step": 773
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.746985684740945e-05,
-      "loss": 1.356,
+      "epoch": 0.08100470957613815,
+      "grad_norm": 2.8296867873205316,
+      "learning_rate": 1.9864053151570305e-05,
+      "loss": 1.1983,
       "step": 774
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.74627627954419e-05,
-      "loss": 1.297,
+      "epoch": 0.08110936682365254,
+      "grad_norm": 2.6932652023688908,
+      "learning_rate": 1.986349555004786e-05,
+      "loss": 0.9712,
       "step": 775
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.745566025687193e-05,
-      "loss": 1.3418,
+      "epoch": 0.08121402407116693,
+      "grad_norm": 2.8908231690550337,
+      "learning_rate": 1.9862936815189778e-05,
+      "loss": 1.0777,
       "step": 776
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.74485492397765e-05,
-      "loss": 1.3253,
+      "epoch": 0.08131868131868132,
+      "grad_norm": 2.7585434811777327,
+      "learning_rate": 1.9862376947060263e-05,
+      "loss": 1.0458,
       "step": 777
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.74414297522422e-05,
-      "loss": 1.3031,
+      "epoch": 0.08142333856619571,
+      "grad_norm": 3.4111164784807535,
+      "learning_rate": 1.9861815945723647e-05,
+      "loss": 1.0959,
       "step": 778
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.7434301802365264e-05,
-      "loss": 1.3136,
+      "epoch": 0.0815279958137101,
+      "grad_norm": 2.794422350270696,
+      "learning_rate": 1.9861253811244383e-05,
+      "loss": 1.0158,
       "step": 779
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.7427165398251557e-05,
-      "loss": 1.2255,
+      "epoch": 0.08163265306122448,
+      "grad_norm": 2.669728034578084,
+      "learning_rate": 1.9860690543687064e-05,
+      "loss": 1.0899,
       "step": 780
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.7420020548016538e-05,
-      "loss": 1.2092,
+      "epoch": 0.08173731030873888,
+      "grad_norm": 2.8079261760048513,
+      "learning_rate": 1.9860126143116412e-05,
+      "loss": 1.1234,
       "step": 781
     },
     {
-      "epoch": 0.51,
-      "learning_rate": 1.7412867259785286e-05,
-      "loss": 1.3789,
+      "epoch": 0.08184196755625327,
+      "grad_norm": 2.926846571046833,
+      "learning_rate": 1.9859560609597282e-05,
+      "loss": 1.0468,
       "step": 782
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7405705541692476e-05,
-      "loss": 1.2396,
+      "epoch": 0.08194662480376766,
+      "grad_norm": 2.8972050163501213,
+      "learning_rate": 1.9858993943194648e-05,
+      "loss": 1.044,
       "step": 783
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7398535401882353e-05,
-      "loss": 1.2194,
+      "epoch": 0.08205128205128205,
+      "grad_norm": 2.718732850064825,
+      "learning_rate": 1.9858426143973623e-05,
+      "loss": 1.0915,
       "step": 784
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.739135684850876e-05,
-      "loss": 1.3393,
+      "epoch": 0.08215593929879644,
+      "grad_norm": 2.509735712880021,
+      "learning_rate": 1.9857857211999452e-05,
+      "loss": 1.1174,
       "step": 785
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7384169889735094e-05,
-      "loss": 1.3454,
+      "epoch": 0.08226059654631084,
+      "grad_norm": 2.7443000185455464,
+      "learning_rate": 1.98572871473375e-05,
+      "loss": 0.8195,
       "step": 786
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7376974533734317e-05,
-      "loss": 1.3465,
+      "epoch": 0.08236525379382523,
+      "grad_norm": 3.129320040379283,
+      "learning_rate": 1.9856715950053272e-05,
+      "loss": 1.0873,
       "step": 787
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7369770788688937e-05,
-      "loss": 1.3755,
+      "epoch": 0.08246991104133962,
+      "grad_norm": 2.734612164978207,
+      "learning_rate": 1.9856143620212403e-05,
+      "loss": 1.1222,
       "step": 788
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7362558662791003e-05,
-      "loss": 1.2685,
+      "epoch": 0.082574568288854,
+      "grad_norm": 3.008168731676326,
+      "learning_rate": 1.9855570157880648e-05,
+      "loss": 0.9387,
       "step": 789
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7355338164242098e-05,
-      "loss": 1.2445,
+      "epoch": 0.08267922553636839,
+      "grad_norm": 3.1280456930254923,
+      "learning_rate": 1.9854995563123904e-05,
+      "loss": 1.1312,
       "step": 790
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.734810930125332e-05,
-      "loss": 1.4126,
+      "epoch": 0.08278388278388278,
+      "grad_norm": 2.8486226186326338,
+      "learning_rate": 1.985441983600819e-05,
+      "loss": 0.9654,
       "step": 791
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.734087208204528e-05,
-      "loss": 1.2169,
+      "epoch": 0.08288854003139717,
+      "grad_norm": 2.880542844835971,
+      "learning_rate": 1.985384297659966e-05,
+      "loss": 1.0636,
       "step": 792
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.73336265148481e-05,
-      "loss": 1.3358,
+      "epoch": 0.08299319727891157,
+      "grad_norm": 2.7235931768471437,
+      "learning_rate": 1.98532649849646e-05,
+      "loss": 1.017,
       "step": 793
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7326372607901388e-05,
-      "loss": 1.364,
+      "epoch": 0.08309785452642596,
+      "grad_norm": 2.968388296598955,
+      "learning_rate": 1.9852685861169415e-05,
+      "loss": 1.2004,
       "step": 794
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7319110369454232e-05,
-      "loss": 1.309,
+      "epoch": 0.08320251177394035,
+      "grad_norm": 2.80143430250923,
+      "learning_rate": 1.9852105605280653e-05,
+      "loss": 0.9673,
       "step": 795
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7311839807765204e-05,
-      "loss": 1.2683,
+      "epoch": 0.08330716902145474,
+      "grad_norm": 2.579406765144872,
+      "learning_rate": 1.9851524217364982e-05,
+      "loss": 0.9932,
       "step": 796
     },
     {
-      "epoch": 0.52,
-      "learning_rate": 1.7304560931102334e-05,
-      "loss": 1.3465,
+      "epoch": 0.08341182626896912,
+      "grad_norm": 2.5521410566455285,
+      "learning_rate": 1.985094169748921e-05,
+      "loss": 1.0031,
       "step": 797
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.729727374774311e-05,
-      "loss": 1.2255,
+      "epoch": 0.08351648351648351,
+      "grad_norm": 2.5987542618084145,
+      "learning_rate": 1.9850358045720268e-05,
+      "loss": 1.066,
       "step": 798
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.7289978265974466e-05,
-      "loss": 1.2879,
+      "epoch": 0.0836211407639979,
+      "grad_norm": 2.6060537256654017,
+      "learning_rate": 1.9849773262125215e-05,
+      "loss": 1.0626,
       "step": 799
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.728267449409278e-05,
-      "loss": 1.28,
+      "epoch": 0.0837257980115123,
+      "grad_norm": 2.7508431371390514,
+      "learning_rate": 1.9849187346771247e-05,
+      "loss": 1.0875,
       "step": 800
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.7275362440403844e-05,
-      "loss": 1.226,
+      "epoch": 0.08383045525902669,
+      "grad_norm": 3.149432837745595,
+      "learning_rate": 1.9848600299725687e-05,
+      "loss": 1.0231,
       "step": 801
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.7268042113222877e-05,
-      "loss": 1.3465,
+      "epoch": 0.08393511250654108,
+      "grad_norm": 3.0521072775807716,
+      "learning_rate": 1.9848012121055987e-05,
+      "loss": 1.0271,
       "step": 802
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.726071352087451e-05,
-      "loss": 1.3905,
+      "epoch": 0.08403976975405547,
+      "grad_norm": 2.5717755802540085,
+      "learning_rate": 1.9847422810829732e-05,
+      "loss": 1.0108,
       "step": 803
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.725337667169276e-05,
-      "loss": 1.3734,
+      "epoch": 0.08414442700156986,
+      "grad_norm": 2.574749430789081,
+      "learning_rate": 1.984683236911463e-05,
+      "loss": 1.0094,
       "step": 804
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.7246031574021047e-05,
-      "loss": 1.3877,
+      "epoch": 0.08424908424908426,
+      "grad_norm": 2.761780907407162,
+      "learning_rate": 1.984624079597853e-05,
+      "loss": 0.9979,
       "step": 805
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.723867823621216e-05,
-      "loss": 1.4236,
+      "epoch": 0.08435374149659863,
+      "grad_norm": 2.2659780949782076,
+      "learning_rate": 1.98456480914894e-05,
+      "loss": 1.0866,
       "step": 806
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.7231316666628268e-05,
-      "loss": 1.4508,
+      "epoch": 0.08445839874411303,
+      "grad_norm": 2.880890790249313,
+      "learning_rate": 1.9845054255715345e-05,
+      "loss": 1.0858,
       "step": 807
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.7223946873640897e-05,
-      "loss": 1.3736,
+      "epoch": 0.08456305599162742,
+      "grad_norm": 3.2637552119352087,
+      "learning_rate": 1.98444592887246e-05,
+      "loss": 1.177,
       "step": 808
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.7216568865630924e-05,
-      "loss": 1.3627,
+      "epoch": 0.08466771323914181,
+      "grad_norm": 2.7805016938861216,
+      "learning_rate": 1.9843863190585527e-05,
+      "loss": 1.1337,
       "step": 809
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.720918265098857e-05,
-      "loss": 1.3579,
+      "epoch": 0.0847723704866562,
+      "grad_norm": 2.9010068085027023,
+      "learning_rate": 1.9843265961366614e-05,
+      "loss": 1.0259,
       "step": 810
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.720178823811338e-05,
-      "loss": 1.3208,
+      "epoch": 0.0848770277341706,
+      "grad_norm": 2.252289860307481,
+      "learning_rate": 1.984266760113649e-05,
+      "loss": 0.7456,
       "step": 811
     },
     {
-      "epoch": 0.53,
-      "learning_rate": 1.7194385635414242e-05,
-      "loss": 1.3973,
+      "epoch": 0.08498168498168499,
+      "grad_norm": 2.80087053004204,
+      "learning_rate": 1.9842068109963904e-05,
+      "loss": 1.1414,
       "step": 812
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.718697485130934e-05,
-      "loss": 1.2564,
+      "epoch": 0.08508634222919938,
+      "grad_norm": 2.9850122512331247,
+      "learning_rate": 1.9841467487917743e-05,
+      "loss": 1.1755,
       "step": 813
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.717955589422616e-05,
-      "loss": 1.1836,
+      "epoch": 0.08519099947671377,
+      "grad_norm": 3.2848807536761595,
+      "learning_rate": 1.9840865735067016e-05,
+      "loss": 1.0185,
       "step": 814
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.7172128772601497e-05,
-      "loss": 1.3372,
+      "epoch": 0.08529565672422815,
+      "grad_norm": 2.8281996236712272,
+      "learning_rate": 1.9840262851480866e-05,
+      "loss": 1.1314,
       "step": 815
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.7164693494881418e-05,
-      "loss": 1.3002,
+      "epoch": 0.08540031397174254,
+      "grad_norm": 2.979553704920702,
+      "learning_rate": 1.9839658837228562e-05,
+      "loss": 1.1369,
       "step": 816
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.715725006952127e-05,
-      "loss": 1.3185,
+      "epoch": 0.08550497121925693,
+      "grad_norm": 2.8697433057340134,
+      "learning_rate": 1.9839053692379516e-05,
+      "loss": 1.0285,
       "step": 817
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.7149798504985665e-05,
-      "loss": 1.2068,
+      "epoch": 0.08560962846677132,
+      "grad_norm": 3.4761142719041906,
+      "learning_rate": 1.9838447417003253e-05,
+      "loss": 1.0726,
       "step": 818
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.7142338809748475e-05,
-      "loss": 1.3255,
+      "epoch": 0.08571428571428572,
+      "grad_norm": 2.6104944016292957,
+      "learning_rate": 1.983784001116944e-05,
+      "loss": 0.9732,
       "step": 819
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.7134870992292808e-05,
-      "loss": 1.3848,
+      "epoch": 0.08581894296180011,
+      "grad_norm": 2.773752386393791,
+      "learning_rate": 1.9837231474947862e-05,
+      "loss": 1.0283,
       "step": 820
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.712739506111102e-05,
-      "loss": 1.2077,
+      "epoch": 0.0859236002093145,
+      "grad_norm": 2.70939361192841,
+      "learning_rate": 1.9836621808408448e-05,
+      "loss": 1.0672,
       "step": 821
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.7119911024704684e-05,
-      "loss": 1.346,
+      "epoch": 0.08602825745682889,
+      "grad_norm": 2.73253043675312,
+      "learning_rate": 1.9836011011621246e-05,
+      "loss": 0.9136,
       "step": 822
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.7112418891584598e-05,
-      "loss": 1.2968,
+      "epoch": 0.08613291470434327,
+      "grad_norm": 2.5285447684085773,
+      "learning_rate": 1.983539908465644e-05,
+      "loss": 1.0069,
       "step": 823
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.7104918670270763e-05,
-      "loss": 1.3476,
+      "epoch": 0.08623757195185766,
+      "grad_norm": 3.039707735225743,
+      "learning_rate": 1.983478602758434e-05,
+      "loss": 1.0909,
       "step": 824
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.7097410369292376e-05,
-      "loss": 1.2234,
+      "epoch": 0.08634222919937205,
+      "grad_norm": 3.0060510240991913,
+      "learning_rate": 1.983417184047539e-05,
+      "loss": 1.0854,
       "step": 825
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.708989399718783e-05,
-      "loss": 1.2995,
+      "epoch": 0.08644688644688645,
+      "grad_norm": 2.575407129448105,
+      "learning_rate": 1.9833556523400156e-05,
+      "loss": 0.9942,
       "step": 826
     },
     {
-      "epoch": 0.54,
-      "learning_rate": 1.708236956250469e-05,
-      "loss": 1.454,
+      "epoch": 0.08655154369440084,
+      "grad_norm": 2.399757949657678,
+      "learning_rate": 1.9832940076429342e-05,
+      "loss": 1.0323,
       "step": 827
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.707483707379969e-05,
-      "loss": 1.1749,
+      "epoch": 0.08665620094191523,
+      "grad_norm": 2.6351017617726935,
+      "learning_rate": 1.9832322499633785e-05,
+      "loss": 1.0229,
       "step": 828
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.706729653963872e-05,
-      "loss": 1.3245,
+      "epoch": 0.08676085818942962,
+      "grad_norm": 2.0539887200521685,
+      "learning_rate": 1.9831703793084436e-05,
+      "loss": 0.8446,
       "step": 829
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.7059747968596836e-05,
-      "loss": 1.3509,
+      "epoch": 0.08686551543694401,
+      "grad_norm": 2.655675530823441,
+      "learning_rate": 1.9831083956852396e-05,
+      "loss": 1.0945,
       "step": 830
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.705219136925821e-05,
-      "loss": 1.3059,
+      "epoch": 0.0869701726844584,
+      "grad_norm": 2.3301289659294224,
+      "learning_rate": 1.9830462991008875e-05,
+      "loss": 0.9982,
       "step": 831
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.704462675021616e-05,
-      "loss": 1.2868,
+      "epoch": 0.08707482993197278,
+      "grad_norm": 1.9878115279486144,
+      "learning_rate": 1.982984089562523e-05,
+      "loss": 0.7846,
       "step": 832
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.7037054120073115e-05,
-      "loss": 1.3464,
+      "epoch": 0.08717948717948718,
+      "grad_norm": 2.7308227759486043,
+      "learning_rate": 1.9829217670772936e-05,
+      "loss": 1.0973,
       "step": 833
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.7029473487440617e-05,
-      "loss": 1.3552,
+      "epoch": 0.08728414442700157,
+      "grad_norm": 3.21302132263838,
+      "learning_rate": 1.9828593316523608e-05,
+      "loss": 1.1387,
       "step": 834
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.7021884860939314e-05,
-      "loss": 1.2263,
+      "epoch": 0.08738880167451596,
+      "grad_norm": 2.580087307714874,
+      "learning_rate": 1.9827967832948985e-05,
+      "loss": 0.9605,
       "step": 835
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.7014288249198937e-05,
-      "loss": 1.3225,
+      "epoch": 0.08749345892203035,
+      "grad_norm": 3.2690704122614505,
+      "learning_rate": 1.9827341220120933e-05,
+      "loss": 1.1851,
       "step": 836
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.70066836608583e-05,
-      "loss": 1.3354,
+      "epoch": 0.08759811616954474,
+      "grad_norm": 2.4671010918720144,
+      "learning_rate": 1.982671347811145e-05,
+      "loss": 1.0144,
       "step": 837
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.699907110456529e-05,
-      "loss": 1.2825,
+      "epoch": 0.08770277341705913,
+      "grad_norm": 2.3586665066642176,
+      "learning_rate": 1.982608460699267e-05,
+      "loss": 0.9452,
       "step": 838
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.6991450588976853e-05,
-      "loss": 1.3966,
+      "epoch": 0.08780743066457353,
+      "grad_norm": 2.827484875847677,
+      "learning_rate": 1.982545460683685e-05,
+      "loss": 1.0465,
       "step": 839
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.698382212275899e-05,
-      "loss": 1.2653,
+      "epoch": 0.08791208791208792,
+      "grad_norm": 2.8254407799166446,
+      "learning_rate": 1.9824823477716374e-05,
+      "loss": 0.9405,
       "step": 840
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.697618571458673e-05,
-      "loss": 1.3246,
+      "epoch": 0.0880167451596023,
+      "grad_norm": 2.828527033050868,
+      "learning_rate": 1.982419121970377e-05,
+      "loss": 1.1592,
       "step": 841
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.6968541373144157e-05,
-      "loss": 1.4638,
+      "epoch": 0.08812140240711669,
+      "grad_norm": 2.958013431535381,
+      "learning_rate": 1.9823557832871675e-05,
+      "loss": 1.1503,
       "step": 842
     },
     {
-      "epoch": 0.55,
-      "learning_rate": 1.696088910712435e-05,
-      "loss": 1.34,
+      "epoch": 0.08822605965463108,
+      "grad_norm": 2.5277477619914617,
+      "learning_rate": 1.982292331729287e-05,
+      "loss": 1.1166,
       "step": 843
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.695322892522943e-05,
-      "loss": 1.3196,
+      "epoch": 0.08833071690214547,
+      "grad_norm": 2.8289824096892353,
+      "learning_rate": 1.982228767304026e-05,
+      "loss": 0.9632,
       "step": 844
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.6945560836170485e-05,
-      "loss": 1.316,
+      "epoch": 0.08843537414965986,
+      "grad_norm": 2.5616715856813195,
+      "learning_rate": 1.9821650900186887e-05,
+      "loss": 0.963,
       "step": 845
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.6937884848667625e-05,
-      "loss": 1.2138,
+      "epoch": 0.08854003139717426,
+      "grad_norm": 2.5986696555097586,
+      "learning_rate": 1.982101299880592e-05,
+      "loss": 1.1771,
       "step": 846
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.6930200971449923e-05,
-      "loss": 1.2587,
+      "epoch": 0.08864468864468865,
+      "grad_norm": 3.051131850126143,
+      "learning_rate": 1.9820373968970642e-05,
+      "loss": 0.9058,
       "step": 847
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.692250921325544e-05,
-      "loss": 1.2763,
+      "epoch": 0.08874934589220304,
+      "grad_norm": 2.7872842033896648,
+      "learning_rate": 1.9819733810754492e-05,
+      "loss": 1.0216,
       "step": 848
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.691480958283118e-05,
-      "loss": 1.3115,
+      "epoch": 0.08885400313971742,
+      "grad_norm": 2.582106404341739,
+      "learning_rate": 1.981909252423102e-05,
+      "loss": 1.0599,
       "step": 849
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.6907102088933118e-05,
-      "loss": 1.6282,
+      "epoch": 0.08895866038723181,
+      "grad_norm": 2.524366926584677,
+      "learning_rate": 1.981845010947391e-05,
+      "loss": 1.0088,
       "step": 850
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.6899386740326164e-05,
-      "loss": 1.3779,
+      "epoch": 0.0890633176347462,
+      "grad_norm": 2.5675683272516445,
+      "learning_rate": 1.9817806566556982e-05,
+      "loss": 0.9454,
       "step": 851
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.6891663545784146e-05,
-      "loss": 1.4226,
+      "epoch": 0.0891679748822606,
+      "grad_norm": 2.522430268689835,
+      "learning_rate": 1.9817161895554173e-05,
+      "loss": 1.0239,
       "step": 852
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.6883932514089842e-05,
-      "loss": 1.2005,
+      "epoch": 0.08927263212977499,
+      "grad_norm": 2.7158212443276897,
+      "learning_rate": 1.9816516096539563e-05,
+      "loss": 0.9493,
       "step": 853
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.6876193654034922e-05,
-      "loss": 1.3838,
+      "epoch": 0.08937728937728938,
+      "grad_norm": 2.6167231825894963,
+      "learning_rate": 1.981586916958735e-05,
+      "loss": 0.9662,
       "step": 854
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.6868446974419962e-05,
-      "loss": 1.1105,
+      "epoch": 0.08948194662480377,
+      "grad_norm": 2.566987568415224,
+      "learning_rate": 1.981522111477187e-05,
+      "loss": 0.8303,
       "step": 855
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.686069248405444e-05,
-      "loss": 1.4163,
+      "epoch": 0.08958660387231816,
+      "grad_norm": 1.9987336493251993,
+      "learning_rate": 1.981457193216759e-05,
+      "loss": 0.8232,
       "step": 856
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.68529301917567e-05,
-      "loss": 1.3145,
+      "epoch": 0.08969126111983255,
+      "grad_norm": 3.165150991848205,
+      "learning_rate": 1.98139216218491e-05,
+      "loss": 1.0353,
       "step": 857
     },
     {
-      "epoch": 0.56,
-      "learning_rate": 1.6845160106353975e-05,
-      "loss": 1.2457,
+      "epoch": 0.08979591836734693,
+      "grad_norm": 2.7844496531584766,
+      "learning_rate": 1.9813270183891117e-05,
+      "loss": 1.1477,
       "step": 858
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.6837382236682347e-05,
-      "loss": 1.2608,
+      "epoch": 0.08990057561486132,
+      "grad_norm": 2.8027853604323747,
+      "learning_rate": 1.98126176183685e-05,
+      "loss": 1.0745,
       "step": 859
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.682959659158676e-05,
-      "loss": 1.1787,
+      "epoch": 0.09000523286237572,
+      "grad_norm": 2.741551828332874,
+      "learning_rate": 1.9811963925356222e-05,
+      "loss": 1.2044,
       "step": 860
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.6821803179920996e-05,
-      "loss": 1.259,
+      "epoch": 0.09010989010989011,
+      "grad_norm": 2.2316560473076015,
+      "learning_rate": 1.98113091049294e-05,
+      "loss": 0.8208,
       "step": 861
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.681400201054766e-05,
-      "loss": 1.3097,
+      "epoch": 0.0902145473574045,
+      "grad_norm": 2.8642542806528026,
+      "learning_rate": 1.9810653157163275e-05,
+      "loss": 1.1049,
       "step": 862
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.68061930923382e-05,
-      "loss": 1.199,
+      "epoch": 0.09031920460491889,
+      "grad_norm": 2.974066935720496,
+      "learning_rate": 1.9809996082133208e-05,
+      "loss": 1.1108,
       "step": 863
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.6798376434172862e-05,
-      "loss": 1.3635,
+      "epoch": 0.09042386185243328,
+      "grad_norm": 3.751229650832022,
+      "learning_rate": 1.980933787991471e-05,
+      "loss": 1.0713,
       "step": 864
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.6790552044940692e-05,
-      "loss": 1.347,
+      "epoch": 0.09052851909994768,
+      "grad_norm": 2.355691840091666,
+      "learning_rate": 1.98086785505834e-05,
+      "loss": 1.0267,
       "step": 865
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.678271993353953e-05,
-      "loss": 1.4459,
+      "epoch": 0.09063317634746205,
+      "grad_norm": 2.4093700564341782,
+      "learning_rate": 1.9808018094215044e-05,
+      "loss": 1.0784,
       "step": 866
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.6774880108876005e-05,
-      "loss": 1.3616,
+      "epoch": 0.09073783359497645,
+      "grad_norm": 2.9768464412651965,
+      "learning_rate": 1.9807356510885526e-05,
+      "loss": 1.0033,
       "step": 867
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.6767032579865503e-05,
-      "loss": 1.096,
+      "epoch": 0.09084249084249084,
+      "grad_norm": 2.930398030142532,
+      "learning_rate": 1.980669380067086e-05,
+      "loss": 1.048,
       "step": 868
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.6759177355432186e-05,
-      "loss": 1.4208,
+      "epoch": 0.09094714809000523,
+      "grad_norm": 2.519451550821377,
+      "learning_rate": 1.9806029963647198e-05,
+      "loss": 1.0554,
       "step": 869
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.6751314444508967e-05,
-      "loss": 1.2412,
+      "epoch": 0.09105180533751962,
+      "grad_norm": 2.7989080225534084,
+      "learning_rate": 1.9805364999890813e-05,
+      "loss": 1.0775,
       "step": 870
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.674344385603748e-05,
-      "loss": 1.3455,
+      "epoch": 0.09115646258503401,
+      "grad_norm": 2.41370990986101,
+      "learning_rate": 1.980469890947811e-05,
+      "loss": 1.0857,
       "step": 871
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.6735565598968114e-05,
-      "loss": 1.2871,
+      "epoch": 0.0912611198325484,
+      "grad_norm": 3.1596287961198146,
+      "learning_rate": 1.9804031692485633e-05,
+      "loss": 1.0765,
       "step": 872
     },
     {
-      "epoch": 0.57,
-      "learning_rate": 1.672767968225997e-05,
-      "loss": 1.2198,
+      "epoch": 0.0913657770800628,
+      "grad_norm": 2.584700642828629,
+      "learning_rate": 1.9803363348990035e-05,
+      "loss": 0.941,
       "step": 873
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.6719786114880846e-05,
-      "loss": 1.2766,
+      "epoch": 0.09147043432757719,
+      "grad_norm": 2.8784994569744193,
+      "learning_rate": 1.9802693879068116e-05,
+      "loss": 1.1304,
       "step": 874
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.6711884905807264e-05,
-      "loss": 1.2012,
+      "epoch": 0.09157509157509157,
+      "grad_norm": 2.849171759445592,
+      "learning_rate": 1.9802023282796797e-05,
+      "loss": 1.0429,
       "step": 875
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.670397606402442e-05,
-      "loss": 1.4071,
+      "epoch": 0.09167974882260596,
+      "grad_norm": 3.3153795943487268,
+      "learning_rate": 1.980135156025313e-05,
+      "loss": 1.1432,
       "step": 876
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.66960595985262e-05,
-      "loss": 1.429,
+      "epoch": 0.09178440607012035,
+      "grad_norm": 2.472442664010389,
+      "learning_rate": 1.9800678711514304e-05,
+      "loss": 1.0624,
       "step": 877
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.6688135518315144e-05,
-      "loss": 1.3494,
+      "epoch": 0.09188906331763474,
+      "grad_norm": 2.7245355328223533,
+      "learning_rate": 1.9800004736657623e-05,
+      "loss": 1.0272,
       "step": 878
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.668020383240247e-05,
-      "loss": 1.2628,
+      "epoch": 0.09199372056514914,
+      "grad_norm": 2.5078380754741847,
+      "learning_rate": 1.9799329635760533e-05,
+      "loss": 1.0269,
       "step": 879
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.6672264549808033e-05,
-      "loss": 1.3206,
+      "epoch": 0.09209837781266353,
+      "grad_norm": 2.9844539652083815,
+      "learning_rate": 1.97986534089006e-05,
+      "loss": 0.9574,
       "step": 880
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.6664317679560333e-05,
-      "loss": 1.3291,
+      "epoch": 0.09220303506017792,
+      "grad_norm": 3.058018108494758,
+      "learning_rate": 1.9797976056155527e-05,
+      "loss": 1.0867,
       "step": 881
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.665636323069649e-05,
-      "loss": 1.4145,
+      "epoch": 0.09230769230769231,
+      "grad_norm": 2.675532794463805,
+      "learning_rate": 1.9797297577603144e-05,
+      "loss": 1.1262,
       "step": 882
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.6648401212262253e-05,
-      "loss": 1.3661,
+      "epoch": 0.0924123495552067,
+      "grad_norm": 2.626017155020485,
+      "learning_rate": 1.979661797332141e-05,
+      "loss": 1.1043,
       "step": 883
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.6640431633311975e-05,
-      "loss": 1.3263,
+      "epoch": 0.09251700680272108,
+      "grad_norm": 2.876484472128455,
+      "learning_rate": 1.9795937243388405e-05,
+      "loss": 1.1571,
       "step": 884
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.6632454502908605e-05,
-      "loss": 1.3406,
+      "epoch": 0.09262166405023547,
+      "grad_norm": 2.5311189083204146,
+      "learning_rate": 1.9795255387882357e-05,
+      "loss": 1.0845,
       "step": 885
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.662446983012368e-05,
-      "loss": 1.3074,
+      "epoch": 0.09272632129774987,
+      "grad_norm": 2.510089613368414,
+      "learning_rate": 1.9794572406881606e-05,
+      "loss": 1.0432,
       "step": 886
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.6616477624037317e-05,
-      "loss": 1.3181,
+      "epoch": 0.09283097854526426,
+      "grad_norm": 3.2255430026028247,
+      "learning_rate": 1.979388830046463e-05,
+      "loss": 1.1879,
       "step": 887
     },
     {
-      "epoch": 0.58,
-      "learning_rate": 1.6608477893738197e-05,
-      "loss": 1.3829,
+      "epoch": 0.09293563579277865,
+      "grad_norm": 2.4203136160149716,
+      "learning_rate": 1.9793203068710036e-05,
+      "loss": 0.8051,
       "step": 888
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6600470648323554e-05,
-      "loss": 1.284,
+      "epoch": 0.09304029304029304,
+      "grad_norm": 2.8289265412585736,
+      "learning_rate": 1.9792516711696555e-05,
+      "loss": 0.9694,
       "step": 889
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6592455896899177e-05,
-      "loss": 1.259,
+      "epoch": 0.09314495028780743,
+      "grad_norm": 2.582322870935106,
+      "learning_rate": 1.9791829229503054e-05,
+      "loss": 1.0845,
       "step": 890
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.658443364857939e-05,
-      "loss": 1.2656,
+      "epoch": 0.09324960753532183,
+      "grad_norm": 2.6892222694722427,
+      "learning_rate": 1.9791140622208523e-05,
+      "loss": 1.111,
       "step": 891
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6576403912487026e-05,
-      "loss": 1.2248,
+      "epoch": 0.0933542647828362,
+      "grad_norm": 3.0266338857528505,
+      "learning_rate": 1.9790450889892082e-05,
+      "loss": 1.0868,
       "step": 892
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.656836669775346e-05,
-      "loss": 1.4117,
+      "epoch": 0.0934589220303506,
+      "grad_norm": 2.966278924481458,
+      "learning_rate": 1.9789760032632993e-05,
+      "loss": 1.1863,
       "step": 893
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6560322013518542e-05,
-      "loss": 1.3967,
+      "epoch": 0.09356357927786499,
+      "grad_norm": 2.6158992441922773,
+      "learning_rate": 1.9789068050510627e-05,
+      "loss": 1.1321,
       "step": 894
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6552269868930646e-05,
-      "loss": 1.3535,
+      "epoch": 0.09366823652537938,
+      "grad_norm": 3.081967659922265,
+      "learning_rate": 1.9788374943604496e-05,
+      "loss": 0.9553,
       "step": 895
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6544210273146608e-05,
-      "loss": 1.3346,
+      "epoch": 0.09377289377289377,
+      "grad_norm": 2.527503447093225,
+      "learning_rate": 1.9787680711994244e-05,
+      "loss": 1.0302,
       "step": 896
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6536143235331743e-05,
-      "loss": 1.4138,
+      "epoch": 0.09387755102040816,
+      "grad_norm": 2.7162559652765466,
+      "learning_rate": 1.9786985355759633e-05,
+      "loss": 1.0795,
       "step": 897
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6528068764659838e-05,
-      "loss": 1.3445,
+      "epoch": 0.09398220826792256,
+      "grad_norm": 2.329247727727232,
+      "learning_rate": 1.9786288874980567e-05,
+      "loss": 0.8319,
       "step": 898
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.651998687031312e-05,
-      "loss": 1.2632,
+      "epoch": 0.09408686551543695,
+      "grad_norm": 2.775539898357401,
+      "learning_rate": 1.978559126973707e-05,
+      "loss": 1.1444,
       "step": 899
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6511897561482263e-05,
-      "loss": 1.1623,
+      "epoch": 0.09419152276295134,
+      "grad_norm": 2.9580563525888635,
+      "learning_rate": 1.97848925401093e-05,
+      "loss": 1.0485,
       "step": 900
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6503800847366373e-05,
-      "loss": 1.265,
+      "epoch": 0.09429618001046572,
+      "grad_norm": 2.5695675385307553,
+      "learning_rate": 1.978419268617754e-05,
+      "loss": 0.9955,
       "step": 901
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6495696737172982e-05,
-      "loss": 1.2231,
+      "epoch": 0.09440083725798011,
+      "grad_norm": 2.2552487789593623,
+      "learning_rate": 1.9783491708022203e-05,
+      "loss": 0.8651,
       "step": 902
     },
     {
-      "epoch": 0.59,
-      "learning_rate": 1.6487585240118025e-05,
-      "loss": 1.3099,
+      "epoch": 0.0945054945054945,
+      "grad_norm": 2.9417126110660554,
+      "learning_rate": 1.978278960572384e-05,
+      "loss": 1.1695,
       "step": 903
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.647946636542584e-05,
-      "loss": 1.3066,
+      "epoch": 0.0946101517530089,
+      "grad_norm": 2.183916665081056,
+      "learning_rate": 1.9782086379363116e-05,
+      "loss": 0.9073,
       "step": 904
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6471340122329158e-05,
-      "loss": 1.2451,
+      "epoch": 0.09471480900052329,
+      "grad_norm": 2.843970018173232,
+      "learning_rate": 1.978138202902084e-05,
+      "loss": 1.1337,
       "step": 905
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6463206520069083e-05,
-      "loss": 1.2643,
+      "epoch": 0.09481946624803768,
+      "grad_norm": 2.5391065676287794,
+      "learning_rate": 1.9780676554777937e-05,
+      "loss": 1.0872,
       "step": 906
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6455065567895098e-05,
-      "loss": 1.3528,
+      "epoch": 0.09492412349555207,
+      "grad_norm": 2.580130909019054,
+      "learning_rate": 1.9779969956715473e-05,
+      "loss": 0.9439,
       "step": 907
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6446917275065032e-05,
-      "loss": 1.3333,
+      "epoch": 0.09502878074306646,
+      "grad_norm": 2.8311688424167016,
+      "learning_rate": 1.9779262234914635e-05,
+      "loss": 1.1965,
       "step": 908
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.643876165084507e-05,
-      "loss": 1.2977,
+      "epoch": 0.09513343799058084,
+      "grad_norm": 3.140721287835419,
+      "learning_rate": 1.977855338945674e-05,
+      "loss": 1.0228,
       "step": 909
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6430598704509723e-05,
-      "loss": 1.3413,
+      "epoch": 0.09523809523809523,
+      "grad_norm": 2.4322100869774976,
+      "learning_rate": 1.977784342042323e-05,
+      "loss": 1.1553,
       "step": 910
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.642242844534185e-05,
-      "loss": 1.2886,
+      "epoch": 0.09534275248560962,
+      "grad_norm": 2.5964369564075196,
+      "learning_rate": 1.9777132327895702e-05,
+      "loss": 1.0501,
       "step": 911
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6414250882632607e-05,
-      "loss": 1.236,
+      "epoch": 0.09544740973312402,
+      "grad_norm": 2.6073741523793186,
+      "learning_rate": 1.9776420111955842e-05,
+      "loss": 1.0065,
       "step": 912
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6406066025681462e-05,
-      "loss": 1.5691,
+      "epoch": 0.09555206698063841,
+      "grad_norm": 2.6220467436425863,
+      "learning_rate": 1.9775706772685495e-05,
+      "loss": 1.1565,
       "step": 913
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6397873883796183e-05,
-      "loss": 1.2697,
+      "epoch": 0.0956567242281528,
+      "grad_norm": 2.923213677461564,
+      "learning_rate": 1.977499231016662e-05,
+      "loss": 1.0593,
       "step": 914
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6389674466292812e-05,
-      "loss": 1.3026,
+      "epoch": 0.09576138147566719,
+      "grad_norm": 2.6380776247624027,
+      "learning_rate": 1.9774276724481316e-05,
+      "loss": 1.0011,
       "step": 915
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.638146778249567e-05,
-      "loss": 1.1728,
+      "epoch": 0.09586603872318158,
+      "grad_norm": 3.8624650361902617,
+      "learning_rate": 1.9773560015711798e-05,
+      "loss": 1.0226,
       "step": 916
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.637325384173734e-05,
-      "loss": 1.3402,
+      "epoch": 0.09597069597069598,
+      "grad_norm": 2.818754439571562,
+      "learning_rate": 1.977284218394042e-05,
+      "loss": 1.1649,
       "step": 917
     },
     {
-      "epoch": 0.6,
-      "learning_rate": 1.6365032653358665e-05,
-      "loss": 1.2947,
+      "epoch": 0.09607535321821035,
+      "grad_norm": 2.4474931176286128,
+      "learning_rate": 1.977212322924967e-05,
+      "loss": 1.0857,
       "step": 918
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.635680422670872e-05,
-      "loss": 1.3534,
+      "epoch": 0.09618001046572475,
+      "grad_norm": 2.8206323427096165,
+      "learning_rate": 1.9771403151722147e-05,
+      "loss": 1.0433,
       "step": 919
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6348568571144816e-05,
-      "loss": 1.3345,
+      "epoch": 0.09628466771323914,
+      "grad_norm": 3.213099908339713,
+      "learning_rate": 1.9770681951440595e-05,
+      "loss": 1.1182,
       "step": 920
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.634032569603248e-05,
-      "loss": 1.3136,
+      "epoch": 0.09638932496075353,
+      "grad_norm": 2.9662481224716974,
+      "learning_rate": 1.9769959628487876e-05,
+      "loss": 0.9635,
       "step": 921
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.633207561074546e-05,
-      "loss": 1.2049,
+      "epoch": 0.09649398220826792,
+      "grad_norm": 4.016583647471507,
+      "learning_rate": 1.9769236182946993e-05,
+      "loss": 0.9521,
       "step": 922
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6323818324665693e-05,
-      "loss": 1.2287,
+      "epoch": 0.09659863945578231,
+      "grad_norm": 3.508939746867643,
+      "learning_rate": 1.9768511614901067e-05,
+      "loss": 1.2703,
       "step": 923
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.63155538471833e-05,
-      "loss": 1.457,
+      "epoch": 0.0967032967032967,
+      "grad_norm": 2.953270773462452,
+      "learning_rate": 1.9767785924433354e-05,
+      "loss": 1.124,
       "step": 924
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6307282187696602e-05,
-      "loss": 1.236,
+      "epoch": 0.0968079539508111,
+      "grad_norm": 2.537786165701537,
+      "learning_rate": 1.9767059111627237e-05,
+      "loss": 1.0141,
       "step": 925
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.629900335561206e-05,
-      "loss": 1.2837,
+      "epoch": 0.09691261119832549,
+      "grad_norm": 3.2226759688842415,
+      "learning_rate": 1.9766331176566226e-05,
+      "loss": 1.0785,
       "step": 926
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6290717360344312e-05,
-      "loss": 1.3094,
+      "epoch": 0.09701726844583987,
+      "grad_norm": 2.8316546956911512,
+      "learning_rate": 1.9765602119333967e-05,
+      "loss": 1.084,
       "step": 927
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6282424211316134e-05,
-      "loss": 1.2838,
+      "epoch": 0.09712192569335426,
+      "grad_norm": 2.8227008106460985,
+      "learning_rate": 1.9764871940014226e-05,
+      "loss": 1.0238,
       "step": 928
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6274123917958438e-05,
-      "loss": 1.2095,
+      "epoch": 0.09722658294086865,
+      "grad_norm": 2.6956185802209323,
+      "learning_rate": 1.97641406386909e-05,
+      "loss": 1.1413,
       "step": 929
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.626581648971026e-05,
-      "loss": 1.2071,
+      "epoch": 0.09733124018838304,
+      "grad_norm": 3.011507745775884,
+      "learning_rate": 1.976340821544803e-05,
+      "loss": 1.0178,
       "step": 930
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.625750193601875e-05,
-      "loss": 1.2638,
+      "epoch": 0.09743589743589744,
+      "grad_norm": 2.7090414787190547,
+      "learning_rate": 1.9762674670369757e-05,
+      "loss": 1.1175,
       "step": 931
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.624918026633916e-05,
-      "loss": 1.2514,
+      "epoch": 0.09754055468341183,
+      "grad_norm": 2.9878646220501337,
+      "learning_rate": 1.9761940003540375e-05,
+      "loss": 1.0434,
       "step": 932
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6240851490134834e-05,
-      "loss": 1.1429,
+      "epoch": 0.09764521193092622,
+      "grad_norm": 2.9207377165620563,
+      "learning_rate": 1.9761204215044295e-05,
+      "loss": 0.9525,
       "step": 933
     },
     {
-      "epoch": 0.61,
-      "learning_rate": 1.6232515616877204e-05,
-      "loss": 1.3468,
+      "epoch": 0.09774986917844061,
+      "grad_norm": 2.7502875366492985,
+      "learning_rate": 1.9760467304966066e-05,
+      "loss": 0.9973,
       "step": 934
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6224172656045767e-05,
-      "loss": 1.3237,
+      "epoch": 0.09785452642595499,
+      "grad_norm": 2.866734557712832,
+      "learning_rate": 1.9759729273390356e-05,
+      "loss": 1.0749,
       "step": 935
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6215822617128075e-05,
-      "loss": 1.2862,
+      "epoch": 0.09795918367346938,
+      "grad_norm": 2.195362172452806,
+      "learning_rate": 1.9758990120401967e-05,
+      "loss": 0.7863,
       "step": 936
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.620746550961974e-05,
-      "loss": 1.2863,
+      "epoch": 0.09806384092098377,
+      "grad_norm": 2.470173351806748,
+      "learning_rate": 1.975824984608583e-05,
+      "loss": 1.0308,
       "step": 937
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6199101343024405e-05,
-      "loss": 1.4152,
+      "epoch": 0.09816849816849817,
+      "grad_norm": 1.9811710309835036,
+      "learning_rate": 1.9757508450527006e-05,
+      "loss": 0.8232,
       "step": 938
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.619073012685374e-05,
-      "loss": 1.4615,
+      "epoch": 0.09827315541601256,
+      "grad_norm": 2.608135387091775,
+      "learning_rate": 1.9756765933810674e-05,
+      "loss": 1.031,
       "step": 939
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.618235187062744e-05,
-      "loss": 1.2388,
+      "epoch": 0.09837781266352695,
+      "grad_norm": 2.57500776512692,
+      "learning_rate": 1.9756022296022164e-05,
+      "loss": 1.1257,
       "step": 940
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6173966583873195e-05,
-      "loss": 1.2309,
+      "epoch": 0.09848246991104134,
+      "grad_norm": 2.976812024734902,
+      "learning_rate": 1.9755277537246914e-05,
+      "loss": 1.1147,
       "step": 941
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6165574276126702e-05,
-      "loss": 1.1803,
+      "epoch": 0.09858712715855573,
+      "grad_norm": 2.6608551364060307,
+      "learning_rate": 1.9754531657570495e-05,
+      "loss": 1.1327,
       "step": 942
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6157174956931627e-05,
-      "loss": 1.3408,
+      "epoch": 0.09869178440607013,
+      "grad_norm": 2.564080580950772,
+      "learning_rate": 1.9753784657078613e-05,
+      "loss": 1.0696,
       "step": 943
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6148768635839623e-05,
-      "loss": 1.3339,
+      "epoch": 0.0987964416535845,
+      "grad_norm": 2.3432953283718327,
+      "learning_rate": 1.9753036535857106e-05,
+      "loss": 1.0246,
       "step": 944
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6140355322410304e-05,
-      "loss": 1.3751,
+      "epoch": 0.0989010989010989,
+      "grad_norm": 2.9307230460788904,
+      "learning_rate": 1.9752287293991927e-05,
+      "loss": 1.0714,
       "step": 945
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6131935026211224e-05,
-      "loss": 1.3877,
+      "epoch": 0.09900575614861329,
+      "grad_norm": 3.2102158887761867,
+      "learning_rate": 1.9751536931569167e-05,
+      "loss": 1.198,
       "step": 946
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.612350775681789e-05,
-      "loss": 1.207,
+      "epoch": 0.09911041339612768,
+      "grad_norm": 2.8495315326753645,
+      "learning_rate": 1.975078544867505e-05,
+      "loss": 1.0824,
       "step": 947
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.611507352381373e-05,
-      "loss": 1.4494,
+      "epoch": 0.09921507064364207,
+      "grad_norm": 2.682593584386815,
+      "learning_rate": 1.9750032845395914e-05,
+      "loss": 1.0324,
       "step": 948
     },
     {
-      "epoch": 0.62,
-      "learning_rate": 1.6106632336790105e-05,
-      "loss": 1.3727,
+      "epoch": 0.09931972789115646,
+      "grad_norm": 2.8693006339538116,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 1.1512,
       "step": 949
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.609818420534627e-05,
-      "loss": 1.3039,
+      "epoch": 0.09942438513867086,
+      "grad_norm": 2.7517932371635774,
+      "learning_rate": 1.974852427802863e-05,
+      "loss": 0.9674,
       "step": 950
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.608972913908938e-05,
-      "loss": 1.1756,
+      "epoch": 0.09952904238618525,
+      "grad_norm": 2.734800109801035,
+      "learning_rate": 1.9747768314113817e-05,
+      "loss": 1.0879,
       "step": 951
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.608126714763448e-05,
-      "loss": 1.2914,
+      "epoch": 0.09963369963369964,
+      "grad_norm": 2.7033090734169893,
+      "learning_rate": 1.9747011230160664e-05,
+      "loss": 1.1549,
       "step": 952
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6072798240604487e-05,
-      "loss": 1.3273,
+      "epoch": 0.09973835688121402,
+      "grad_norm": 2.6253103443822683,
+      "learning_rate": 1.9746253026256164e-05,
+      "loss": 1.0787,
       "step": 953
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6064322427630186e-05,
-      "loss": 1.322,
+      "epoch": 0.09984301412872841,
+      "grad_norm": 3.125722150105614,
+      "learning_rate": 1.974549370248743e-05,
+      "loss": 1.1259,
       "step": 954
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6055839718350207e-05,
-      "loss": 1.1453,
+      "epoch": 0.0999476713762428,
+      "grad_norm": 2.7448374047244326,
+      "learning_rate": 1.9744733258941717e-05,
+      "loss": 0.9836,
       "step": 955
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6047350122411038e-05,
-      "loss": 1.3025,
+      "epoch": 0.1000523286237572,
+      "grad_norm": 2.955943924297256,
+      "learning_rate": 1.9743971695706398e-05,
+      "loss": 1.1255,
       "step": 956
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6038853649466983e-05,
-      "loss": 1.1693,
+      "epoch": 0.10015698587127159,
+      "grad_norm": 2.7395719342747897,
+      "learning_rate": 1.9743209012868977e-05,
+      "loss": 1.042,
       "step": 957
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6030350309180173e-05,
-      "loss": 1.2767,
+      "epoch": 0.10026164311878598,
+      "grad_norm": 2.5186535770163103,
+      "learning_rate": 1.974244521051709e-05,
+      "loss": 0.7976,
       "step": 958
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6021840111220548e-05,
-      "loss": 1.4647,
+      "epoch": 0.10036630036630037,
+      "grad_norm": 2.499288564217357,
+      "learning_rate": 1.9741680288738495e-05,
+      "loss": 1.0014,
       "step": 959
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.6013323065265848e-05,
-      "loss": 1.2958,
+      "epoch": 0.10047095761381476,
+      "grad_norm": 2.8901978870749008,
+      "learning_rate": 1.9740914247621088e-05,
+      "loss": 1.074,
       "step": 960
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.60047991810016e-05,
-      "loss": 1.3274,
+      "epoch": 0.10057561486132914,
+      "grad_norm": 2.624567677834044,
+      "learning_rate": 1.9740147087252887e-05,
+      "loss": 1.0408,
       "step": 961
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.59962684681211e-05,
-      "loss": 1.325,
+      "epoch": 0.10068027210884353,
+      "grad_norm": 2.1866362133414827,
+      "learning_rate": 1.973937880772204e-05,
+      "loss": 1.0002,
       "step": 962
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.5987730936325424e-05,
-      "loss": 1.3704,
+      "epoch": 0.10078492935635792,
+      "grad_norm": 2.3411382143212793,
+      "learning_rate": 1.9738609409116824e-05,
+      "loss": 0.8504,
       "step": 963
     },
     {
-      "epoch": 0.63,
-      "learning_rate": 1.597918659532339e-05,
-      "loss": 1.3307,
+      "epoch": 0.10088958660387232,
+      "grad_norm": 2.9089697665795864,
+      "learning_rate": 1.9737838891525647e-05,
+      "loss": 1.0074,
       "step": 964
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5970635454831565e-05,
-      "loss": 1.2188,
+      "epoch": 0.10099424385138671,
+      "grad_norm": 2.532467662978773,
+      "learning_rate": 1.973706725503704e-05,
+      "loss": 1.0635,
       "step": 965
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5962077524574248e-05,
-      "loss": 1.2714,
+      "epoch": 0.1010989010989011,
+      "grad_norm": 2.6762533062616125,
+      "learning_rate": 1.9736294499739667e-05,
+      "loss": 0.9866,
       "step": 966
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.595351281428345e-05,
-      "loss": 1.2796,
+      "epoch": 0.10120355834641549,
+      "grad_norm": 3.0420458033024658,
+      "learning_rate": 1.9735520625722312e-05,
+      "loss": 1.0371,
       "step": 967
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5944941333698912e-05,
-      "loss": 1.2497,
+      "epoch": 0.10130821559392988,
+      "grad_norm": 2.5023937580187985,
+      "learning_rate": 1.9734745633073908e-05,
+      "loss": 1.1425,
       "step": 968
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5936363092568058e-05,
-      "loss": 1.244,
+      "epoch": 0.10141287284144428,
+      "grad_norm": 2.638941018679516,
+      "learning_rate": 1.9733969521883494e-05,
+      "loss": 1.0191,
       "step": 969
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5927778100646007e-05,
-      "loss": 1.212,
+      "epoch": 0.10151753008895865,
+      "grad_norm": 3.0955666704793576,
+      "learning_rate": 1.9733192292240252e-05,
+      "loss": 1.013,
       "step": 970
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5919186367695546e-05,
-      "loss": 1.3979,
+      "epoch": 0.10162218733647305,
+      "grad_norm": 2.3039556868285223,
+      "learning_rate": 1.973241394423348e-05,
+      "loss": 0.8819,
       "step": 971
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5910587903487142e-05,
-      "loss": 1.3326,
+      "epoch": 0.10172684458398744,
+      "grad_norm": 2.808971945311129,
+      "learning_rate": 1.973163447795262e-05,
+      "loss": 1.2111,
       "step": 972
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5901982717798903e-05,
-      "loss": 1.3511,
+      "epoch": 0.10183150183150183,
+      "grad_norm": 2.1370050360048927,
+      "learning_rate": 1.973085389348723e-05,
+      "loss": 0.7627,
       "step": 973
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5893370820416594e-05,
-      "loss": 1.3456,
+      "epoch": 0.10193615907901622,
+      "grad_norm": 2.597731049761627,
+      "learning_rate": 1.9730072190926996e-05,
+      "loss": 1.1669,
       "step": 974
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5884752221133597e-05,
-      "loss": 1.3462,
+      "epoch": 0.10204081632653061,
+      "grad_norm": 3.395661508370253,
+      "learning_rate": 1.9729289370361745e-05,
+      "loss": 0.9587,
       "step": 975
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.587612692975093e-05,
-      "loss": 1.2286,
+      "epoch": 0.102145473574045,
+      "grad_norm": 3.0895550900923294,
+      "learning_rate": 1.9728505431881425e-05,
+      "loss": 0.9229,
       "step": 976
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.5867494956077215e-05,
-      "loss": 1.3136,
+      "epoch": 0.1022501308215594,
+      "grad_norm": 2.458084346576055,
+      "learning_rate": 1.972772037557611e-05,
+      "loss": 1.0632,
       "step": 977
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.585885630992867e-05,
-      "loss": 1.3715,
+      "epoch": 0.10235478806907378,
+      "grad_norm": 3.0516732311665313,
+      "learning_rate": 1.9726934201535998e-05,
+      "loss": 1.1159,
       "step": 978
     },
     {
-      "epoch": 0.64,
-      "learning_rate": 1.58502110011291e-05,
-      "loss": 1.2604,
+      "epoch": 0.10245944531658817,
+      "grad_norm": 2.354235000879386,
+      "learning_rate": 1.972614690985143e-05,
+      "loss": 0.9929,
       "step": 979
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5841559039509898e-05,
-      "loss": 1.1808,
+      "epoch": 0.10256410256410256,
+      "grad_norm": 3.2111492690142995,
+      "learning_rate": 1.9725358500612865e-05,
+      "loss": 0.9502,
       "step": 980
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5832900434910014e-05,
-      "loss": 1.2897,
+      "epoch": 0.10266875981161695,
+      "grad_norm": 3.0248285962435895,
+      "learning_rate": 1.9724568973910893e-05,
+      "loss": 1.0119,
       "step": 981
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5824235197175954e-05,
-      "loss": 1.2339,
+      "epoch": 0.10277341705913134,
+      "grad_norm": 2.770623468886491,
+      "learning_rate": 1.9723778329836235e-05,
+      "loss": 1.0712,
       "step": 982
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5815563336161756e-05,
-      "loss": 1.2399,
+      "epoch": 0.10287807430664574,
+      "grad_norm": 2.5692351103675364,
+      "learning_rate": 1.9722986568479734e-05,
+      "loss": 1.0985,
       "step": 983
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.580688486172902e-05,
-      "loss": 1.3925,
+      "epoch": 0.10298273155416013,
+      "grad_norm": 2.8212239759961104,
+      "learning_rate": 1.9722193689932364e-05,
+      "loss": 1.0989,
       "step": 984
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5798199783746828e-05,
-      "loss": 1.3116,
+      "epoch": 0.10308738880167452,
+      "grad_norm": 2.7660547575806187,
+      "learning_rate": 1.9721399694285227e-05,
+      "loss": 1.0853,
       "step": 985
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5789508112091804e-05,
-      "loss": 1.4549,
+      "epoch": 0.10319204604918891,
+      "grad_norm": 2.7676555235602995,
+      "learning_rate": 1.9720604581629562e-05,
+      "loss": 1.1553,
       "step": 986
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.578080985664805e-05,
-      "loss": 1.2233,
+      "epoch": 0.10329670329670329,
+      "grad_norm": 2.487671204295441,
+      "learning_rate": 1.9719808352056728e-05,
+      "loss": 0.9583,
       "step": 987
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5772105027307168e-05,
-      "loss": 1.1387,
+      "epoch": 0.10340136054421768,
+      "grad_norm": 2.637848541738491,
+      "learning_rate": 1.9719011005658203e-05,
+      "loss": 1.1547,
       "step": 988
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5763393633968223e-05,
-      "loss": 1.2636,
+      "epoch": 0.10350601779173207,
+      "grad_norm": 2.6148860251102524,
+      "learning_rate": 1.9718212542525616e-05,
+      "loss": 1.0602,
       "step": 989
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5754675686537756e-05,
-      "loss": 1.155,
+      "epoch": 0.10361067503924647,
+      "grad_norm": 2.512048499705755,
+      "learning_rate": 1.9717412962750704e-05,
+      "loss": 1.1129,
       "step": 990
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5745951194929756e-05,
-      "loss": 1.098,
+      "epoch": 0.10371533228676086,
+      "grad_norm": 2.9398186150069945,
+      "learning_rate": 1.9716612266425343e-05,
+      "loss": 1.0064,
       "step": 991
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5737220169065656e-05,
-      "loss": 1.1932,
+      "epoch": 0.10381998953427525,
+      "grad_norm": 2.5999608156502245,
+      "learning_rate": 1.971581045364154e-05,
+      "loss": 1.0484,
       "step": 992
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5728482618874314e-05,
-      "loss": 1.1976,
+      "epoch": 0.10392464678178964,
+      "grad_norm": 2.6014924463154743,
+      "learning_rate": 1.9715007524491413e-05,
+      "loss": 0.9897,
       "step": 993
     },
     {
-      "epoch": 0.65,
-      "learning_rate": 1.5719738554292017e-05,
-      "loss": 1.2005,
+      "epoch": 0.10402930402930403,
+      "grad_norm": 2.361645305711183,
+      "learning_rate": 1.9714203479067232e-05,
+      "loss": 1.1281,
       "step": 994
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.571098798526245e-05,
-      "loss": 1.5259,
+      "epoch": 0.10413396127681843,
+      "grad_norm": 2.5112783788806685,
+      "learning_rate": 1.9713398317461377e-05,
+      "loss": 0.8673,
       "step": 995
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.5702230921736705e-05,
-      "loss": 1.2184,
+      "epoch": 0.1042386185243328,
+      "grad_norm": 2.7601681785587373,
+      "learning_rate": 1.9712592039766364e-05,
+      "loss": 1.1427,
       "step": 996
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.569346737367325e-05,
-      "loss": 1.3392,
+      "epoch": 0.1043432757718472,
+      "grad_norm": 2.9435423523320696,
+      "learning_rate": 1.9711784646074837e-05,
+      "loss": 1.0373,
       "step": 997
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.5684697351037935e-05,
-      "loss": 1.3473,
+      "epoch": 0.10444793301936159,
+      "grad_norm": 2.513526513093683,
+      "learning_rate": 1.9710976136479565e-05,
+      "loss": 1.0433,
       "step": 998
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.5675920863803967e-05,
-      "loss": 1.2805,
+      "epoch": 0.10455259026687598,
+      "grad_norm": 2.8188327082080575,
+      "learning_rate": 1.971016651107345e-05,
+      "loss": 1.0745,
       "step": 999
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.566713792195191e-05,
-      "loss": 1.1129,
+      "epoch": 0.10465724751439037,
+      "grad_norm": 2.713657193992208,
+      "learning_rate": 1.9709355769949515e-05,
+      "loss": 1.1284,
       "step": 1000
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.565834853546966e-05,
-      "loss": 1.2706,
+      "epoch": 0.10476190476190476,
+      "grad_norm": 3.0415166367388573,
+      "learning_rate": 1.9708543913200925e-05,
+      "loss": 1.0857,
       "step": 1001
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.5649552714352448e-05,
-      "loss": 1.323,
+      "epoch": 0.10486656200941916,
+      "grad_norm": 3.0280839509415074,
+      "learning_rate": 1.9707730940920955e-05,
+      "loss": 1.1141,
       "step": 1002
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.5640750468602822e-05,
-      "loss": 1.3073,
+      "epoch": 0.10497121925693355,
+      "grad_norm": 2.94662887338583,
+      "learning_rate": 1.970691685320302e-05,
+      "loss": 1.0994,
       "step": 1003
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.563194180823064e-05,
-      "loss": 1.2237,
+      "epoch": 0.10507587650444793,
+      "grad_norm": 2.610110149296357,
+      "learning_rate": 1.9706101650140657e-05,
+      "loss": 1.1039,
       "step": 1004
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.562312674325304e-05,
-      "loss": 1.3814,
+      "epoch": 0.10518053375196232,
+      "grad_norm": 3.192863824515813,
+      "learning_rate": 1.9705285331827544e-05,
+      "loss": 1.0594,
       "step": 1005
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.5614305283694462e-05,
-      "loss": 1.3482,
+      "epoch": 0.10528519099947671,
+      "grad_norm": 3.0083558422084393,
+      "learning_rate": 1.9704467898357473e-05,
+      "loss": 0.9451,
       "step": 1006
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.56054774395866e-05,
-      "loss": 1.2636,
+      "epoch": 0.1053898482469911,
+      "grad_norm": 2.692443979020099,
+      "learning_rate": 1.970364934982436e-05,
+      "loss": 1.0166,
       "step": 1007
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.559664322096843e-05,
-      "loss": 1.2905,
+      "epoch": 0.1054945054945055,
+      "grad_norm": 2.473816517340895,
+      "learning_rate": 1.9702829686322272e-05,
+      "loss": 1.0046,
       "step": 1008
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.5587802637886156e-05,
-      "loss": 1.2747,
+      "epoch": 0.10559916274201989,
+      "grad_norm": 2.5772774995170136,
+      "learning_rate": 1.970200890794538e-05,
+      "loss": 1.0931,
       "step": 1009
     },
     {
-      "epoch": 0.66,
-      "learning_rate": 1.557895570039323e-05,
-      "loss": 1.2749,
+      "epoch": 0.10570381998953428,
+      "grad_norm": 2.4344331305938423,
+      "learning_rate": 1.9701187014787994e-05,
+      "loss": 0.9071,
       "step": 1010
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.5570102418550327e-05,
-      "loss": 1.2037,
+      "epoch": 0.10580847723704867,
+      "grad_norm": 2.49913692348178,
+      "learning_rate": 1.970036400694456e-05,
+      "loss": 1.0385,
       "step": 1011
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.5561242802425337e-05,
-      "loss": 1.3082,
+      "epoch": 0.10591313448456306,
+      "grad_norm": 2.7676304166369197,
+      "learning_rate": 1.9699539884509632e-05,
+      "loss": 0.9947,
       "step": 1012
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.5552376862093357e-05,
-      "loss": 1.1666,
+      "epoch": 0.10601779173207744,
+      "grad_norm": 2.653669677176504,
+      "learning_rate": 1.969871464757791e-05,
+      "loss": 1.0598,
       "step": 1013
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.554350460763667e-05,
-      "loss": 1.1587,
+      "epoch": 0.10612244897959183,
+      "grad_norm": 3.1238605727049733,
+      "learning_rate": 1.9697888296244214e-05,
+      "loss": 1.1389,
       "step": 1014
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.5534626049144753e-05,
-      "loss": 1.3303,
+      "epoch": 0.10622710622710622,
+      "grad_norm": 2.9772018563723672,
+      "learning_rate": 1.9697060830603495e-05,
+      "loss": 1.0753,
       "step": 1015
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.552574119671423e-05,
-      "loss": 1.254,
+      "epoch": 0.10633176347462062,
+      "grad_norm": 2.668071509791539,
+      "learning_rate": 1.9696232250750823e-05,
+      "loss": 1.0608,
       "step": 1016
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.55168500604489e-05,
-      "loss": 1.2916,
+      "epoch": 0.10643642072213501,
+      "grad_norm": 2.589387863961408,
+      "learning_rate": 1.9695402556781414e-05,
+      "loss": 1.0848,
       "step": 1017
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.5507952650459697e-05,
-      "loss": 1.2834,
+      "epoch": 0.1065410779696494,
+      "grad_norm": 2.575997267076478,
+      "learning_rate": 1.9694571748790593e-05,
+      "loss": 1.0157,
       "step": 1018
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.54990489768647e-05,
-      "loss": 1.2852,
+      "epoch": 0.10664573521716379,
+      "grad_norm": 2.4804280288072222,
+      "learning_rate": 1.9693739826873828e-05,
+      "loss": 0.9616,
       "step": 1019
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.5490139049789097e-05,
-      "loss": 1.37,
+      "epoch": 0.10675039246467818,
+      "grad_norm": 2.8490573537756854,
+      "learning_rate": 1.9692906791126708e-05,
+      "loss": 1.0175,
       "step": 1020
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.5481222879365205e-05,
-      "loss": 1.2925,
+      "epoch": 0.10685504971219256,
+      "grad_norm": 2.7446287985468367,
+      "learning_rate": 1.9692072641644945e-05,
+      "loss": 1.0971,
       "step": 1021
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.5472300475732427e-05,
-      "loss": 1.3541,
+      "epoch": 0.10695970695970695,
+      "grad_norm": 3.0448502861545292,
+      "learning_rate": 1.9691237378524387e-05,
+      "loss": 1.014,
       "step": 1022
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.546337184903726e-05,
-      "loss": 1.3664,
+      "epoch": 0.10706436420722135,
+      "grad_norm": 2.6319835236235622,
+      "learning_rate": 1.9690401001861006e-05,
+      "loss": 1.0825,
       "step": 1023
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.5454437009433277e-05,
-      "loss": 1.2974,
+      "epoch": 0.10716902145473574,
+      "grad_norm": 2.4456143448333547,
+      "learning_rate": 1.9689563511750907e-05,
+      "loss": 1.1329,
       "step": 1024
     },
     {
-      "epoch": 0.67,
-      "learning_rate": 1.5445495967081114e-05,
-      "loss": 1.3451,
+      "epoch": 0.10727367870225013,
+      "grad_norm": 3.059839362265981,
+      "learning_rate": 1.9688724908290318e-05,
+      "loss": 1.1094,
       "step": 1025
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.543654873214846e-05,
-      "loss": 1.4349,
+      "epoch": 0.10737833594976452,
+      "grad_norm": 2.4535767216133797,
+      "learning_rate": 1.9687885191575597e-05,
+      "loss": 0.8073,
       "step": 1026
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.542759531481006e-05,
-      "loss": 1.2617,
+      "epoch": 0.10748299319727891,
+      "grad_norm": 2.2330132410114656,
+      "learning_rate": 1.9687044361703228e-05,
+      "loss": 0.9921,
       "step": 1027
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.5418635725247666e-05,
-      "loss": 1.2663,
+      "epoch": 0.1075876504447933,
+      "grad_norm": 2.6139103762274876,
+      "learning_rate": 1.9686202418769823e-05,
+      "loss": 1.0012,
       "step": 1028
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.540966997365007e-05,
-      "loss": 1.2458,
+      "epoch": 0.1076923076923077,
+      "grad_norm": 2.7091370591165576,
+      "learning_rate": 1.9685359362872124e-05,
+      "loss": 1.2318,
       "step": 1029
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.540069807021306e-05,
-      "loss": 1.3168,
+      "epoch": 0.10779696493982208,
+      "grad_norm": 2.414700209712605,
+      "learning_rate": 1.9684515194107004e-05,
+      "loss": 1.0775,
       "step": 1030
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.539172002513942e-05,
-      "loss": 1.286,
+      "epoch": 0.10790162218733647,
+      "grad_norm": 2.6192205052365236,
+      "learning_rate": 1.9683669912571452e-05,
+      "loss": 1.0712,
       "step": 1031
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.538273584863893e-05,
-      "loss": 1.233,
+      "epoch": 0.10800627943485086,
+      "grad_norm": 2.737341412891626,
+      "learning_rate": 1.96828235183626e-05,
+      "loss": 1.0652,
       "step": 1032
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.537374555092832e-05,
-      "loss": 1.3463,
+      "epoch": 0.10811093668236525,
+      "grad_norm": 2.465865978038424,
+      "learning_rate": 1.968197601157769e-05,
+      "loss": 1.124,
       "step": 1033
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.5364749142231302e-05,
-      "loss": 1.3144,
+      "epoch": 0.10821559392987964,
+      "grad_norm": 2.4388907774406574,
+      "learning_rate": 1.968112739231411e-05,
+      "loss": 1.0664,
       "step": 1034
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.5355746632778533e-05,
-      "loss": 1.2864,
+      "epoch": 0.10832025117739404,
+      "grad_norm": 2.412351278224597,
+      "learning_rate": 1.968027766066937e-05,
+      "loss": 1.0165,
       "step": 1035
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.5346738032807603e-05,
-      "loss": 1.352,
+      "epoch": 0.10842490842490843,
+      "grad_norm": 3.061388366092205,
+      "learning_rate": 1.9679426816741102e-05,
+      "loss": 1.2182,
       "step": 1036
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.5337723352563028e-05,
-      "loss": 1.2151,
+      "epoch": 0.10852956567242282,
+      "grad_norm": 2.674591671542277,
+      "learning_rate": 1.967857486062707e-05,
+      "loss": 0.8857,
       "step": 1037
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.5328702602296238e-05,
-      "loss": 1.2613,
+      "epoch": 0.10863422291993721,
+      "grad_norm": 2.6778191321571,
+      "learning_rate": 1.9677721792425167e-05,
+      "loss": 1.1491,
       "step": 1038
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.531967579226557e-05,
-      "loss": 1.1542,
+      "epoch": 0.10873888016745159,
+      "grad_norm": 2.591446579098128,
+      "learning_rate": 1.967686761223341e-05,
+      "loss": 0.9744,
       "step": 1039
     },
     {
-      "epoch": 0.68,
-      "learning_rate": 1.5310642932736253e-05,
-      "loss": 1.2407,
+      "epoch": 0.10884353741496598,
+      "grad_norm": 2.7917814478780314,
+      "learning_rate": 1.967601232014995e-05,
+      "loss": 0.9032,
       "step": 1040
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.5301604033980387e-05,
-      "loss": 1.2442,
+      "epoch": 0.10894819466248037,
+      "grad_norm": 2.2711122801371517,
+      "learning_rate": 1.9675155916273057e-05,
+      "loss": 1.0988,
       "step": 1041
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.529255910627695e-05,
-      "loss": 1.203,
+      "epoch": 0.10905285190999477,
+      "grad_norm": 2.6881776652627645,
+      "learning_rate": 1.9674298400701137e-05,
+      "loss": 0.997,
       "step": 1042
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.5283508159911776e-05,
-      "loss": 1.3566,
+      "epoch": 0.10915750915750916,
+      "grad_norm": 2.4975909079275502,
+      "learning_rate": 1.9673439773532714e-05,
+      "loss": 0.9299,
       "step": 1043
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.5274451205177526e-05,
-      "loss": 1.2837,
+      "epoch": 0.10926216640502355,
+      "grad_norm": 2.710402657049504,
+      "learning_rate": 1.9672580034866455e-05,
+      "loss": 1.0081,
       "step": 1044
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.5265388252373712e-05,
-      "loss": 1.3364,
+      "epoch": 0.10936682365253794,
+      "grad_norm": 2.643549698589489,
+      "learning_rate": 1.9671719184801144e-05,
+      "loss": 1.0695,
       "step": 1045
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.525631931180667e-05,
-      "loss": 1.2052,
+      "epoch": 0.10947148090005233,
+      "grad_norm": 2.7508161610052246,
+      "learning_rate": 1.9670857223435688e-05,
+      "loss": 1.112,
       "step": 1046
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.524724439378953e-05,
-      "loss": 1.3054,
+      "epoch": 0.10957613814756671,
+      "grad_norm": 2.5165912885846113,
+      "learning_rate": 1.9669994150869134e-05,
+      "loss": 0.9325,
       "step": 1047
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.5238163508642219e-05,
-      "loss": 1.224,
+      "epoch": 0.1096807953950811,
+      "grad_norm": 2.4710904066047616,
+      "learning_rate": 1.9669129967200648e-05,
+      "loss": 1.1537,
       "step": 1048
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.5229076666691471e-05,
-      "loss": 1.4566,
+      "epoch": 0.1097854526425955,
+      "grad_norm": 3.0975055499064514,
+      "learning_rate": 1.9668264672529528e-05,
+      "loss": 1.0,
       "step": 1049
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.5219983878270767e-05,
-      "loss": 1.3532,
+      "epoch": 0.10989010989010989,
+      "grad_norm": 2.253295288672016,
+      "learning_rate": 1.9667398266955195e-05,
+      "loss": 1.0463,
       "step": 1050
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.5210885153720374e-05,
-      "loss": 1.26,
+      "epoch": 0.10999476713762428,
+      "grad_norm": 2.6293412785828014,
+      "learning_rate": 1.9666530750577207e-05,
+      "loss": 0.9453,
       "step": 1051
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.520178050338729e-05,
-      "loss": 1.1311,
+      "epoch": 0.11009942438513867,
+      "grad_norm": 2.1033760444602727,
+      "learning_rate": 1.966566212349524e-05,
+      "loss": 0.9526,
       "step": 1052
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.519266993762527e-05,
-      "loss": 1.3461,
+      "epoch": 0.11020408163265306,
+      "grad_norm": 2.5610804233496127,
+      "learning_rate": 1.9664792385809094e-05,
+      "loss": 0.9974,
       "step": 1053
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.5183553466794778e-05,
-      "loss": 1.2327,
+      "epoch": 0.11030873888016746,
+      "grad_norm": 2.5769705235076934,
+      "learning_rate": 1.9663921537618717e-05,
+      "loss": 1.0322,
       "step": 1054
     },
     {
-      "epoch": 0.69,
-      "learning_rate": 1.5174431101263004e-05,
-      "loss": 1.4256,
+      "epoch": 0.11041339612768185,
+      "grad_norm": 2.4860849473694935,
+      "learning_rate": 1.9663049579024162e-05,
+      "loss": 1.0821,
       "step": 1055
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5165302851403843e-05,
-      "loss": 1.3932,
+      "epoch": 0.11051805337519623,
+      "grad_norm": 2.556645999664551,
+      "learning_rate": 1.966217651012562e-05,
+      "loss": 0.9631,
       "step": 1056
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5156168727597873e-05,
-      "loss": 1.288,
+      "epoch": 0.11062271062271062,
+      "grad_norm": 2.69622491307049,
+      "learning_rate": 1.9661302331023413e-05,
+      "loss": 0.963,
       "step": 1057
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5147028740232361e-05,
-      "loss": 1.3192,
+      "epoch": 0.11072736787022501,
+      "grad_norm": 2.322114587301298,
+      "learning_rate": 1.9660427041817982e-05,
+      "loss": 0.77,
       "step": 1058
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5137882899701234e-05,
-      "loss": 1.27,
+      "epoch": 0.1108320251177394,
+      "grad_norm": 2.086861967831109,
+      "learning_rate": 1.9659550642609898e-05,
+      "loss": 0.8174,
       "step": 1059
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5128731216405084e-05,
-      "loss": 1.2837,
+      "epoch": 0.1109366823652538,
+      "grad_norm": 2.5465767908188,
+      "learning_rate": 1.9658673133499868e-05,
+      "loss": 1.1169,
       "step": 1060
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5119573700751133e-05,
-      "loss": 1.4091,
+      "epoch": 0.11104133961276819,
+      "grad_norm": 2.3306210424010865,
+      "learning_rate": 1.9657794514588708e-05,
+      "loss": 1.0342,
       "step": 1061
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5110410363153256e-05,
-      "loss": 1.2189,
+      "epoch": 0.11114599686028258,
+      "grad_norm": 4.169058819176632,
+      "learning_rate": 1.9656914785977386e-05,
+      "loss": 1.0828,
       "step": 1062
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5101241214031928e-05,
-      "loss": 1.3311,
+      "epoch": 0.11125065410779697,
+      "grad_norm": 2.974406996118035,
+      "learning_rate": 1.9656033947766974e-05,
+      "loss": 1.1862,
       "step": 1063
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5092066263814245e-05,
-      "loss": 1.2795,
+      "epoch": 0.11135531135531136,
+      "grad_norm": 2.263717528826466,
+      "learning_rate": 1.965515200005869e-05,
+      "loss": 0.8976,
       "step": 1064
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5082885522933901e-05,
-      "loss": 1.4888,
+      "epoch": 0.11145996860282574,
+      "grad_norm": 2.1583188268439297,
+      "learning_rate": 1.9654268942953867e-05,
+      "loss": 0.9869,
       "step": 1065
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5073699001831168e-05,
-      "loss": 1.1529,
+      "epoch": 0.11156462585034013,
+      "grad_norm": 2.6538687038608715,
+      "learning_rate": 1.965338477655397e-05,
+      "loss": 1.015,
       "step": 1066
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.506450671095289e-05,
-      "loss": 1.2273,
+      "epoch": 0.11166928309785452,
+      "grad_norm": 3.088871449306695,
+      "learning_rate": 1.9652499500960594e-05,
+      "loss": 1.0569,
       "step": 1067
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5055308660752486e-05,
-      "loss": 1.4096,
+      "epoch": 0.11177394034536892,
+      "grad_norm": 2.576832115600818,
+      "learning_rate": 1.9651613116275457e-05,
+      "loss": 0.8432,
       "step": 1068
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5046104861689911e-05,
-      "loss": 1.2175,
+      "epoch": 0.11187859759288331,
+      "grad_norm": 2.110328498949912,
+      "learning_rate": 1.9650725622600407e-05,
+      "loss": 0.9884,
       "step": 1069
     },
     {
-      "epoch": 0.7,
-      "learning_rate": 1.5036895324231662e-05,
-      "loss": 1.2721,
+      "epoch": 0.1119832548403977,
+      "grad_norm": 2.790857879613387,
+      "learning_rate": 1.964983702003742e-05,
+      "loss": 1.074,
       "step": 1070
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.5027680058850757e-05,
-      "loss": 1.2757,
+      "epoch": 0.11208791208791209,
+      "grad_norm": 2.9160190675466158,
+      "learning_rate": 1.9648947308688594e-05,
+      "loss": 1.1648,
       "step": 1071
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.501845907602674e-05,
-      "loss": 1.306,
+      "epoch": 0.11219256933542648,
+      "grad_norm": 2.122470818651794,
+      "learning_rate": 1.9648056488656166e-05,
+      "loss": 0.8107,
       "step": 1072
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.5009232386245644e-05,
-      "loss": 1.3438,
+      "epoch": 0.11229722658294086,
+      "grad_norm": 2.579496972713126,
+      "learning_rate": 1.9647164560042486e-05,
+      "loss": 1.1268,
       "step": 1073
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.5000000000000002e-05,
-      "loss": 1.2436,
+      "epoch": 0.11240188383045525,
+      "grad_norm": 3.122189191320992,
+      "learning_rate": 1.9646271522950043e-05,
+      "loss": 1.0926,
       "step": 1074
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4990761927788815e-05,
-      "loss": 1.333,
+      "epoch": 0.11250654107796965,
+      "grad_norm": 2.596905120190331,
+      "learning_rate": 1.9645377377481446e-05,
+      "loss": 1.0616,
       "step": 1075
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4981518180117558e-05,
-      "loss": 1.402,
+      "epoch": 0.11261119832548404,
+      "grad_norm": 2.851456857123207,
+      "learning_rate": 1.9644482123739434e-05,
+      "loss": 1.1331,
       "step": 1076
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.497226876749815e-05,
-      "loss": 1.3782,
+      "epoch": 0.11271585557299843,
+      "grad_norm": 2.4375447703917223,
+      "learning_rate": 1.964358576182688e-05,
+      "loss": 0.9973,
       "step": 1077
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4963013700448969e-05,
-      "loss": 1.3064,
+      "epoch": 0.11282051282051282,
+      "grad_norm": 2.388922166429233,
+      "learning_rate": 1.9642688291846762e-05,
+      "loss": 0.911,
       "step": 1078
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4953752989494815e-05,
-      "loss": 1.2067,
+      "epoch": 0.11292517006802721,
+      "grad_norm": 2.6304965592524328,
+      "learning_rate": 1.9641789713902223e-05,
+      "loss": 1.0654,
       "step": 1079
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4944486645166892e-05,
-      "loss": 1.3255,
+      "epoch": 0.1130298273155416,
+      "grad_norm": 2.4302290806051277,
+      "learning_rate": 1.9640890028096492e-05,
+      "loss": 1.0569,
       "step": 1080
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4935214678002834e-05,
-      "loss": 1.3169,
+      "epoch": 0.113134484563056,
+      "grad_norm": 2.2285317412672745,
+      "learning_rate": 1.9639989234532955e-05,
+      "loss": 1.1056,
       "step": 1081
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4925937098546652e-05,
-      "loss": 1.2642,
+      "epoch": 0.11323914181057038,
+      "grad_norm": 2.5935496575040693,
+      "learning_rate": 1.9639087333315113e-05,
+      "loss": 1.0176,
       "step": 1082
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4916653917348751e-05,
-      "loss": 1.1457,
+      "epoch": 0.11334379905808477,
+      "grad_norm": 2.699922891159306,
+      "learning_rate": 1.9638184324546597e-05,
+      "loss": 1.1053,
       "step": 1083
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4907365144965896e-05,
-      "loss": 1.297,
+      "epoch": 0.11344845630559916,
+      "grad_norm": 3.2198916620549354,
+      "learning_rate": 1.963728020833116e-05,
+      "loss": 1.1036,
       "step": 1084
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4898070791961215e-05,
-      "loss": 1.2597,
+      "epoch": 0.11355311355311355,
+      "grad_norm": 2.6375927221507145,
+      "learning_rate": 1.963637498477269e-05,
+      "loss": 0.9988,
       "step": 1085
     },
     {
-      "epoch": 0.71,
-      "learning_rate": 1.4888770868904185e-05,
-      "loss": 1.3395,
+      "epoch": 0.11365777080062794,
+      "grad_norm": 2.938305615487851,
+      "learning_rate": 1.9635468653975203e-05,
+      "loss": 1.166,
       "step": 1086
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.4879465386370613e-05,
-      "loss": 1.2254,
+      "epoch": 0.11376242804814234,
+      "grad_norm": 2.2156017547233295,
+      "learning_rate": 1.9634561216042834e-05,
+      "loss": 1.0057,
       "step": 1087
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.487015435494263e-05,
-      "loss": 1.1784,
+      "epoch": 0.11386708529565673,
+      "grad_norm": 2.4105372096002755,
+      "learning_rate": 1.9633652671079854e-05,
+      "loss": 0.9225,
       "step": 1088
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.4860837785208678e-05,
-      "loss": 1.2387,
+      "epoch": 0.11397174254317112,
+      "grad_norm": 2.412529782609449,
+      "learning_rate": 1.9632743019190646e-05,
+      "loss": 0.9356,
       "step": 1089
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.4851515687763492e-05,
-      "loss": 1.2022,
+      "epoch": 0.1140763997906855,
+      "grad_norm": 2.738234258928923,
+      "learning_rate": 1.9631832260479736e-05,
+      "loss": 1.1259,
       "step": 1090
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.48421880732081e-05,
-      "loss": 1.3362,
+      "epoch": 0.11418105703819989,
+      "grad_norm": 2.295890829748179,
+      "learning_rate": 1.963092039505178e-05,
+      "loss": 1.0476,
       "step": 1091
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.48328549521498e-05,
-      "loss": 1.2714,
+      "epoch": 0.11428571428571428,
+      "grad_norm": 2.4190921434998574,
+      "learning_rate": 1.9630007423011543e-05,
+      "loss": 1.0351,
       "step": 1092
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.4823516335202152e-05,
-      "loss": 1.2764,
+      "epoch": 0.11439037153322867,
+      "grad_norm": 2.5227441453952033,
+      "learning_rate": 1.962909334446393e-05,
+      "loss": 1.0712,
       "step": 1093
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.4814172232984968e-05,
-      "loss": 1.1313,
+      "epoch": 0.11449502878074307,
+      "grad_norm": 2.5782201716213775,
+      "learning_rate": 1.9628178159513976e-05,
+      "loss": 0.8372,
       "step": 1094
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.4804822656124302e-05,
-      "loss": 1.3175,
+      "epoch": 0.11459968602825746,
+      "grad_norm": 2.4279239527241367,
+      "learning_rate": 1.962726186826683e-05,
+      "loss": 1.0871,
       "step": 1095
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.479546761525242e-05,
-      "loss": 1.2485,
+      "epoch": 0.11470434327577185,
+      "grad_norm": 2.2289981767762734,
+      "learning_rate": 1.962634447082778e-05,
+      "loss": 1.0551,
       "step": 1096
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.4786107121007813e-05,
-      "loss": 1.3056,
+      "epoch": 0.11480900052328624,
+      "grad_norm": 2.7919927641730387,
+      "learning_rate": 1.9625425967302232e-05,
+      "loss": 1.0695,
       "step": 1097
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.4776741184035171e-05,
-      "loss": 1.1867,
+      "epoch": 0.11491365777080063,
+      "grad_norm": 2.567716154128025,
+      "learning_rate": 1.962450635779573e-05,
+      "loss": 1.1617,
       "step": 1098
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.476736981498537e-05,
-      "loss": 1.266,
+      "epoch": 0.11501831501831501,
+      "grad_norm": 2.449051302816616,
+      "learning_rate": 1.962358564241394e-05,
+      "loss": 1.0829,
       "step": 1099
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.475799302451547e-05,
-      "loss": 1.227,
+      "epoch": 0.1151229722658294,
+      "grad_norm": 2.4898532560578728,
+      "learning_rate": 1.9622663821262647e-05,
+      "loss": 1.1857,
       "step": 1100
     },
     {
-      "epoch": 0.72,
-      "learning_rate": 1.4748610823288693e-05,
-      "loss": 1.3661,
+      "epoch": 0.1152276295133438,
+      "grad_norm": 2.508714185849526,
+      "learning_rate": 1.9621740894447776e-05,
+      "loss": 0.9801,
       "step": 1101
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4739223221974415e-05,
-      "loss": 1.3033,
+      "epoch": 0.11533228676085819,
+      "grad_norm": 2.864088004419296,
+      "learning_rate": 1.9620816862075365e-05,
+      "loss": 0.9741,
       "step": 1102
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4729830231248143e-05,
-      "loss": 1.2722,
+      "epoch": 0.11543694400837258,
+      "grad_norm": 2.2451938017369457,
+      "learning_rate": 1.9619891724251602e-05,
+      "loss": 0.9874,
       "step": 1103
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4720431861791531e-05,
-      "loss": 1.2511,
+      "epoch": 0.11554160125588697,
+      "grad_norm": 2.43522131047157,
+      "learning_rate": 1.961896548108277e-05,
+      "loss": 1.0198,
       "step": 1104
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4711028124292332e-05,
-      "loss": 1.2971,
+      "epoch": 0.11564625850340136,
+      "grad_norm": 2.6482111396981196,
+      "learning_rate": 1.961803813267531e-05,
+      "loss": 1.0065,
       "step": 1105
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4701619029444419e-05,
-      "loss": 1.348,
+      "epoch": 0.11575091575091576,
+      "grad_norm": 2.663006273262908,
+      "learning_rate": 1.961710967913577e-05,
+      "loss": 1.0731,
       "step": 1106
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4692204587947745e-05,
-      "loss": 1.3339,
+      "epoch": 0.11585557299843015,
+      "grad_norm": 2.2592261853498745,
+      "learning_rate": 1.9616180120570833e-05,
+      "loss": 1.0667,
       "step": 1107
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.468278481050835e-05,
-      "loss": 1.1522,
+      "epoch": 0.11596023024594453,
+      "grad_norm": 2.5579103444641587,
+      "learning_rate": 1.9615249457087302e-05,
+      "loss": 1.0649,
       "step": 1108
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4673359707838338e-05,
-      "loss": 1.1803,
+      "epoch": 0.11606488749345892,
+      "grad_norm": 2.5681709092759846,
+      "learning_rate": 1.9614317688792118e-05,
+      "loss": 1.0629,
       "step": 1109
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4663929290655872e-05,
-      "loss": 1.1292,
+      "epoch": 0.11616954474097331,
+      "grad_norm": 2.7062909391638468,
+      "learning_rate": 1.9613384815792343e-05,
+      "loss": 1.0538,
       "step": 1110
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4654493569685157e-05,
-      "loss": 1.3618,
+      "epoch": 0.1162742019884877,
+      "grad_norm": 3.1506469607183423,
+      "learning_rate": 1.9612450838195164e-05,
+      "loss": 0.9414,
       "step": 1111
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.464505255565643e-05,
-      "loss": 1.369,
+      "epoch": 0.1163788592360021,
+      "grad_norm": 2.771093053485509,
+      "learning_rate": 1.9611515756107895e-05,
+      "loss": 1.0721,
       "step": 1112
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4635606259305948e-05,
-      "loss": 1.2768,
+      "epoch": 0.11648351648351649,
+      "grad_norm": 2.6299240759159033,
+      "learning_rate": 1.9610579569637982e-05,
+      "loss": 1.0506,
       "step": 1113
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4626154691375974e-05,
-      "loss": 1.2776,
+      "epoch": 0.11658817373103088,
+      "grad_norm": 2.718685548324005,
+      "learning_rate": 1.9609642278892992e-05,
+      "loss": 1.0423,
       "step": 1114
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4616697862614766e-05,
-      "loss": 1.2922,
+      "epoch": 0.11669283097854527,
+      "grad_norm": 2.617103354933396,
+      "learning_rate": 1.9608703883980625e-05,
+      "loss": 1.135,
       "step": 1115
     },
     {
-      "epoch": 0.73,
-      "learning_rate": 1.4607235783776563e-05,
-      "loss": 1.3472,
+      "epoch": 0.11679748822605965,
+      "grad_norm": 2.8284262061059775,
+      "learning_rate": 1.9607764385008706e-05,
+      "loss": 0.9975,
       "step": 1116
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4597768465621575e-05,
-      "loss": 1.2253,
+      "epoch": 0.11690214547357404,
+      "grad_norm": 2.422530563865842,
+      "learning_rate": 1.9606823782085178e-05,
+      "loss": 0.9965,
       "step": 1117
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4588295918915978e-05,
-      "loss": 1.303,
+      "epoch": 0.11700680272108843,
+      "grad_norm": 2.7309355633540022,
+      "learning_rate": 1.9605882075318124e-05,
+      "loss": 1.0416,
       "step": 1118
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4578818154431878e-05,
-      "loss": 1.2062,
+      "epoch": 0.11711145996860282,
+      "grad_norm": 2.41740256806887,
+      "learning_rate": 1.9604939264815746e-05,
+      "loss": 1.091,
       "step": 1119
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4569335182947325e-05,
-      "loss": 1.3157,
+      "epoch": 0.11721611721611722,
+      "grad_norm": 2.6049047174023565,
+      "learning_rate": 1.960399535068638e-05,
+      "loss": 0.9203,
       "step": 1120
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4559847015246294e-05,
-      "loss": 1.3139,
+      "epoch": 0.11732077446363161,
+      "grad_norm": 2.352711031496301,
+      "learning_rate": 1.9603050333038473e-05,
+      "loss": 1.0542,
       "step": 1121
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4550353662118659e-05,
-      "loss": 1.423,
+      "epoch": 0.117425431711146,
+      "grad_norm": 3.3295305172916727,
+      "learning_rate": 1.9602104211980614e-05,
+      "loss": 0.9534,
       "step": 1122
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4540855134360198e-05,
-      "loss": 1.2684,
+      "epoch": 0.11753008895866039,
+      "grad_norm": 3.1546428345111197,
+      "learning_rate": 1.960115698762152e-05,
+      "loss": 1.0338,
       "step": 1123
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.453135144277257e-05,
-      "loss": 1.4146,
+      "epoch": 0.11763474620617478,
+      "grad_norm": 2.728833024559098,
+      "learning_rate": 1.9600208660070022e-05,
+      "loss": 1.131,
       "step": 1124
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4521842598163309e-05,
-      "loss": 1.3044,
+      "epoch": 0.11773940345368916,
+      "grad_norm": 2.7185354106465236,
+      "learning_rate": 1.9599259229435088e-05,
+      "loss": 0.9256,
       "step": 1125
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4512328611345808e-05,
-      "loss": 1.2642,
+      "epoch": 0.11784406070120355,
+      "grad_norm": 2.498618364520084,
+      "learning_rate": 1.959830869582581e-05,
+      "loss": 1.0662,
       "step": 1126
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4502809493139305e-05,
-      "loss": 1.3398,
+      "epoch": 0.11794871794871795,
+      "grad_norm": 2.495040333087842,
+      "learning_rate": 1.9597357059351404e-05,
+      "loss": 1.0717,
       "step": 1127
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4493285254368878e-05,
-      "loss": 1.1626,
+      "epoch": 0.11805337519623234,
+      "grad_norm": 2.623375696161012,
+      "learning_rate": 1.9596404320121217e-05,
+      "loss": 1.0103,
       "step": 1128
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4483755905865423e-05,
-      "loss": 1.2641,
+      "epoch": 0.11815803244374673,
+      "grad_norm": 2.4391065346519496,
+      "learning_rate": 1.9595450478244718e-05,
+      "loss": 0.9587,
       "step": 1129
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4474221458465651e-05,
-      "loss": 1.3286,
+      "epoch": 0.11826268969126112,
+      "grad_norm": 2.663286662491619,
+      "learning_rate": 1.9594495533831507e-05,
+      "loss": 1.1976,
       "step": 1130
     },
     {
-      "epoch": 0.74,
-      "learning_rate": 1.4464681923012072e-05,
-      "loss": 1.4002,
+      "epoch": 0.11836734693877551,
+      "grad_norm": 2.5842407873088766,
+      "learning_rate": 1.959353948699131e-05,
+      "loss": 1.1649,
       "step": 1131
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4455137310352978e-05,
-      "loss": 1.1133,
+      "epoch": 0.1184720041862899,
+      "grad_norm": 2.5723829154298534,
+      "learning_rate": 1.959258233783398e-05,
+      "loss": 1.1408,
       "step": 1132
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4445587631342433e-05,
-      "loss": 1.3146,
+      "epoch": 0.11857666143380428,
+      "grad_norm": 2.378313766984001,
+      "learning_rate": 1.959162408646949e-05,
+      "loss": 1.0323,
       "step": 1133
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4436032896840271e-05,
-      "loss": 1.2476,
+      "epoch": 0.11868131868131868,
+      "grad_norm": 2.561592915095796,
+      "learning_rate": 1.9590664733007947e-05,
+      "loss": 1.0473,
       "step": 1134
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4426473117712071e-05,
-      "loss": 1.3524,
+      "epoch": 0.11878597592883307,
+      "grad_norm": 2.6784137838787765,
+      "learning_rate": 1.9589704277559586e-05,
+      "loss": 0.9144,
       "step": 1135
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4416908304829142e-05,
-      "loss": 1.2622,
+      "epoch": 0.11889063317634746,
+      "grad_norm": 2.4050436698772057,
+      "learning_rate": 1.958874272023476e-05,
+      "loss": 0.9927,
       "step": 1136
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4407338469068532e-05,
-      "loss": 1.2244,
+      "epoch": 0.11899529042386185,
+      "grad_norm": 2.787218387154808,
+      "learning_rate": 1.9587780061143965e-05,
+      "loss": 1.112,
       "step": 1137
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4397763621312985e-05,
-      "loss": 1.3472,
+      "epoch": 0.11909994767137624,
+      "grad_norm": 2.467244551948357,
+      "learning_rate": 1.9586816300397796e-05,
+      "loss": 1.0967,
       "step": 1138
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4388183772450957e-05,
-      "loss": 1.1446,
+      "epoch": 0.11920460491889064,
+      "grad_norm": 2.974817620992224,
+      "learning_rate": 1.9585851438107007e-05,
+      "loss": 1.0886,
       "step": 1139
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.437859893337658e-05,
-      "loss": 1.2744,
+      "epoch": 0.11930926216640503,
+      "grad_norm": 2.7737239161164906,
+      "learning_rate": 1.958488547438245e-05,
+      "loss": 0.9261,
       "step": 1140
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4369009114989671e-05,
-      "loss": 1.2349,
+      "epoch": 0.11941391941391942,
+      "grad_norm": 3.0863679881725132,
+      "learning_rate": 1.958391840933512e-05,
+      "loss": 0.8894,
       "step": 1141
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4359414328195704e-05,
-      "loss": 1.2209,
+      "epoch": 0.1195185766614338,
+      "grad_norm": 2.2653693939706403,
+      "learning_rate": 1.958295024307614e-05,
+      "loss": 1.0348,
       "step": 1142
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4349814583905807e-05,
-      "loss": 1.3382,
+      "epoch": 0.11962323390894819,
+      "grad_norm": 2.754842275125979,
+      "learning_rate": 1.958198097571675e-05,
+      "loss": 1.075,
       "step": 1143
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.434020989303674e-05,
-      "loss": 1.2584,
+      "epoch": 0.11972789115646258,
+      "grad_norm": 2.832206040807847,
+      "learning_rate": 1.9581010607368324e-05,
+      "loss": 1.104,
       "step": 1144
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4330600266510895e-05,
-      "loss": 1.3664,
+      "epoch": 0.11983254840397697,
+      "grad_norm": 2.6220443857258044,
+      "learning_rate": 1.958003913814235e-05,
+      "loss": 1.0461,
       "step": 1145
     },
     {
-      "epoch": 0.75,
-      "learning_rate": 1.4320985715256269e-05,
-      "loss": 1.162,
+      "epoch": 0.11993720565149137,
+      "grad_norm": 3.167797177471765,
+      "learning_rate": 1.957906656815046e-05,
+      "loss": 1.2079,
       "step": 1146
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4311366250206464e-05,
-      "loss": 1.3295,
+      "epoch": 0.12004186289900576,
+      "grad_norm": 2.5145650658616647,
+      "learning_rate": 1.9578092897504404e-05,
+      "loss": 1.1371,
       "step": 1147
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4301741882300673e-05,
-      "loss": 1.3086,
+      "epoch": 0.12014652014652015,
+      "grad_norm": 2.532990354557582,
+      "learning_rate": 1.957711812631606e-05,
+      "loss": 0.847,
       "step": 1148
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4292112622483653e-05,
-      "loss": 1.278,
+      "epoch": 0.12025117739403454,
+      "grad_norm": 2.498723144014304,
+      "learning_rate": 1.9576142254697422e-05,
+      "loss": 1.0768,
       "step": 1149
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4282478481705742e-05,
-      "loss": 1.2891,
+      "epoch": 0.12035583464154893,
+      "grad_norm": 2.8848332638507608,
+      "learning_rate": 1.957516528276063e-05,
+      "loss": 1.0935,
       "step": 1150
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4272839470922812e-05,
-      "loss": 1.2409,
+      "epoch": 0.12046049188906331,
+      "grad_norm": 2.407079189345069,
+      "learning_rate": 1.9574187210617935e-05,
+      "loss": 0.9884,
       "step": 1151
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.426319560109628e-05,
-      "loss": 1.2973,
+      "epoch": 0.1205651491365777,
+      "grad_norm": 2.4789862581848325,
+      "learning_rate": 1.9573208038381722e-05,
+      "loss": 1.0612,
       "step": 1152
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4253546883193093e-05,
-      "loss": 1.2486,
+      "epoch": 0.1206698063840921,
+      "grad_norm": 2.634050308370591,
+      "learning_rate": 1.95722277661645e-05,
+      "loss": 1.055,
       "step": 1153
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.42438933281857e-05,
-      "loss": 1.2693,
+      "epoch": 0.12077446363160649,
+      "grad_norm": 2.646401341815591,
+      "learning_rate": 1.95712463940789e-05,
+      "loss": 1.1474,
       "step": 1154
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4234234947052063e-05,
-      "loss": 1.2249,
+      "epoch": 0.12087912087912088,
+      "grad_norm": 2.8047167361076957,
+      "learning_rate": 1.9570263922237686e-05,
+      "loss": 1.0503,
       "step": 1155
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4224571750775625e-05,
-      "loss": 1.3395,
+      "epoch": 0.12098377812663527,
+      "grad_norm": 2.5741451106686317,
+      "learning_rate": 1.956928035075375e-05,
+      "loss": 0.9393,
       "step": 1156
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4214903750345298e-05,
-      "loss": 1.449,
+      "epoch": 0.12108843537414966,
+      "grad_norm": 2.5973811680350045,
+      "learning_rate": 1.9568295679740098e-05,
+      "loss": 1.1236,
       "step": 1157
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4205230956755479e-05,
-      "loss": 1.1161,
+      "epoch": 0.12119309262166406,
+      "grad_norm": 2.6113678867025616,
+      "learning_rate": 1.956730990930988e-05,
+      "loss": 1.0036,
       "step": 1158
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.419555338100599e-05,
-      "loss": 1.3085,
+      "epoch": 0.12129774986917843,
+      "grad_norm": 2.763270013817754,
+      "learning_rate": 1.9566323039576353e-05,
+      "loss": 1.1911,
       "step": 1159
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4185871034102117e-05,
-      "loss": 1.351,
+      "epoch": 0.12140240711669283,
+      "grad_norm": 2.705626562419259,
+      "learning_rate": 1.956533507065292e-05,
+      "loss": 1.2495,
       "step": 1160
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4176183927054545e-05,
-      "loss": 1.3234,
+      "epoch": 0.12150706436420722,
+      "grad_norm": 2.5695027901999175,
+      "learning_rate": 1.9564346002653094e-05,
+      "loss": 1.0895,
       "step": 1161
     },
     {
-      "epoch": 0.76,
-      "learning_rate": 1.4166492070879389e-05,
-      "loss": 1.2037,
+      "epoch": 0.12161172161172161,
+      "grad_norm": 2.688239436865601,
+      "learning_rate": 1.9563355835690526e-05,
+      "loss": 1.066,
       "step": 1162
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4156795476598166e-05,
-      "loss": 1.3678,
+      "epoch": 0.121716378859236,
+      "grad_norm": 2.4721295753539847,
+      "learning_rate": 1.9562364569878985e-05,
+      "loss": 0.9474,
       "step": 1163
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4147094155237772e-05,
-      "loss": 1.3293,
+      "epoch": 0.1218210361067504,
+      "grad_norm": 2.9695025162758064,
+      "learning_rate": 1.9561372205332366e-05,
+      "loss": 1.1938,
       "step": 1164
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4137388117830486e-05,
-      "loss": 1.2505,
+      "epoch": 0.12192569335426479,
+      "grad_norm": 2.3922931991769927,
+      "learning_rate": 1.95603787421647e-05,
+      "loss": 0.931,
       "step": 1165
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4127677375413943e-05,
-      "loss": 1.1536,
+      "epoch": 0.12203035060177918,
+      "grad_norm": 2.533155414113076,
+      "learning_rate": 1.9559384180490138e-05,
+      "loss": 1.0848,
       "step": 1166
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4117961939031135e-05,
-      "loss": 1.2298,
+      "epoch": 0.12213500784929357,
+      "grad_norm": 2.7804617711740462,
+      "learning_rate": 1.9558388520422955e-05,
+      "loss": 1.1158,
       "step": 1167
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4108241819730384e-05,
-      "loss": 1.2324,
+      "epoch": 0.12223966509680795,
+      "grad_norm": 3.07185941955718,
+      "learning_rate": 1.955739176207755e-05,
+      "loss": 1.1685,
       "step": 1168
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.409851702856535e-05,
-      "loss": 1.3091,
+      "epoch": 0.12234432234432234,
+      "grad_norm": 2.632035250919555,
+      "learning_rate": 1.955639390556846e-05,
+      "loss": 1.0409,
       "step": 1169
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4088787576594996e-05,
-      "loss": 1.2983,
+      "epoch": 0.12244897959183673,
+      "grad_norm": 2.573941413049372,
+      "learning_rate": 1.9555394951010337e-05,
+      "loss": 1.2533,
       "step": 1170
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4079053474883584e-05,
-      "loss": 1.1055,
+      "epoch": 0.12255363683935112,
+      "grad_norm": 2.5319580043800687,
+      "learning_rate": 1.9554394898517962e-05,
+      "loss": 1.1681,
       "step": 1171
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4069314734500674e-05,
-      "loss": 1.2983,
+      "epoch": 0.12265829408686552,
+      "grad_norm": 2.983393734826706,
+      "learning_rate": 1.9553393748206245e-05,
+      "loss": 1.0353,
       "step": 1172
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.405957136652109e-05,
-      "loss": 1.4468,
+      "epoch": 0.12276295133437991,
+      "grad_norm": 2.5654991130101776,
+      "learning_rate": 1.955239150019022e-05,
+      "loss": 1.1485,
       "step": 1173
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4049823382024922e-05,
-      "loss": 1.3416,
+      "epoch": 0.1228676085818943,
+      "grad_norm": 2.577167856594292,
+      "learning_rate": 1.9551388154585045e-05,
+      "loss": 1.0843,
       "step": 1174
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4040070792097515e-05,
-      "loss": 1.3257,
+      "epoch": 0.12297226582940869,
+      "grad_norm": 2.672800031307482,
+      "learning_rate": 1.955038371150601e-05,
+      "loss": 1.0571,
       "step": 1175
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4030313607829444e-05,
-      "loss": 1.3353,
+      "epoch": 0.12307692307692308,
+      "grad_norm": 2.2255667081208133,
+      "learning_rate": 1.954937817106853e-05,
+      "loss": 0.9892,
       "step": 1176
     },
     {
-      "epoch": 0.77,
-      "learning_rate": 1.4020551840316515e-05,
-      "loss": 1.1875,
+      "epoch": 0.12318158032443746,
+      "grad_norm": 2.5034867735183304,
+      "learning_rate": 1.9548371533388134e-05,
+      "loss": 1.1986,
       "step": 1177
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.4010785500659735e-05,
-      "loss": 1.3478,
+      "epoch": 0.12328623757195185,
+      "grad_norm": 2.8720211636190043,
+      "learning_rate": 1.954736379858049e-05,
+      "loss": 1.2097,
       "step": 1178
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.4001014599965327e-05,
-      "loss": 1.3157,
+      "epoch": 0.12339089481946625,
+      "grad_norm": 2.8330847231206047,
+      "learning_rate": 1.9546354966761397e-05,
+      "loss": 1.115,
       "step": 1179
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3991239149344686e-05,
-      "loss": 1.1173,
+      "epoch": 0.12349555206698064,
+      "grad_norm": 2.4195461792808306,
+      "learning_rate": 1.9545345038046762e-05,
+      "loss": 1.0036,
       "step": 1180
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3981459159914392e-05,
-      "loss": 1.2161,
+      "epoch": 0.12360020931449503,
+      "grad_norm": 2.906126283855808,
+      "learning_rate": 1.9544334012552633e-05,
+      "loss": 1.0183,
       "step": 1181
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3971674642796178e-05,
-      "loss": 1.2952,
+      "epoch": 0.12370486656200942,
+      "grad_norm": 2.5436235281065698,
+      "learning_rate": 1.954332189039518e-05,
+      "loss": 0.9942,
       "step": 1182
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.396188560911693e-05,
-      "loss": 1.2907,
+      "epoch": 0.12380952380952381,
+      "grad_norm": 2.826315997445539,
+      "learning_rate": 1.954230867169069e-05,
+      "loss": 1.2136,
       "step": 1183
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3952092070008669e-05,
-      "loss": 1.4054,
+      "epoch": 0.1239141810570382,
+      "grad_norm": 2.7260954242001514,
+      "learning_rate": 1.9541294356555594e-05,
+      "loss": 0.9941,
       "step": 1184
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3942294036608544e-05,
-      "loss": 1.2853,
+      "epoch": 0.12401883830455258,
+      "grad_norm": 3.1932984486388234,
+      "learning_rate": 1.9540278945106433e-05,
+      "loss": 1.065,
       "step": 1185
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3932491520058807e-05,
-      "loss": 1.2316,
+      "epoch": 0.12412349555206698,
+      "grad_norm": 2.3449880675934884,
+      "learning_rate": 1.953926243745988e-05,
+      "loss": 0.965,
       "step": 1186
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.392268453150681e-05,
-      "loss": 1.3066,
+      "epoch": 0.12422815279958137,
+      "grad_norm": 2.475504471283617,
+      "learning_rate": 1.9538244833732737e-05,
+      "loss": 1.0389,
       "step": 1187
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3912873082105002e-05,
-      "loss": 1.2377,
+      "epoch": 0.12433281004709576,
+      "grad_norm": 2.648680996485242,
+      "learning_rate": 1.9537226134041925e-05,
+      "loss": 1.1165,
       "step": 1188
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3903057183010887e-05,
-      "loss": 1.3614,
+      "epoch": 0.12443746729461015,
+      "grad_norm": 2.658874814055567,
+      "learning_rate": 1.9536206338504494e-05,
+      "loss": 1.0662,
       "step": 1189
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3893236845387041e-05,
-      "loss": 1.3397,
+      "epoch": 0.12454212454212454,
+      "grad_norm": 2.486917672270425,
+      "learning_rate": 1.9535185447237626e-05,
+      "loss": 1.0065,
       "step": 1190
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3883412080401083e-05,
-      "loss": 1.1668,
+      "epoch": 0.12464678178963894,
+      "grad_norm": 2.522352267870716,
+      "learning_rate": 1.953416346035862e-05,
+      "loss": 1.0536,
       "step": 1191
     },
     {
-      "epoch": 0.78,
-      "learning_rate": 1.3873582899225672e-05,
-      "loss": 1.1924,
+      "epoch": 0.12475143903715333,
+      "grad_norm": 2.595719707912744,
+      "learning_rate": 1.9533140377984906e-05,
+      "loss": 1.0279,
       "step": 1192
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3863749313038484e-05,
-      "loss": 1.3731,
+      "epoch": 0.12485609628466772,
+      "grad_norm": 2.7586476277587315,
+      "learning_rate": 1.9532116200234034e-05,
+      "loss": 0.997,
       "step": 1193
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3853911333022199e-05,
-      "loss": 1.1934,
+      "epoch": 0.1249607535321821,
+      "grad_norm": 2.323954362886985,
+      "learning_rate": 1.9531090927223687e-05,
+      "loss": 1.0201,
       "step": 1194
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3844068970364506e-05,
-      "loss": 1.1635,
+      "epoch": 0.1250654107796965,
+      "grad_norm": 2.3865025964537656,
+      "learning_rate": 1.9530064559071672e-05,
+      "loss": 1.0528,
       "step": 1195
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.383422223625807e-05,
-      "loss": 1.3358,
+      "epoch": 0.1251700680272109,
+      "grad_norm": 2.4307960677955838,
+      "learning_rate": 1.952903709589592e-05,
+      "loss": 1.0709,
       "step": 1196
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3824371141900533e-05,
-      "loss": 1.2897,
+      "epoch": 0.12527472527472527,
+      "grad_norm": 2.51314339549336,
+      "learning_rate": 1.9528008537814488e-05,
+      "loss": 1.1309,
       "step": 1197
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3814515698494484e-05,
-      "loss": 1.3208,
+      "epoch": 0.12537938252223965,
+      "grad_norm": 2.4769743325065594,
+      "learning_rate": 1.952697888494556e-05,
+      "loss": 1.144,
       "step": 1198
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3804655917247471e-05,
-      "loss": 1.2866,
+      "epoch": 0.12548403976975406,
+      "grad_norm": 2.358721280245034,
+      "learning_rate": 1.9525948137407443e-05,
+      "loss": 0.9865,
       "step": 1199
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3794791809371971e-05,
-      "loss": 1.0992,
+      "epoch": 0.12558869701726844,
+      "grad_norm": 2.3206179689553723,
+      "learning_rate": 1.9524916295318576e-05,
+      "loss": 1.0968,
       "step": 1200
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3784923386085373e-05,
-      "loss": 1.1884,
+      "epoch": 0.12569335426478284,
+      "grad_norm": 2.316497811222147,
+      "learning_rate": 1.9523883358797517e-05,
+      "loss": 0.9474,
       "step": 1201
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3775050658609988e-05,
-      "loss": 1.2635,
+      "epoch": 0.12579801151229722,
+      "grad_norm": 2.706485454764553,
+      "learning_rate": 1.9522849327962954e-05,
+      "loss": 1.0489,
       "step": 1202
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3765173638173004e-05,
-      "loss": 1.3466,
+      "epoch": 0.12590266875981163,
+      "grad_norm": 2.838226256388987,
+      "learning_rate": 1.9521814202933697e-05,
+      "loss": 1.1056,
       "step": 1203
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3755292336006512e-05,
-      "loss": 1.2894,
+      "epoch": 0.126007326007326,
+      "grad_norm": 2.6389275404903327,
+      "learning_rate": 1.952077798382869e-05,
+      "loss": 1.0294,
       "step": 1204
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3745406763347446e-05,
-      "loss": 1.237,
+      "epoch": 0.1261119832548404,
+      "grad_norm": 2.24218821930294,
+      "learning_rate": 1.9519740670766985e-05,
+      "loss": 1.059,
       "step": 1205
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3735516931437625e-05,
-      "loss": 1.1691,
+      "epoch": 0.1262166405023548,
+      "grad_norm": 2.3594941049558176,
+      "learning_rate": 1.951870226386778e-05,
+      "loss": 1.1141,
       "step": 1206
     },
     {
-      "epoch": 0.79,
-      "learning_rate": 1.3725622851523692e-05,
-      "loss": 1.367,
+      "epoch": 0.12632129774986917,
+      "grad_norm": 2.867410423081424,
+      "learning_rate": 1.951766276325039e-05,
+      "loss": 1.1167,
       "step": 1207
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3715724534857127e-05,
-      "loss": 1.1355,
+      "epoch": 0.12642595499738357,
+      "grad_norm": 2.360366489000132,
+      "learning_rate": 1.9516622169034253e-05,
+      "loss": 1.0121,
       "step": 1208
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3705821992694228e-05,
-      "loss": 1.4048,
+      "epoch": 0.12653061224489795,
+      "grad_norm": 2.8918870260240634,
+      "learning_rate": 1.951558048133894e-05,
+      "loss": 0.9879,
       "step": 1209
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3695915236296095e-05,
-      "loss": 1.225,
+      "epoch": 0.12663526949241236,
+      "grad_norm": 2.470183042311598,
+      "learning_rate": 1.951453770028414e-05,
+      "loss": 1.0716,
       "step": 1210
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3686004276928627e-05,
-      "loss": 1.2157,
+      "epoch": 0.12673992673992673,
+      "grad_norm": 2.5896489389181374,
+      "learning_rate": 1.9513493825989664e-05,
+      "loss": 1.078,
       "step": 1211
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3676089125862495e-05,
-      "loss": 1.2612,
+      "epoch": 0.12684458398744114,
+      "grad_norm": 2.476294549537629,
+      "learning_rate": 1.9512448858575466e-05,
+      "loss": 0.9433,
       "step": 1212
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3666169794373141e-05,
-      "loss": 1.2672,
+      "epoch": 0.12694924123495552,
+      "grad_norm": 2.678105912674675,
+      "learning_rate": 1.951140279816161e-05,
+      "loss": 1.0967,
       "step": 1213
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3656246293740767e-05,
-      "loss": 1.1888,
+      "epoch": 0.12705389848246992,
+      "grad_norm": 2.837977652142853,
+      "learning_rate": 1.951035564486829e-05,
+      "loss": 1.0051,
       "step": 1214
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.36463186352503e-05,
-      "loss": 1.3704,
+      "epoch": 0.1271585557299843,
+      "grad_norm": 2.836479859442291,
+      "learning_rate": 1.950930739881583e-05,
+      "loss": 1.0707,
       "step": 1215
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3636386830191409e-05,
-      "loss": 1.2579,
+      "epoch": 0.12726321297749868,
+      "grad_norm": 2.2675840070291633,
+      "learning_rate": 1.9508258060124668e-05,
+      "loss": 1.0395,
       "step": 1216
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3626450889858476e-05,
-      "loss": 1.3641,
+      "epoch": 0.12736787022501309,
+      "grad_norm": 3.2573289605017974,
+      "learning_rate": 1.950720762891538e-05,
+      "loss": 1.1698,
       "step": 1217
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3616510825550582e-05,
-      "loss": 1.2377,
+      "epoch": 0.12747252747252746,
+      "grad_norm": 2.252961963832606,
+      "learning_rate": 1.9506156105308665e-05,
+      "loss": 0.8967,
       "step": 1218
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.36065666485715e-05,
-      "loss": 1.3744,
+      "epoch": 0.12757718472004187,
+      "grad_norm": 1.8906880106435673,
+      "learning_rate": 1.950510348942534e-05,
+      "loss": 0.8001,
       "step": 1219
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3596618370229679e-05,
-      "loss": 1.3327,
+      "epoch": 0.12768184196755625,
+      "grad_norm": 2.5679268387367373,
+      "learning_rate": 1.9504049781386358e-05,
+      "loss": 1.0456,
       "step": 1220
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3586666001838231e-05,
-      "loss": 1.2221,
+      "epoch": 0.12778649921507065,
+      "grad_norm": 2.742942369429627,
+      "learning_rate": 1.9502994981312785e-05,
+      "loss": 1.0444,
       "step": 1221
     },
     {
-      "epoch": 0.8,
-      "learning_rate": 1.3576709554714923e-05,
-      "loss": 1.27,
+      "epoch": 0.12789115646258503,
+      "grad_norm": 2.5837062790653302,
+      "learning_rate": 1.950193908932583e-05,
+      "loss": 1.0361,
       "step": 1222
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.356674904018216e-05,
-      "loss": 1.3184,
+      "epoch": 0.12799581371009944,
+      "grad_norm": 2.601717403532944,
+      "learning_rate": 1.9500882105546806e-05,
+      "loss": 0.9211,
       "step": 1223
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.3556784469566968e-05,
-      "loss": 1.3058,
+      "epoch": 0.12810047095761382,
+      "grad_norm": 2.3596656749420317,
+      "learning_rate": 1.9499824030097173e-05,
+      "loss": 0.9949,
       "step": 1224
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.3546815854200987e-05,
-      "loss": 1.297,
+      "epoch": 0.1282051282051282,
+      "grad_norm": 2.431051750501933,
+      "learning_rate": 1.9498764863098494e-05,
+      "loss": 1.111,
       "step": 1225
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.353684320542046e-05,
-      "loss": 1.3612,
+      "epoch": 0.1283097854526426,
+      "grad_norm": 2.465871960006348,
+      "learning_rate": 1.949770460467248e-05,
+      "loss": 1.0887,
       "step": 1226
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.3526866534566213e-05,
-      "loss": 1.2144,
+      "epoch": 0.12841444270015698,
+      "grad_norm": 2.3804735150810883,
+      "learning_rate": 1.9496643254940952e-05,
+      "loss": 1.0723,
       "step": 1227
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.3516885852983645e-05,
-      "loss": 1.3782,
+      "epoch": 0.12851909994767138,
+      "grad_norm": 2.5375697941496265,
+      "learning_rate": 1.9495580814025864e-05,
+      "loss": 0.9679,
       "step": 1228
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.3506901172022724e-05,
-      "loss": 1.3937,
+      "epoch": 0.12862375719518576,
+      "grad_norm": 2.3461082467161565,
+      "learning_rate": 1.949451728204929e-05,
+      "loss": 0.9353,
       "step": 1229
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.3496912503037959e-05,
-      "loss": 1.3548,
+      "epoch": 0.12872841444270017,
+      "grad_norm": 2.770975522225647,
+      "learning_rate": 1.9493452659133437e-05,
+      "loss": 1.0304,
       "step": 1230
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.3486919857388392e-05,
-      "loss": 1.3579,
+      "epoch": 0.12883307169021455,
+      "grad_norm": 2.3083829511711396,
+      "learning_rate": 1.9492386945400623e-05,
+      "loss": 0.9771,
       "step": 1231
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.347692324643759e-05,
-      "loss": 1.2221,
+      "epoch": 0.12893772893772895,
+      "grad_norm": 2.697784666924789,
+      "learning_rate": 1.9491320140973305e-05,
+      "loss": 1.1808,
       "step": 1232
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.3466922681553636e-05,
-      "loss": 1.2839,
+      "epoch": 0.12904238618524333,
+      "grad_norm": 2.8520274235768532,
+      "learning_rate": 1.9490252245974062e-05,
+      "loss": 1.0209,
       "step": 1233
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.3456918174109102e-05,
-      "loss": 1.2777,
+      "epoch": 0.1291470434327577,
+      "grad_norm": 2.2734770595974276,
+      "learning_rate": 1.9489183260525603e-05,
+      "loss": 1.0474,
       "step": 1234
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.344690973548104e-05,
-      "loss": 1.2537,
+      "epoch": 0.1292517006802721,
+      "grad_norm": 2.645718364092188,
+      "learning_rate": 1.9488113184750747e-05,
+      "loss": 1.064,
       "step": 1235
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.3436897377050982e-05,
-      "loss": 1.3699,
+      "epoch": 0.1293563579277865,
+      "grad_norm": 2.41791085736045,
+      "learning_rate": 1.9487042018772452e-05,
+      "loss": 1.0731,
       "step": 1236
     },
     {
-      "epoch": 0.81,
-      "learning_rate": 1.342688111020491e-05,
-      "loss": 1.248,
+      "epoch": 0.1294610151753009,
+      "grad_norm": 2.527589108825978,
+      "learning_rate": 1.9485969762713797e-05,
+      "loss": 1.082,
       "step": 1237
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3416860946333256e-05,
-      "loss": 1.3886,
+      "epoch": 0.12956567242281528,
+      "grad_norm": 2.779752862819606,
+      "learning_rate": 1.9484896416697983e-05,
+      "loss": 0.9216,
       "step": 1238
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3406836896830884e-05,
-      "loss": 1.1917,
+      "epoch": 0.12967032967032968,
+      "grad_norm": 2.732412932188138,
+      "learning_rate": 1.9483821980848346e-05,
+      "loss": 1.1465,
       "step": 1239
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3396808973097066e-05,
-      "loss": 1.2582,
+      "epoch": 0.12977498691784406,
+      "grad_norm": 2.4729282241151935,
+      "learning_rate": 1.948274645528834e-05,
+      "loss": 0.9567,
       "step": 1240
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3386777186535495e-05,
-      "loss": 1.2666,
+      "epoch": 0.12987964416535844,
+      "grad_norm": 2.6626696165070056,
+      "learning_rate": 1.948166984014154e-05,
+      "loss": 0.9809,
       "step": 1241
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3376741548554248e-05,
-      "loss": 1.1695,
+      "epoch": 0.12998430141287284,
+      "grad_norm": 2.6256425374078676,
+      "learning_rate": 1.9480592135531654e-05,
+      "loss": 1.0356,
       "step": 1242
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3366702070565775e-05,
-      "loss": 1.2851,
+      "epoch": 0.13008895866038722,
+      "grad_norm": 2.692113072642092,
+      "learning_rate": 1.9479513341582513e-05,
+      "loss": 1.028,
       "step": 1243
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3356658763986919e-05,
-      "loss": 1.1393,
+      "epoch": 0.13019361590790163,
+      "grad_norm": 2.834736069467216,
+      "learning_rate": 1.947843345841807e-05,
+      "loss": 0.9586,
       "step": 1244
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.334661164023884e-05,
-      "loss": 1.3551,
+      "epoch": 0.130298273155416,
+      "grad_norm": 2.7981103330661568,
+      "learning_rate": 1.947735248616241e-05,
+      "loss": 1.0169,
       "step": 1245
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3336560710747073e-05,
-      "loss": 1.2921,
+      "epoch": 0.1304029304029304,
+      "grad_norm": 2.6541080453909545,
+      "learning_rate": 1.9476270424939736e-05,
+      "loss": 1.0977,
       "step": 1246
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3326505986941456e-05,
-      "loss": 1.2649,
+      "epoch": 0.1305075876504448,
+      "grad_norm": 2.6432248199471005,
+      "learning_rate": 1.9475187274874382e-05,
+      "loss": 1.1162,
       "step": 1247
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3316447480256153e-05,
-      "loss": 1.4216,
+      "epoch": 0.1306122448979592,
+      "grad_norm": 2.655070618581444,
+      "learning_rate": 1.9474103036090795e-05,
+      "loss": 1.0413,
       "step": 1248
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3306385202129633e-05,
-      "loss": 1.2133,
+      "epoch": 0.13071690214547357,
+      "grad_norm": 3.197415853957612,
+      "learning_rate": 1.947301770871357e-05,
+      "loss": 1.0412,
       "step": 1249
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3296319164004646e-05,
-      "loss": 1.1427,
+      "epoch": 0.13082155939298795,
+      "grad_norm": 2.592069054029112,
+      "learning_rate": 1.9471931292867405e-05,
+      "loss": 1.0165,
       "step": 1250
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3286249377328222e-05,
-      "loss": 1.0525,
+      "epoch": 0.13092621664050236,
+      "grad_norm": 3.655152675264706,
+      "learning_rate": 1.9470843788677132e-05,
+      "loss": 0.9054,
       "step": 1251
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.327617585355165e-05,
-      "loss": 1.2209,
+      "epoch": 0.13103087388801674,
+      "grad_norm": 2.882892532973182,
+      "learning_rate": 1.9469755196267706e-05,
+      "loss": 1.1792,
       "step": 1252
     },
     {
-      "epoch": 0.82,
-      "learning_rate": 1.3266098604130477e-05,
-      "loss": 1.2439,
+      "epoch": 0.13113553113553114,
+      "grad_norm": 2.6595064843196954,
+      "learning_rate": 1.9468665515764216e-05,
+      "loss": 1.0301,
       "step": 1253
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.325601764052448e-05,
-      "loss": 1.3988,
+      "epoch": 0.13124018838304552,
+      "grad_norm": 2.3865070700069158,
+      "learning_rate": 1.9467574747291862e-05,
+      "loss": 0.9721,
       "step": 1254
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.324593297419766e-05,
-      "loss": 1.3372,
+      "epoch": 0.13134484563055993,
+      "grad_norm": 2.723979611241111,
+      "learning_rate": 1.9466482890975975e-05,
+      "loss": 1.1416,
       "step": 1255
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.323584461661823e-05,
-      "loss": 1.1898,
+      "epoch": 0.1314495028780743,
+      "grad_norm": 2.605292005226839,
+      "learning_rate": 1.9465389946942013e-05,
+      "loss": 1.1899,
       "step": 1256
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3225752579258599e-05,
-      "loss": 1.2429,
+      "epoch": 0.1315541601255887,
+      "grad_norm": 2.5605584808263675,
+      "learning_rate": 1.9464295915315555e-05,
+      "loss": 1.139,
       "step": 1257
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3215656873595366e-05,
-      "loss": 1.2746,
+      "epoch": 0.1316588173731031,
+      "grad_norm": 2.6247600589750046,
+      "learning_rate": 1.9463200796222318e-05,
+      "loss": 1.1418,
       "step": 1258
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3205557511109298e-05,
-      "loss": 1.1461,
+      "epoch": 0.13176347462061747,
+      "grad_norm": 2.0595672766120896,
+      "learning_rate": 1.946210458978812e-05,
+      "loss": 1.0889,
       "step": 1259
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.319545450328532e-05,
-      "loss": 1.1718,
+      "epoch": 0.13186813186813187,
+      "grad_norm": 2.3474913936763313,
+      "learning_rate": 1.9461007296138925e-05,
+      "loss": 1.0599,
       "step": 1260
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3185347861612501e-05,
-      "loss": 1.2531,
+      "epoch": 0.13197278911564625,
+      "grad_norm": 2.9581647619226685,
+      "learning_rate": 1.945990891540081e-05,
+      "loss": 0.9018,
       "step": 1261
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3175237597584045e-05,
-      "loss": 1.2999,
+      "epoch": 0.13207744636316066,
+      "grad_norm": 2.984957639865946,
+      "learning_rate": 1.9458809447699985e-05,
+      "loss": 1.1799,
       "step": 1262
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3165123722697276e-05,
-      "loss": 1.2824,
+      "epoch": 0.13218210361067503,
+      "grad_norm": 2.772705342436505,
+      "learning_rate": 1.945770889316278e-05,
+      "loss": 1.0051,
       "step": 1263
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3155006248453615e-05,
-      "loss": 1.16,
+      "epoch": 0.13228676085818944,
+      "grad_norm": 2.6058433960187894,
+      "learning_rate": 1.945660725191565e-05,
+      "loss": 1.0885,
       "step": 1264
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3144885186358596e-05,
-      "loss": 1.3179,
+      "epoch": 0.13239141810570382,
+      "grad_norm": 2.7920782381998035,
+      "learning_rate": 1.9455504524085174e-05,
+      "loss": 0.9356,
       "step": 1265
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3134760547921814e-05,
-      "loss": 1.234,
+      "epoch": 0.13249607535321822,
+      "grad_norm": 2.6926709248661966,
+      "learning_rate": 1.9454400709798063e-05,
+      "loss": 1.0235,
       "step": 1266
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3124632344656943e-05,
-      "loss": 1.3048,
+      "epoch": 0.1326007326007326,
+      "grad_norm": 2.306933420813203,
+      "learning_rate": 1.9453295809181144e-05,
+      "loss": 1.0709,
       "step": 1267
     },
     {
-      "epoch": 0.83,
-      "learning_rate": 1.3114500588081698e-05,
-      "loss": 1.2577,
+      "epoch": 0.13270538984824698,
+      "grad_norm": 2.400935059767989,
+      "learning_rate": 1.9452189822361368e-05,
+      "loss": 1.0141,
       "step": 1268
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3104365289717849e-05,
-      "loss": 1.2236,
+      "epoch": 0.13281004709576139,
+      "grad_norm": 2.706220647073549,
+      "learning_rate": 1.9451082749465822e-05,
+      "loss": 0.9398,
       "step": 1269
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3094226461091186e-05,
-      "loss": 1.2877,
+      "epoch": 0.13291470434327576,
+      "grad_norm": 2.5158658128425015,
+      "learning_rate": 1.9449974590621713e-05,
+      "loss": 1.1766,
       "step": 1270
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3084084113731514e-05,
-      "loss": 1.3343,
+      "epoch": 0.13301936159079017,
+      "grad_norm": 2.506227273861366,
+      "learning_rate": 1.9448865345956364e-05,
+      "loss": 1.0084,
       "step": 1271
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3073938259172645e-05,
-      "loss": 1.242,
+      "epoch": 0.13312401883830455,
+      "grad_norm": 2.2919348376307402,
+      "learning_rate": 1.944775501559723e-05,
+      "loss": 0.9608,
       "step": 1272
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3063788908952368e-05,
-      "loss": 1.3352,
+      "epoch": 0.13322867608581895,
+      "grad_norm": 2.9193041380716545,
+      "learning_rate": 1.9446643599671895e-05,
+      "loss": 0.8943,
       "step": 1273
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3053636074612457e-05,
-      "loss": 1.3005,
+      "epoch": 0.13333333333333333,
+      "grad_norm": 2.615394555513494,
+      "learning_rate": 1.944553109830806e-05,
+      "loss": 1.2087,
       "step": 1274
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3043479767698646e-05,
-      "loss": 1.1926,
+      "epoch": 0.13343799058084774,
+      "grad_norm": 2.489254352916976,
+      "learning_rate": 1.944441751163355e-05,
+      "loss": 0.9713,
       "step": 1275
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3033319999760619e-05,
-      "loss": 1.2242,
+      "epoch": 0.13354264782836212,
+      "grad_norm": 2.841255305838923,
+      "learning_rate": 1.9443302839776327e-05,
+      "loss": 1.0022,
       "step": 1276
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3023156782351989e-05,
-      "loss": 1.2421,
+      "epoch": 0.1336473050758765,
+      "grad_norm": 2.5946261317323507,
+      "learning_rate": 1.944218708286446e-05,
+      "loss": 1.0554,
       "step": 1277
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.30129901270303e-05,
-      "loss": 1.158,
+      "epoch": 0.1337519623233909,
+      "grad_norm": 2.284752575634108,
+      "learning_rate": 1.944107024102616e-05,
+      "loss": 1.0401,
       "step": 1278
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.3002820045357005e-05,
-      "loss": 1.2962,
+      "epoch": 0.13385661957090528,
+      "grad_norm": 2.9886804389402735,
+      "learning_rate": 1.943995231438975e-05,
+      "loss": 1.0555,
       "step": 1279
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.2992646548897442e-05,
-      "loss": 1.1178,
+      "epoch": 0.13396127681841968,
+      "grad_norm": 2.602163456273472,
+      "learning_rate": 1.9438833303083677e-05,
+      "loss": 1.0488,
       "step": 1280
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.2982469649220854e-05,
-      "loss": 1.3417,
+      "epoch": 0.13406593406593406,
+      "grad_norm": 2.485141727539193,
+      "learning_rate": 1.9437713207236525e-05,
+      "loss": 1.04,
       "step": 1281
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.2972289357900332e-05,
-      "loss": 1.2901,
+      "epoch": 0.13417059131344847,
+      "grad_norm": 2.727354509959165,
+      "learning_rate": 1.9436592026976994e-05,
+      "loss": 1.0357,
       "step": 1282
     },
     {
-      "epoch": 0.84,
-      "learning_rate": 1.2962105686512834e-05,
-      "loss": 1.2904,
+      "epoch": 0.13427524856096285,
+      "grad_norm": 2.3638486724054513,
+      "learning_rate": 1.943546976243391e-05,
+      "loss": 1.1289,
       "step": 1283
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2951918646639163e-05,
-      "loss": 1.3182,
+      "epoch": 0.13437990580847722,
+      "grad_norm": 2.717669536906904,
+      "learning_rate": 1.943434641373622e-05,
+      "loss": 1.0574,
       "step": 1284
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.294172824986395e-05,
-      "loss": 1.226,
+      "epoch": 0.13448456305599163,
+      "grad_norm": 2.2734518432998914,
+      "learning_rate": 1.9433221981013007e-05,
+      "loss": 1.0728,
       "step": 1285
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.293153450777564e-05,
-      "loss": 1.2812,
+      "epoch": 0.134589220303506,
+      "grad_norm": 2.6630965958749533,
+      "learning_rate": 1.9432096464393463e-05,
+      "loss": 1.061,
       "step": 1286
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2921337431966494e-05,
-      "loss": 1.4035,
+      "epoch": 0.1346938775510204,
+      "grad_norm": 3.1093231373052923,
+      "learning_rate": 1.943096986400691e-05,
+      "loss": 1.0387,
       "step": 1287
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2911137034032553e-05,
-      "loss": 1.3059,
+      "epoch": 0.1347985347985348,
+      "grad_norm": 2.6785202972118682,
+      "learning_rate": 1.942984217998281e-05,
+      "loss": 1.021,
       "step": 1288
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2900933325573636e-05,
-      "loss": 1.3659,
+      "epoch": 0.1349031920460492,
+      "grad_norm": 2.5540937962722543,
+      "learning_rate": 1.9428713412450718e-05,
+      "loss": 1.0165,
       "step": 1289
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2890726318193332e-05,
-      "loss": 1.2342,
+      "epoch": 0.13500784929356358,
+      "grad_norm": 2.3166372688832726,
+      "learning_rate": 1.9427583561540344e-05,
+      "loss": 1.0847,
       "step": 1290
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2880516023498981e-05,
-      "loss": 1.4161,
+      "epoch": 0.13511250654107798,
+      "grad_norm": 2.301232179651965,
+      "learning_rate": 1.942645262738151e-05,
+      "loss": 1.1781,
       "step": 1291
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2870302453101657e-05,
-      "loss": 1.2108,
+      "epoch": 0.13521716378859236,
+      "grad_norm": 2.3252095241169077,
+      "learning_rate": 1.9425320610104157e-05,
+      "loss": 1.0947,
       "step": 1292
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2860085618616169e-05,
-      "loss": 1.3012,
+      "epoch": 0.13532182103610674,
+      "grad_norm": 2.493316820257742,
+      "learning_rate": 1.942418750983836e-05,
+      "loss": 1.0104,
       "step": 1293
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.284986553166102e-05,
-      "loss": 1.2797,
+      "epoch": 0.13542647828362114,
+      "grad_norm": 2.3545726405840153,
+      "learning_rate": 1.942305332671431e-05,
+      "loss": 0.9365,
       "step": 1294
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2839642203858433e-05,
-      "loss": 1.2384,
+      "epoch": 0.13553113553113552,
+      "grad_norm": 2.3550828896798324,
+      "learning_rate": 1.9421918060862333e-05,
+      "loss": 1.1441,
       "step": 1295
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2829415646834295e-05,
-      "loss": 1.3128,
+      "epoch": 0.13563579277864993,
+      "grad_norm": 2.6556094154616146,
+      "learning_rate": 1.942078171241287e-05,
+      "loss": 1.0687,
       "step": 1296
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2819185872218184e-05,
-      "loss": 1.2666,
+      "epoch": 0.1357404500261643,
+      "grad_norm": 2.5889079401056017,
+      "learning_rate": 1.9419644281496492e-05,
+      "loss": 1.036,
       "step": 1297
     },
     {
-      "epoch": 0.85,
-      "learning_rate": 1.2808952891643327e-05,
-      "loss": 1.3178,
+      "epoch": 0.1358451072736787,
+      "grad_norm": 2.4413416204425222,
+      "learning_rate": 1.9418505768243888e-05,
+      "loss": 1.1142,
       "step": 1298
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2798716716746602e-05,
-      "loss": 1.2278,
+      "epoch": 0.1359497645211931,
+      "grad_norm": 2.2226669500183753,
+      "learning_rate": 1.941736617278588e-05,
+      "loss": 1.0729,
       "step": 1299
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2788477359168512e-05,
-      "loss": 1.3281,
+      "epoch": 0.1360544217687075,
+      "grad_norm": 2.3250136278743234,
+      "learning_rate": 1.9416225495253406e-05,
+      "loss": 1.0012,
       "step": 1300
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2778234830553183e-05,
-      "loss": 1.1671,
+      "epoch": 0.13615907901622187,
+      "grad_norm": 2.6455616428189743,
+      "learning_rate": 1.9415083735777535e-05,
+      "loss": 0.9987,
       "step": 1301
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.276798914254835e-05,
-      "loss": 1.3176,
+      "epoch": 0.13626373626373625,
+      "grad_norm": 2.515570541189039,
+      "learning_rate": 1.9413940894489456e-05,
+      "loss": 0.9546,
       "step": 1302
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2757740306805345e-05,
-      "loss": 1.2479,
+      "epoch": 0.13636839351125066,
+      "grad_norm": 2.3610892292812014,
+      "learning_rate": 1.9412796971520486e-05,
+      "loss": 1.002,
       "step": 1303
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2747488334979064e-05,
-      "loss": 1.176,
+      "epoch": 0.13647305075876504,
+      "grad_norm": 2.351486484199385,
+      "learning_rate": 1.9411651967002057e-05,
+      "loss": 0.952,
       "step": 1304
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2737233238727982e-05,
-      "loss": 1.3775,
+      "epoch": 0.13657770800627944,
+      "grad_norm": 2.2806059320082377,
+      "learning_rate": 1.9410505881065745e-05,
+      "loss": 1.0238,
       "step": 1305
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2726975029714124e-05,
-      "loss": 1.3324,
+      "epoch": 0.13668236525379382,
+      "grad_norm": 2.8037668411236947,
+      "learning_rate": 1.9409358713843226e-05,
+      "loss": 0.8635,
       "step": 1306
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2716713719603053e-05,
-      "loss": 1.23,
+      "epoch": 0.13678702250130823,
+      "grad_norm": 2.354476426007749,
+      "learning_rate": 1.9408210465466315e-05,
+      "loss": 1.0029,
       "step": 1307
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2706449320063866e-05,
-      "loss": 1.21,
+      "epoch": 0.1368916797488226,
+      "grad_norm": 2.58570996843689,
+      "learning_rate": 1.9407061136066952e-05,
+      "loss": 1.1598,
       "step": 1308
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2696181842769161e-05,
-      "loss": 1.3184,
+      "epoch": 0.136996336996337,
+      "grad_norm": 2.2297849493454174,
+      "learning_rate": 1.940591072577719e-05,
+      "loss": 1.0518,
       "step": 1309
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2685911299395046e-05,
-      "loss": 1.1431,
+      "epoch": 0.1371009942438514,
+      "grad_norm": 2.4744470445442968,
+      "learning_rate": 1.9404759234729224e-05,
+      "loss": 1.0939,
       "step": 1310
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2675637701621111e-05,
-      "loss": 1.2324,
+      "epoch": 0.13720565149136577,
+      "grad_norm": 2.597406306328376,
+      "learning_rate": 1.940360666305535e-05,
+      "loss": 1.122,
       "step": 1311
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2665361061130416e-05,
-      "loss": 1.4506,
+      "epoch": 0.13731030873888017,
+      "grad_norm": 2.24297119060127,
+      "learning_rate": 1.9402453010888013e-05,
+      "loss": 0.9809,
       "step": 1312
     },
     {
-      "epoch": 0.86,
-      "learning_rate": 1.2655081389609496e-05,
-      "loss": 1.2632,
+      "epoch": 0.13741496598639455,
+      "grad_norm": 2.3878513385258633,
+      "learning_rate": 1.940129827835976e-05,
+      "loss": 1.1173,
       "step": 1313
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2644798698748314e-05,
-      "loss": 1.2199,
+      "epoch": 0.13751962323390896,
+      "grad_norm": 2.195222848508457,
+      "learning_rate": 1.940014246560328e-05,
+      "loss": 1.0293,
       "step": 1314
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2634513000240277e-05,
-      "loss": 1.1961,
+      "epoch": 0.13762428048142333,
+      "grad_norm": 2.266044481844879,
+      "learning_rate": 1.939898557275137e-05,
+      "loss": 0.9389,
       "step": 1315
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2624224305782215e-05,
-      "loss": 1.2649,
+      "epoch": 0.13772893772893774,
+      "grad_norm": 2.227326516279883,
+      "learning_rate": 1.9397827599936967e-05,
+      "loss": 1.0587,
       "step": 1316
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2613932627074356e-05,
-      "loss": 1.3104,
+      "epoch": 0.13783359497645212,
+      "grad_norm": 2.3264930899908216,
+      "learning_rate": 1.9396668547293123e-05,
+      "loss": 1.0047,
       "step": 1317
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2603637975820324e-05,
-      "loss": 1.0901,
+      "epoch": 0.13793825222396652,
+      "grad_norm": 2.7272328130888543,
+      "learning_rate": 1.9395508414953014e-05,
+      "loss": 1.0649,
       "step": 1318
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2593340363727131e-05,
-      "loss": 1.3713,
+      "epoch": 0.1380429094714809,
+      "grad_norm": 2.317310994073378,
+      "learning_rate": 1.9394347203049935e-05,
+      "loss": 0.8952,
       "step": 1319
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2583039802505148e-05,
-      "loss": 1.2994,
+      "epoch": 0.13814756671899528,
+      "grad_norm": 2.7425995244638863,
+      "learning_rate": 1.9393184911717325e-05,
+      "loss": 1.0672,
       "step": 1320
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2572736303868102e-05,
-      "loss": 1.307,
+      "epoch": 0.13825222396650969,
+      "grad_norm": 3.3758460756291897,
+      "learning_rate": 1.9392021541088723e-05,
+      "loss": 1.133,
       "step": 1321
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.256242987953306e-05,
-      "loss": 1.2154,
+      "epoch": 0.13835688121402406,
+      "grad_norm": 2.316620489795767,
+      "learning_rate": 1.939085709129781e-05,
+      "loss": 0.9964,
       "step": 1322
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2552120541220419e-05,
-      "loss": 1.1019,
+      "epoch": 0.13846153846153847,
+      "grad_norm": 2.3989333216450355,
+      "learning_rate": 1.9389691562478375e-05,
+      "loss": 1.111,
       "step": 1323
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2541808300653883e-05,
-      "loss": 1.2299,
+      "epoch": 0.13856619570905285,
+      "grad_norm": 2.562438640327413,
+      "learning_rate": 1.9388524954764345e-05,
+      "loss": 0.9959,
       "step": 1324
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.253149316956047e-05,
-      "loss": 1.1303,
+      "epoch": 0.13867085295656725,
+      "grad_norm": 2.4165917099961782,
+      "learning_rate": 1.938735726828977e-05,
+      "loss": 1.1408,
       "step": 1325
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2521175159670464e-05,
-      "loss": 1.3876,
+      "epoch": 0.13877551020408163,
+      "grad_norm": 2.5972911347326297,
+      "learning_rate": 1.938618850318881e-05,
+      "loss": 1.0121,
       "step": 1326
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2510854282717444e-05,
-      "loss": 1.2807,
+      "epoch": 0.138880167451596,
+      "grad_norm": 2.178384853061503,
+      "learning_rate": 1.9385018659595768e-05,
+      "loss": 0.9381,
       "step": 1327
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2500530550438232e-05,
-      "loss": 1.1717,
+      "epoch": 0.13898482469911042,
+      "grad_norm": 2.341869108186974,
+      "learning_rate": 1.9383847737645053e-05,
+      "loss": 1.0985,
       "step": 1328
     },
     {
-      "epoch": 0.87,
-      "learning_rate": 1.2490203974572915e-05,
-      "loss": 1.3065,
+      "epoch": 0.1390894819466248,
+      "grad_norm": 2.211733194769639,
+      "learning_rate": 1.938267573747121e-05,
+      "loss": 0.9999,
       "step": 1329
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2479874566864799e-05,
-      "loss": 1.2305,
+      "epoch": 0.1391941391941392,
+      "grad_norm": 2.4888824976978463,
+      "learning_rate": 1.9381502659208903e-05,
+      "loss": 1.1316,
       "step": 1330
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2469542339060414e-05,
-      "loss": 1.0857,
+      "epoch": 0.13929879644165358,
+      "grad_norm": 2.5487250145244755,
+      "learning_rate": 1.9380328502992926e-05,
+      "loss": 1.1004,
       "step": 1331
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2459207302909499e-05,
-      "loss": 1.1372,
+      "epoch": 0.13940345368916798,
+      "grad_norm": 2.8605454756354027,
+      "learning_rate": 1.9379153268958183e-05,
+      "loss": 0.86,
       "step": 1332
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2448869470164988e-05,
-      "loss": 1.4061,
+      "epoch": 0.13950811093668236,
+      "grad_norm": 2.275799199333925,
+      "learning_rate": 1.937797695723972e-05,
+      "loss": 1.0387,
       "step": 1333
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.243852885258299e-05,
-      "loss": 1.3289,
+      "epoch": 0.13961276818419677,
+      "grad_norm": 2.256873395213135,
+      "learning_rate": 1.9376799567972692e-05,
+      "loss": 0.9848,
       "step": 1334
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2428185461922783e-05,
-      "loss": 1.3146,
+      "epoch": 0.13971742543171115,
+      "grad_norm": 2.664507783387787,
+      "learning_rate": 1.9375621101292386e-05,
+      "loss": 1.0539,
       "step": 1335
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2417839309946806e-05,
-      "loss": 1.3308,
+      "epoch": 0.13982208267922552,
+      "grad_norm": 2.571479041795175,
+      "learning_rate": 1.9374441557334206e-05,
+      "loss": 1.0578,
       "step": 1336
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2407490408420629e-05,
-      "loss": 1.1638,
+      "epoch": 0.13992673992673993,
+      "grad_norm": 2.6718984503316623,
+      "learning_rate": 1.9373260936233692e-05,
+      "loss": 1.0356,
       "step": 1337
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2397138769112949e-05,
-      "loss": 1.1167,
+      "epoch": 0.1400313971742543,
+      "grad_norm": 3.01919139078155,
+      "learning_rate": 1.937207923812649e-05,
+      "loss": 1.0371,
       "step": 1338
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2386784403795577e-05,
-      "loss": 1.1957,
+      "epoch": 0.1401360544217687,
+      "grad_norm": 2.267242964618876,
+      "learning_rate": 1.9370896463148387e-05,
+      "loss": 0.9883,
       "step": 1339
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2376427324243432e-05,
-      "loss": 1.2913,
+      "epoch": 0.1402407116692831,
+      "grad_norm": 2.513036526104354,
+      "learning_rate": 1.9369712611435285e-05,
+      "loss": 1.073,
       "step": 1340
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2366067542234512e-05,
-      "loss": 1.3546,
+      "epoch": 0.1403453689167975,
+      "grad_norm": 2.3727052030885663,
+      "learning_rate": 1.9368527683123205e-05,
+      "loss": 0.8568,
       "step": 1341
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2355705069549885e-05,
-      "loss": 1.2994,
+      "epoch": 0.14045002616431188,
+      "grad_norm": 2.3202546866710585,
+      "learning_rate": 1.9367341678348307e-05,
+      "loss": 1.0071,
       "step": 1342
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2345339917973688e-05,
-      "loss": 1.4036,
+      "epoch": 0.14055468341182628,
+      "grad_norm": 2.689452076940995,
+      "learning_rate": 1.9366154597246856e-05,
+      "loss": 1.1804,
       "step": 1343
     },
     {
-      "epoch": 0.88,
-      "learning_rate": 1.2334972099293102e-05,
-      "loss": 1.2087,
+      "epoch": 0.14065934065934066,
+      "grad_norm": 2.742100535424126,
+      "learning_rate": 1.936496643995526e-05,
+      "loss": 0.9589,
       "step": 1344
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2324601625298331e-05,
-      "loss": 1.4056,
+      "epoch": 0.14076399790685504,
+      "grad_norm": 2.481796517485657,
+      "learning_rate": 1.936377720661003e-05,
+      "loss": 1.0471,
       "step": 1345
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2314228507782614e-05,
-      "loss": 1.2082,
+      "epoch": 0.14086865515436944,
+      "grad_norm": 2.5501058876317995,
+      "learning_rate": 1.936258689734782e-05,
+      "loss": 0.9166,
       "step": 1346
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2303852758542187e-05,
-      "loss": 1.4642,
+      "epoch": 0.14097331240188382,
+      "grad_norm": 2.336496862440958,
+      "learning_rate": 1.9361395512305395e-05,
+      "loss": 1.088,
       "step": 1347
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2293474389376281e-05,
-      "loss": 1.2091,
+      "epoch": 0.14107796964939823,
+      "grad_norm": 2.8026304750796487,
+      "learning_rate": 1.9360203051619648e-05,
+      "loss": 1.084,
       "step": 1348
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2283093412087099e-05,
-      "loss": 1.2325,
+      "epoch": 0.1411826268969126,
+      "grad_norm": 3.299010180332248,
+      "learning_rate": 1.9359009515427595e-05,
+      "loss": 1.1847,
       "step": 1349
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2272709838479824e-05,
-      "loss": 1.2466,
+      "epoch": 0.141287284144427,
+      "grad_norm": 2.7958593548788655,
+      "learning_rate": 1.9357814903866376e-05,
+      "loss": 1.0987,
       "step": 1350
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2262323680362587e-05,
-      "loss": 1.2338,
+      "epoch": 0.1413919413919414,
+      "grad_norm": 2.3205373831985274,
+      "learning_rate": 1.9356619217073252e-05,
+      "loss": 0.9035,
       "step": 1351
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2251934949546446e-05,
-      "loss": 1.2539,
+      "epoch": 0.1414965986394558,
+      "grad_norm": 2.761571592559715,
+      "learning_rate": 1.9355422455185614e-05,
+      "loss": 1.1353,
       "step": 1352
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2241543657845403e-05,
-      "loss": 1.1242,
+      "epoch": 0.14160125588697017,
+      "grad_norm": 2.7278975042129443,
+      "learning_rate": 1.9354224618340974e-05,
+      "loss": 1.0695,
       "step": 1353
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2231149817076358e-05,
-      "loss": 1.2605,
+      "epoch": 0.14170591313448455,
+      "grad_norm": 2.261842313891641,
+      "learning_rate": 1.9353025706676956e-05,
+      "loss": 1.0403,
       "step": 1354
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2220753439059116e-05,
-      "loss": 1.3588,
+      "epoch": 0.14181057038199896,
+      "grad_norm": 2.375471536045439,
+      "learning_rate": 1.9351825720331328e-05,
+      "loss": 1.0732,
       "step": 1355
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2210354535616366e-05,
-      "loss": 1.2776,
+      "epoch": 0.14191522762951334,
+      "grad_norm": 2.27300544182062,
+      "learning_rate": 1.9350624659441966e-05,
+      "loss": 1.0484,
       "step": 1356
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2199953118573674e-05,
-      "loss": 1.241,
+      "epoch": 0.14201988487702774,
+      "grad_norm": 2.316150199671341,
+      "learning_rate": 1.9349422524146876e-05,
+      "loss": 1.0243,
       "step": 1357
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2189549199759454e-05,
-      "loss": 1.1881,
+      "epoch": 0.14212454212454212,
+      "grad_norm": 2.1366142849380494,
+      "learning_rate": 1.9348219314584182e-05,
+      "loss": 0.9372,
       "step": 1358
     },
     {
-      "epoch": 0.89,
-      "learning_rate": 1.2179142791004974e-05,
-      "loss": 1.2892,
+      "epoch": 0.14222919937205653,
+      "grad_norm": 2.4024552323411545,
+      "learning_rate": 1.934701503089214e-05,
+      "loss": 0.9742,
       "step": 1359
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2168733904144328e-05,
-      "loss": 1.1443,
+      "epoch": 0.1423338566195709,
+      "grad_norm": 2.7893632727077557,
+      "learning_rate": 1.934580967320912e-05,
+      "loss": 0.9427,
       "step": 1360
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2158322551014431e-05,
-      "loss": 1.1217,
+      "epoch": 0.1424385138670853,
+      "grad_norm": 2.7934280403563143,
+      "learning_rate": 1.9344603241673624e-05,
+      "loss": 1.0359,
       "step": 1361
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2147908743455005e-05,
-      "loss": 1.3153,
+      "epoch": 0.1425431711145997,
+      "grad_norm": 2.4453534181554044,
+      "learning_rate": 1.9343395736424273e-05,
+      "loss": 0.915,
       "step": 1362
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2137492493308558e-05,
-      "loss": 1.0986,
+      "epoch": 0.14264782836211407,
+      "grad_norm": 2.643045406854803,
+      "learning_rate": 1.934218715759981e-05,
+      "loss": 1.0316,
       "step": 1363
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2127073812420376e-05,
-      "loss": 1.0793,
+      "epoch": 0.14275248560962847,
+      "grad_norm": 2.6295938809870023,
+      "learning_rate": 1.9340977505339105e-05,
+      "loss": 1.0912,
       "step": 1364
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2116652712638513e-05,
-      "loss": 1.326,
+      "epoch": 0.14285714285714285,
+      "grad_norm": 2.34279514669924,
+      "learning_rate": 1.9339766779781145e-05,
+      "loss": 0.9535,
       "step": 1365
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.210622920581377e-05,
-      "loss": 1.3537,
+      "epoch": 0.14296180010465726,
+      "grad_norm": 2.9096664201728535,
+      "learning_rate": 1.9338554981065055e-05,
+      "loss": 1.0789,
       "step": 1366
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2095803303799693e-05,
-      "loss": 1.2164,
+      "epoch": 0.14306645735217163,
+      "grad_norm": 1.7847384845546328,
+      "learning_rate": 1.9337342109330063e-05,
+      "loss": 0.8184,
       "step": 1367
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2085375018452541e-05,
-      "loss": 1.0878,
+      "epoch": 0.14317111459968604,
+      "grad_norm": 4.650586633766713,
+      "learning_rate": 1.933612816471553e-05,
+      "loss": 0.9754,
       "step": 1368
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2074944361631288e-05,
-      "loss": 1.2595,
+      "epoch": 0.14327577184720042,
+      "grad_norm": 2.4222691647243484,
+      "learning_rate": 1.9334913147360947e-05,
+      "loss": 1.1472,
       "step": 1369
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2064511345197606e-05,
-      "loss": 1.2288,
+      "epoch": 0.1433804290947148,
+      "grad_norm": 2.8646659408254034,
+      "learning_rate": 1.9333697057405923e-05,
+      "loss": 1.0426,
       "step": 1370
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.205407598101585e-05,
-      "loss": 1.2282,
+      "epoch": 0.1434850863422292,
+      "grad_norm": 2.755651171543178,
+      "learning_rate": 1.933247989499018e-05,
+      "loss": 0.9983,
       "step": 1371
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2043638280953043e-05,
-      "loss": 1.3438,
+      "epoch": 0.14358974358974358,
+      "grad_norm": 2.823997144325609,
+      "learning_rate": 1.9331261660253582e-05,
+      "loss": 1.0014,
       "step": 1372
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2033198256878867e-05,
-      "loss": 1.271,
+      "epoch": 0.14369440083725799,
+      "grad_norm": 1.7965865119756081,
+      "learning_rate": 1.9330042353336105e-05,
+      "loss": 0.8996,
       "step": 1373
     },
     {
-      "epoch": 0.9,
-      "learning_rate": 1.2022755920665648e-05,
-      "loss": 1.2748,
+      "epoch": 0.14379905808477236,
+      "grad_norm": 2.530518145697574,
+      "learning_rate": 1.9328821974377843e-05,
+      "loss": 1.0131,
       "step": 1374
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.201231128418833e-05,
-      "loss": 1.2505,
+      "epoch": 0.14390371533228677,
+      "grad_norm": 2.5598798754427587,
+      "learning_rate": 1.9327600523519024e-05,
+      "loss": 0.8846,
       "step": 1375
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.200186435932449e-05,
-      "loss": 1.2923,
+      "epoch": 0.14400837257980115,
+      "grad_norm": 2.404146251680647,
+      "learning_rate": 1.93263780009e-05,
+      "loss": 0.9166,
       "step": 1376
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1991415157954293e-05,
-      "loss": 1.1689,
+      "epoch": 0.14411302982731555,
+      "grad_norm": 2.5090164360878915,
+      "learning_rate": 1.9325154406661236e-05,
+      "loss": 1.0343,
       "step": 1377
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1980963691960508e-05,
-      "loss": 1.2749,
+      "epoch": 0.14421768707482993,
+      "grad_norm": 2.613140816630469,
+      "learning_rate": 1.9323929740943327e-05,
+      "loss": 1.0372,
       "step": 1378
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1970509973228459e-05,
-      "loss": 1.1473,
+      "epoch": 0.1443223443223443,
+      "grad_norm": 2.7666096244065805,
+      "learning_rate": 1.9322704003886988e-05,
+      "loss": 1.0378,
       "step": 1379
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1960054013646048e-05,
-      "loss": 1.2292,
+      "epoch": 0.14442700156985872,
+      "grad_norm": 3.1803876980173023,
+      "learning_rate": 1.932147719563306e-05,
+      "loss": 1.0994,
       "step": 1380
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1949595825103721e-05,
-      "loss": 1.2904,
+      "epoch": 0.1445316588173731,
+      "grad_norm": 2.0896284201518567,
+      "learning_rate": 1.9320249316322505e-05,
+      "loss": 1.0265,
       "step": 1381
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1939135419494456e-05,
-      "loss": 1.2193,
+      "epoch": 0.1446363160648875,
+      "grad_norm": 2.322244889283912,
+      "learning_rate": 1.931902036609641e-05,
+      "loss": 0.9989,
       "step": 1382
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.192867280871376e-05,
-      "loss": 1.0881,
+      "epoch": 0.14474097331240188,
+      "grad_norm": 2.752401702892132,
+      "learning_rate": 1.9317790345095984e-05,
+      "loss": 1.0343,
       "step": 1383
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1918208004659632e-05,
-      "loss": 1.3105,
+      "epoch": 0.14484563055991628,
+      "grad_norm": 2.6752134593317454,
+      "learning_rate": 1.9316559253462556e-05,
+      "loss": 1.0981,
       "step": 1384
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1907741019232584e-05,
-      "loss": 1.2666,
+      "epoch": 0.14495028780743066,
+      "grad_norm": 2.234026157936921,
+      "learning_rate": 1.9315327091337587e-05,
+      "loss": 0.9987,
       "step": 1385
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1897271864335594e-05,
-      "loss": 1.2536,
+      "epoch": 0.14505494505494507,
+      "grad_norm": 2.825030948671621,
+      "learning_rate": 1.931409385886265e-05,
+      "loss": 1.1929,
       "step": 1386
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1886800551874111e-05,
-      "loss": 1.3282,
+      "epoch": 0.14515960230245945,
+      "grad_norm": 2.4650646815837676,
+      "learning_rate": 1.9312859556179445e-05,
+      "loss": 1.0682,
       "step": 1387
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1876327093756048e-05,
-      "loss": 1.1443,
+      "epoch": 0.14526425954997382,
+      "grad_norm": 2.7590671185555267,
+      "learning_rate": 1.9311624183429795e-05,
+      "loss": 1.0708,
       "step": 1388
     },
     {
-      "epoch": 0.91,
-      "learning_rate": 1.1865851501891741e-05,
-      "loss": 1.1805,
+      "epoch": 0.14536891679748823,
+      "grad_norm": 2.5815894772069954,
+      "learning_rate": 1.9310387740755657e-05,
+      "loss": 0.9889,
       "step": 1389
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1855373788193969e-05,
-      "loss": 1.1864,
+      "epoch": 0.1454735740450026,
+      "grad_norm": 2.6236133860227637,
+      "learning_rate": 1.930915022829909e-05,
+      "loss": 1.071,
       "step": 1390
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1844893964577906e-05,
-      "loss": 1.2669,
+      "epoch": 0.145578231292517,
+      "grad_norm": 2.7247337786787638,
+      "learning_rate": 1.9307911646202288e-05,
+      "loss": 1.1056,
       "step": 1391
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1834412042961143e-05,
-      "loss": 1.3911,
+      "epoch": 0.1456828885400314,
+      "grad_norm": 2.862234904110278,
+      "learning_rate": 1.930667199460757e-05,
+      "loss": 1.0049,
       "step": 1392
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1823928035263646e-05,
-      "loss": 1.3024,
+      "epoch": 0.1457875457875458,
+      "grad_norm": 2.4645303878971156,
+      "learning_rate": 1.9305431273657373e-05,
+      "loss": 1.005,
       "step": 1393
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1813441953407754e-05,
-      "loss": 1.1906,
+      "epoch": 0.14589220303506018,
+      "grad_norm": 2.213086769666513,
+      "learning_rate": 1.9304189483494264e-05,
+      "loss": 0.8771,
       "step": 1394
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1802953809318173e-05,
-      "loss": 1.2009,
+      "epoch": 0.14599686028257458,
+      "grad_norm": 2.743298220409371,
+      "learning_rate": 1.9302946624260914e-05,
+      "loss": 1.2425,
       "step": 1395
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1792463614921945e-05,
-      "loss": 1.1585,
+      "epoch": 0.14610151753008896,
+      "grad_norm": 2.6622529241671757,
+      "learning_rate": 1.930170269610014e-05,
+      "loss": 1.0216,
       "step": 1396
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1781971382148446e-05,
-      "loss": 1.3063,
+      "epoch": 0.14620617477760334,
+      "grad_norm": 2.7147812441457515,
+      "learning_rate": 1.9300457699154874e-05,
+      "loss": 0.9964,
       "step": 1397
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.177147712292937e-05,
-      "loss": 1.3428,
+      "epoch": 0.14631083202511774,
+      "grad_norm": 2.393611225263065,
+      "learning_rate": 1.929921163356816e-05,
+      "loss": 1.0048,
       "step": 1398
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1760980849198717e-05,
-      "loss": 1.2162,
+      "epoch": 0.14641548927263212,
+      "grad_norm": 2.7084391545813573,
+      "learning_rate": 1.9297964499483178e-05,
+      "loss": 1.069,
       "step": 1399
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1750482572892781e-05,
-      "loss": 1.2148,
+      "epoch": 0.14652014652014653,
+      "grad_norm": 4.288573871222411,
+      "learning_rate": 1.929671629704323e-05,
+      "loss": 1.0899,
       "step": 1400
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1739982305950124e-05,
-      "loss": 1.292,
+      "epoch": 0.1466248037676609,
+      "grad_norm": 2.317166017560044,
+      "learning_rate": 1.929546702639173e-05,
+      "loss": 0.9723,
       "step": 1401
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.172948006031158e-05,
-      "loss": 1.2038,
+      "epoch": 0.1467294610151753,
+      "grad_norm": 2.0977663597117373,
+      "learning_rate": 1.9294216687672222e-05,
+      "loss": 1.0926,
       "step": 1402
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1718975847920229e-05,
-      "loss": 1.2464,
+      "epoch": 0.1468341182626897,
+      "grad_norm": 2.494438503271147,
+      "learning_rate": 1.929296528102838e-05,
+      "loss": 0.8911,
       "step": 1403
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.170846968072139e-05,
-      "loss": 1.1437,
+      "epoch": 0.1469387755102041,
+      "grad_norm": 2.718143112765956,
+      "learning_rate": 1.9291712806603987e-05,
+      "loss": 1.0866,
       "step": 1404
     },
     {
-      "epoch": 0.92,
-      "learning_rate": 1.1697961570662606e-05,
-      "loss": 1.2991,
+      "epoch": 0.14704343275771847,
+      "grad_norm": 2.2131296320523797,
+      "learning_rate": 1.9290459264542957e-05,
+      "loss": 1.0103,
       "step": 1405
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1687451529693626e-05,
-      "loss": 1.2902,
+      "epoch": 0.14714809000523285,
+      "grad_norm": 2.9091331858933347,
+      "learning_rate": 1.9289204654989324e-05,
+      "loss": 1.1178,
       "step": 1406
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.167693956976639e-05,
-      "loss": 1.2128,
+      "epoch": 0.14725274725274726,
+      "grad_norm": 2.6753705952770708,
+      "learning_rate": 1.928794897808724e-05,
+      "loss": 1.0762,
       "step": 1407
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1666425702835033e-05,
-      "loss": 1.1326,
+      "epoch": 0.14735740450026164,
+      "grad_norm": 2.000163233169189,
+      "learning_rate": 1.9286692233980994e-05,
+      "loss": 1.0423,
       "step": 1408
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1655909940855852e-05,
-      "loss": 1.1265,
+      "epoch": 0.14746206174777604,
+      "grad_norm": 2.293295761305028,
+      "learning_rate": 1.9285434422814984e-05,
+      "loss": 1.0555,
       "step": 1409
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1645392295787299e-05,
-      "loss": 1.3404,
+      "epoch": 0.14756671899529042,
+      "grad_norm": 2.2949310514737262,
+      "learning_rate": 1.9284175544733735e-05,
+      "loss": 1.0947,
       "step": 1410
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1634872779589966e-05,
-      "loss": 1.2424,
+      "epoch": 0.14767137624280482,
+      "grad_norm": 2.5294931864353485,
+      "learning_rate": 1.9282915599881895e-05,
+      "loss": 1.1798,
       "step": 1411
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1624351404226573e-05,
-      "loss": 1.1055,
+      "epoch": 0.1477760334903192,
+      "grad_norm": 2.3551404864975822,
+      "learning_rate": 1.9281654588404233e-05,
+      "loss": 1.0175,
       "step": 1412
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1613828181661957e-05,
-      "loss": 1.3506,
+      "epoch": 0.1478806907378336,
+      "grad_norm": 1.9955894483022132,
+      "learning_rate": 1.9280392510445646e-05,
+      "loss": 1.0019,
       "step": 1413
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1603303123863054e-05,
-      "loss": 1.2615,
+      "epoch": 0.147985347985348,
+      "grad_norm": 2.1950167870687385,
+      "learning_rate": 1.9279129366151143e-05,
+      "loss": 0.8811,
       "step": 1414
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1592776242798887e-05,
-      "loss": 1.2522,
+      "epoch": 0.14809000523286237,
+      "grad_norm": 2.21385771131273,
+      "learning_rate": 1.9277865155665867e-05,
+      "loss": 0.9541,
       "step": 1415
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1582247550440555e-05,
-      "loss": 1.2939,
+      "epoch": 0.14819466248037677,
+      "grad_norm": 3.1210671096804017,
+      "learning_rate": 1.9276599879135074e-05,
+      "loss": 1.3079,
       "step": 1416
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1571717058761211e-05,
-      "loss": 1.2789,
+      "epoch": 0.14829931972789115,
+      "grad_norm": 2.629699340784082,
+      "learning_rate": 1.927533353670415e-05,
+      "loss": 0.9163,
       "step": 1417
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1561184779736061e-05,
-      "loss": 1.2443,
+      "epoch": 0.14840397697540555,
+      "grad_norm": 2.77960958263951,
+      "learning_rate": 1.92740661285186e-05,
+      "loss": 1.0696,
       "step": 1418
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1550650725342337e-05,
-      "loss": 1.3018,
+      "epoch": 0.14850863422291993,
+      "grad_norm": 2.5708441812829177,
+      "learning_rate": 1.9272797654724052e-05,
+      "loss": 1.1057,
       "step": 1419
     },
     {
-      "epoch": 0.93,
-      "learning_rate": 1.1540114907559294e-05,
-      "loss": 1.3083,
+      "epoch": 0.14861329147043434,
+      "grad_norm": 2.1036647647909303,
+      "learning_rate": 1.9271528115466257e-05,
+      "loss": 0.9253,
       "step": 1420
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1529577338368191e-05,
-      "loss": 1.0566,
+      "epoch": 0.14871794871794872,
+      "grad_norm": 2.2534281596027483,
+      "learning_rate": 1.9270257510891083e-05,
+      "loss": 0.9099,
       "step": 1421
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1519038029752281e-05,
-      "loss": 1.2826,
+      "epoch": 0.1488226059654631,
+      "grad_norm": 2.1743546862854366,
+      "learning_rate": 1.926898584114453e-05,
+      "loss": 1.047,
       "step": 1422
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1508496993696791e-05,
-      "loss": 1.1012,
+      "epoch": 0.1489272632129775,
+      "grad_norm": 2.810151030114405,
+      "learning_rate": 1.9267713106372716e-05,
+      "loss": 0.9191,
       "step": 1423
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1497954242188913e-05,
-      "loss": 1.1776,
+      "epoch": 0.14903192046049188,
+      "grad_norm": 2.2964956304821578,
+      "learning_rate": 1.9266439306721874e-05,
+      "loss": 1.0545,
       "step": 1424
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.14874097872178e-05,
-      "loss": 1.462,
+      "epoch": 0.14913657770800628,
+      "grad_norm": 2.3401825418391407,
+      "learning_rate": 1.9265164442338374e-05,
+      "loss": 1.0503,
       "step": 1425
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1476863640774523e-05,
-      "loss": 1.2547,
+      "epoch": 0.14924123495552066,
+      "grad_norm": 2.2508646328782818,
+      "learning_rate": 1.9263888513368693e-05,
+      "loss": 1.0283,
       "step": 1426
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1466315814852094e-05,
-      "loss": 1.2628,
+      "epoch": 0.14934589220303507,
+      "grad_norm": 2.6552965793248364,
+      "learning_rate": 1.9262611519959446e-05,
+      "loss": 1.0631,
       "step": 1427
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.145576632144542e-05,
-      "loss": 1.2201,
+      "epoch": 0.14945054945054945,
+      "grad_norm": 2.532689406015064,
+      "learning_rate": 1.9261333462257357e-05,
+      "loss": 1.1762,
       "step": 1428
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.144521517255132e-05,
-      "loss": 1.2119,
+      "epoch": 0.14955520669806385,
+      "grad_norm": 2.5552203568036758,
+      "learning_rate": 1.9260054340409277e-05,
+      "loss": 1.0354,
       "step": 1429
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1434662380168487e-05,
-      "loss": 1.2235,
+      "epoch": 0.14965986394557823,
+      "grad_norm": 2.3797872928219026,
+      "learning_rate": 1.9258774154562183e-05,
+      "loss": 1.044,
       "step": 1430
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1424107956297479e-05,
-      "loss": 1.2132,
+      "epoch": 0.1497645211930926,
+      "grad_norm": 2.5188480453800097,
+      "learning_rate": 1.9257492904863168e-05,
+      "loss": 1.0178,
       "step": 1431
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1413551912940721e-05,
-      "loss": 1.2307,
+      "epoch": 0.14986917844060701,
+      "grad_norm": 1.9882868472731003,
+      "learning_rate": 1.9256210591459452e-05,
+      "loss": 0.7584,
       "step": 1432
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1402994262102468e-05,
-      "loss": 1.2471,
+      "epoch": 0.1499738356881214,
+      "grad_norm": 2.4171327636272837,
+      "learning_rate": 1.9254927214498376e-05,
+      "loss": 1.0275,
       "step": 1433
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.1392435015788805e-05,
-      "loss": 1.1631,
+      "epoch": 0.1500784929356358,
+      "grad_norm": 2.6335229959993693,
+      "learning_rate": 1.92536427741274e-05,
+      "loss": 1.157,
       "step": 1434
     },
     {
-      "epoch": 0.94,
-      "learning_rate": 1.138187418600764e-05,
-      "loss": 1.3305,
+      "epoch": 0.15018315018315018,
+      "grad_norm": 2.418758003326527,
+      "learning_rate": 1.9252357270494108e-05,
+      "loss": 1.107,
       "step": 1435
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1371311784768672e-05,
-      "loss": 1.1862,
+      "epoch": 0.15028780743066458,
+      "grad_norm": 2.4515624975379073,
+      "learning_rate": 1.9251070703746212e-05,
+      "loss": 1.0629,
       "step": 1436
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1360747824083397e-05,
-      "loss": 1.1721,
+      "epoch": 0.15039246467817896,
+      "grad_norm": 2.5078647866604364,
+      "learning_rate": 1.9249783074031537e-05,
+      "loss": 0.9575,
       "step": 1437
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1350182315965071e-05,
-      "loss": 1.2731,
+      "epoch": 0.15049712192569337,
+      "grad_norm": 2.05009431610471,
+      "learning_rate": 1.9248494381498036e-05,
+      "loss": 0.9304,
       "step": 1438
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1339615272428722e-05,
-      "loss": 1.0712,
+      "epoch": 0.15060177917320774,
+      "grad_norm": 2.2150107036134288,
+      "learning_rate": 1.924720462629378e-05,
+      "loss": 1.0904,
       "step": 1439
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1329046705491111e-05,
-      "loss": 1.2964,
+      "epoch": 0.15070643642072212,
+      "grad_norm": 2.147637671766143,
+      "learning_rate": 1.9245913808566972e-05,
+      "loss": 0.8344,
       "step": 1440
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1318476627170751e-05,
-      "loss": 1.2767,
+      "epoch": 0.15081109366823653,
+      "grad_norm": 2.883011617888164,
+      "learning_rate": 1.9244621928465922e-05,
+      "loss": 1.1529,
       "step": 1441
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1307905049487855e-05,
-      "loss": 1.3459,
+      "epoch": 0.1509157509157509,
+      "grad_norm": 2.5500903724133543,
+      "learning_rate": 1.9243328986139067e-05,
+      "loss": 1.0428,
       "step": 1442
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.129733198446435e-05,
-      "loss": 1.3206,
+      "epoch": 0.1510204081632653,
+      "grad_norm": 2.4159647382548393,
+      "learning_rate": 1.9242034981734977e-05,
+      "loss": 1.0373,
       "step": 1443
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1286757444123854e-05,
-      "loss": 1.2021,
+      "epoch": 0.1511250654107797,
+      "grad_norm": 2.454453579212534,
+      "learning_rate": 1.924073991540233e-05,
+      "loss": 1.0861,
       "step": 1444
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1276181440491658e-05,
-      "loss": 1.302,
+      "epoch": 0.1512297226582941,
+      "grad_norm": 1.8614495838101486,
+      "learning_rate": 1.9239443787289937e-05,
+      "loss": 0.821,
       "step": 1445
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1265603985594723e-05,
-      "loss": 1.3366,
+      "epoch": 0.15133437990580847,
+      "grad_norm": 2.450919159468918,
+      "learning_rate": 1.923814659754672e-05,
+      "loss": 0.9241,
       "step": 1446
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1255025091461657e-05,
-      "loss": 1.3548,
+      "epoch": 0.15143903715332288,
+      "grad_norm": 2.3140383964591784,
+      "learning_rate": 1.9236848346321733e-05,
+      "loss": 1.09,
       "step": 1447
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1244444770122707e-05,
-      "loss": 1.2291,
+      "epoch": 0.15154369440083726,
+      "grad_norm": 2.4237734839061624,
+      "learning_rate": 1.9235549033764146e-05,
+      "loss": 0.944,
       "step": 1448
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.123386303360974e-05,
-      "loss": 1.459,
+      "epoch": 0.15164835164835164,
+      "grad_norm": 2.748321943938643,
+      "learning_rate": 1.923424866002325e-05,
+      "loss": 1.1062,
       "step": 1449
     },
     {
-      "epoch": 0.95,
-      "learning_rate": 1.1223279893956234e-05,
-      "loss": 1.2989,
+      "epoch": 0.15175300889586604,
+      "grad_norm": 3.0209245770619453,
+      "learning_rate": 1.9232947225248465e-05,
+      "loss": 1.0515,
       "step": 1450
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1212695363197262e-05,
-      "loss": 1.2016,
+      "epoch": 0.15185766614338042,
+      "grad_norm": 2.5606861904785476,
+      "learning_rate": 1.9231644729589326e-05,
+      "loss": 1.1291,
       "step": 1451
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.120210945336948e-05,
-      "loss": 1.3596,
+      "epoch": 0.15196232339089483,
+      "grad_norm": 2.581511623994528,
+      "learning_rate": 1.9230341173195493e-05,
+      "loss": 1.0573,
       "step": 1452
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1191522176511112e-05,
-      "loss": 1.2772,
+      "epoch": 0.1520669806384092,
+      "grad_norm": 2.158586432143417,
+      "learning_rate": 1.9229036556216748e-05,
+      "loss": 0.9755,
       "step": 1453
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1180933544661936e-05,
-      "loss": 1.1584,
+      "epoch": 0.1521716378859236,
+      "grad_norm": 2.4855105857033926,
+      "learning_rate": 1.9227730878802996e-05,
+      "loss": 1.023,
       "step": 1454
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1170343569863274e-05,
-      "loss": 1.0758,
+      "epoch": 0.152276295133438,
+      "grad_norm": 2.5999432367590316,
+      "learning_rate": 1.9226424141104252e-05,
+      "loss": 1.1008,
       "step": 1455
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1159752264157963e-05,
-      "loss": 1.3042,
+      "epoch": 0.1523809523809524,
+      "grad_norm": 2.3917057103613075,
+      "learning_rate": 1.9225116343270677e-05,
+      "loss": 1.0145,
       "step": 1456
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1149159639590377e-05,
-      "loss": 1.1603,
+      "epoch": 0.15248560962846677,
+      "grad_norm": 2.2293516130088795,
+      "learning_rate": 1.9223807485452532e-05,
+      "loss": 1.019,
       "step": 1457
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1138565708206364e-05,
-      "loss": 1.3142,
+      "epoch": 0.15259026687598115,
+      "grad_norm": 2.42242544414817,
+      "learning_rate": 1.9222497567800208e-05,
+      "loss": 0.9934,
       "step": 1458
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1127970482053274e-05,
-      "loss": 1.0895,
+      "epoch": 0.15269492412349556,
+      "grad_norm": 2.209915153802859,
+      "learning_rate": 1.9221186590464214e-05,
+      "loss": 1.0644,
       "step": 1459
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1117373973179925e-05,
-      "loss": 1.1043,
+      "epoch": 0.15279958137100993,
+      "grad_norm": 2.2671395202419773,
+      "learning_rate": 1.9219874553595192e-05,
+      "loss": 0.98,
       "step": 1460
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1106776193636595e-05,
-      "loss": 1.3214,
+      "epoch": 0.15290423861852434,
+      "grad_norm": 2.6320601579269725,
+      "learning_rate": 1.9218561457343892e-05,
+      "loss": 0.9559,
       "step": 1461
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1096177155475006e-05,
-      "loss": 1.1407,
+      "epoch": 0.15300889586603872,
+      "grad_norm": 3.128326326853589,
+      "learning_rate": 1.9217247301861192e-05,
+      "loss": 1.057,
       "step": 1462
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1085576870748307e-05,
-      "loss": 1.2877,
+      "epoch": 0.15311355311355312,
+      "grad_norm": 2.6110874383167224,
+      "learning_rate": 1.9215932087298093e-05,
+      "loss": 1.1651,
       "step": 1463
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1074975351511075e-05,
-      "loss": 1.1184,
+      "epoch": 0.1532182103610675,
+      "grad_norm": 2.899961166471011,
+      "learning_rate": 1.9214615813805713e-05,
+      "loss": 1.0419,
       "step": 1464
     },
     {
-      "epoch": 0.96,
-      "learning_rate": 1.1064372609819279e-05,
-      "loss": 1.2395,
+      "epoch": 0.15332286760858188,
+      "grad_norm": 2.484808931850369,
+      "learning_rate": 1.9213298481535295e-05,
+      "loss": 0.9486,
       "step": 1465
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.1053768657730284e-05,
-      "loss": 1.2231,
+      "epoch": 0.1534275248560963,
+      "grad_norm": 2.4399470897359095,
+      "learning_rate": 1.9211980090638204e-05,
+      "loss": 0.9143,
       "step": 1466
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.1043163507302835e-05,
-      "loss": 1.2085,
+      "epoch": 0.15353218210361066,
+      "grad_norm": 2.6027448143774756,
+      "learning_rate": 1.9210660641265926e-05,
+      "loss": 1.0988,
       "step": 1467
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.1032557170597034e-05,
-      "loss": 1.1212,
+      "epoch": 0.15363683935112507,
+      "grad_norm": 2.806968895627165,
+      "learning_rate": 1.920934013357007e-05,
+      "loss": 1.1467,
       "step": 1468
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.1021949659674337e-05,
-      "loss": 1.0985,
+      "epoch": 0.15374149659863945,
+      "grad_norm": 2.5531294035484446,
+      "learning_rate": 1.9208018567702365e-05,
+      "loss": 1.1388,
       "step": 1469
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.1011340986597528e-05,
-      "loss": 1.2408,
+      "epoch": 0.15384615384615385,
+      "grad_norm": 2.811782089564992,
+      "learning_rate": 1.9206695943814657e-05,
+      "loss": 0.9528,
       "step": 1470
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.100073116343072e-05,
-      "loss": 1.2321,
+      "epoch": 0.15395081109366823,
+      "grad_norm": 2.3041792729089683,
+      "learning_rate": 1.9205372262058923e-05,
+      "loss": 0.9661,
       "step": 1471
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.0990120202239324e-05,
-      "loss": 1.3038,
+      "epoch": 0.15405546834118264,
+      "grad_norm": 2.106029830838967,
+      "learning_rate": 1.9204047522587256e-05,
+      "loss": 0.8363,
       "step": 1472
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.097950811509006e-05,
-      "loss": 1.2647,
+      "epoch": 0.15416012558869702,
+      "grad_norm": 2.417394330181495,
+      "learning_rate": 1.9202721725551864e-05,
+      "loss": 1.1144,
       "step": 1473
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.096889491405092e-05,
-      "loss": 1.2462,
+      "epoch": 0.1542647828362114,
+      "grad_norm": 2.046523358752946,
+      "learning_rate": 1.9201394871105098e-05,
+      "loss": 0.9455,
       "step": 1474
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.0958280611191162e-05,
-      "loss": 1.1796,
+      "epoch": 0.1543694400837258,
+      "grad_norm": 2.6640103245094946,
+      "learning_rate": 1.9200066959399403e-05,
+      "loss": 1.007,
       "step": 1475
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.0947665218581293e-05,
-      "loss": 1.2872,
+      "epoch": 0.15447409733124018,
+      "grad_norm": 2.583568326692352,
+      "learning_rate": 1.9198737990587367e-05,
+      "loss": 1.0767,
       "step": 1476
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.0937048748293068e-05,
-      "loss": 1.3001,
+      "epoch": 0.15457875457875458,
+      "grad_norm": 2.5142524512605644,
+      "learning_rate": 1.9197407964821684e-05,
+      "loss": 1.1079,
       "step": 1477
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.0926431212399467e-05,
-      "loss": 1.2903,
+      "epoch": 0.15468341182626896,
+      "grad_norm": 2.608398316355045,
+      "learning_rate": 1.919607688225518e-05,
+      "loss": 1.0722,
       "step": 1478
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.0915812622974673e-05,
-      "loss": 1.1658,
+      "epoch": 0.15478806907378337,
+      "grad_norm": 2.575108862444703,
+      "learning_rate": 1.9194744743040806e-05,
+      "loss": 1.1298,
       "step": 1479
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.0905192992094075e-05,
-      "loss": 1.2011,
+      "epoch": 0.15489272632129775,
+      "grad_norm": 2.548511459425752,
+      "learning_rate": 1.9193411547331618e-05,
+      "loss": 1.0149,
       "step": 1480
     },
     {
-      "epoch": 0.97,
-      "learning_rate": 1.0894572331834248e-05,
-      "loss": 1.1497,
+      "epoch": 0.15499738356881215,
+      "grad_norm": 2.8655016663149366,
+      "learning_rate": 1.9192077295280804e-05,
+      "loss": 0.9912,
       "step": 1481
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.0883950654272925e-05,
-      "loss": 1.3064,
+      "epoch": 0.15510204081632653,
+      "grad_norm": 2.302307429570692,
+      "learning_rate": 1.9190741987041677e-05,
+      "loss": 1.1227,
       "step": 1482
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.0873327971489015e-05,
-      "loss": 1.1824,
+      "epoch": 0.1552066980638409,
+      "grad_norm": 2.709474213319208,
+      "learning_rate": 1.9189405622767666e-05,
+      "loss": 1.1448,
       "step": 1483
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.086270429556255e-05,
-      "loss": 1.145,
+      "epoch": 0.15531135531135531,
+      "grad_norm": 1.9449774082886633,
+      "learning_rate": 1.9188068202612317e-05,
+      "loss": 0.8605,
       "step": 1484
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.0852079638574708e-05,
-      "loss": 1.1142,
+      "epoch": 0.1554160125588697,
+      "grad_norm": 2.217490324767267,
+      "learning_rate": 1.9186729726729308e-05,
+      "loss": 1.0131,
       "step": 1485
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.084145401260777e-05,
-      "loss": 1.4423,
+      "epoch": 0.1555206698063841,
+      "grad_norm": 2.3193422710271605,
+      "learning_rate": 1.9185390195272428e-05,
+      "loss": 1.0585,
       "step": 1486
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.083082742974513e-05,
-      "loss": 1.3293,
+      "epoch": 0.15562532705389848,
+      "grad_norm": 2.4865611357772432,
+      "learning_rate": 1.9184049608395596e-05,
+      "loss": 1.0997,
       "step": 1487
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.082019990207126e-05,
-      "loss": 1.2088,
+      "epoch": 0.15572998430141288,
+      "grad_norm": 2.783780276185395,
+      "learning_rate": 1.9182707966252842e-05,
+      "loss": 0.9589,
       "step": 1488
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.0809571441671716e-05,
-      "loss": 1.3175,
+      "epoch": 0.15583464154892726,
+      "grad_norm": 1.8816997343630404,
+      "learning_rate": 1.9181365268998328e-05,
+      "loss": 0.8778,
       "step": 1489
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.0798942060633108e-05,
-      "loss": 1.1579,
+      "epoch": 0.15593929879644167,
+      "grad_norm": 2.341369035073116,
+      "learning_rate": 1.9180021516786334e-05,
+      "loss": 1.0037,
       "step": 1490
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.0788311771043092e-05,
-      "loss": 1.2273,
+      "epoch": 0.15604395604395604,
+      "grad_norm": 2.7516857057542397,
+      "learning_rate": 1.917867670977126e-05,
+      "loss": 1.1602,
       "step": 1491
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.0777680584990366e-05,
-      "loss": 1.4225,
+      "epoch": 0.15614861329147042,
+      "grad_norm": 2.5272703478492127,
+      "learning_rate": 1.917733084810762e-05,
+      "loss": 1.0647,
       "step": 1492
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.0767048514564633e-05,
-      "loss": 1.345,
+      "epoch": 0.15625327053898483,
+      "grad_norm": 2.4404363392464723,
+      "learning_rate": 1.9175983931950064e-05,
+      "loss": 0.9985,
       "step": 1493
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.0756415571856617e-05,
-      "loss": 0.9828,
+      "epoch": 0.1563579277864992,
+      "grad_norm": 2.9891735698022908,
+      "learning_rate": 1.9174635961453353e-05,
+      "loss": 1.1115,
       "step": 1494
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.074578176895803e-05,
-      "loss": 1.276,
+      "epoch": 0.1564625850340136,
+      "grad_norm": 2.67141152526144,
+      "learning_rate": 1.9173286936772368e-05,
+      "loss": 1.1347,
       "step": 1495
     },
     {
-      "epoch": 0.98,
-      "learning_rate": 1.073514711796155e-05,
-      "loss": 1.2913,
+      "epoch": 0.156567242281528,
+      "grad_norm": 2.4091308182799223,
+      "learning_rate": 1.917193685806212e-05,
+      "loss": 1.1213,
       "step": 1496
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0724511630960837e-05,
-      "loss": 1.2195,
+      "epoch": 0.1566718995290424,
+      "grad_norm": 2.126176198100282,
+      "learning_rate": 1.9170585725477734e-05,
+      "loss": 1.0569,
       "step": 1497
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0713875320050487e-05,
-      "loss": 1.2488,
+      "epoch": 0.15677655677655677,
+      "grad_norm": 2.729603104944734,
+      "learning_rate": 1.9169233539174458e-05,
+      "loss": 1.0782,
       "step": 1498
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0703238197326042e-05,
-      "loss": 1.1805,
+      "epoch": 0.15688121402407118,
+      "grad_norm": 2.513953681189516,
+      "learning_rate": 1.916788029930766e-05,
+      "loss": 1.0887,
       "step": 1499
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0692600274883969e-05,
-      "loss": 1.2677,
+      "epoch": 0.15698587127158556,
+      "grad_norm": 2.6538867773454715,
+      "learning_rate": 1.9166526006032828e-05,
+      "loss": 1.0529,
       "step": 1500
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0681961564821632e-05,
-      "loss": 1.2276,
+      "epoch": 0.15709052851909994,
+      "grad_norm": 2.7220788186480744,
+      "learning_rate": 1.9165170659505577e-05,
+      "loss": 1.108,
       "step": 1501
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0671322079237307e-05,
-      "loss": 1.1109,
+      "epoch": 0.15719518576661434,
+      "grad_norm": 2.4347603808435676,
+      "learning_rate": 1.916381425988164e-05,
+      "loss": 1.0695,
       "step": 1502
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0660681830230137e-05,
-      "loss": 1.2044,
+      "epoch": 0.15729984301412872,
+      "grad_norm": 2.8818520587577425,
+      "learning_rate": 1.916245680731686e-05,
+      "loss": 1.1366,
       "step": 1503
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0650040829900143e-05,
-      "loss": 1.2021,
+      "epoch": 0.15740450026164313,
+      "grad_norm": 2.3660768203543325,
+      "learning_rate": 1.916109830196723e-05,
+      "loss": 1.1054,
       "step": 1504
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.06393990903482e-05,
-      "loss": 1.0864,
+      "epoch": 0.1575091575091575,
+      "grad_norm": 2.0199292578050114,
+      "learning_rate": 1.9159738743988824e-05,
+      "loss": 0.9812,
       "step": 1505
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0628756623676016e-05,
-      "loss": 1.2712,
+      "epoch": 0.1576138147566719,
+      "grad_norm": 2.846624791338478,
+      "learning_rate": 1.915837813353787e-05,
+      "loss": 1.0785,
       "step": 1506
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.061811344198613e-05,
-      "loss": 1.3277,
+      "epoch": 0.1577184720041863,
+      "grad_norm": 2.84528056057021,
+      "learning_rate": 1.9157016470770704e-05,
+      "loss": 1.0292,
       "step": 1507
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.06074695573819e-05,
-      "loss": 1.1496,
+      "epoch": 0.15782312925170067,
+      "grad_norm": 2.9007238219313765,
+      "learning_rate": 1.915565375584378e-05,
+      "loss": 1.0423,
       "step": 1508
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0596824981967475e-05,
-      "loss": 1.2536,
+      "epoch": 0.15792778649921507,
+      "grad_norm": 2.2980335605739857,
+      "learning_rate": 1.9154289988913684e-05,
+      "loss": 1.0505,
       "step": 1509
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0586179727847791e-05,
-      "loss": 1.3152,
+      "epoch": 0.15803244374672945,
+      "grad_norm": 2.5527713053393626,
+      "learning_rate": 1.9152925170137107e-05,
+      "loss": 0.9359,
       "step": 1510
     },
     {
-      "epoch": 0.99,
-      "learning_rate": 1.0575533807128567e-05,
-      "loss": 1.2341,
+      "epoch": 0.15813710099424386,
+      "grad_norm": 2.8078774178279953,
+      "learning_rate": 1.915155929967087e-05,
+      "loss": 0.941,
       "step": 1511
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0564887231916256e-05,
-      "loss": 1.2395,
+      "epoch": 0.15824175824175823,
+      "grad_norm": 2.5590185945658077,
+      "learning_rate": 1.9150192377671923e-05,
+      "loss": 1.0774,
       "step": 1512
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0554240014318085e-05,
-      "loss": 1.2996,
+      "epoch": 0.15834641548927264,
+      "grad_norm": 2.6027080227490953,
+      "learning_rate": 1.914882440429732e-05,
+      "loss": 0.9899,
       "step": 1513
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0543592166441983e-05,
-      "loss": 1.1862,
+      "epoch": 0.15845107273678702,
+      "grad_norm": 2.6337545570436927,
+      "learning_rate": 1.9147455379704246e-05,
+      "loss": 1.0319,
       "step": 1514
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0532943700396616e-05,
-      "loss": 1.3286,
+      "epoch": 0.15855572998430142,
+      "grad_norm": 2.502047847027918,
+      "learning_rate": 1.9146085304050007e-05,
+      "loss": 0.8999,
       "step": 1515
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0522294628291347e-05,
-      "loss": 1.2093,
+      "epoch": 0.1586603872318158,
+      "grad_norm": 2.327612116726649,
+      "learning_rate": 1.9144714177492022e-05,
+      "loss": 1.0622,
       "step": 1516
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0511644962236228e-05,
-      "loss": 1.2798,
+      "epoch": 0.15876504447933018,
+      "grad_norm": 2.5795756527864837,
+      "learning_rate": 1.9143342000187843e-05,
+      "loss": 1.0127,
       "step": 1517
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0500994714341984e-05,
-      "loss": 1.2026,
+      "epoch": 0.1588697017268446,
+      "grad_norm": 2.440086755209261,
+      "learning_rate": 1.914196877229513e-05,
+      "loss": 0.9672,
       "step": 1518
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.049034389672e-05,
-      "loss": 1.1517,
+      "epoch": 0.15897435897435896,
+      "grad_norm": 2.6625915906187654,
+      "learning_rate": 1.9140594493971676e-05,
+      "loss": 1.0476,
       "step": 1519
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0479692521482316e-05,
-      "loss": 0.7474,
+      "epoch": 0.15907901622187337,
+      "grad_norm": 2.2315198666216443,
+      "learning_rate": 1.9139219165375383e-05,
+      "loss": 1.0311,
       "step": 1520
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0469040600741603e-05,
-      "loss": 0.8453,
+      "epoch": 0.15918367346938775,
+      "grad_norm": 2.4701713494311592,
+      "learning_rate": 1.913784278666428e-05,
+      "loss": 1.0938,
       "step": 1521
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0458388146611151e-05,
-      "loss": 0.7812,
+      "epoch": 0.15928833071690215,
+      "grad_norm": 2.377821429542283,
+      "learning_rate": 1.9136465357996517e-05,
+      "loss": 1.0208,
       "step": 1522
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.044773517120486e-05,
-      "loss": 0.7486,
+      "epoch": 0.15939298796441653,
+      "grad_norm": 2.4194712197927974,
+      "learning_rate": 1.9135086879530363e-05,
+      "loss": 1.0568,
       "step": 1523
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0437081686637216e-05,
-      "loss": 0.8325,
+      "epoch": 0.15949764521193094,
+      "grad_norm": 2.3390951758786005,
+      "learning_rate": 1.9133707351424204e-05,
+      "loss": 1.0695,
       "step": 1524
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0426427705023289e-05,
-      "loss": 0.8364,
+      "epoch": 0.15960230245944532,
+      "grad_norm": 2.530783648332205,
+      "learning_rate": 1.913232677383656e-05,
+      "loss": 0.9741,
       "step": 1525
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 1.0415773238478715e-05,
-      "loss": 0.8639,
+      "epoch": 0.1597069597069597,
+      "grad_norm": 2.5392759403405254,
+      "learning_rate": 1.9130945146926054e-05,
+      "loss": 1.0942,
       "step": 1526
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0405118299119682e-05,
-      "loss": 0.9195,
+      "epoch": 0.1598116169544741,
+      "grad_norm": 2.3718557105745073,
+      "learning_rate": 1.912956247085144e-05,
+      "loss": 0.9684,
       "step": 1527
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.039446289906291e-05,
-      "loss": 0.8059,
+      "epoch": 0.15991627420198848,
+      "grad_norm": 2.260719250710198,
+      "learning_rate": 1.9128178745771592e-05,
+      "loss": 0.9989,
       "step": 1528
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0383807050425648e-05,
-      "loss": 0.7991,
+      "epoch": 0.16002093144950288,
+      "grad_norm": 2.739167566121222,
+      "learning_rate": 1.91267939718455e-05,
+      "loss": 1.0678,
       "step": 1529
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.037315076532565e-05,
-      "loss": 0.8506,
+      "epoch": 0.16012558869701726,
+      "grad_norm": 2.6646863118785613,
+      "learning_rate": 1.912540814923228e-05,
+      "loss": 1.0639,
       "step": 1530
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0362494055881172e-05,
-      "loss": 0.798,
+      "epoch": 0.16023024594453167,
+      "grad_norm": 2.194868214474638,
+      "learning_rate": 1.912402127809116e-05,
+      "loss": 0.874,
       "step": 1531
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0351836934210957e-05,
-      "loss": 0.7271,
+      "epoch": 0.16033490319204605,
+      "grad_norm": 2.3676540984204992,
+      "learning_rate": 1.9122633358581507e-05,
+      "loss": 1.1603,
       "step": 1532
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.03411794124342e-05,
-      "loss": 0.7752,
+      "epoch": 0.16043956043956045,
+      "grad_norm": 2.6218767772877634,
+      "learning_rate": 1.912124439086278e-05,
+      "loss": 1.0975,
       "step": 1533
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0330521502670574e-05,
-      "loss": 0.7374,
+      "epoch": 0.16054421768707483,
+      "grad_norm": 2.5083643616431206,
+      "learning_rate": 1.9119854375094586e-05,
+      "loss": 0.9346,
       "step": 1534
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0319863217040166e-05,
-      "loss": 0.8382,
+      "epoch": 0.1606488749345892,
+      "grad_norm": 2.721016148141621,
+      "learning_rate": 1.9118463311436636e-05,
+      "loss": 1.0989,
       "step": 1535
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0309204567663513e-05,
-      "loss": 0.8403,
+      "epoch": 0.16075353218210361,
+      "grad_norm": 2.473905042239637,
+      "learning_rate": 1.9117071200048766e-05,
+      "loss": 1.1074,
       "step": 1536
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0298545566661558e-05,
-      "loss": 0.7714,
+      "epoch": 0.160858189429618,
+      "grad_norm": 2.265192355726035,
+      "learning_rate": 1.9115678041090934e-05,
+      "loss": 0.9356,
       "step": 1537
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.028788622615564e-05,
-      "loss": 0.7968,
+      "epoch": 0.1609628466771324,
+      "grad_norm": 2.8019320178423563,
+      "learning_rate": 1.911428383472321e-05,
+      "loss": 1.1416,
       "step": 1538
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0277226558267494e-05,
-      "loss": 0.7852,
+      "epoch": 0.16106750392464678,
+      "grad_norm": 2.339428603113728,
+      "learning_rate": 1.9112888581105803e-05,
+      "loss": 1.0687,
       "step": 1539
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0266566575119213e-05,
-      "loss": 0.7557,
+      "epoch": 0.16117216117216118,
+      "grad_norm": 2.832854657617948,
+      "learning_rate": 1.9111492280399022e-05,
+      "loss": 1.056,
       "step": 1540
     },
     {
-      "epoch": 1.01,
-      "learning_rate": 1.0255906288833259e-05,
-      "loss": 0.7614,
+      "epoch": 0.16127681841967556,
+      "grad_norm": 2.230222759725198,
+      "learning_rate": 1.911009493276331e-05,
+      "loss": 1.0508,
       "step": 1541
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0245245711532437e-05,
-      "loss": 0.8238,
+      "epoch": 0.16138147566718997,
+      "grad_norm": 2.3090795673914215,
+      "learning_rate": 1.9108696538359218e-05,
+      "loss": 1.1137,
       "step": 1542
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0234584855339884e-05,
-      "loss": 0.6947,
+      "epoch": 0.16148613291470434,
+      "grad_norm": 2.613787477378895,
+      "learning_rate": 1.910729709734743e-05,
+      "loss": 1.0693,
       "step": 1543
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0223923732379049e-05,
-      "loss": 0.7356,
+      "epoch": 0.16159079016221872,
+      "grad_norm": 2.7873744759292243,
+      "learning_rate": 1.9105896609888742e-05,
+      "loss": 1.0308,
       "step": 1544
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0213262354773689e-05,
-      "loss": 0.6816,
+      "epoch": 0.16169544740973313,
+      "grad_norm": 2.2750511261647026,
+      "learning_rate": 1.9104495076144077e-05,
+      "loss": 0.9986,
       "step": 1545
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.020260073464785e-05,
-      "loss": 0.6796,
+      "epoch": 0.1618001046572475,
+      "grad_norm": 2.2610184840565863,
+      "learning_rate": 1.9103092496274467e-05,
+      "loss": 0.9975,
       "step": 1546
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0191938884125847e-05,
-      "loss": 0.7914,
+      "epoch": 0.1619047619047619,
+      "grad_norm": 2.6441259800557275,
+      "learning_rate": 1.910168887044108e-05,
+      "loss": 1.0092,
       "step": 1547
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0181276815332273e-05,
-      "loss": 0.7402,
+      "epoch": 0.1620094191522763,
+      "grad_norm": 2.4288506046325096,
+      "learning_rate": 1.910028419880519e-05,
+      "loss": 1.1243,
       "step": 1548
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0170614540391952e-05,
-      "loss": 0.6915,
+      "epoch": 0.1621140763997907,
+      "grad_norm": 2.4032519124937473,
+      "learning_rate": 1.9098878481528192e-05,
+      "loss": 0.9376,
       "step": 1549
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0159952071429953e-05,
-      "loss": 0.7947,
+      "epoch": 0.16221873364730507,
+      "grad_norm": 2.6930599410682805,
+      "learning_rate": 1.9097471718771615e-05,
+      "loss": 1.0011,
       "step": 1550
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0149289420571558e-05,
-      "loss": 0.7117,
+      "epoch": 0.16232339089481945,
+      "grad_norm": 2.944692543573583,
+      "learning_rate": 1.9096063910697096e-05,
+      "loss": 1.1354,
       "step": 1551
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0138626599942263e-05,
-      "loss": 0.7479,
+      "epoch": 0.16242804814233386,
+      "grad_norm": 2.1076650082395787,
+      "learning_rate": 1.9094655057466394e-05,
+      "loss": 0.8077,
       "step": 1552
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0127963621667754e-05,
-      "loss": 0.6167,
+      "epoch": 0.16253270538984824,
+      "grad_norm": 2.633417159588095,
+      "learning_rate": 1.9093245159241386e-05,
+      "loss": 0.9511,
       "step": 1553
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0117300497873897e-05,
-      "loss": 0.6633,
+      "epoch": 0.16263736263736264,
+      "grad_norm": 2.4640899726417986,
+      "learning_rate": 1.9091834216184078e-05,
+      "loss": 1.1126,
       "step": 1554
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.0106637240686723e-05,
-      "loss": 0.7158,
+      "epoch": 0.16274201988487702,
+      "grad_norm": 2.309637404087591,
+      "learning_rate": 1.9090422228456586e-05,
+      "loss": 1.0576,
       "step": 1555
     },
     {
-      "epoch": 1.02,
-      "learning_rate": 1.009597386223241e-05,
-      "loss": 0.6957,
+      "epoch": 0.16284667713239143,
+      "grad_norm": 2.4383894345913655,
+      "learning_rate": 1.9089009196221153e-05,
+      "loss": 0.7495,
       "step": 1556
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 1.0085310374637288e-05,
-      "loss": 0.8345,
+      "epoch": 0.1629513343799058,
+      "grad_norm": 2.445885248005966,
+      "learning_rate": 1.9087595119640137e-05,
+      "loss": 1.0271,
       "step": 1557
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 1.0074646790027792e-05,
-      "loss": 0.8166,
+      "epoch": 0.1630559916274202,
+      "grad_norm": 2.5458426627911757,
+      "learning_rate": 1.908617999887602e-05,
+      "loss": 1.0411,
       "step": 1558
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 1.006398312053048e-05,
-      "loss": 0.7793,
+      "epoch": 0.1631606488749346,
+      "grad_norm": 2.1937796229963213,
+      "learning_rate": 1.90847638340914e-05,
+      "loss": 1.0347,
       "step": 1559
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 1.0053319378272008e-05,
-      "loss": 0.7357,
+      "epoch": 0.16326530612244897,
+      "grad_norm": 2.537434579639156,
+      "learning_rate": 1.9083346625449e-05,
+      "loss": 1.0541,
       "step": 1560
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 1.0042655575379103e-05,
-      "loss": 0.7202,
+      "epoch": 0.16336996336996337,
+      "grad_norm": 2.144339908895929,
+      "learning_rate": 1.908192837311166e-05,
+      "loss": 0.9906,
       "step": 1561
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 1.0031991723978574e-05,
-      "loss": 0.8471,
+      "epoch": 0.16347462061747775,
+      "grad_norm": 2.7950297429331843,
+      "learning_rate": 1.9080509077242336e-05,
+      "loss": 1.0121,
       "step": 1562
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 1.0021327836197274e-05,
-      "loss": 0.7022,
+      "epoch": 0.16357927786499216,
+      "grad_norm": 2.146536389927275,
+      "learning_rate": 1.9079088738004115e-05,
+      "loss": 0.8715,
       "step": 1563
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 1.001066392416211e-05,
-      "loss": 0.7363,
+      "epoch": 0.16368393511250653,
+      "grad_norm": 2.342813449285308,
+      "learning_rate": 1.907766735556019e-05,
+      "loss": 0.9457,
       "step": 1564
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 1e-05,
-      "loss": 0.8191,
+      "epoch": 0.16378859236002094,
+      "grad_norm": 2.423858021588017,
+      "learning_rate": 1.9076244930073885e-05,
+      "loss": 0.9405,
       "step": 1565
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.989336075837893e-06,
-      "loss": 0.831,
+      "epoch": 0.16389324960753532,
+      "grad_norm": 2.2113311521403514,
+      "learning_rate": 1.9074821461708633e-05,
+      "loss": 1.032,
       "step": 1566
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.978672163802726e-06,
-      "loss": 0.846,
+      "epoch": 0.16399790685504972,
+      "grad_norm": 2.5538309174750347,
+      "learning_rate": 1.9073396950628006e-05,
+      "loss": 1.078,
       "step": 1567
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.96800827602143e-06,
-      "loss": 0.8937,
+      "epoch": 0.1641025641025641,
+      "grad_norm": 2.116186550938168,
+      "learning_rate": 1.9071971396995673e-05,
+      "loss": 0.9762,
       "step": 1568
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.9573444246209e-06,
-      "loss": 0.7693,
+      "epoch": 0.16420722135007848,
+      "grad_norm": 2.787785331892462,
+      "learning_rate": 1.9070544800975433e-05,
+      "loss": 1.1774,
       "step": 1569
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.946680621727995e-06,
-      "loss": 0.5848,
+      "epoch": 0.1643118785975929,
+      "grad_norm": 2.4908693717866295,
+      "learning_rate": 1.906911716273121e-05,
+      "loss": 1.0658,
       "step": 1570
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.936016879469522e-06,
-      "loss": 0.76,
+      "epoch": 0.16441653584510726,
+      "grad_norm": 2.267571098697772,
+      "learning_rate": 1.9067688482427037e-05,
+      "loss": 0.9325,
       "step": 1571
     },
     {
-      "epoch": 1.03,
-      "learning_rate": 9.925353209972213e-06,
-      "loss": 0.6796,
+      "epoch": 0.16452119309262167,
+      "grad_norm": 2.3281432184570665,
+      "learning_rate": 1.906625876022708e-05,
+      "loss": 1.0401,
       "step": 1572
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.914689625362717e-06,
-      "loss": 0.7551,
+      "epoch": 0.16462585034013605,
+      "grad_norm": 2.2909902279482703,
+      "learning_rate": 1.906482799629561e-05,
+      "loss": 1.0412,
       "step": 1573
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.904026137767591e-06,
-      "loss": 0.7341,
+      "epoch": 0.16473050758765045,
+      "grad_norm": 2.2388140959121916,
+      "learning_rate": 1.9063396190797024e-05,
+      "loss": 1.0802,
       "step": 1574
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.89336275931328e-06,
-      "loss": 0.6881,
+      "epoch": 0.16483516483516483,
+      "grad_norm": 2.4751194312927973,
+      "learning_rate": 1.9061963343895848e-05,
+      "loss": 0.9623,
       "step": 1575
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.882699502126103e-06,
-      "loss": 0.7877,
+      "epoch": 0.16493982208267924,
+      "grad_norm": 2.2766727680557963,
+      "learning_rate": 1.9060529455756707e-05,
+      "loss": 1.078,
       "step": 1576
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.872036378332251e-06,
-      "loss": 0.6821,
+      "epoch": 0.16504447933019362,
+      "grad_norm": 2.2759067164759785,
+      "learning_rate": 1.905909452654437e-05,
+      "loss": 1.0286,
       "step": 1577
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.86137340005774e-06,
-      "loss": 0.6926,
+      "epoch": 0.165149136577708,
+      "grad_norm": 2.366060725452454,
+      "learning_rate": 1.9057658556423702e-05,
+      "loss": 0.9825,
       "step": 1578
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.850710579428445e-06,
-      "loss": 0.7044,
+      "epoch": 0.1652537938252224,
+      "grad_norm": 2.3640547803376646,
+      "learning_rate": 1.9056221545559704e-05,
+      "loss": 0.9975,
       "step": 1579
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.84004792857005e-06,
-      "loss": 0.6913,
+      "epoch": 0.16535845107273678,
+      "grad_norm": 2.4952266534643512,
+      "learning_rate": 1.9054783494117496e-05,
+      "loss": 0.9696,
       "step": 1580
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.82938545960805e-06,
-      "loss": 0.7573,
+      "epoch": 0.16546310832025118,
+      "grad_norm": 2.566903238353252,
+      "learning_rate": 1.9053344402262306e-05,
+      "loss": 1.1521,
       "step": 1581
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.818723184667732e-06,
-      "loss": 0.7585,
+      "epoch": 0.16556776556776556,
+      "grad_norm": 2.39022571551939,
+      "learning_rate": 1.9051904270159492e-05,
+      "loss": 0.8817,
       "step": 1582
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.808061115874156e-06,
-      "loss": 0.7767,
+      "epoch": 0.16567242281527997,
+      "grad_norm": 2.461629879445559,
+      "learning_rate": 1.9050463097974528e-05,
+      "loss": 0.9734,
       "step": 1583
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.797399265352154e-06,
-      "loss": 0.721,
+      "epoch": 0.16577708006279435,
+      "grad_norm": 2.398269719344242,
+      "learning_rate": 1.9049020885873008e-05,
+      "loss": 1.0474,
       "step": 1584
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.786737645226313e-06,
-      "loss": 0.7766,
+      "epoch": 0.16588173731030875,
+      "grad_norm": 1.9737021667497825,
+      "learning_rate": 1.904757763402064e-05,
+      "loss": 0.9153,
       "step": 1585
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.776076267620955e-06,
-      "loss": 0.697,
+      "epoch": 0.16598639455782313,
+      "grad_norm": 2.4557990474197258,
+      "learning_rate": 1.904613334258326e-05,
+      "loss": 1.0526,
       "step": 1586
     },
     {
-      "epoch": 1.04,
-      "learning_rate": 9.76541514466012e-06,
-      "loss": 0.7165,
+      "epoch": 0.1660910518053375,
+      "grad_norm": 2.296988769284568,
+      "learning_rate": 1.9044688011726823e-05,
+      "loss": 0.9095,
       "step": 1587
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.754754288467564e-06,
-      "loss": 0.821,
+      "epoch": 0.16619570905285191,
+      "grad_norm": 2.261789866599913,
+      "learning_rate": 1.9043241641617398e-05,
+      "loss": 1.0134,
       "step": 1588
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.744093711166745e-06,
-      "loss": 0.732,
+      "epoch": 0.1663003663003663,
+      "grad_norm": 2.457617607787954,
+      "learning_rate": 1.9041794232421177e-05,
+      "loss": 1.0589,
       "step": 1589
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.733433424880789e-06,
-      "loss": 0.705,
+      "epoch": 0.1664050235478807,
+      "grad_norm": 2.5050727103817696,
+      "learning_rate": 1.9040345784304467e-05,
+      "loss": 1.0968,
       "step": 1590
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.72277344173251e-06,
-      "loss": 0.7957,
+      "epoch": 0.16650968079539508,
+      "grad_norm": 2.6614296904360106,
+      "learning_rate": 1.90388962974337e-05,
+      "loss": 0.9425,
       "step": 1591
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.712113773844361e-06,
-      "loss": 0.7128,
+      "epoch": 0.16661433804290948,
+      "grad_norm": 2.388506412574016,
+      "learning_rate": 1.9037445771975425e-05,
+      "loss": 1.0221,
       "step": 1592
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.701454433338445e-06,
-      "loss": 0.7031,
+      "epoch": 0.16671899529042386,
+      "grad_norm": 2.410859511215317,
+      "learning_rate": 1.903599420809631e-05,
+      "loss": 0.9942,
       "step": 1593
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.690795432336489e-06,
-      "loss": 0.7189,
+      "epoch": 0.16682365253793824,
+      "grad_norm": 2.263106423709346,
+      "learning_rate": 1.9034541605963143e-05,
+      "loss": 1.0328,
       "step": 1594
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.680136782959835e-06,
-      "loss": 0.7433,
+      "epoch": 0.16692830978545264,
+      "grad_norm": 2.322273159353024,
+      "learning_rate": 1.9033087965742826e-05,
+      "loss": 1.0696,
       "step": 1595
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.669478497329431e-06,
-      "loss": 0.7002,
+      "epoch": 0.16703296703296702,
+      "grad_norm": 2.5405277930274197,
+      "learning_rate": 1.9031633287602396e-05,
+      "loss": 1.0605,
       "step": 1596
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.658820587565801e-06,
-      "loss": 0.7529,
+      "epoch": 0.16713762428048143,
+      "grad_norm": 2.2074468419651474,
+      "learning_rate": 1.903017757170899e-05,
+      "loss": 1.0402,
       "step": 1597
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.648163065789046e-06,
-      "loss": 0.6715,
+      "epoch": 0.1672422815279958,
+      "grad_norm": 2.754455496878057,
+      "learning_rate": 1.9028720818229874e-05,
+      "loss": 1.0416,
       "step": 1598
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.637505944118828e-06,
-      "loss": 0.7375,
+      "epoch": 0.1673469387755102,
+      "grad_norm": 2.3580569683993198,
+      "learning_rate": 1.9027263027332434e-05,
+      "loss": 1.1088,
       "step": 1599
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.626849234674356e-06,
-      "loss": 0.7689,
+      "epoch": 0.1674515960230246,
+      "grad_norm": 2.3717864565978757,
+      "learning_rate": 1.9025804199184173e-05,
+      "loss": 1.0257,
       "step": 1600
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.616192949574357e-06,
-      "loss": 0.7926,
+      "epoch": 0.167556253270539,
+      "grad_norm": 2.314198126343136,
+      "learning_rate": 1.9024344333952708e-05,
+      "loss": 1.1035,
       "step": 1601
     },
     {
-      "epoch": 1.05,
-      "learning_rate": 9.605537100937094e-06,
-      "loss": 0.7911,
+      "epoch": 0.16766091051805337,
+      "grad_norm": 2.442992823002754,
+      "learning_rate": 1.902288343180579e-05,
+      "loss": 0.9067,
       "step": 1602
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.59488170088032e-06,
-      "loss": 0.7955,
+      "epoch": 0.16776556776556775,
+      "grad_norm": 2.3303308497808963,
+      "learning_rate": 1.9021421492911272e-05,
+      "loss": 1.0525,
       "step": 1603
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.584226761521285e-06,
-      "loss": 0.8466,
+      "epoch": 0.16787022501308216,
+      "grad_norm": 2.3600681521712255,
+      "learning_rate": 1.901995851743714e-05,
+      "loss": 1.0627,
       "step": 1604
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.573572294976714e-06,
-      "loss": 0.6859,
+      "epoch": 0.16797488226059654,
+      "grad_norm": 2.5485749167168077,
+      "learning_rate": 1.9018494505551484e-05,
+      "loss": 1.0949,
       "step": 1605
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.562918313362787e-06,
-      "loss": 0.6251,
+      "epoch": 0.16807953950811094,
+      "grad_norm": 2.403018203528686,
+      "learning_rate": 1.9017029457422525e-05,
+      "loss": 0.9727,
       "step": 1606
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.552264828795142e-06,
-      "loss": 0.7441,
+      "epoch": 0.16818419675562532,
+      "grad_norm": 2.186348329353992,
+      "learning_rate": 1.9015563373218607e-05,
+      "loss": 1.0965,
       "step": 1607
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.541611853388849e-06,
-      "loss": 0.7742,
+      "epoch": 0.16828885400313973,
+      "grad_norm": 2.5369378525468216,
+      "learning_rate": 1.901409625310818e-05,
+      "loss": 0.9336,
       "step": 1608
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.530959399258397e-06,
-      "loss": 0.755,
+      "epoch": 0.1683935112506541,
+      "grad_norm": 2.406260744207432,
+      "learning_rate": 1.901262809725982e-05,
+      "loss": 0.9296,
       "step": 1609
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.520307478517686e-06,
-      "loss": 0.6872,
+      "epoch": 0.1684981684981685,
+      "grad_norm": 2.229664380822881,
+      "learning_rate": 1.9011158905842218e-05,
+      "loss": 1.0721,
       "step": 1610
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.509656103280004e-06,
-      "loss": 0.7239,
+      "epoch": 0.1686028257456829,
+      "grad_norm": 2.588043339747656,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.9903,
       "step": 1611
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.499005285658021e-06,
-      "loss": 0.8755,
+      "epoch": 0.16870748299319727,
+      "grad_norm": 2.4100976190366903,
+      "learning_rate": 1.9008217416974674e-05,
+      "loss": 0.9882,
       "step": 1612
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.488355037763774e-06,
-      "loss": 0.8583,
+      "epoch": 0.16881214024071167,
+      "grad_norm": 2.2572340824619013,
+      "learning_rate": 1.9006745119862713e-05,
+      "loss": 1.0155,
       "step": 1613
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.477705371708657e-06,
-      "loss": 0.7734,
+      "epoch": 0.16891679748822605,
+      "grad_norm": 2.5087512390175117,
+      "learning_rate": 1.9005271787857477e-05,
+      "loss": 1.0246,
       "step": 1614
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.467056299603388e-06,
-      "loss": 0.7673,
+      "epoch": 0.16902145473574046,
+      "grad_norm": 2.644963375872955,
+      "learning_rate": 1.900379742112826e-05,
+      "loss": 1.0464,
       "step": 1615
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.456407833558019e-06,
-      "loss": 0.7102,
+      "epoch": 0.16912611198325483,
+      "grad_norm": 2.4254460348690987,
+      "learning_rate": 1.900232201984446e-05,
+      "loss": 1.0792,
       "step": 1616
     },
     {
-      "epoch": 1.06,
-      "learning_rate": 9.445759985681919e-06,
-      "loss": 0.7643,
+      "epoch": 0.16923076923076924,
+      "grad_norm": 2.3127114426824975,
+      "learning_rate": 1.9000845584175617e-05,
+      "loss": 0.9903,
       "step": 1617
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.435112768083742e-06,
-      "loss": 0.7464,
+      "epoch": 0.16933542647828362,
+      "grad_norm": 2.412862064453713,
+      "learning_rate": 1.8999368114291367e-05,
+      "loss": 0.9951,
       "step": 1618
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.42446619287144e-06,
-      "loss": 0.6291,
+      "epoch": 0.16944008372579802,
+      "grad_norm": 2.469305422395179,
+      "learning_rate": 1.899788961036148e-05,
+      "loss": 0.918,
       "step": 1619
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.41382027215221e-06,
-      "loss": 0.8474,
+      "epoch": 0.1695447409733124,
+      "grad_norm": 2.1595304334372347,
+      "learning_rate": 1.8996410072555833e-05,
+      "loss": 1.0967,
       "step": 1620
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.403175018032528e-06,
-      "loss": 0.824,
+      "epoch": 0.16964939822082678,
+      "grad_norm": 2.7358337011342106,
+      "learning_rate": 1.8994929501044433e-05,
+      "loss": 1.0458,
       "step": 1621
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.3925304426181e-06,
-      "loss": 0.67,
+      "epoch": 0.1697540554683412,
+      "grad_norm": 2.5015188075028747,
+      "learning_rate": 1.8993447895997396e-05,
+      "loss": 1.0962,
       "step": 1622
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.38188655801387e-06,
-      "loss": 0.78,
+      "epoch": 0.16985871271585556,
+      "grad_norm": 2.746712031614558,
+      "learning_rate": 1.8991965257584967e-05,
+      "loss": 0.9008,
       "step": 1623
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.37124337632399e-06,
-      "loss": 0.6538,
+      "epoch": 0.16996336996336997,
+      "grad_norm": 2.308046183330425,
+      "learning_rate": 1.8990481585977503e-05,
+      "loss": 0.8214,
       "step": 1624
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.360600909651804e-06,
-      "loss": 0.6635,
+      "epoch": 0.17006802721088435,
+      "grad_norm": 2.241008498260066,
+      "learning_rate": 1.8988996881345472e-05,
+      "loss": 0.9915,
       "step": 1625
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.349959170099859e-06,
-      "loss": 0.7027,
+      "epoch": 0.17017268445839875,
+      "grad_norm": 2.336157225073833,
+      "learning_rate": 1.8987511143859484e-05,
+      "loss": 1.0093,
       "step": 1626
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.339318169769865e-06,
-      "loss": 0.7704,
+      "epoch": 0.17027734170591313,
+      "grad_norm": 2.456597311443299,
+      "learning_rate": 1.898602437369024e-05,
+      "loss": 1.0753,
       "step": 1627
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.328677920762698e-06,
-      "loss": 0.7284,
+      "epoch": 0.17038199895342754,
+      "grad_norm": 2.1653784471734583,
+      "learning_rate": 1.8984536571008585e-05,
+      "loss": 0.9081,
       "step": 1628
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.318038435178371e-06,
-      "loss": 0.7295,
+      "epoch": 0.17048665620094192,
+      "grad_norm": 2.1955935295728515,
+      "learning_rate": 1.8983047735985464e-05,
+      "loss": 0.9734,
       "step": 1629
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.307399725116035e-06,
-      "loss": 0.7458,
+      "epoch": 0.1705913134484563,
+      "grad_norm": 3.0667450842861457,
+      "learning_rate": 1.8981557868791943e-05,
+      "loss": 1.1295,
       "step": 1630
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.29676180267396e-06,
-      "loss": 0.701,
+      "epoch": 0.1706959706959707,
+      "grad_norm": 2.457644941062628,
+      "learning_rate": 1.8980066969599216e-05,
+      "loss": 1.1109,
       "step": 1631
     },
     {
-      "epoch": 1.07,
-      "learning_rate": 9.286124679949515e-06,
-      "loss": 0.7679,
+      "epoch": 0.17080062794348508,
+      "grad_norm": 2.5249379238888627,
+      "learning_rate": 1.8978575038578593e-05,
+      "loss": 1.0207,
       "step": 1632
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.275488369039169e-06,
-      "loss": 0.7512,
+      "epoch": 0.17090528519099948,
+      "grad_norm": 2.3825489629259966,
+      "learning_rate": 1.8977082075901493e-05,
+      "loss": 1.0369,
       "step": 1633
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.264852882038453e-06,
-      "loss": 0.7408,
+      "epoch": 0.17100994243851386,
+      "grad_norm": 2.2994283024432063,
+      "learning_rate": 1.8975588081739468e-05,
+      "loss": 0.9943,
       "step": 1634
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.254218231041973e-06,
-      "loss": 0.7379,
+      "epoch": 0.17111459968602827,
+      "grad_norm": 2.7924613134451888,
+      "learning_rate": 1.8974093056264173e-05,
+      "loss": 1.1996,
       "step": 1635
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.243584428143383e-06,
-      "loss": 0.74,
+      "epoch": 0.17121925693354265,
+      "grad_norm": 2.44507996655918,
+      "learning_rate": 1.8972596999647398e-05,
+      "loss": 1.1193,
       "step": 1636
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.232951485435367e-06,
-      "loss": 0.7516,
+      "epoch": 0.17132391418105705,
+      "grad_norm": 2.16180218326001,
+      "learning_rate": 1.897109991206103e-05,
+      "loss": 0.8592,
       "step": 1637
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.222319415009639e-06,
-      "loss": 0.8017,
+      "epoch": 0.17142857142857143,
+      "grad_norm": 2.699376290036355,
+      "learning_rate": 1.8969601793677107e-05,
+      "loss": 1.1523,
       "step": 1638
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.21168822895691e-06,
-      "loss": 0.7807,
+      "epoch": 0.1715332286760858,
+      "grad_norm": 2.735100934448089,
+      "learning_rate": 1.8968102644667748e-05,
+      "loss": 1.0564,
       "step": 1639
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.201057939366896e-06,
-      "loss": 0.7438,
+      "epoch": 0.17163788592360021,
+      "grad_norm": 2.573807990283748,
+      "learning_rate": 1.8966602465205214e-05,
+      "loss": 1.0115,
       "step": 1640
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.190428558328285e-06,
-      "loss": 0.6397,
+      "epoch": 0.1717425431711146,
+      "grad_norm": 2.7249740908747784,
+      "learning_rate": 1.8965101255461883e-05,
+      "loss": 0.9658,
       "step": 1641
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.179800097928743e-06,
-      "loss": 0.7843,
+      "epoch": 0.171847200418629,
+      "grad_norm": 2.9345488789227785,
+      "learning_rate": 1.896359901561024e-05,
+      "loss": 1.0128,
       "step": 1642
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.169172570254874e-06,
-      "loss": 0.7219,
+      "epoch": 0.17195185766614338,
+      "grad_norm": 2.889098265415476,
+      "learning_rate": 1.89620957458229e-05,
+      "loss": 1.0816,
       "step": 1643
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.158545987392233e-06,
-      "loss": 1.1518,
+      "epoch": 0.17205651491365778,
+      "grad_norm": 2.627888084723054,
+      "learning_rate": 1.8960591446272595e-05,
+      "loss": 1.0394,
       "step": 1644
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.147920361425295e-06,
-      "loss": 0.8327,
+      "epoch": 0.17216117216117216,
+      "grad_norm": 2.2397484008487663,
+      "learning_rate": 1.8959086117132162e-05,
+      "loss": 0.9653,
       "step": 1645
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.13729570443745e-06,
-      "loss": 0.7701,
+      "epoch": 0.17226582940868654,
+      "grad_norm": 1.9966793466082513,
+      "learning_rate": 1.8957579758574575e-05,
+      "loss": 0.8132,
       "step": 1646
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.12667202851099e-06,
-      "loss": 0.693,
+      "epoch": 0.17237048665620094,
+      "grad_norm": 3.0317885783655307,
+      "learning_rate": 1.8956072370772914e-05,
+      "loss": 1.0076,
       "step": 1647
     },
     {
-      "epoch": 1.08,
-      "learning_rate": 9.116049345727076e-06,
-      "loss": 0.6348,
+      "epoch": 0.17247514390371532,
+      "grad_norm": 2.6374048369635483,
+      "learning_rate": 1.8954563953900386e-05,
+      "loss": 1.0983,
       "step": 1648
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.105427668165755e-06,
-      "loss": 0.7202,
+      "epoch": 0.17257980115122973,
+      "grad_norm": 2.96669464371244,
+      "learning_rate": 1.8953054508130302e-05,
+      "loss": 0.9889,
       "step": 1649
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.094807007905927e-06,
-      "loss": 0.8613,
+      "epoch": 0.1726844583987441,
+      "grad_norm": 2.4723284751273322,
+      "learning_rate": 1.8951544033636105e-05,
+      "loss": 1.1264,
       "step": 1650
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.084187377025334e-06,
-      "loss": 0.6858,
+      "epoch": 0.1727891156462585,
+      "grad_norm": 2.622144198286044,
+      "learning_rate": 1.8950032530591356e-05,
+      "loss": 1.0118,
       "step": 1651
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.073568787600538e-06,
-      "loss": 0.7561,
+      "epoch": 0.1728937728937729,
+      "grad_norm": 2.891941422685323,
+      "learning_rate": 1.8948519999169723e-05,
+      "loss": 0.9545,
       "step": 1652
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.062951251706933e-06,
-      "loss": 0.5943,
+      "epoch": 0.1729984301412873,
+      "grad_norm": 2.91494413227998,
+      "learning_rate": 1.8947006439545e-05,
+      "loss": 1.0054,
       "step": 1653
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.052334781418708e-06,
-      "loss": 0.9754,
+      "epoch": 0.17310308738880167,
+      "grad_norm": 2.3618205093150344,
+      "learning_rate": 1.89454918518911e-05,
+      "loss": 1.0833,
       "step": 1654
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.04171938880884e-06,
-      "loss": 0.8839,
+      "epoch": 0.17320774463631605,
+      "grad_norm": 2.381058104001439,
+      "learning_rate": 1.8943976236382055e-05,
+      "loss": 0.7995,
       "step": 1655
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.031105085949083e-06,
-      "loss": 0.7718,
+      "epoch": 0.17331240188383046,
+      "grad_norm": 2.4725029468294357,
+      "learning_rate": 1.8942459593192008e-05,
+      "loss": 1.1094,
       "step": 1656
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.020491884909941e-06,
-      "loss": 0.7039,
+      "epoch": 0.17341705913134484,
+      "grad_norm": 2.5977724802112045,
+      "learning_rate": 1.8940941922495223e-05,
+      "loss": 0.9014,
       "step": 1657
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 9.009879797760679e-06,
-      "loss": 0.8086,
+      "epoch": 0.17352171637885924,
+      "grad_norm": 3.566756798991803,
+      "learning_rate": 1.8939423224466086e-05,
+      "loss": 0.9807,
       "step": 1658
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 8.999268836569284e-06,
-      "loss": 0.8561,
+      "epoch": 0.17362637362637362,
+      "grad_norm": 2.5787026088981637,
+      "learning_rate": 1.8937903499279104e-05,
+      "loss": 1.0571,
       "step": 1659
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 8.988659013402474e-06,
-      "loss": 0.7165,
+      "epoch": 0.17373103087388803,
+      "grad_norm": 2.5611787827337182,
+      "learning_rate": 1.8936382747108885e-05,
+      "loss": 1.0353,
       "step": 1660
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 8.978050340325668e-06,
-      "loss": 0.6422,
+      "epoch": 0.1738356881214024,
+      "grad_norm": 2.5012911002756417,
+      "learning_rate": 1.8934860968130176e-05,
+      "loss": 1.1265,
       "step": 1661
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 8.967442829402968e-06,
-      "loss": 0.8316,
+      "epoch": 0.1739403453689168,
+      "grad_norm": 2.2836015231005033,
+      "learning_rate": 1.8933338162517824e-05,
+      "loss": 1.026,
       "step": 1662
     },
     {
-      "epoch": 1.09,
-      "learning_rate": 8.956836492697167e-06,
-      "loss": 0.7368,
+      "epoch": 0.1740450026164312,
+      "grad_norm": 2.8317634648824486,
+      "learning_rate": 1.8931814330446813e-05,
+      "loss": 1.1476,
       "step": 1663
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.94623134226972e-06,
-      "loss": 0.7332,
+      "epoch": 0.17414965986394557,
+      "grad_norm": 2.5276867878076907,
+      "learning_rate": 1.8930289472092225e-05,
+      "loss": 1.0794,
       "step": 1664
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.935627390180728e-06,
-      "loss": 0.7195,
+      "epoch": 0.17425431711145997,
+      "grad_norm": 2.6172530432146957,
+      "learning_rate": 1.8928763587629273e-05,
+      "loss": 1.1507,
       "step": 1665
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.925024648488932e-06,
-      "loss": 0.7806,
+      "epoch": 0.17435897435897435,
+      "grad_norm": 3.298918918566836,
+      "learning_rate": 1.8927236677233286e-05,
+      "loss": 1.0261,
       "step": 1666
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.914423129251697e-06,
-      "loss": 0.6617,
+      "epoch": 0.17446363160648876,
+      "grad_norm": 2.0906785380095956,
+      "learning_rate": 1.8925708741079706e-05,
+      "loss": 0.9586,
       "step": 1667
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.903822844524999e-06,
-      "loss": 0.6465,
+      "epoch": 0.17456828885400313,
+      "grad_norm": 2.3161353864865806,
+      "learning_rate": 1.89241797793441e-05,
+      "loss": 1.0169,
       "step": 1668
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.893223806363407e-06,
-      "loss": 0.7953,
+      "epoch": 0.17467294610151754,
+      "grad_norm": 2.7515446593384065,
+      "learning_rate": 1.8922649792202145e-05,
+      "loss": 1.1237,
       "step": 1669
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.882626026820078e-06,
-      "loss": 0.7214,
+      "epoch": 0.17477760334903192,
+      "grad_norm": 2.5581343380268833,
+      "learning_rate": 1.892111877982964e-05,
+      "loss": 0.9509,
       "step": 1670
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.872029517946729e-06,
-      "loss": 0.754,
+      "epoch": 0.17488226059654632,
+      "grad_norm": 2.215250689710175,
+      "learning_rate": 1.8919586742402504e-05,
+      "loss": 1.0509,
       "step": 1671
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.861434291793639e-06,
-      "loss": 0.7778,
+      "epoch": 0.1749869178440607,
+      "grad_norm": 2.2557081770607454,
+      "learning_rate": 1.891805368009677e-05,
+      "loss": 0.9262,
       "step": 1672
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.850840360409627e-06,
-      "loss": 0.8362,
+      "epoch": 0.17509157509157508,
+      "grad_norm": 2.6566875549500626,
+      "learning_rate": 1.8916519593088586e-05,
+      "loss": 1.0298,
       "step": 1673
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.840247735842037e-06,
-      "loss": 0.7572,
+      "epoch": 0.17519623233908949,
+      "grad_norm": 2.5841170671199993,
+      "learning_rate": 1.8914984481554226e-05,
+      "loss": 1.1961,
       "step": 1674
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.829656430136732e-06,
-      "loss": 0.7194,
+      "epoch": 0.17530088958660386,
+      "grad_norm": 2.401181500742945,
+      "learning_rate": 1.8913448345670075e-05,
+      "loss": 0.8559,
       "step": 1675
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.819066455338067e-06,
-      "loss": 0.6776,
+      "epoch": 0.17540554683411827,
+      "grad_norm": 2.7236964577084275,
+      "learning_rate": 1.891191118561264e-05,
+      "loss": 0.9921,
       "step": 1676
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.80847782348889e-06,
-      "loss": 0.8073,
+      "epoch": 0.17551020408163265,
+      "grad_norm": 2.279719174576562,
+      "learning_rate": 1.8910373001558548e-05,
+      "loss": 1.1413,
       "step": 1677
     },
     {
-      "epoch": 1.1,
-      "learning_rate": 8.797890546630522e-06,
-      "loss": 0.6929,
+      "epoch": 0.17561486132914705,
+      "grad_norm": 2.4343550584438725,
+      "learning_rate": 1.890883379368453e-05,
+      "loss": 1.0764,
       "step": 1678
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.787304636802743e-06,
-      "loss": 0.7384,
+      "epoch": 0.17571951857666143,
+      "grad_norm": 2.6430739908289986,
+      "learning_rate": 1.890729356216745e-05,
+      "loss": 0.9843,
       "step": 1679
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.77672010604377e-06,
-      "loss": 0.7707,
+      "epoch": 0.17582417582417584,
+      "grad_norm": 2.2188927123894544,
+      "learning_rate": 1.8905752307184287e-05,
+      "loss": 1.0551,
       "step": 1680
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.766136966390263e-06,
-      "loss": 0.6968,
+      "epoch": 0.17592883307169022,
+      "grad_norm": 2.674634884089362,
+      "learning_rate": 1.8904210028912126e-05,
+      "loss": 1.0451,
       "step": 1681
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.755555229877295e-06,
-      "loss": 0.8593,
+      "epoch": 0.1760334903192046,
+      "grad_norm": 2.604704932941921,
+      "learning_rate": 1.8902666727528184e-05,
+      "loss": 1.0089,
       "step": 1682
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.744974908538345e-06,
-      "loss": 0.7993,
+      "epoch": 0.176138147566719,
+      "grad_norm": 2.542591420241676,
+      "learning_rate": 1.8901122403209785e-05,
+      "loss": 1.0962,
       "step": 1683
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.734396014405282e-06,
-      "loss": 0.7381,
+      "epoch": 0.17624280481423338,
+      "grad_norm": 2.6574054799605755,
+      "learning_rate": 1.889957705613438e-05,
+      "loss": 1.1428,
       "step": 1684
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.723818559508345e-06,
-      "loss": 0.7781,
+      "epoch": 0.17634746206174778,
+      "grad_norm": 2.486565966246706,
+      "learning_rate": 1.8898030686479526e-05,
+      "loss": 1.0292,
       "step": 1685
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.713242555876149e-06,
-      "loss": 0.7275,
+      "epoch": 0.17645211930926216,
+      "grad_norm": 2.4219588038934443,
+      "learning_rate": 1.889648329442291e-05,
+      "loss": 1.0308,
       "step": 1686
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.702668015535651e-06,
-      "loss": 0.6601,
+      "epoch": 0.17655677655677657,
+      "grad_norm": 2.503380792899425,
+      "learning_rate": 1.889493488014233e-05,
+      "loss": 1.0214,
       "step": 1687
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.692094950512145e-06,
-      "loss": 0.665,
+      "epoch": 0.17666143380429095,
+      "grad_norm": 2.346582662942175,
+      "learning_rate": 1.88933854438157e-05,
+      "loss": 1.0546,
       "step": 1688
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.681523372829254e-06,
-      "loss": 0.8076,
+      "epoch": 0.17676609105180532,
+      "grad_norm": 2.460045482771163,
+      "learning_rate": 1.889183498562105e-05,
+      "loss": 1.1341,
       "step": 1689
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.670953294508892e-06,
-      "loss": 0.7575,
+      "epoch": 0.17687074829931973,
+      "grad_norm": 2.5881691646400653,
+      "learning_rate": 1.889028350573654e-05,
+      "loss": 1.1007,
       "step": 1690
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.660384727571282e-06,
-      "loss": 0.7832,
+      "epoch": 0.1769754055468341,
+      "grad_norm": 2.622909655326263,
+      "learning_rate": 1.888873100434043e-05,
+      "loss": 1.069,
       "step": 1691
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.64981768403493e-06,
-      "loss": 0.7434,
+      "epoch": 0.17708006279434851,
+      "grad_norm": 2.5204132573206692,
+      "learning_rate": 1.8887177481611107e-05,
+      "loss": 0.885,
       "step": 1692
     },
     {
-      "epoch": 1.11,
-      "learning_rate": 8.639252175916608e-06,
-      "loss": 0.6912,
+      "epoch": 0.1771847200418629,
+      "grad_norm": 2.510954873616835,
+      "learning_rate": 1.888562293772708e-05,
+      "loss": 1.0541,
       "step": 1693
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.62868821523133e-06,
-      "loss": 0.7807,
+      "epoch": 0.1772893772893773,
+      "grad_norm": 2.7873358491588167,
+      "learning_rate": 1.888406737286696e-05,
+      "loss": 0.9605,
       "step": 1694
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.618125813992364e-06,
-      "loss": 0.6817,
+      "epoch": 0.17739403453689168,
+      "grad_norm": 2.326692971407233,
+      "learning_rate": 1.888251078720949e-05,
+      "loss": 0.967,
       "step": 1695
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.607564984211199e-06,
-      "loss": 0.7088,
+      "epoch": 0.17749869178440608,
+      "grad_norm": 2.106615127470322,
+      "learning_rate": 1.8880953180933524e-05,
+      "loss": 0.8576,
       "step": 1696
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.597005737897534e-06,
-      "loss": 0.7503,
+      "epoch": 0.17760334903192046,
+      "grad_norm": 2.4795735591289487,
+      "learning_rate": 1.8879394554218033e-05,
+      "loss": 1.1236,
       "step": 1697
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.586448087059284e-06,
-      "loss": 0.701,
+      "epoch": 0.17770800627943484,
+      "grad_norm": 3.0042166107472825,
+      "learning_rate": 1.887783490724211e-05,
+      "loss": 1.0146,
       "step": 1698
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.575892043702523e-06,
-      "loss": 0.83,
+      "epoch": 0.17781266352694924,
+      "grad_norm": 2.492060495608788,
+      "learning_rate": 1.8876274240184954e-05,
+      "loss": 0.9161,
       "step": 1699
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.565337619831516e-06,
-      "loss": 0.6545,
+      "epoch": 0.17791732077446362,
+      "grad_norm": 2.777606937433633,
+      "learning_rate": 1.88747125532259e-05,
+      "loss": 1.0153,
       "step": 1700
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.554784827448681e-06,
-      "loss": 0.6521,
+      "epoch": 0.17802197802197803,
+      "grad_norm": 2.460229108054716,
+      "learning_rate": 1.8873149846544377e-05,
+      "loss": 0.9356,
       "step": 1701
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.544233678554581e-06,
-      "loss": 0.8336,
+      "epoch": 0.1781266352694924,
+      "grad_norm": 2.5144312422866197,
+      "learning_rate": 1.8871586120319952e-05,
+      "loss": 1.0638,
       "step": 1702
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.533684185147913e-06,
-      "loss": 0.802,
+      "epoch": 0.1782312925170068,
+      "grad_norm": 2.1719985008734466,
+      "learning_rate": 1.8870021374732297e-05,
+      "loss": 1.009,
       "step": 1703
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.52313635922548e-06,
-      "loss": 0.6752,
+      "epoch": 0.1783359497645212,
+      "grad_norm": 2.464059885032523,
+      "learning_rate": 1.8868455609961206e-05,
+      "loss": 1.1706,
       "step": 1704
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.512590212782203e-06,
-      "loss": 0.6158,
+      "epoch": 0.1784406070120356,
+      "grad_norm": 2.7220400341035917,
+      "learning_rate": 1.8866888826186583e-05,
+      "loss": 1.0106,
       "step": 1705
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.502045757811087e-06,
-      "loss": 0.8078,
+      "epoch": 0.17854526425954997,
+      "grad_norm": 2.3596603917693,
+      "learning_rate": 1.886532102358846e-05,
+      "loss": 1.0622,
       "step": 1706
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.491503006303212e-06,
-      "loss": 0.7619,
+      "epoch": 0.17864992150706435,
+      "grad_norm": 2.4967392208163317,
+      "learning_rate": 1.8863752202346982e-05,
+      "loss": 1.0043,
       "step": 1707
     },
     {
-      "epoch": 1.12,
-      "learning_rate": 8.480961970247722e-06,
-      "loss": 0.7769,
+      "epoch": 0.17875457875457876,
+      "grad_norm": 1.937686380270267,
+      "learning_rate": 1.8862182362642406e-05,
+      "loss": 0.8468,
       "step": 1708
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.47042266163181e-06,
-      "loss": 0.6736,
+      "epoch": 0.17885923600209314,
+      "grad_norm": 2.4387935220753048,
+      "learning_rate": 1.8860611504655112e-05,
+      "loss": 1.0241,
       "step": 1709
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.459885092440708e-06,
-      "loss": 0.8792,
+      "epoch": 0.17896389324960754,
+      "grad_norm": 2.3758956185372755,
+      "learning_rate": 1.885903962856559e-05,
+      "loss": 1.1111,
       "step": 1710
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.449349274657665e-06,
-      "loss": 0.7375,
+      "epoch": 0.17906855049712192,
+      "grad_norm": 2.21312437965851,
+      "learning_rate": 1.885746673455446e-05,
+      "loss": 1.0037,
       "step": 1711
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.438815220263942e-06,
-      "loss": 0.6718,
+      "epoch": 0.17917320774463633,
+      "grad_norm": 2.5602801001146673,
+      "learning_rate": 1.8855892822802443e-05,
+      "loss": 1.1503,
       "step": 1712
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.42828294123879e-06,
-      "loss": 0.7983,
+      "epoch": 0.1792778649921507,
+      "grad_norm": 2.558534262130665,
+      "learning_rate": 1.885431789349039e-05,
+      "loss": 1.1362,
       "step": 1713
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.417752449559447e-06,
-      "loss": 0.8696,
+      "epoch": 0.1793825222396651,
+      "grad_norm": 2.074103983805813,
+      "learning_rate": 1.885274194679926e-05,
+      "loss": 1.017,
       "step": 1714
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.407223757201113e-06,
-      "loss": 0.6301,
+      "epoch": 0.1794871794871795,
+      "grad_norm": 3.144604908715539,
+      "learning_rate": 1.8851164982910135e-05,
+      "loss": 1.1212,
       "step": 1715
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.396696876136948e-06,
-      "loss": 0.6936,
+      "epoch": 0.17959183673469387,
+      "grad_norm": 2.142185971803945,
+      "learning_rate": 1.884958700200421e-05,
+      "loss": 1.1587,
       "step": 1716
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.386171818338048e-06,
-      "loss": 0.768,
+      "epoch": 0.17969649398220827,
+      "grad_norm": 2.0371324766241394,
+      "learning_rate": 1.88480080042628e-05,
+      "loss": 1.0488,
       "step": 1717
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.37564859577343e-06,
-      "loss": 0.82,
+      "epoch": 0.17980115122972265,
+      "grad_norm": 2.5493299229194855,
+      "learning_rate": 1.8846427989867334e-05,
+      "loss": 1.1183,
       "step": 1718
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.365127220410037e-06,
-      "loss": 0.7359,
+      "epoch": 0.17990580847723706,
+      "grad_norm": 2.1735484655699477,
+      "learning_rate": 1.8844846958999355e-05,
+      "loss": 1.0369,
       "step": 1719
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.354607704212703e-06,
-      "loss": 0.7153,
+      "epoch": 0.18001046572475143,
+      "grad_norm": 2.084854109587084,
+      "learning_rate": 1.884326491184053e-05,
+      "loss": 0.9337,
       "step": 1720
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.344090059144151e-06,
-      "loss": 0.8518,
+      "epoch": 0.18011512297226584,
+      "grad_norm": 2.6326212200262202,
+      "learning_rate": 1.884168184857264e-05,
+      "loss": 1.0429,
       "step": 1721
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.333574297164969e-06,
-      "loss": 0.6675,
+      "epoch": 0.18021978021978022,
+      "grad_norm": 2.4562822207892756,
+      "learning_rate": 1.8840097769377585e-05,
+      "loss": 1.0598,
       "step": 1722
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.323060430233613e-06,
-      "loss": 0.8015,
+      "epoch": 0.18032443746729462,
+      "grad_norm": 2.4277508413270272,
+      "learning_rate": 1.8838512674437372e-05,
+      "loss": 0.9412,
       "step": 1723
     },
     {
-      "epoch": 1.13,
-      "learning_rate": 8.312548470306379e-06,
-      "loss": 0.7307,
+      "epoch": 0.180429094714809,
+      "grad_norm": 2.8821956418407804,
+      "learning_rate": 1.8836926563934137e-05,
+      "loss": 0.8627,
       "step": 1724
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.302038429337396e-06,
-      "loss": 0.722,
+      "epoch": 0.18053375196232338,
+      "grad_norm": 2.709669048226828,
+      "learning_rate": 1.8835339438050125e-05,
+      "loss": 0.9982,
       "step": 1725
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.291530319278614e-06,
-      "loss": 0.8462,
+      "epoch": 0.18063840920983779,
+      "grad_norm": 2.111424162303935,
+      "learning_rate": 1.88337512969677e-05,
+      "loss": 1.0113,
       "step": 1726
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.281024152079775e-06,
-      "loss": 0.8232,
+      "epoch": 0.18074306645735216,
+      "grad_norm": 2.234604237746023,
+      "learning_rate": 1.8832162140869343e-05,
+      "loss": 1.0126,
       "step": 1727
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.270519939688421e-06,
-      "loss": 0.7586,
+      "epoch": 0.18084772370486657,
+      "grad_norm": 2.524183949936875,
+      "learning_rate": 1.883057196993765e-05,
+      "loss": 1.2232,
       "step": 1728
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.260017694049878e-06,
-      "loss": 0.8015,
+      "epoch": 0.18095238095238095,
+      "grad_norm": 2.130924634431564,
+      "learning_rate": 1.882898078435534e-05,
+      "loss": 0.9923,
       "step": 1729
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.249517427107225e-06,
-      "loss": 0.7307,
+      "epoch": 0.18105703819989535,
+      "grad_norm": 2.4549286849941385,
+      "learning_rate": 1.8827388584305237e-05,
+      "loss": 1.1306,
       "step": 1730
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.239019150801286e-06,
-      "loss": 0.6664,
+      "epoch": 0.18116169544740973,
+      "grad_norm": 2.698247562016706,
+      "learning_rate": 1.882579536997029e-05,
+      "loss": 1.0832,
       "step": 1731
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.228522877070636e-06,
-      "loss": 0.7064,
+      "epoch": 0.1812663526949241,
+      "grad_norm": 2.4883956970188463,
+      "learning_rate": 1.8824201141533565e-05,
+      "loss": 1.0421,
       "step": 1732
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.21802861785156e-06,
-      "loss": 0.758,
+      "epoch": 0.18137100994243852,
+      "grad_norm": 2.205789516439095,
+      "learning_rate": 1.8822605899178242e-05,
+      "loss": 0.8382,
       "step": 1733
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.207536385078057e-06,
-      "loss": 0.8387,
+      "epoch": 0.1814756671899529,
+      "grad_norm": 2.043592458587824,
+      "learning_rate": 1.8821009643087613e-05,
+      "loss": 1.0106,
       "step": 1734
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.197046190681832e-06,
-      "loss": 0.6874,
+      "epoch": 0.1815803244374673,
+      "grad_norm": 2.678564167470479,
+      "learning_rate": 1.8819412373445094e-05,
+      "loss": 1.0847,
       "step": 1735
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.186558046592247e-06,
-      "loss": 0.8579,
+      "epoch": 0.18168498168498168,
+      "grad_norm": 2.29945001233077,
+      "learning_rate": 1.8817814090434218e-05,
+      "loss": 0.9098,
       "step": 1736
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.176071964736358e-06,
-      "loss": 0.6297,
+      "epoch": 0.18178963893249608,
+      "grad_norm": 2.294178577646669,
+      "learning_rate": 1.8816214794238623e-05,
+      "loss": 1.0782,
       "step": 1737
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.16558795703886e-06,
-      "loss": 0.6555,
+      "epoch": 0.18189429618001046,
+      "grad_norm": 2.420407699481665,
+      "learning_rate": 1.8814614485042077e-05,
+      "loss": 1.1683,
       "step": 1738
     },
     {
-      "epoch": 1.14,
-      "learning_rate": 8.155106035422095e-06,
-      "loss": 0.6886,
+      "epoch": 0.18199895342752487,
+      "grad_norm": 2.389361847046125,
+      "learning_rate": 1.881301316302846e-05,
+      "loss": 0.9132,
       "step": 1739
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.144626211806038e-06,
-      "loss": 0.7178,
+      "epoch": 0.18210361067503925,
+      "grad_norm": 2.3666157461593653,
+      "learning_rate": 1.881141082838176e-05,
+      "loss": 1.0545,
       "step": 1740
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.13414849810826e-06,
-      "loss": 0.7144,
+      "epoch": 0.18220826792255362,
+      "grad_norm": 2.3334015197345024,
+      "learning_rate": 1.8809807481286096e-05,
+      "loss": 1.0485,
       "step": 1741
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.123672906243955e-06,
-      "loss": 0.7969,
+      "epoch": 0.18231292517006803,
+      "grad_norm": 2.0751756735643223,
+      "learning_rate": 1.8808203121925692e-05,
+      "loss": 0.8661,
       "step": 1742
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.11319944812589e-06,
-      "loss": 0.7901,
+      "epoch": 0.1824175824175824,
+      "grad_norm": 2.138459888794792,
+      "learning_rate": 1.8806597750484895e-05,
+      "loss": 0.9413,
       "step": 1743
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.102728135664411e-06,
-      "loss": 0.7181,
+      "epoch": 0.1825222396650968,
+      "grad_norm": 2.3884065014316707,
+      "learning_rate": 1.8804991367148165e-05,
+      "loss": 1.0748,
       "step": 1744
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.09225898076742e-06,
-      "loss": 0.8622,
+      "epoch": 0.1826268969126112,
+      "grad_norm": 2.3903789475716826,
+      "learning_rate": 1.8803383972100073e-05,
+      "loss": 1.2001,
       "step": 1745
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.08179199534037e-06,
-      "loss": 0.8609,
+      "epoch": 0.1827315541601256,
+      "grad_norm": 2.431230215643124,
+      "learning_rate": 1.880177556552532e-05,
+      "loss": 1.097,
       "step": 1746
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.071327191286244e-06,
-      "loss": 0.7806,
+      "epoch": 0.18283621140763998,
+      "grad_norm": 2.6229790482474016,
+      "learning_rate": 1.880016614760871e-05,
+      "loss": 1.1281,
       "step": 1747
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.060864580505542e-06,
-      "loss": 0.6724,
+      "epoch": 0.18294086865515438,
+      "grad_norm": 2.3850699291301,
+      "learning_rate": 1.879855571853517e-05,
+      "loss": 1.0819,
       "step": 1748
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.050404174896282e-06,
-      "loss": 0.7419,
+      "epoch": 0.18304552590266876,
+      "grad_norm": 2.876444507419557,
+      "learning_rate": 1.879694427848974e-05,
+      "loss": 1.0471,
       "step": 1749
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.039945986353954e-06,
-      "loss": 0.7907,
+      "epoch": 0.18315018315018314,
+      "grad_norm": 2.316504236046675,
+      "learning_rate": 1.879533182765758e-05,
+      "loss": 1.0519,
       "step": 1750
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.029490026771545e-06,
-      "loss": 0.6673,
+      "epoch": 0.18325484039769754,
+      "grad_norm": 2.2030334955552413,
+      "learning_rate": 1.8793718366223963e-05,
+      "loss": 0.928,
       "step": 1751
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.019036308039497e-06,
-      "loss": 0.8518,
+      "epoch": 0.18335949764521192,
+      "grad_norm": 2.3511232102886495,
+      "learning_rate": 1.879210389437428e-05,
+      "loss": 0.9923,
       "step": 1752
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 8.008584842045706e-06,
-      "loss": 0.7744,
+      "epoch": 0.18346415489272633,
+      "grad_norm": 2.4416831019547907,
+      "learning_rate": 1.8790488412294035e-05,
+      "loss": 1.0361,
       "step": 1753
     },
     {
-      "epoch": 1.15,
-      "learning_rate": 7.998135640675514e-06,
-      "loss": 0.8429,
+      "epoch": 0.1835688121402407,
+      "grad_norm": 2.4333832677997034,
+      "learning_rate": 1.8788871920168855e-05,
+      "loss": 1.0897,
       "step": 1754
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.987688715811672e-06,
-      "loss": 0.7407,
+      "epoch": 0.1836734693877551,
+      "grad_norm": 2.2799904335051915,
+      "learning_rate": 1.878725441818447e-05,
+      "loss": 1.0096,
       "step": 1755
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.977244079334357e-06,
-      "loss": 0.7023,
+      "epoch": 0.1837781266352695,
+      "grad_norm": 2.462858451122516,
+      "learning_rate": 1.8785635906526737e-05,
+      "loss": 1.0372,
       "step": 1756
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.966801743121133e-06,
-      "loss": 0.7511,
+      "epoch": 0.1838827838827839,
+      "grad_norm": 2.3533243746357146,
+      "learning_rate": 1.8784016385381633e-05,
+      "loss": 1.0798,
       "step": 1757
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.95636171904696e-06,
-      "loss": 0.6814,
+      "epoch": 0.18398744113029827,
+      "grad_norm": 2.718496586207958,
+      "learning_rate": 1.8782395854935237e-05,
+      "loss": 1.0633,
       "step": 1758
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.945924018984155e-06,
-      "loss": 0.8168,
+      "epoch": 0.18409209837781265,
+      "grad_norm": 2.652705423511989,
+      "learning_rate": 1.878077431537375e-05,
+      "loss": 1.0664,
       "step": 1759
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.935488654802395e-06,
-      "loss": 0.6846,
+      "epoch": 0.18419675562532706,
+      "grad_norm": 2.586313483256547,
+      "learning_rate": 1.8779151766883502e-05,
+      "loss": 1.0038,
       "step": 1760
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.925055638368713e-06,
-      "loss": 0.6312,
+      "epoch": 0.18430141287284144,
+      "grad_norm": 2.5041990988530514,
+      "learning_rate": 1.877752820965091e-05,
+      "loss": 1.0798,
       "step": 1761
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.914624981547459e-06,
-      "loss": 0.712,
+      "epoch": 0.18440607012035584,
+      "grad_norm": 2.53966882445818,
+      "learning_rate": 1.877590364386254e-05,
+      "loss": 1.1485,
       "step": 1762
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.90419669620031e-06,
-      "loss": 0.8883,
+      "epoch": 0.18451072736787022,
+      "grad_norm": 2.5828686624066695,
+      "learning_rate": 1.877427806970505e-05,
+      "loss": 1.0969,
       "step": 1763
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.893770794186231e-06,
-      "loss": 0.7941,
+      "epoch": 0.18461538461538463,
+      "grad_norm": 2.3393212197615343,
+      "learning_rate": 1.8772651487365217e-05,
+      "loss": 1.1038,
       "step": 1764
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.88334728736149e-06,
-      "loss": 0.7611,
+      "epoch": 0.184720041862899,
+      "grad_norm": 2.7443600527578393,
+      "learning_rate": 1.8771023897029948e-05,
+      "loss": 0.8974,
       "step": 1765
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.872926187579625e-06,
-      "loss": 0.729,
+      "epoch": 0.1848246991104134,
+      "grad_norm": 2.5532153964529893,
+      "learning_rate": 1.876939529888625e-05,
+      "loss": 0.9866,
       "step": 1766
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.862507506691444e-06,
-      "loss": 0.7939,
+      "epoch": 0.1849293563579278,
+      "grad_norm": 2.4029344349509194,
+      "learning_rate": 1.8767765693121258e-05,
+      "loss": 1.0403,
       "step": 1767
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.852091256545e-06,
-      "loss": 0.713,
+      "epoch": 0.18503401360544217,
+      "grad_norm": 2.176455654469618,
+      "learning_rate": 1.8766135079922213e-05,
+      "loss": 0.9654,
       "step": 1768
     },
     {
-      "epoch": 1.16,
-      "learning_rate": 7.841677448985572e-06,
-      "loss": 0.8089,
+      "epoch": 0.18513867085295657,
+      "grad_norm": 2.396055407588553,
+      "learning_rate": 1.8764503459476476e-05,
+      "loss": 0.8856,
       "step": 1769
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.831266095855675e-06,
-      "loss": 0.6924,
+      "epoch": 0.18524332810047095,
+      "grad_norm": 2.8406525852271165,
+      "learning_rate": 1.8762870831971523e-05,
+      "loss": 1.1081,
       "step": 1770
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.82085720899503e-06,
-      "loss": 0.6898,
+      "epoch": 0.18534798534798536,
+      "grad_norm": 2.9851011238261362,
+      "learning_rate": 1.8761237197594945e-05,
+      "loss": 1.0719,
       "step": 1771
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.810450800240549e-06,
-      "loss": 0.7316,
+      "epoch": 0.18545264259549973,
+      "grad_norm": 1.982828765335705,
+      "learning_rate": 1.8759602556534458e-05,
+      "loss": 0.9932,
       "step": 1772
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.800046881426331e-06,
-      "loss": 0.7067,
+      "epoch": 0.18555729984301414,
+      "grad_norm": 2.7241455065505287,
+      "learning_rate": 1.8757966908977875e-05,
+      "loss": 1.2454,
       "step": 1773
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.789645464383635e-06,
-      "loss": 0.6733,
+      "epoch": 0.18566195709052852,
+      "grad_norm": 2.252662593453829,
+      "learning_rate": 1.875633025511314e-05,
+      "loss": 1.0011,
       "step": 1774
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.779246560940887e-06,
-      "loss": 0.8138,
+      "epoch": 0.1857666143380429,
+      "grad_norm": 2.5528950049004204,
+      "learning_rate": 1.8754692595128313e-05,
+      "loss": 0.9809,
       "step": 1775
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.768850182923646e-06,
-      "loss": 0.7343,
+      "epoch": 0.1858712715855573,
+      "grad_norm": 2.0122663818965134,
+      "learning_rate": 1.8753053929211555e-05,
+      "loss": 0.9965,
       "step": 1776
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.758456342154602e-06,
-      "loss": 0.8757,
+      "epoch": 0.18597592883307168,
+      "grad_norm": 2.1764615737427997,
+      "learning_rate": 1.875141425755116e-05,
+      "loss": 1.0969,
       "step": 1777
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.748065050453557e-06,
-      "loss": 0.7713,
+      "epoch": 0.18608058608058609,
+      "grad_norm": 1.8948836535370301,
+      "learning_rate": 1.874977358033552e-05,
+      "loss": 0.8658,
       "step": 1778
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.737676319637417e-06,
-      "loss": 0.7135,
+      "epoch": 0.18618524332810046,
+      "grad_norm": 2.361535893832336,
+      "learning_rate": 1.874813189775316e-05,
+      "loss": 0.964,
       "step": 1779
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.727290161520176e-06,
-      "loss": 0.8717,
+      "epoch": 0.18628990057561487,
+      "grad_norm": 2.3456335496105707,
+      "learning_rate": 1.8746489209992713e-05,
+      "loss": 0.924,
       "step": 1780
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.716906587912903e-06,
-      "loss": 0.6306,
+      "epoch": 0.18639455782312925,
+      "grad_norm": 2.423670459664142,
+      "learning_rate": 1.874484551724292e-05,
+      "loss": 0.9615,
       "step": 1781
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.706525610623726e-06,
-      "loss": 0.7388,
+      "epoch": 0.18649921507064365,
+      "grad_norm": 2.2890242877296125,
+      "learning_rate": 1.8743200819692652e-05,
+      "loss": 1.0306,
       "step": 1782
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.696147241457817e-06,
-      "loss": 0.6993,
+      "epoch": 0.18660387231815803,
+      "grad_norm": 2.5596300069718505,
+      "learning_rate": 1.8741555117530887e-05,
+      "loss": 0.9138,
       "step": 1783
     },
     {
-      "epoch": 1.17,
-      "learning_rate": 7.685771492217387e-06,
-      "loss": 0.6875,
+      "epoch": 0.1867085295656724,
+      "grad_norm": 2.129042464863284,
+      "learning_rate": 1.873990841094672e-05,
+      "loss": 1.1452,
       "step": 1784
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.675398374701669e-06,
-      "loss": 0.821,
+      "epoch": 0.18681318681318682,
+      "grad_norm": 3.064823960289263,
+      "learning_rate": 1.8738260700129354e-05,
+      "loss": 0.9969,
       "step": 1785
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.665027900706902e-06,
-      "loss": 0.7293,
+      "epoch": 0.1869178440607012,
+      "grad_norm": 2.1130779511781688,
+      "learning_rate": 1.8736611985268124e-05,
+      "loss": 0.9711,
       "step": 1786
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.654660082026313e-06,
-      "loss": 0.7566,
+      "epoch": 0.1870225013082156,
+      "grad_norm": 2.0389255541139994,
+      "learning_rate": 1.873496226655246e-05,
+      "loss": 0.8438,
       "step": 1787
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.644294930450117e-06,
-      "loss": 0.8534,
+      "epoch": 0.18712715855572998,
+      "grad_norm": 2.2033567685248947,
+      "learning_rate": 1.873331154417193e-05,
+      "loss": 1.0108,
       "step": 1788
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.633932457765491e-06,
-      "loss": 0.7217,
+      "epoch": 0.18723181580324438,
+      "grad_norm": 2.15080379319512,
+      "learning_rate": 1.8731659818316196e-05,
+      "loss": 0.8633,
       "step": 1789
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.623572675756568e-06,
-      "loss": 0.7808,
+      "epoch": 0.18733647305075876,
+      "grad_norm": 2.2191227916720004,
+      "learning_rate": 1.8730007089175047e-05,
+      "loss": 0.9503,
       "step": 1790
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.613215596204425e-06,
-      "loss": 0.7292,
+      "epoch": 0.18744113029827317,
+      "grad_norm": 2.3128407690400814,
+      "learning_rate": 1.8728353356938386e-05,
+      "loss": 0.9191,
       "step": 1791
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.602861230887055e-06,
-      "loss": 0.73,
+      "epoch": 0.18754578754578755,
+      "grad_norm": 2.3372349993067782,
+      "learning_rate": 1.872669862179623e-05,
+      "loss": 1.0454,
       "step": 1792
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.592509591579373e-06,
-      "loss": 0.8007,
+      "epoch": 0.18765044479330192,
+      "grad_norm": 2.4025198105450607,
+      "learning_rate": 1.8725042883938714e-05,
+      "loss": 1.0697,
       "step": 1793
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.582160690053194e-06,
-      "loss": 0.9258,
+      "epoch": 0.18775510204081633,
+      "grad_norm": 2.2588303051011196,
+      "learning_rate": 1.8723386143556085e-05,
+      "loss": 1.21,
       "step": 1794
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.571814538077216e-06,
-      "loss": 0.7619,
+      "epoch": 0.1878597592883307,
+      "grad_norm": 2.1758809138656834,
+      "learning_rate": 1.87217284008387e-05,
+      "loss": 1.017,
       "step": 1795
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.5614711474170155e-06,
-      "loss": 0.7193,
+      "epoch": 0.1879644165358451,
+      "grad_norm": 2.505013897983491,
+      "learning_rate": 1.872006965597704e-05,
+      "loss": 0.96,
       "step": 1796
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.551130529835016e-06,
-      "loss": 0.7072,
+      "epoch": 0.1880690737833595,
+      "grad_norm": 2.5795773446705303,
+      "learning_rate": 1.8718409909161704e-05,
+      "loss": 1.0457,
       "step": 1797
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.540792697090502e-06,
-      "loss": 0.6891,
+      "epoch": 0.1881737310308739,
+      "grad_norm": 2.5053247104374945,
+      "learning_rate": 1.8716749160583393e-05,
+      "loss": 1.0533,
       "step": 1798
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.530457660939588e-06,
-      "loss": 0.7172,
+      "epoch": 0.18827838827838828,
+      "grad_norm": 2.5122626396928633,
+      "learning_rate": 1.871508741043293e-05,
+      "loss": 0.9278,
       "step": 1799
     },
     {
-      "epoch": 1.18,
-      "learning_rate": 7.5201254331352056e-06,
-      "loss": 0.7301,
+      "epoch": 0.18838304552590268,
+      "grad_norm": 2.443610775616308,
+      "learning_rate": 1.871342465890126e-05,
+      "loss": 0.85,
       "step": 1800
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.509796025427088e-06,
-      "loss": 0.7413,
+      "epoch": 0.18848770277341706,
+      "grad_norm": 2.515901355227164,
+      "learning_rate": 1.8711760906179428e-05,
+      "loss": 1.0415,
       "step": 1801
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.49946944956177e-06,
-      "loss": 0.7004,
+      "epoch": 0.18859236002093144,
+      "grad_norm": 2.6696223601207714,
+      "learning_rate": 1.871009615245861e-05,
+      "loss": 1.0236,
       "step": 1802
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.489145717282559e-06,
-      "loss": 0.5519,
+      "epoch": 0.18869701726844584,
+      "grad_norm": 2.4784833366393553,
+      "learning_rate": 1.870843039793009e-05,
+      "loss": 0.9525,
       "step": 1803
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.478824840329537e-06,
-      "loss": 0.7054,
+      "epoch": 0.18880167451596022,
+      "grad_norm": 2.4542775584185264,
+      "learning_rate": 1.870676364278526e-05,
+      "loss": 1.1093,
       "step": 1804
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.4685068304395366e-06,
-      "loss": 0.8361,
+      "epoch": 0.18890633176347463,
+      "grad_norm": 2.1766923572797774,
+      "learning_rate": 1.8705095887215636e-05,
+      "loss": 0.9177,
       "step": 1805
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.45819169934612e-06,
-      "loss": 0.7112,
+      "epoch": 0.189010989010989,
+      "grad_norm": 2.486937565003481,
+      "learning_rate": 1.870342713141285e-05,
+      "loss": 1.0531,
       "step": 1806
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.4478794587795855e-06,
-      "loss": 0.813,
+      "epoch": 0.1891156462585034,
+      "grad_norm": 2.5489906996516605,
+      "learning_rate": 1.8701757375568642e-05,
+      "loss": 1.1089,
       "step": 1807
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.437570120466943e-06,
-      "loss": 0.7203,
+      "epoch": 0.1892203035060178,
+      "grad_norm": 1.940507425147104,
+      "learning_rate": 1.870008661987487e-05,
+      "loss": 0.8917,
       "step": 1808
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.427263696131903e-06,
-      "loss": 0.7893,
+      "epoch": 0.1893249607535322,
+      "grad_norm": 2.0972924062895464,
+      "learning_rate": 1.8698414864523512e-05,
+      "loss": 1.0538,
       "step": 1809
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.416960197494856e-06,
-      "loss": 0.7818,
+      "epoch": 0.18942961800104657,
+      "grad_norm": 2.2496046336500166,
+      "learning_rate": 1.869674210970665e-05,
+      "loss": 1.0835,
       "step": 1810
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.406659636272872e-06,
-      "loss": 0.6633,
+      "epoch": 0.18953427524856095,
+      "grad_norm": 2.5621615108753035,
+      "learning_rate": 1.869506835561649e-05,
+      "loss": 1.0739,
       "step": 1811
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.396362024179677e-06,
-      "loss": 0.7338,
+      "epoch": 0.18963893249607536,
+      "grad_norm": 2.440505472169808,
+      "learning_rate": 1.869339360244535e-05,
+      "loss": 1.0871,
       "step": 1812
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.386067372925647e-06,
-      "loss": 0.735,
+      "epoch": 0.18974358974358974,
+      "grad_norm": 2.403241452428988,
+      "learning_rate": 1.869171785038566e-05,
+      "loss": 0.9661,
       "step": 1813
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.375775694217788e-06,
-      "loss": 0.81,
+      "epoch": 0.18984824699110414,
+      "grad_norm": 2.469688713891638,
+      "learning_rate": 1.869004109962997e-05,
+      "loss": 1.0122,
       "step": 1814
     },
     {
-      "epoch": 1.19,
-      "learning_rate": 7.365486999759723e-06,
-      "loss": 0.7522,
+      "epoch": 0.18995290423861852,
+      "grad_norm": 2.333884875941877,
+      "learning_rate": 1.868836335037094e-05,
+      "loss": 1.1133,
       "step": 1815
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.355201301251689e-06,
-      "loss": 0.6648,
+      "epoch": 0.19005756148613293,
+      "grad_norm": 2.3621677518785376,
+      "learning_rate": 1.868668460280135e-05,
+      "loss": 0.8132,
       "step": 1816
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.344918610390506e-06,
-      "loss": 0.6333,
+      "epoch": 0.1901622187336473,
+      "grad_norm": 2.7072055929293235,
+      "learning_rate": 1.8685004857114085e-05,
+      "loss": 1.0276,
       "step": 1817
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.334638938869582e-06,
-      "loss": 0.68,
+      "epoch": 0.19026687598116168,
+      "grad_norm": 2.645013242609602,
+      "learning_rate": 1.8683324113502158e-05,
+      "loss": 1.0181,
       "step": 1818
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.324362298378893e-06,
-      "loss": 0.7931,
+      "epoch": 0.1903715332286761,
+      "grad_norm": 2.5368104479142692,
+      "learning_rate": 1.868164237215869e-05,
+      "loss": 0.919,
       "step": 1819
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.314088700604958e-06,
-      "loss": 0.7953,
+      "epoch": 0.19047619047619047,
+      "grad_norm": 1.9021982130849477,
+      "learning_rate": 1.867995963327691e-05,
+      "loss": 0.8938,
       "step": 1820
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.303818157230841e-06,
-      "loss": 0.6957,
+      "epoch": 0.19058084772370487,
+      "grad_norm": 2.33319178257821,
+      "learning_rate": 1.8678275897050177e-05,
+      "loss": 1.0466,
       "step": 1821
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.293550679936135e-06,
-      "loss": 0.7854,
+      "epoch": 0.19068550497121925,
+      "grad_norm": 2.7008377094227676,
+      "learning_rate": 1.8676591163671948e-05,
+      "loss": 1.1263,
       "step": 1822
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.283286280396948e-06,
-      "loss": 0.6751,
+      "epoch": 0.19079016221873366,
+      "grad_norm": 2.1650716424421073,
+      "learning_rate": 1.8674905433335805e-05,
+      "loss": 0.9968,
       "step": 1823
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.27302497028588e-06,
-      "loss": 0.7078,
+      "epoch": 0.19089481946624803,
+      "grad_norm": 2.398261119939507,
+      "learning_rate": 1.867321870623544e-05,
+      "loss": 1.1226,
       "step": 1824
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.262766761272022e-06,
-      "loss": 0.7975,
+      "epoch": 0.19099947671376244,
+      "grad_norm": 2.312460276704168,
+      "learning_rate": 1.8671530982564664e-05,
+      "loss": 0.9971,
       "step": 1825
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.252511665020939e-06,
-      "loss": 0.6619,
+      "epoch": 0.19110413396127682,
+      "grad_norm": 2.647523183773923,
+      "learning_rate": 1.86698422625174e-05,
+      "loss": 1.1726,
       "step": 1826
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.242259693194658e-06,
-      "loss": 0.7909,
+      "epoch": 0.1912087912087912,
+      "grad_norm": 2.6076487342169603,
+      "learning_rate": 1.8668152546287686e-05,
+      "loss": 1.0306,
       "step": 1827
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.232010857451653e-06,
-      "loss": 0.6512,
+      "epoch": 0.1913134484563056,
+      "grad_norm": 2.3115074361083883,
+      "learning_rate": 1.8666461834069672e-05,
+      "loss": 0.8432,
       "step": 1828
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.221765169446819e-06,
-      "loss": 0.748,
+      "epoch": 0.19141810570381998,
+      "grad_norm": 2.1836521898238197,
+      "learning_rate": 1.866477012605762e-05,
+      "loss": 1.0452,
       "step": 1829
     },
     {
-      "epoch": 1.2,
-      "learning_rate": 7.211522640831492e-06,
-      "loss": 0.6781,
+      "epoch": 0.19152276295133439,
+      "grad_norm": 2.330317681639118,
+      "learning_rate": 1.866307742244592e-05,
+      "loss": 1.0984,
       "step": 1830
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.2012832832534e-06,
-      "loss": 0.7517,
+      "epoch": 0.19162742019884876,
+      "grad_norm": 2.457940283244116,
+      "learning_rate": 1.8661383723429062e-05,
+      "loss": 1.108,
       "step": 1831
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.1910471083566725e-06,
-      "loss": 0.7194,
+      "epoch": 0.19173207744636317,
+      "grad_norm": 2.4031558334493757,
+      "learning_rate": 1.8659689029201654e-05,
+      "loss": 1.0354,
       "step": 1832
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.180814127781819e-06,
-      "loss": 0.658,
+      "epoch": 0.19183673469387755,
+      "grad_norm": 1.9518351072841722,
+      "learning_rate": 1.8657993339958422e-05,
+      "loss": 0.8381,
       "step": 1833
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.170584353165708e-06,
-      "loss": 0.7403,
+      "epoch": 0.19194139194139195,
+      "grad_norm": 2.140948339516903,
+      "learning_rate": 1.8656296655894205e-05,
+      "loss": 0.9547,
       "step": 1834
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.1603577961415704e-06,
-      "loss": 0.7236,
+      "epoch": 0.19204604918890633,
+      "grad_norm": 2.248185211385205,
+      "learning_rate": 1.865459897720395e-05,
+      "loss": 0.8359,
       "step": 1835
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.15013446833898e-06,
-      "loss": 0.8157,
+      "epoch": 0.1921507064364207,
+      "grad_norm": 1.9412091916767238,
+      "learning_rate": 1.865290030408273e-05,
+      "loss": 0.7927,
       "step": 1836
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.139914381383837e-06,
-      "loss": 0.7308,
+      "epoch": 0.19225536368393512,
+      "grad_norm": 2.7448554100886717,
+      "learning_rate": 1.8651200636725727e-05,
+      "loss": 1.1132,
       "step": 1837
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.129697546898344e-06,
-      "loss": 0.7194,
+      "epoch": 0.1923600209314495,
+      "grad_norm": 2.2079122058545533,
+      "learning_rate": 1.864949997532823e-05,
+      "loss": 0.9166,
       "step": 1838
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.119483976501021e-06,
-      "loss": 0.8648,
+      "epoch": 0.1924646781789639,
+      "grad_norm": 2.512086470901001,
+      "learning_rate": 1.864779832008565e-05,
+      "loss": 1.0589,
       "step": 1839
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.10927368180667e-06,
-      "loss": 0.6839,
+      "epoch": 0.19256933542647828,
+      "grad_norm": 2.232551449072093,
+      "learning_rate": 1.8646095671193512e-05,
+      "loss": 1.0556,
       "step": 1840
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.0990666744263646e-06,
-      "loss": 0.7048,
+      "epoch": 0.19267399267399268,
+      "grad_norm": 2.5120233437446577,
+      "learning_rate": 1.8644392028847457e-05,
+      "loss": 1.0832,
       "step": 1841
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.0888629659674525e-06,
-      "loss": 0.695,
+      "epoch": 0.19277864992150706,
+      "grad_norm": 2.4799212636866454,
+      "learning_rate": 1.8642687393243233e-05,
+      "loss": 0.9932,
       "step": 1842
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.07866256803351e-06,
-      "loss": 0.7182,
+      "epoch": 0.19288330716902147,
+      "grad_norm": 2.0756667968361073,
+      "learning_rate": 1.8640981764576707e-05,
+      "loss": 1.0258,
       "step": 1843
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.068465492224361e-06,
-      "loss": 0.6605,
+      "epoch": 0.19298796441653585,
+      "grad_norm": 2.570142389257925,
+      "learning_rate": 1.863927514304386e-05,
+      "loss": 1.179,
       "step": 1844
     },
     {
-      "epoch": 1.21,
-      "learning_rate": 7.058271750136054e-06,
-      "loss": 0.725,
+      "epoch": 0.19309262166405022,
+      "grad_norm": 2.3453819079115314,
+      "learning_rate": 1.8637567528840784e-05,
+      "loss": 0.9876,
       "step": 1845
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 7.04808135336084e-06,
-      "loss": 0.7677,
+      "epoch": 0.19319727891156463,
+      "grad_norm": 2.1997745972557015,
+      "learning_rate": 1.863585892216369e-05,
+      "loss": 0.9367,
       "step": 1846
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 7.03789431348717e-06,
-      "loss": 0.6982,
+      "epoch": 0.193301936159079,
+      "grad_norm": 2.2836486587979543,
+      "learning_rate": 1.8634149323208896e-05,
+      "loss": 1.0386,
       "step": 1847
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 7.027710642099672e-06,
-      "loss": 0.6249,
+      "epoch": 0.1934065934065934,
+      "grad_norm": 2.21044201206823,
+      "learning_rate": 1.8632438732172846e-05,
+      "loss": 1.0342,
       "step": 1848
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 7.017530350779148e-06,
-      "loss": 0.8421,
+      "epoch": 0.1935112506541078,
+      "grad_norm": 2.9097569831352352,
+      "learning_rate": 1.8630727149252082e-05,
+      "loss": 0.8409,
       "step": 1849
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 7.007353451102557e-06,
-      "loss": 0.8239,
+      "epoch": 0.1936159079016222,
+      "grad_norm": 2.498032621838584,
+      "learning_rate": 1.8629014574643275e-05,
+      "loss": 1.032,
       "step": 1850
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 6.997179954642998e-06,
-      "loss": 0.6682,
+      "epoch": 0.19372056514913658,
+      "grad_norm": 2.3432769230009196,
+      "learning_rate": 1.8627301008543198e-05,
+      "loss": 1.0969,
       "step": 1851
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 6.9870098729697015e-06,
-      "loss": 0.7499,
+      "epoch": 0.19382522239665098,
+      "grad_norm": 2.241916204274371,
+      "learning_rate": 1.8625586451148747e-05,
+      "loss": 0.8316,
       "step": 1852
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 6.976843217648013e-06,
-      "loss": 0.6502,
+      "epoch": 0.19392987964416536,
+      "grad_norm": 2.5512406560898055,
+      "learning_rate": 1.8623870902656928e-05,
+      "loss": 1.0418,
       "step": 1853
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 6.966680000239384e-06,
-      "loss": 0.6084,
+      "epoch": 0.19403453689167974,
+      "grad_norm": 2.3680570078020233,
+      "learning_rate": 1.862215436326486e-05,
+      "loss": 0.8752,
       "step": 1854
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 6.956520232301355e-06,
-      "loss": 0.762,
+      "epoch": 0.19413919413919414,
+      "grad_norm": 2.416117696667423,
+      "learning_rate": 1.8620436833169773e-05,
+      "loss": 0.972,
       "step": 1855
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 6.946363925387546e-06,
-      "loss": 0.7446,
+      "epoch": 0.19424385138670852,
+      "grad_norm": 2.2006979425801783,
+      "learning_rate": 1.861871831256902e-05,
+      "loss": 1.0598,
       "step": 1856
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 6.936211091047636e-06,
-      "loss": 0.8434,
+      "epoch": 0.19434850863422293,
+      "grad_norm": 2.1767343281841884,
+      "learning_rate": 1.861699880166006e-05,
+      "loss": 0.9034,
       "step": 1857
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 6.926061740827359e-06,
-      "loss": 0.8147,
+      "epoch": 0.1944531658817373,
+      "grad_norm": 2.4534072009756973,
+      "learning_rate": 1.8615278300640475e-05,
+      "loss": 0.9428,
       "step": 1858
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 6.915915886268486e-06,
-      "loss": 0.7941,
+      "epoch": 0.1945578231292517,
+      "grad_norm": 2.1602024149446355,
+      "learning_rate": 1.8613556809707943e-05,
+      "loss": 1.0906,
       "step": 1859
     },
     {
-      "epoch": 1.22,
-      "learning_rate": 6.905773538908814e-06,
-      "loss": 0.7455,
+      "epoch": 0.1946624803767661,
+      "grad_norm": 2.169341746326727,
+      "learning_rate": 1.8611834329060272e-05,
+      "loss": 0.9131,
       "step": 1860
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.895634710282153e-06,
-      "loss": 0.8107,
+      "epoch": 0.1947671376242805,
+      "grad_norm": 2.9643059788895476,
+      "learning_rate": 1.8610110858895383e-05,
+      "loss": 0.9651,
       "step": 1861
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.885499411918304e-06,
-      "loss": 0.7882,
+      "epoch": 0.19487179487179487,
+      "grad_norm": 2.213393165596908,
+      "learning_rate": 1.8608386399411296e-05,
+      "loss": 1.1012,
       "step": 1862
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.8753676553430606e-06,
-      "loss": 0.8008,
+      "epoch": 0.19497645211930925,
+      "grad_norm": 2.530920660576334,
+      "learning_rate": 1.8606660950806165e-05,
+      "loss": 1.0807,
       "step": 1863
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.8652394520781865e-06,
-      "loss": 0.7703,
+      "epoch": 0.19508110936682366,
+      "grad_norm": 2.5971872408801744,
+      "learning_rate": 1.860493451327824e-05,
+      "loss": 1.0592,
       "step": 1864
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.855114813641408e-06,
-      "loss": 0.8034,
+      "epoch": 0.19518576661433804,
+      "grad_norm": 2.9737379735677725,
+      "learning_rate": 1.8603207087025897e-05,
+      "loss": 1.0376,
       "step": 1865
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.844993751546388e-06,
-      "loss": 0.7733,
+      "epoch": 0.19529042386185244,
+      "grad_norm": 2.3943885247243006,
+      "learning_rate": 1.8601478672247622e-05,
+      "loss": 0.8863,
       "step": 1866
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.8348762773027285e-06,
-      "loss": 0.6668,
+      "epoch": 0.19539508110936682,
+      "grad_norm": 2.4125949201995787,
+      "learning_rate": 1.8599749269142007e-05,
+      "loss": 1.0162,
       "step": 1867
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.824762402415958e-06,
-      "loss": 0.8248,
+      "epoch": 0.19549973835688123,
+      "grad_norm": 2.3700879805758883,
+      "learning_rate": 1.859801887790777e-05,
+      "loss": 0.9566,
       "step": 1868
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.814652138387501e-06,
-      "loss": 0.7283,
+      "epoch": 0.1956043956043956,
+      "grad_norm": 2.17186126800691,
+      "learning_rate": 1.859628749874373e-05,
+      "loss": 1.0475,
       "step": 1869
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.8045454967146854e-06,
-      "loss": 0.7483,
+      "epoch": 0.19570905285190998,
+      "grad_norm": 2.0282801776801715,
+      "learning_rate": 1.8594555131848834e-05,
+      "loss": 0.9321,
       "step": 1870
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.794442488890705e-06,
-      "loss": 0.7233,
+      "epoch": 0.1958137100994244,
+      "grad_norm": 2.4375974168550423,
+      "learning_rate": 1.8592821777422126e-05,
+      "loss": 0.9614,
       "step": 1871
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.784343126404636e-06,
-      "loss": 0.8548,
+      "epoch": 0.19591836734693877,
+      "grad_norm": 2.2740457885998624,
+      "learning_rate": 1.859108743566278e-05,
+      "loss": 1.0135,
       "step": 1872
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.774247420741401e-06,
-      "loss": 0.7449,
+      "epoch": 0.19602302459445317,
+      "grad_norm": 2.417294338783456,
+      "learning_rate": 1.8589352106770072e-05,
+      "loss": 0.9173,
       "step": 1873
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.764155383381772e-06,
-      "loss": 0.8191,
+      "epoch": 0.19612768184196755,
+      "grad_norm": 2.2618143493988363,
+      "learning_rate": 1.858761579094339e-05,
+      "loss": 0.8704,
       "step": 1874
     },
     {
-      "epoch": 1.23,
-      "learning_rate": 6.754067025802346e-06,
-      "loss": 0.7457,
+      "epoch": 0.19623233908948196,
+      "grad_norm": 2.3808284708358363,
+      "learning_rate": 1.858587848838225e-05,
+      "loss": 1.0471,
       "step": 1875
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.743982359475525e-06,
-      "loss": 0.647,
+      "epoch": 0.19633699633699633,
+      "grad_norm": 2.210022086562613,
+      "learning_rate": 1.8584140199286263e-05,
+      "loss": 0.9631,
       "step": 1876
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.733901395869527e-06,
-      "loss": 0.8944,
+      "epoch": 0.19644165358451074,
+      "grad_norm": 2.3840130902341916,
+      "learning_rate": 1.8582400923855166e-05,
+      "loss": 1.045,
       "step": 1877
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.723824146448352e-06,
-      "loss": 0.695,
+      "epoch": 0.19654631083202512,
+      "grad_norm": 2.606315391047431,
+      "learning_rate": 1.8580660662288807e-05,
+      "loss": 1.0785,
       "step": 1878
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.713750622671784e-06,
-      "loss": 0.596,
+      "epoch": 0.1966509680795395,
+      "grad_norm": 2.6364699756222194,
+      "learning_rate": 1.857891941478714e-05,
+      "loss": 0.9784,
       "step": 1879
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.7036808359953585e-06,
-      "loss": 0.5971,
+      "epoch": 0.1967556253270539,
+      "grad_norm": 2.072482657011719,
+      "learning_rate": 1.8577177181550243e-05,
+      "loss": 0.8879,
       "step": 1880
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.693614797870371e-06,
-      "loss": 0.6719,
+      "epoch": 0.19686028257456828,
+      "grad_norm": 2.2034184725027237,
+      "learning_rate": 1.8575433962778297e-05,
+      "loss": 1.0744,
       "step": 1881
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.68355251974385e-06,
-      "loss": 0.7233,
+      "epoch": 0.19696493982208269,
+      "grad_norm": 2.5086871489753526,
+      "learning_rate": 1.8573689758671607e-05,
+      "loss": 1.0613,
       "step": 1882
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.673494013058548e-06,
-      "loss": 0.7284,
+      "epoch": 0.19706959706959706,
+      "grad_norm": 2.1761373108687416,
+      "learning_rate": 1.8571944569430582e-05,
+      "loss": 1.0279,
       "step": 1883
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.663439289252932e-06,
-      "loss": 0.8362,
+      "epoch": 0.19717425431711147,
+      "grad_norm": 2.3377334894353488,
+      "learning_rate": 1.857019839525575e-05,
+      "loss": 1.0278,
       "step": 1884
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.653388359761161e-06,
-      "loss": 0.8038,
+      "epoch": 0.19727891156462585,
+      "grad_norm": 2.4580887107257685,
+      "learning_rate": 1.856845123634775e-05,
+      "loss": 1.019,
       "step": 1885
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.6433412360130855e-06,
-      "loss": 0.7687,
+      "epoch": 0.19738356881214025,
+      "grad_norm": 2.5860343029123913,
+      "learning_rate": 1.856670309290733e-05,
+      "loss": 1.0096,
       "step": 1886
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.633297929434224e-06,
-      "loss": 0.6707,
+      "epoch": 0.19748822605965463,
+      "grad_norm": 2.960738910419292,
+      "learning_rate": 1.8564953965135358e-05,
+      "loss": 0.9591,
       "step": 1887
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.623258451445758e-06,
-      "loss": 0.7155,
+      "epoch": 0.197592883307169,
+      "grad_norm": 2.3795107266092264,
+      "learning_rate": 1.8563203853232812e-05,
+      "loss": 0.8907,
       "step": 1888
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.613222813464508e-06,
-      "loss": 0.6336,
+      "epoch": 0.19769754055468342,
+      "grad_norm": 2.2435840544300336,
+      "learning_rate": 1.8561452757400785e-05,
+      "loss": 0.9413,
       "step": 1889
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.603191026902936e-06,
-      "loss": 0.7285,
+      "epoch": 0.1978021978021978,
+      "grad_norm": 2.068348693567875,
+      "learning_rate": 1.8559700677840482e-05,
+      "loss": 1.1047,
       "step": 1890
     },
     {
-      "epoch": 1.24,
-      "learning_rate": 6.59316310316912e-06,
-      "loss": 0.6758,
+      "epoch": 0.1979068550497122,
+      "grad_norm": 2.114604247450884,
+      "learning_rate": 1.855794761475322e-05,
+      "loss": 1.0478,
       "step": 1891
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.583139053666745e-06,
-      "loss": 0.7682,
+      "epoch": 0.19801151229722658,
+      "grad_norm": 2.3715850812930466,
+      "learning_rate": 1.8556193568340423e-05,
+      "loss": 1.0699,
       "step": 1892
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.573118889795093e-06,
-      "loss": 0.6451,
+      "epoch": 0.19811616954474098,
+      "grad_norm": 2.037660362224576,
+      "learning_rate": 1.855443853880364e-05,
+      "loss": 0.9552,
       "step": 1893
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.563102622949021e-06,
-      "loss": 0.733,
+      "epoch": 0.19822082679225536,
+      "grad_norm": 2.472478303294875,
+      "learning_rate": 1.8552682526344532e-05,
+      "loss": 1.0812,
       "step": 1894
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.553090264518962e-06,
-      "loss": 0.7325,
+      "epoch": 0.19832548403976977,
+      "grad_norm": 2.174274765095585,
+      "learning_rate": 1.855092553116486e-05,
+      "loss": 1.0514,
       "step": 1895
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.543081825890901e-06,
-      "loss": 0.7829,
+      "epoch": 0.19843014128728415,
+      "grad_norm": 2.418758882311152,
+      "learning_rate": 1.8549167553466515e-05,
+      "loss": 1.1054,
       "step": 1896
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.533077318446365e-06,
-      "loss": 0.7278,
+      "epoch": 0.19853479853479852,
+      "grad_norm": 2.6391811792480926,
+      "learning_rate": 1.8547408593451483e-05,
+      "loss": 0.8834,
       "step": 1897
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.523076753562411e-06,
-      "loss": 0.7916,
+      "epoch": 0.19863945578231293,
+      "grad_norm": 2.362589627641238,
+      "learning_rate": 1.8545648651321874e-05,
+      "loss": 1.049,
       "step": 1898
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.513080142611612e-06,
-      "loss": 0.7904,
+      "epoch": 0.1987441130298273,
+      "grad_norm": 3.6366633137355113,
+      "learning_rate": 1.8543887727279915e-05,
+      "loss": 0.8768,
       "step": 1899
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.503087496962044e-06,
-      "loss": 0.6089,
+      "epoch": 0.1988487702773417,
+      "grad_norm": 2.5316763474698423,
+      "learning_rate": 1.8542125821527933e-05,
+      "loss": 0.95,
       "step": 1900
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.493098827977278e-06,
-      "loss": 0.7025,
+      "epoch": 0.1989534275248561,
+      "grad_norm": 2.4002232910837322,
+      "learning_rate": 1.8540362934268378e-05,
+      "loss": 1.1223,
       "step": 1901
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.483114147016358e-06,
-      "loss": 0.7289,
+      "epoch": 0.1990580847723705,
+      "grad_norm": 2.4136377650468477,
+      "learning_rate": 1.853859906570381e-05,
+      "loss": 1.0736,
       "step": 1902
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.473133465433793e-06,
-      "loss": 0.754,
+      "epoch": 0.19916274201988488,
+      "grad_norm": 2.5540511046453362,
+      "learning_rate": 1.85368342160369e-05,
+      "loss": 1.1074,
       "step": 1903
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.463156794579543e-06,
-      "loss": 0.7923,
+      "epoch": 0.19926739926739928,
+      "grad_norm": 2.2183989268155706,
+      "learning_rate": 1.8535068385470434e-05,
+      "loss": 0.9569,
       "step": 1904
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.453184145799015e-06,
-      "loss": 0.7206,
+      "epoch": 0.19937205651491366,
+      "grad_norm": 2.2251618210625463,
+      "learning_rate": 1.8533301574207305e-05,
+      "loss": 1.024,
       "step": 1905
     },
     {
-      "epoch": 1.25,
-      "learning_rate": 6.4432155304330335e-06,
-      "loss": 0.6077,
+      "epoch": 0.19947671376242804,
+      "grad_norm": 2.4725959487205005,
+      "learning_rate": 1.853153378245053e-05,
+      "loss": 0.9864,
       "step": 1906
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.433250959817845e-06,
-      "loss": 0.6976,
+      "epoch": 0.19958137100994244,
+      "grad_norm": 2.426105671626237,
+      "learning_rate": 1.8529765010403226e-05,
+      "loss": 1.1186,
       "step": 1907
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.42329044528508e-06,
-      "loss": 0.7523,
+      "epoch": 0.19968602825745682,
+      "grad_norm": 2.2610224754205457,
+      "learning_rate": 1.852799525826863e-05,
+      "loss": 1.0378,
       "step": 1908
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.413333998161772e-06,
-      "loss": 0.6929,
+      "epoch": 0.19979068550497123,
+      "grad_norm": 2.1525597269244683,
+      "learning_rate": 1.852622452625009e-05,
+      "loss": 0.9537,
       "step": 1909
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.4033816297703245e-06,
-      "loss": 0.8903,
+      "epoch": 0.1998953427524856,
+      "grad_norm": 2.251421695737976,
+      "learning_rate": 1.8524452814551067e-05,
+      "loss": 1.0182,
       "step": 1910
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.393433351428502e-06,
-      "loss": 0.6549,
+      "epoch": 0.2,
+      "grad_norm": 2.4742262214471458,
+      "learning_rate": 1.852268012337514e-05,
+      "loss": 0.9118,
       "step": 1911
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.383489174449422e-06,
-      "loss": 0.6684,
+      "epoch": 0.2001046572475144,
+      "grad_norm": 2.351956722228832,
+      "learning_rate": 1.8520906452925983e-05,
+      "loss": 0.9906,
       "step": 1912
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.373549110141528e-06,
-      "loss": 0.6565,
+      "epoch": 0.20020931449502877,
+      "grad_norm": 2.5457326056054304,
+      "learning_rate": 1.8519131803407405e-05,
+      "loss": 0.9223,
       "step": 1913
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.363613169808593e-06,
-      "loss": 0.6944,
+      "epoch": 0.20031397174254317,
+      "grad_norm": 2.3125295104481087,
+      "learning_rate": 1.8517356175023312e-05,
+      "loss": 1.0705,
       "step": 1914
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.353681364749703e-06,
-      "loss": 0.7965,
+      "epoch": 0.20041862899005755,
+      "grad_norm": 2.4672166374713655,
+      "learning_rate": 1.8515579567977728e-05,
+      "loss": 1.003,
       "step": 1915
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.343753706259239e-06,
-      "loss": 0.7501,
+      "epoch": 0.20052328623757196,
+      "grad_norm": 2.1521918308626193,
+      "learning_rate": 1.8513801982474788e-05,
+      "loss": 0.9784,
       "step": 1916
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.333830205626861e-06,
-      "loss": 0.7404,
+      "epoch": 0.20062794348508634,
+      "grad_norm": 2.4414899299046295,
+      "learning_rate": 1.8512023418718745e-05,
+      "loss": 1.0942,
       "step": 1917
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.323910874137509e-06,
-      "loss": 0.7629,
+      "epoch": 0.20073260073260074,
+      "grad_norm": 2.3650808987908105,
+      "learning_rate": 1.851024387691395e-05,
+      "loss": 1.047,
       "step": 1918
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.313995723071377e-06,
-      "loss": 0.8053,
+      "epoch": 0.20083725798011512,
+      "grad_norm": 2.5314876863007703,
+      "learning_rate": 1.8508463357264883e-05,
+      "loss": 1.0066,
       "step": 1919
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.3040847637039075e-06,
-      "loss": 0.6441,
+      "epoch": 0.20094191522762953,
+      "grad_norm": 2.0756056456744156,
+      "learning_rate": 1.8506681859976127e-05,
+      "loss": 0.8131,
       "step": 1920
     },
     {
-      "epoch": 1.26,
-      "learning_rate": 6.294178007305776e-06,
-      "loss": 0.718,
+      "epoch": 0.2010465724751439,
+      "grad_norm": 2.4985838715578894,
+      "learning_rate": 1.850489938525238e-05,
+      "loss": 0.9835,
       "step": 1921
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.284275465142875e-06,
-      "loss": 0.7312,
+      "epoch": 0.20115122972265828,
+      "grad_norm": 2.272801695052019,
+      "learning_rate": 1.8503115933298455e-05,
+      "loss": 1.0649,
       "step": 1922
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.27437714847631e-06,
-      "loss": 0.7502,
+      "epoch": 0.2012558869701727,
+      "grad_norm": 2.0961137864615265,
+      "learning_rate": 1.850133150431927e-05,
+      "loss": 1.0169,
       "step": 1923
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.2644830685623756e-06,
-      "loss": 0.8266,
+      "epoch": 0.20136054421768707,
+      "grad_norm": 2.2133469820080234,
+      "learning_rate": 1.8499546098519863e-05,
+      "loss": 0.7855,
       "step": 1924
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.254593236652554e-06,
-      "loss": 0.7382,
+      "epoch": 0.20146520146520147,
+      "grad_norm": 2.272863601676403,
+      "learning_rate": 1.8497759716105376e-05,
+      "loss": 1.1192,
       "step": 1925
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.244707663993494e-06,
-      "loss": 0.7314,
+      "epoch": 0.20156985871271585,
+      "grad_norm": 2.3222789724527915,
+      "learning_rate": 1.8495972357281073e-05,
+      "loss": 1.1352,
       "step": 1926
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.2348263618269975e-06,
-      "loss": 0.7311,
+      "epoch": 0.20167451596023026,
+      "grad_norm": 2.3537538388981933,
+      "learning_rate": 1.8494184022252324e-05,
+      "loss": 1.0501,
       "step": 1927
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.224949341390017e-06,
-      "loss": 0.7672,
+      "epoch": 0.20177917320774463,
+      "grad_norm": 2.7018467868355485,
+      "learning_rate": 1.849239471122461e-05,
+      "loss": 1.1058,
       "step": 1928
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.215076613914628e-06,
-      "loss": 0.7134,
+      "epoch": 0.20188383045525904,
+      "grad_norm": 2.004307635291416,
+      "learning_rate": 1.8490604424403527e-05,
+      "loss": 0.874,
       "step": 1929
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.205208190628034e-06,
-      "loss": 0.7885,
+      "epoch": 0.20198848770277342,
+      "grad_norm": 2.096808325868463,
+      "learning_rate": 1.8488813161994784e-05,
+      "loss": 1.0896,
       "step": 1930
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.1953440827525306e-06,
-      "loss": 0.7421,
+      "epoch": 0.2020931449502878,
+      "grad_norm": 2.0352763262829705,
+      "learning_rate": 1.84870209242042e-05,
+      "loss": 0.8975,
       "step": 1931
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.185484301505518e-06,
-      "loss": 0.7605,
+      "epoch": 0.2021978021978022,
+      "grad_norm": 2.6788994218915705,
+      "learning_rate": 1.8485227711237707e-05,
+      "loss": 1.0147,
       "step": 1932
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.175628858099471e-06,
-      "loss": 0.7185,
+      "epoch": 0.20230245944531658,
+      "grad_norm": 2.462010742069801,
+      "learning_rate": 1.848343352330135e-05,
+      "loss": 1.1053,
       "step": 1933
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.165777763741932e-06,
-      "loss": 0.7015,
+      "epoch": 0.20240711669283099,
+      "grad_norm": 2.50095298478946,
+      "learning_rate": 1.848163836060128e-05,
+      "loss": 1.0792,
       "step": 1934
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.1559310296355005e-06,
-      "loss": 0.6552,
+      "epoch": 0.20251177394034536,
+      "grad_norm": 2.49507265553553,
+      "learning_rate": 1.847984222334377e-05,
+      "loss": 0.9941,
       "step": 1935
     },
     {
-      "epoch": 1.27,
-      "learning_rate": 6.146088666977804e-06,
-      "loss": 0.7147,
+      "epoch": 0.20261643118785977,
+      "grad_norm": 2.669346708093879,
+      "learning_rate": 1.84780451117352e-05,
+      "loss": 1.1946,
       "step": 1936
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.1362506869615205e-06,
-      "loss": 0.6034,
+      "epoch": 0.20272108843537415,
+      "grad_norm": 3.0351707992452344,
+      "learning_rate": 1.8476247025982058e-05,
+      "loss": 1.0224,
       "step": 1937
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.126417100774329e-06,
-      "loss": 0.6316,
+      "epoch": 0.20282574568288855,
+      "grad_norm": 2.1301304657947058,
+      "learning_rate": 1.847444796629095e-05,
+      "loss": 0.9222,
       "step": 1938
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.116587919598915e-06,
-      "loss": 0.6678,
+      "epoch": 0.20293040293040293,
+      "grad_norm": 2.1876447064810467,
+      "learning_rate": 1.847264793286859e-05,
+      "loss": 0.9753,
       "step": 1939
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.106763154612963e-06,
-      "loss": 0.6918,
+      "epoch": 0.2030350601779173,
+      "grad_norm": 2.2563332049617006,
+      "learning_rate": 1.8470846925921807e-05,
+      "loss": 1.0181,
       "step": 1940
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.096942816989116e-06,
-      "loss": 0.7406,
+      "epoch": 0.20313971742543172,
+      "grad_norm": 2.3267222100748826,
+      "learning_rate": 1.8469044945657543e-05,
+      "loss": 1.015,
       "step": 1941
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.0871269178949986e-06,
-      "loss": 0.7056,
+      "epoch": 0.2032443746729461,
+      "grad_norm": 2.441159690044786,
+      "learning_rate": 1.8467241992282842e-05,
+      "loss": 1.1735,
       "step": 1942
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.0773154684931875e-06,
-      "loss": 0.6214,
+      "epoch": 0.2033490319204605,
+      "grad_norm": 2.1163326570343806,
+      "learning_rate": 1.8465438066004875e-05,
+      "loss": 1.0618,
       "step": 1943
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.0675084799411975e-06,
-      "loss": 0.648,
+      "epoch": 0.20345368916797488,
+      "grad_norm": 2.117403160909713,
+      "learning_rate": 1.846363316703091e-05,
+      "loss": 1.0112,
       "step": 1944
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.057705963391461e-06,
-      "loss": 0.6987,
+      "epoch": 0.20355834641548928,
+      "grad_norm": 2.597051152568622,
+      "learning_rate": 1.846182729556834e-05,
+      "loss": 0.9412,
       "step": 1945
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.047907929991333e-06,
-      "loss": 0.7254,
+      "epoch": 0.20366300366300366,
+      "grad_norm": 2.3342553808450024,
+      "learning_rate": 1.8460020451824658e-05,
+      "loss": 0.8846,
       "step": 1946
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.038114390883073e-06,
-      "loss": 0.7628,
+      "epoch": 0.20376766091051807,
+      "grad_norm": 2.493805789843552,
+      "learning_rate": 1.8458212636007474e-05,
+      "loss": 1.0156,
       "step": 1947
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.028325357203822e-06,
-      "loss": 0.6533,
+      "epoch": 0.20387231815803245,
+      "grad_norm": 2.597283250350886,
+      "learning_rate": 1.8456403848324513e-05,
+      "loss": 1.0745,
       "step": 1948
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.018540840085611e-06,
-      "loss": 0.9035,
+      "epoch": 0.20397697540554682,
+      "grad_norm": 2.178100429523685,
+      "learning_rate": 1.8454594088983608e-05,
+      "loss": 1.0338,
       "step": 1949
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 6.008760850655316e-06,
-      "loss": 0.7052,
+      "epoch": 0.20408163265306123,
+      "grad_norm": 2.627673207192234,
+      "learning_rate": 1.8452783358192697e-05,
+      "loss": 1.1234,
       "step": 1950
     },
     {
-      "epoch": 1.28,
-      "learning_rate": 5.998985400034676e-06,
-      "loss": 0.6983,
+      "epoch": 0.2041862899005756,
+      "grad_norm": 2.6660377019985315,
+      "learning_rate": 1.845097165615985e-05,
+      "loss": 1.0812,
       "step": 1951
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.989214499340267e-06,
-      "loss": 0.6291,
+      "epoch": 0.20429094714809,
+      "grad_norm": 2.4399501701622492,
+      "learning_rate": 1.8449158983093226e-05,
+      "loss": 1.1253,
       "step": 1952
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.979448159683491e-06,
-      "loss": 0.7766,
+      "epoch": 0.2043956043956044,
+      "grad_norm": 2.078040777525875,
+      "learning_rate": 1.8447345339201105e-05,
+      "loss": 0.9986,
       "step": 1953
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.96968639217056e-06,
-      "loss": 0.6956,
+      "epoch": 0.2045002616431188,
+      "grad_norm": 2.344622782021923,
+      "learning_rate": 1.844553072469188e-05,
+      "loss": 1.0179,
       "step": 1954
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.959929207902488e-06,
-      "loss": 0.742,
+      "epoch": 0.20460491889063318,
+      "grad_norm": 2.3007348240880807,
+      "learning_rate": 1.8443715139774055e-05,
+      "loss": 1.0646,
       "step": 1955
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.950176617975079e-06,
-      "loss": 0.7034,
+      "epoch": 0.20470957613814755,
+      "grad_norm": 2.281518769125784,
+      "learning_rate": 1.844189858465624e-05,
+      "loss": 1.0593,
       "step": 1956
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.9404286334789115e-06,
-      "loss": 0.6267,
+      "epoch": 0.20481423338566196,
+      "grad_norm": 2.4164773491310765,
+      "learning_rate": 1.844008105954717e-05,
+      "loss": 1.078,
       "step": 1957
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.930685265499329e-06,
-      "loss": 0.589,
+      "epoch": 0.20491889063317634,
+      "grad_norm": 2.3330463609538232,
+      "learning_rate": 1.843826256465567e-05,
+      "loss": 1.0492,
       "step": 1958
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.9209465251164175e-06,
-      "loss": 0.8011,
+      "epoch": 0.20502354788069074,
+      "grad_norm": 2.2742931414383167,
+      "learning_rate": 1.8436443100190702e-05,
+      "loss": 0.9409,
       "step": 1959
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.911212423405008e-06,
-      "loss": 0.6309,
+      "epoch": 0.20512820512820512,
+      "grad_norm": 4.341203039506652,
+      "learning_rate": 1.8434622666361316e-05,
+      "loss": 1.0529,
       "step": 1960
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.901482971434652e-06,
-      "loss": 0.6323,
+      "epoch": 0.20523286237571953,
+      "grad_norm": 2.293281560740709,
+      "learning_rate": 1.8432801263376686e-05,
+      "loss": 1.1591,
       "step": 1961
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.891758180269618e-06,
-      "loss": 0.71,
+      "epoch": 0.2053375196232339,
+      "grad_norm": 2.0799628461136614,
+      "learning_rate": 1.8430978891446097e-05,
+      "loss": 1.009,
       "step": 1962
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.882038060968871e-06,
-      "loss": 0.6835,
+      "epoch": 0.2054421768707483,
+      "grad_norm": 2.34597923911034,
+      "learning_rate": 1.8429155550778942e-05,
+      "loss": 1.0635,
       "step": 1963
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.872322624586061e-06,
-      "loss": 0.7589,
+      "epoch": 0.2055468341182627,
+      "grad_norm": 2.525872889513285,
+      "learning_rate": 1.8427331241584724e-05,
+      "loss": 0.9838,
       "step": 1964
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.862611882169518e-06,
-      "loss": 0.7705,
+      "epoch": 0.20565149136577707,
+      "grad_norm": 2.178144919085616,
+      "learning_rate": 1.8425505964073065e-05,
+      "loss": 1.0336,
       "step": 1965
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.852905844762229e-06,
-      "loss": 0.7749,
+      "epoch": 0.20575614861329147,
+      "grad_norm": 2.8651481979144036,
+      "learning_rate": 1.8423679718453692e-05,
+      "loss": 1.1158,
       "step": 1966
     },
     {
-      "epoch": 1.29,
-      "learning_rate": 5.8432045234018365e-06,
-      "loss": 0.7817,
+      "epoch": 0.20586080586080585,
+      "grad_norm": 2.555904916520759,
+      "learning_rate": 1.842185250493644e-05,
+      "loss": 1.0651,
       "step": 1967
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.833507929120612e-06,
-      "loss": 0.6737,
+      "epoch": 0.20596546310832026,
+      "grad_norm": 2.140596234465768,
+      "learning_rate": 1.842002432373126e-05,
+      "loss": 0.8645,
       "step": 1968
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.823816072945458e-06,
-      "loss": 0.6447,
+      "epoch": 0.20607012035583464,
+      "grad_norm": 2.089196272187266,
+      "learning_rate": 1.8418195175048217e-05,
+      "loss": 1.0134,
       "step": 1969
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.814128965897887e-06,
-      "loss": 0.7379,
+      "epoch": 0.20617477760334904,
+      "grad_norm": 2.6661677515756597,
+      "learning_rate": 1.841636505909749e-05,
+      "loss": 1.0963,
       "step": 1970
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.804446618994009e-06,
-      "loss": 0.613,
+      "epoch": 0.20627943485086342,
+      "grad_norm": 2.023006671314738,
+      "learning_rate": 1.841453397608935e-05,
+      "loss": 0.9889,
       "step": 1971
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.794769043244527e-06,
-      "loss": 0.5713,
+      "epoch": 0.20638409209837782,
+      "grad_norm": 2.4607506296207267,
+      "learning_rate": 1.8412701926234197e-05,
+      "loss": 0.9246,
       "step": 1972
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.785096249654705e-06,
-      "loss": 0.8384,
+      "epoch": 0.2064887493458922,
+      "grad_norm": 2.0748117406848228,
+      "learning_rate": 1.841086890974254e-05,
+      "loss": 1.0495,
       "step": 1973
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.775428249224382e-06,
-      "loss": 0.7849,
+      "epoch": 0.20659340659340658,
+      "grad_norm": 2.293516807467432,
+      "learning_rate": 1.8409034926824995e-05,
+      "loss": 1.0593,
       "step": 1974
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.765765052947941e-06,
-      "loss": 0.9222,
+      "epoch": 0.206698063840921,
+      "grad_norm": 2.2095392100107003,
+      "learning_rate": 1.8407199977692292e-05,
+      "loss": 0.9036,
       "step": 1975
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.756106671814301e-06,
-      "loss": 0.8544,
+      "epoch": 0.20680272108843537,
+      "grad_norm": 2.0148849689379587,
+      "learning_rate": 1.8405364062555263e-05,
+      "loss": 0.9078,
       "step": 1976
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.7464531168069096e-06,
-      "loss": 0.7115,
+      "epoch": 0.20690737833594977,
+      "grad_norm": 2.3688402106868907,
+      "learning_rate": 1.840352718162487e-05,
+      "loss": 1.0614,
       "step": 1977
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.7368043989037206e-06,
-      "loss": 0.7258,
+      "epoch": 0.20701203558346415,
+      "grad_norm": 2.316317970644639,
+      "learning_rate": 1.8401689335112164e-05,
+      "loss": 0.9362,
       "step": 1978
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.7271605290771895e-06,
-      "loss": 0.7261,
+      "epoch": 0.20711669283097855,
+      "grad_norm": 2.16166387910284,
+      "learning_rate": 1.8399850523228325e-05,
+      "loss": 0.9942,
       "step": 1979
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.717521518294259e-06,
-      "loss": 0.6235,
+      "epoch": 0.20722135007849293,
+      "grad_norm": 2.6063947890703902,
+      "learning_rate": 1.8398010746184628e-05,
+      "loss": 1.1124,
       "step": 1980
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.70788737751635e-06,
-      "loss": 0.6413,
+      "epoch": 0.20732600732600734,
+      "grad_norm": 2.47106442178898,
+      "learning_rate": 1.8396170004192474e-05,
+      "loss": 1.1101,
       "step": 1981
     },
     {
-      "epoch": 1.3,
-      "learning_rate": 5.698258117699334e-06,
-      "loss": 0.7574,
+      "epoch": 0.20743066457352172,
+      "grad_norm": 2.2549560321044875,
+      "learning_rate": 1.8394328297463366e-05,
+      "loss": 1.0547,
       "step": 1982
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.68863374979354e-06,
-      "loss": 0.7668,
+      "epoch": 0.2075353218210361,
+      "grad_norm": 2.0122298157744796,
+      "learning_rate": 1.839248562620892e-05,
+      "loss": 0.8095,
       "step": 1983
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.6790142847437355e-06,
-      "loss": 0.7314,
+      "epoch": 0.2076399790685505,
+      "grad_norm": 2.1855154297200494,
+      "learning_rate": 1.839064199064086e-05,
+      "loss": 0.9771,
       "step": 1984
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.669399733489108e-06,
-      "loss": 0.7755,
+      "epoch": 0.20774463631606488,
+      "grad_norm": 1.9322206736673364,
+      "learning_rate": 1.8388797390971026e-05,
+      "loss": 0.8952,
       "step": 1985
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.659790106963261e-06,
-      "loss": 0.7053,
+      "epoch": 0.20784929356357928,
+      "grad_norm": 2.414532007703879,
+      "learning_rate": 1.838695182741137e-05,
+      "loss": 0.9121,
       "step": 1986
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.650185416094194e-06,
-      "loss": 0.6335,
+      "epoch": 0.20795395081109366,
+      "grad_norm": 2.186308171958817,
+      "learning_rate": 1.8385105300173943e-05,
+      "loss": 1.0025,
       "step": 1987
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.640585671804296e-06,
-      "loss": 0.6977,
+      "epoch": 0.20805860805860807,
+      "grad_norm": 2.2280850457274455,
+      "learning_rate": 1.838325780947092e-05,
+      "loss": 1.0759,
       "step": 1988
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.63099088501033e-06,
-      "loss": 0.7254,
+      "epoch": 0.20816326530612245,
+      "grad_norm": 2.2909334704230195,
+      "learning_rate": 1.838140935551458e-05,
+      "loss": 0.9594,
       "step": 1989
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.621401066623421e-06,
-      "loss": 0.75,
+      "epoch": 0.20826792255363685,
+      "grad_norm": 2.104015005202415,
+      "learning_rate": 1.8379559938517314e-05,
+      "loss": 0.8966,
       "step": 1990
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.611816227549048e-06,
-      "loss": 0.7286,
+      "epoch": 0.20837257980115123,
+      "grad_norm": 3.017180350499111,
+      "learning_rate": 1.8377709558691622e-05,
+      "loss": 1.0479,
       "step": 1991
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.602236378687018e-06,
-      "loss": 0.7389,
+      "epoch": 0.2084772370486656,
+      "grad_norm": 2.1719617305672356,
+      "learning_rate": 1.837585821625012e-05,
+      "loss": 1.131,
       "step": 1992
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.59266153093147e-06,
-      "loss": 0.7002,
+      "epoch": 0.20858189429618001,
+      "grad_norm": 2.8996497541322994,
+      "learning_rate": 1.8374005911405528e-05,
+      "loss": 0.9922,
       "step": 1993
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.583091695170857e-06,
-      "loss": 0.8015,
+      "epoch": 0.2086865515436944,
+      "grad_norm": 2.2079598409798527,
+      "learning_rate": 1.8372152644370684e-05,
+      "loss": 1.007,
       "step": 1994
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.5735268822879344e-06,
-      "loss": 0.6768,
+      "epoch": 0.2087912087912088,
+      "grad_norm": 2.1831162058992684,
+      "learning_rate": 1.8370298415358527e-05,
+      "loss": 1.0665,
       "step": 1995
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.563967103159733e-06,
-      "loss": 0.7159,
+      "epoch": 0.20889586603872318,
+      "grad_norm": 2.164569022998782,
+      "learning_rate": 1.8368443224582115e-05,
+      "loss": 1.1039,
       "step": 1996
     },
     {
-      "epoch": 1.31,
-      "learning_rate": 5.554412368657572e-06,
-      "loss": 0.6989,
+      "epoch": 0.20900052328623758,
+      "grad_norm": 3.8082638015537533,
+      "learning_rate": 1.836658707225461e-05,
+      "loss": 1.1108,
       "step": 1997
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.544862689647024e-06,
-      "loss": 0.7128,
+      "epoch": 0.20910518053375196,
+      "grad_norm": 2.252635657133077,
+      "learning_rate": 1.836472995858929e-05,
+      "loss": 1.0929,
       "step": 1998
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.535318076987928e-06,
-      "loss": 0.5933,
+      "epoch": 0.20920983778126634,
+      "grad_norm": 2.159315517759604,
+      "learning_rate": 1.836287188379954e-05,
+      "loss": 1.0949,
       "step": 1999
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.52577854153435e-06,
-      "loss": 0.7736,
+      "epoch": 0.20931449502878074,
+      "grad_norm": 2.3732269117605327,
+      "learning_rate": 1.836101284809886e-05,
+      "loss": 1.2197,
       "step": 2000
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.516244094134578e-06,
-      "loss": 0.6531,
+      "epoch": 0.20941915227629512,
+      "grad_norm": 2.062458615626669,
+      "learning_rate": 1.835915285170085e-05,
+      "loss": 0.8862,
       "step": 2001
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.506714745631124e-06,
-      "loss": 0.8514,
+      "epoch": 0.20952380952380953,
+      "grad_norm": 2.3583708531763636,
+      "learning_rate": 1.8357291894819236e-05,
+      "loss": 1.0029,
       "step": 2002
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.497190506860695e-06,
-      "loss": 0.8415,
+      "epoch": 0.2096284667713239,
+      "grad_norm": 2.284742186222721,
+      "learning_rate": 1.8355429977667837e-05,
+      "loss": 0.9637,
       "step": 2003
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.487671388654192e-06,
-      "loss": 0.7159,
+      "epoch": 0.2097331240188383,
+      "grad_norm": 2.2878915237206865,
+      "learning_rate": 1.83535671004606e-05,
+      "loss": 1.0314,
       "step": 2004
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.478157401836694e-06,
-      "loss": 0.7302,
+      "epoch": 0.2098377812663527,
+      "grad_norm": 2.380158906701526,
+      "learning_rate": 1.8351703263411567e-05,
+      "loss": 1.0208,
       "step": 2005
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.468648557227434e-06,
-      "loss": 0.6675,
+      "epoch": 0.2099424385138671,
+      "grad_norm": 2.3836744458989756,
+      "learning_rate": 1.83498384667349e-05,
+      "loss": 1.0668,
       "step": 2006
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.459144865639806e-06,
-      "loss": 0.7683,
+      "epoch": 0.21004709576138147,
+      "grad_norm": 2.1286045732612147,
+      "learning_rate": 1.8347972710644862e-05,
+      "loss": 0.9946,
       "step": 2007
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.449646337881345e-06,
-      "loss": 0.7269,
+      "epoch": 0.21015175300889585,
+      "grad_norm": 2.400087509008151,
+      "learning_rate": 1.8346105995355837e-05,
+      "loss": 0.9493,
       "step": 2008
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.440152984753709e-06,
-      "loss": 0.7207,
+      "epoch": 0.21025641025641026,
+      "grad_norm": 2.128726329762142,
+      "learning_rate": 1.8344238321082316e-05,
+      "loss": 0.9336,
       "step": 2009
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.430664817052678e-06,
-      "loss": 0.7395,
+      "epoch": 0.21036106750392464,
+      "grad_norm": 2.1830248786492956,
+      "learning_rate": 1.8342369688038894e-05,
+      "loss": 1.0607,
       "step": 2010
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.421181845568125e-06,
-      "loss": 0.7995,
+      "epoch": 0.21046572475143904,
+      "grad_norm": 2.2689021885521576,
+      "learning_rate": 1.8340500096440286e-05,
+      "loss": 1.069,
       "step": 2011
     },
     {
-      "epoch": 1.32,
-      "learning_rate": 5.411704081084025e-06,
-      "loss": 0.666,
+      "epoch": 0.21057038199895342,
+      "grad_norm": 2.5029531826193057,
+      "learning_rate": 1.833862954650131e-05,
+      "loss": 1.1992,
       "step": 2012
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.402231534378424e-06,
-      "loss": 0.6999,
+      "epoch": 0.21067503924646783,
+      "grad_norm": 2.69872427128542,
+      "learning_rate": 1.833675803843689e-05,
+      "loss": 1.048,
       "step": 2013
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.392764216223441e-06,
-      "loss": 0.7382,
+      "epoch": 0.2107796964939822,
+      "grad_norm": 2.6551070968845094,
+      "learning_rate": 1.8334885572462076e-05,
+      "loss": 1.0161,
       "step": 2014
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.383302137385238e-06,
-      "loss": 0.8523,
+      "epoch": 0.2108843537414966,
+      "grad_norm": 2.5294187266588293,
+      "learning_rate": 1.8333012148792008e-05,
+      "loss": 1.0735,
       "step": 2015
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.373845308624028e-06,
-      "loss": 0.6393,
+      "epoch": 0.210989010989011,
+      "grad_norm": 2.1783913112001754,
+      "learning_rate": 1.8331137767641958e-05,
+      "loss": 1.0417,
       "step": 2016
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.364393740694054e-06,
-      "loss": 0.6108,
+      "epoch": 0.21109366823652537,
+      "grad_norm": 2.369045148316055,
+      "learning_rate": 1.8329262429227285e-05,
+      "loss": 1.1235,
       "step": 2017
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.3549474443435726e-06,
-      "loss": 0.6351,
+      "epoch": 0.21119832548403977,
+      "grad_norm": 2.117286187563206,
+      "learning_rate": 1.8327386133763475e-05,
+      "loss": 0.8765,
       "step": 2018
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.345506430314845e-06,
-      "loss": 0.6475,
+      "epoch": 0.21130298273155415,
+      "grad_norm": 2.481083285401919,
+      "learning_rate": 1.8325508881466115e-05,
+      "loss": 1.0116,
       "step": 2019
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.33607070934413e-06,
-      "loss": 0.6502,
+      "epoch": 0.21140763997906856,
+      "grad_norm": 2.2535978341456753,
+      "learning_rate": 1.832363067255091e-05,
+      "loss": 1.0478,
       "step": 2020
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.326640292161665e-06,
-      "loss": 0.6687,
+      "epoch": 0.21151229722658293,
+      "grad_norm": 2.1891174065268824,
+      "learning_rate": 1.8321751507233663e-05,
+      "loss": 0.9707,
       "step": 2021
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.317215189491651e-06,
-      "loss": 0.5928,
+      "epoch": 0.21161695447409734,
+      "grad_norm": 2.29735895919819,
+      "learning_rate": 1.8319871385730298e-05,
+      "loss": 1.0669,
       "step": 2022
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.307795412052259e-06,
-      "loss": 0.6536,
+      "epoch": 0.21172161172161172,
+      "grad_norm": 2.147861663463576,
+      "learning_rate": 1.831799030825685e-05,
+      "loss": 1.0486,
       "step": 2023
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.298380970555584e-06,
-      "loss": 0.6886,
+      "epoch": 0.21182626896912612,
+      "grad_norm": 2.0405385220270973,
+      "learning_rate": 1.831610827502945e-05,
+      "loss": 0.9038,
       "step": 2024
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.288971875707669e-06,
-      "loss": 0.6748,
+      "epoch": 0.2119309262166405,
+      "grad_norm": 2.5478447806636657,
+      "learning_rate": 1.8314225286264353e-05,
+      "loss": 1.0453,
       "step": 2025
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.279568138208473e-06,
-      "loss": 0.7961,
+      "epoch": 0.21203558346415488,
+      "grad_norm": 2.3690660543652475,
+      "learning_rate": 1.8312341342177912e-05,
+      "loss": 1.0493,
       "step": 2026
     },
     {
-      "epoch": 1.33,
-      "learning_rate": 5.270169768751858e-06,
-      "loss": 0.756,
+      "epoch": 0.2121402407116693,
+      "grad_norm": 2.2032721351975995,
+      "learning_rate": 1.83104564429866e-05,
+      "loss": 0.9498,
       "step": 2027
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.260776778025591e-06,
-      "loss": 0.7973,
+      "epoch": 0.21224489795918366,
+      "grad_norm": 2.609109823726276,
+      "learning_rate": 1.8308570588906996e-05,
+      "loss": 1.1299,
       "step": 2028
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.251389176711307e-06,
-      "loss": 0.7368,
+      "epoch": 0.21234955520669807,
+      "grad_norm": 1.8610292340166064,
+      "learning_rate": 1.8306683780155792e-05,
+      "loss": 0.9029,
       "step": 2029
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.242006975484528e-06,
-      "loss": 0.706,
+      "epoch": 0.21245421245421245,
+      "grad_norm": 2.4304964979220025,
+      "learning_rate": 1.830479601694978e-05,
+      "loss": 1.087,
       "step": 2030
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.2326301850146285e-06,
-      "loss": 0.7257,
+      "epoch": 0.21255886970172685,
+      "grad_norm": 2.381417687058148,
+      "learning_rate": 1.830290729950587e-05,
+      "loss": 0.9628,
       "step": 2031
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.223258815964833e-06,
-      "loss": 0.7576,
+      "epoch": 0.21266352694924123,
+      "grad_norm": 2.017480163965436,
+      "learning_rate": 1.8301017628041076e-05,
+      "loss": 0.799,
       "step": 2032
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.213892878992191e-06,
-      "loss": 0.7511,
+      "epoch": 0.21276818419675564,
+      "grad_norm": 2.935623238435619,
+      "learning_rate": 1.829912700277253e-05,
+      "loss": 0.9128,
       "step": 2033
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.204532384747584e-06,
-      "loss": 0.7854,
+      "epoch": 0.21287284144427002,
+      "grad_norm": 2.2265140475206384,
+      "learning_rate": 1.8297235423917473e-05,
+      "loss": 0.9985,
       "step": 2034
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.195177343875701e-06,
-      "loss": 0.6626,
+      "epoch": 0.2129774986917844,
+      "grad_norm": 2.188420784579257,
+      "learning_rate": 1.829534289169324e-05,
+      "loss": 1.0615,
       "step": 2035
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.185827767015031e-06,
-      "loss": 0.6775,
+      "epoch": 0.2130821559392988,
+      "grad_norm": 1.9589933092767846,
+      "learning_rate": 1.8293449406317294e-05,
+      "loss": 0.9486,
       "step": 2036
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.176483664797852e-06,
-      "loss": 0.721,
+      "epoch": 0.21318681318681318,
+      "grad_norm": 2.7569189446285796,
+      "learning_rate": 1.82915549680072e-05,
+      "loss": 1.0995,
       "step": 2037
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.1671450478502064e-06,
-      "loss": 0.7386,
+      "epoch": 0.21329147043432758,
+      "grad_norm": 1.9836569561149964,
+      "learning_rate": 1.828965957698063e-05,
+      "loss": 1.0582,
       "step": 2038
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.157811926791905e-06,
-      "loss": 0.738,
+      "epoch": 0.21339612768184196,
+      "grad_norm": 2.6141067722134266,
+      "learning_rate": 1.8287763233455365e-05,
+      "loss": 0.9942,
       "step": 2039
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.1484843122365125e-06,
-      "loss": 0.8146,
+      "epoch": 0.21350078492935637,
+      "grad_norm": 2.6804666482513726,
+      "learning_rate": 1.8285865937649313e-05,
+      "loss": 1.0738,
       "step": 2040
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.139162214791322e-06,
-      "loss": 0.5877,
+      "epoch": 0.21360544217687075,
+      "grad_norm": 2.129743160129942,
+      "learning_rate": 1.8283967689780464e-05,
+      "loss": 0.81,
       "step": 2041
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.129845645057372e-06,
-      "loss": 0.7367,
+      "epoch": 0.21371009942438512,
+      "grad_norm": 2.363273343477441,
+      "learning_rate": 1.8282068490066932e-05,
+      "loss": 0.8765,
       "step": 2042
     },
     {
-      "epoch": 1.34,
-      "learning_rate": 5.120534613629389e-06,
-      "loss": 0.7743,
+      "epoch": 0.21381475667189953,
+      "grad_norm": 2.476072286035289,
+      "learning_rate": 1.828016833872694e-05,
+      "loss": 1.0747,
       "step": 2043
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.1112291310958166e-06,
-      "loss": 0.8649,
+      "epoch": 0.2139194139194139,
+      "grad_norm": 2.2832920524718343,
+      "learning_rate": 1.8278267235978823e-05,
+      "loss": 1.0481,
       "step": 2044
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.101929208038786e-06,
-      "loss": 0.7196,
+      "epoch": 0.21402407116692831,
+      "grad_norm": 2.0256536513257473,
+      "learning_rate": 1.8276365182041015e-05,
+      "loss": 0.9437,
       "step": 2045
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.09263485503411e-06,
-      "loss": 0.5693,
+      "epoch": 0.2141287284144427,
+      "grad_norm": 2.0349450996429237,
+      "learning_rate": 1.8274462177132074e-05,
+      "loss": 0.9745,
       "step": 2046
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.083346082651255e-06,
-      "loss": 0.7333,
+      "epoch": 0.2142333856619571,
+      "grad_norm": 2.3558277924424154,
+      "learning_rate": 1.8272558221470652e-05,
+      "loss": 0.9551,
       "step": 2047
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.074062901453352e-06,
-      "loss": 0.7562,
+      "epoch": 0.21433804290947148,
+      "grad_norm": 2.2290513676067962,
+      "learning_rate": 1.8270653315275525e-05,
+      "loss": 1.0465,
       "step": 2048
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.06478532199717e-06,
-      "loss": 0.7743,
+      "epoch": 0.21444270015698588,
+      "grad_norm": 2.258045963609993,
+      "learning_rate": 1.826874745876556e-05,
+      "loss": 1.0306,
       "step": 2049
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.05551335483311e-06,
-      "loss": 0.7186,
+      "epoch": 0.21454735740450026,
+      "grad_norm": 3.527078541148376,
+      "learning_rate": 1.8266840652159752e-05,
+      "loss": 0.8868,
       "step": 2050
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.04624701050519e-06,
-      "loss": 0.6196,
+      "epoch": 0.21465201465201464,
+      "grad_norm": 2.2609749036853226,
+      "learning_rate": 1.8264932895677195e-05,
+      "loss": 0.9106,
       "step": 2051
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.03698629955103e-06,
-      "loss": 0.6444,
+      "epoch": 0.21475667189952904,
+      "grad_norm": 2.5062029257028215,
+      "learning_rate": 1.826302418953709e-05,
+      "loss": 1.1274,
       "step": 2052
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.027731232501849e-06,
-      "loss": 0.8377,
+      "epoch": 0.21486132914704342,
+      "grad_norm": 2.353992741301644,
+      "learning_rate": 1.8261114533958757e-05,
+      "loss": 1.0853,
       "step": 2053
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.018481819882446e-06,
-      "loss": 0.807,
+      "epoch": 0.21496598639455783,
+      "grad_norm": 2.2428620293641144,
+      "learning_rate": 1.825920392916162e-05,
+      "loss": 0.9662,
       "step": 2054
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.009238072211187e-06,
-      "loss": 0.7461,
+      "epoch": 0.2150706436420722,
+      "grad_norm": 2.305344851187627,
+      "learning_rate": 1.8257292375365202e-05,
+      "loss": 1.1007,
       "step": 2055
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 5.000000000000003e-06,
-      "loss": 0.6753,
+      "epoch": 0.2151753008895866,
+      "grad_norm": 2.2656413224426033,
+      "learning_rate": 1.825537987278916e-05,
+      "loss": 0.9697,
       "step": 2056
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 4.990767613754358e-06,
-      "loss": 0.8889,
+      "epoch": 0.215279958137101,
+      "grad_norm": 2.148064446805488,
+      "learning_rate": 1.8253466421653226e-05,
+      "loss": 0.9534,
       "step": 2057
     },
     {
-      "epoch": 1.35,
-      "learning_rate": 4.981540923973263e-06,
-      "loss": 0.7549,
+      "epoch": 0.2153846153846154,
+      "grad_norm": 2.1646722395898714,
+      "learning_rate": 1.8251552022177273e-05,
+      "loss": 0.9849,
       "step": 2058
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.972319941149245e-06,
-      "loss": 0.745,
+      "epoch": 0.21548927263212977,
+      "grad_norm": 2.2230725886492007,
+      "learning_rate": 1.8249636674581265e-05,
+      "loss": 1.0095,
       "step": 2059
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.963104675768345e-06,
-      "loss": 0.6283,
+      "epoch": 0.21559392987964415,
+      "grad_norm": 2.284428221986189,
+      "learning_rate": 1.824772037908528e-05,
+      "loss": 0.987,
       "step": 2060
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.953895138310091e-06,
-      "loss": 0.6988,
+      "epoch": 0.21569858712715856,
+      "grad_norm": 2.5819764848015256,
+      "learning_rate": 1.8245803135909498e-05,
+      "loss": 1.0183,
       "step": 2061
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.944691339247515e-06,
-      "loss": 0.7217,
+      "epoch": 0.21580324437467294,
+      "grad_norm": 2.3210577007465423,
+      "learning_rate": 1.8243884945274227e-05,
+      "loss": 0.9843,
       "step": 2062
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.9354932890471096e-06,
-      "loss": 0.654,
+      "epoch": 0.21590790162218734,
+      "grad_norm": 2.5362607817744625,
+      "learning_rate": 1.824196580739986e-05,
+      "loss": 1.1811,
       "step": 2063
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.926300998168835e-06,
-      "loss": 0.7342,
+      "epoch": 0.21601255886970172,
+      "grad_norm": 2.3962630982553397,
+      "learning_rate": 1.8240045722506915e-05,
+      "loss": 1.0299,
       "step": 2064
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.917114477066104e-06,
-      "loss": 0.6831,
+      "epoch": 0.21611721611721613,
+      "grad_norm": 1.988482806508578,
+      "learning_rate": 1.823812469081601e-05,
+      "loss": 1.0407,
       "step": 2065
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.907933736185757e-06,
-      "loss": 0.5879,
+      "epoch": 0.2162218733647305,
+      "grad_norm": 2.727138846671659,
+      "learning_rate": 1.823620271254788e-05,
+      "loss": 0.8944,
       "step": 2066
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.898758785968075e-06,
-      "loss": 0.8035,
+      "epoch": 0.2163265306122449,
+      "grad_norm": 2.2182850283211475,
+      "learning_rate": 1.8234279787923358e-05,
+      "loss": 1.0132,
       "step": 2067
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.889589636846747e-06,
-      "loss": 0.7861,
+      "epoch": 0.2164311878597593,
+      "grad_norm": 2.4303463162856365,
+      "learning_rate": 1.82323559171634e-05,
+      "loss": 1.0625,
       "step": 2068
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.880426299248867e-06,
-      "loss": 0.7847,
+      "epoch": 0.21653584510727367,
+      "grad_norm": 2.419684174616418,
+      "learning_rate": 1.8230431100489057e-05,
+      "loss": 0.89,
       "step": 2069
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.871268783594922e-06,
-      "loss": 0.6659,
+      "epoch": 0.21664050235478807,
+      "grad_norm": 2.4048668926385885,
+      "learning_rate": 1.8228505338121496e-05,
+      "loss": 1.0583,
       "step": 2070
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.86211710029877e-06,
-      "loss": 0.7981,
+      "epoch": 0.21674515960230245,
+      "grad_norm": 2.342058187371194,
+      "learning_rate": 1.822657863028199e-05,
+      "loss": 0.9905,
       "step": 2071
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.852971259767642e-06,
-      "loss": 0.6935,
+      "epoch": 0.21684981684981686,
+      "grad_norm": 2.4230482675249716,
+      "learning_rate": 1.822465097719192e-05,
+      "loss": 1.0081,
       "step": 2072
     },
     {
-      "epoch": 1.36,
-      "learning_rate": 4.843831272402126e-06,
-      "loss": 0.7619,
+      "epoch": 0.21695447409733123,
+      "grad_norm": 2.72537241732261,
+      "learning_rate": 1.8222722379072785e-05,
+      "loss": 1.0245,
       "step": 2073
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.83469714859616e-06,
-      "loss": 0.6715,
+      "epoch": 0.21705913134484564,
+      "grad_norm": 2.4296676623937556,
+      "learning_rate": 1.8220792836146177e-05,
+      "loss": 1.0386,
       "step": 2074
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.825568898736999e-06,
-      "loss": 0.7134,
+      "epoch": 0.21716378859236002,
+      "grad_norm": 2.5375888498111343,
+      "learning_rate": 1.8218862348633806e-05,
+      "loss": 1.0651,
       "step": 2075
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.816446533205225e-06,
-      "loss": 0.741,
+      "epoch": 0.21726844583987442,
+      "grad_norm": 2.3320537485576804,
+      "learning_rate": 1.821693091675749e-05,
+      "loss": 1.0419,
       "step": 2076
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.807330062374733e-06,
-      "loss": 0.7401,
+      "epoch": 0.2173731030873888,
+      "grad_norm": 2.2920845320825793,
+      "learning_rate": 1.8214998540739154e-05,
+      "loss": 1.0797,
       "step": 2077
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.79821949661271e-06,
-      "loss": 0.7026,
+      "epoch": 0.21747776033490318,
+      "grad_norm": 2.156245726685389,
+      "learning_rate": 1.8213065220800833e-05,
+      "loss": 1.0072,
       "step": 2078
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.789114846279631e-06,
-      "loss": 0.6982,
+      "epoch": 0.2175824175824176,
+      "grad_norm": 2.4898649858067117,
+      "learning_rate": 1.821113095716467e-05,
+      "loss": 1.1149,
       "step": 2079
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.780016121729236e-06,
-      "loss": 0.7182,
+      "epoch": 0.21768707482993196,
+      "grad_norm": 2.0771189974183586,
+      "learning_rate": 1.820919575005291e-05,
+      "loss": 1.0435,
       "step": 2080
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.7709233333085345e-06,
-      "loss": 0.7178,
+      "epoch": 0.21779173207744637,
+      "grad_norm": 2.277727814797236,
+      "learning_rate": 1.820725959968792e-05,
+      "loss": 1.0899,
       "step": 2081
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.761836491357782e-06,
-      "loss": 0.7042,
+      "epoch": 0.21789638932496075,
+      "grad_norm": 2.373876155395553,
+      "learning_rate": 1.8205322506292162e-05,
+      "loss": 1.0877,
       "step": 2082
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.7527556062104756e-06,
-      "loss": 0.6837,
+      "epoch": 0.21800104657247515,
+      "grad_norm": 2.1588302308233525,
+      "learning_rate": 1.8203384470088217e-05,
+      "loss": 0.8801,
       "step": 2083
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.7436806881933304e-06,
-      "loss": 0.7406,
+      "epoch": 0.21810570381998953,
+      "grad_norm": 2.28595739064826,
+      "learning_rate": 1.8201445491298765e-05,
+      "loss": 1.051,
       "step": 2084
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.734611747626285e-06,
-      "loss": 0.7689,
+      "epoch": 0.21821036106750394,
+      "grad_norm": 2.1087266644570923,
+      "learning_rate": 1.81995055701466e-05,
+      "loss": 1.1129,
       "step": 2085
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.7255487948224754e-06,
-      "loss": 0.7852,
+      "epoch": 0.21831501831501832,
+      "grad_norm": 2.425503184353989,
+      "learning_rate": 1.8197564706854624e-05,
+      "loss": 0.855,
       "step": 2086
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.716491840088227e-06,
-      "loss": 0.7435,
+      "epoch": 0.2184196755625327,
+      "grad_norm": 2.2595365972197294,
+      "learning_rate": 1.8195622901645843e-05,
+      "loss": 0.8961,
       "step": 2087
     },
     {
-      "epoch": 1.37,
-      "learning_rate": 4.707440893723052e-06,
-      "loss": 0.6964,
+      "epoch": 0.2185243328100471,
+      "grad_norm": 2.514090138405137,
+      "learning_rate": 1.8193680154743375e-05,
+      "loss": 1.0657,
       "step": 2088
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.698395966019615e-06,
-      "loss": 0.7188,
+      "epoch": 0.21862899005756148,
+      "grad_norm": 2.4118987132654266,
+      "learning_rate": 1.819173646637045e-05,
+      "loss": 1.0297,
       "step": 2089
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.689357067263751e-06,
-      "loss": 0.7081,
+      "epoch": 0.21873364730507588,
+      "grad_norm": 2.1575019695204127,
+      "learning_rate": 1.8189791836750396e-05,
+      "loss": 0.9463,
       "step": 2090
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.680324207734434e-06,
-      "loss": 0.813,
+      "epoch": 0.21883830455259026,
+      "grad_norm": 2.408277886570118,
+      "learning_rate": 1.818784626610666e-05,
+      "loss": 1.1036,
       "step": 2091
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.671297397703766e-06,
-      "loss": 0.7185,
+      "epoch": 0.21894296180010467,
+      "grad_norm": 1.795388244350893,
+      "learning_rate": 1.8185899754662787e-05,
+      "loss": 0.8641,
       "step": 2092
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.662276647436977e-06,
-      "loss": 0.5856,
+      "epoch": 0.21904761904761905,
+      "grad_norm": 2.581772835215814,
+      "learning_rate": 1.818395230264244e-05,
+      "loss": 1.061,
       "step": 2093
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.653261967192398e-06,
-      "loss": 0.6918,
+      "epoch": 0.21915227629513342,
+      "grad_norm": 2.259983033421856,
+      "learning_rate": 1.8182003910269382e-05,
+      "loss": 1.0749,
       "step": 2094
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.644253367221467e-06,
-      "loss": 0.7439,
+      "epoch": 0.21925693354264783,
+      "grad_norm": 2.3449988057297206,
+      "learning_rate": 1.818005457776749e-05,
+      "loss": 1.1159,
       "step": 2095
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.635250857768696e-06,
-      "loss": 0.7696,
+      "epoch": 0.2193615907901622,
+      "grad_norm": 2.2960373512554746,
+      "learning_rate": 1.8178104305360743e-05,
+      "loss": 1.155,
       "step": 2096
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.62625444907168e-06,
-      "loss": 0.7711,
+      "epoch": 0.21946624803767661,
+      "grad_norm": 2.2977900952956687,
+      "learning_rate": 1.817615309327323e-05,
+      "loss": 1.0825,
       "step": 2097
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.6172641513610765e-06,
-      "loss": 0.771,
+      "epoch": 0.219570905285191,
+      "grad_norm": 1.9351630635888202,
+      "learning_rate": 1.8174200941729153e-05,
+      "loss": 0.818,
       "step": 2098
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.608279974860581e-06,
-      "loss": 0.8323,
+      "epoch": 0.2196755625327054,
+      "grad_norm": 1.9567019706927198,
+      "learning_rate": 1.8172247850952816e-05,
+      "loss": 0.9481,
       "step": 2099
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.599301929786943e-06,
-      "loss": 0.6977,
+      "epoch": 0.21978021978021978,
+      "grad_norm": 2.431540987817807,
+      "learning_rate": 1.817029382116864e-05,
+      "loss": 1.1475,
       "step": 2100
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.5903300263499315e-06,
-      "loss": 0.7317,
+      "epoch": 0.21988487702773418,
+      "grad_norm": 2.6689599398800925,
+      "learning_rate": 1.8168338852601136e-05,
+      "loss": 0.9702,
       "step": 2101
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.581364274752338e-06,
-      "loss": 0.7067,
+      "epoch": 0.21998953427524856,
+      "grad_norm": 2.1586431953787004,
+      "learning_rate": 1.8166382945474937e-05,
+      "loss": 0.9888,
       "step": 2102
     },
     {
-      "epoch": 1.38,
-      "learning_rate": 4.572404685189946e-06,
-      "loss": 0.6092,
+      "epoch": 0.22009419152276294,
+      "grad_norm": 2.0960254129246865,
+      "learning_rate": 1.8164426100014787e-05,
+      "loss": 0.8704,
       "step": 2103
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.563451267851542e-06,
-      "loss": 0.6164,
+      "epoch": 0.22019884877027734,
+      "grad_norm": 2.717246868399569,
+      "learning_rate": 1.8162468316445526e-05,
+      "loss": 0.7871,
       "step": 2104
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.554504032918888e-06,
-      "loss": 0.7474,
+      "epoch": 0.22030350601779172,
+      "grad_norm": 2.2870173127883255,
+      "learning_rate": 1.8160509594992105e-05,
+      "loss": 1.0826,
       "step": 2105
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.545562990566725e-06,
-      "loss": 0.6677,
+      "epoch": 0.22040816326530613,
+      "grad_norm": 2.383105180002082,
+      "learning_rate": 1.815854993587959e-05,
+      "loss": 1.017,
       "step": 2106
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.536628150962744e-06,
-      "loss": 0.6281,
+      "epoch": 0.2205128205128205,
+      "grad_norm": 2.1939394100255596,
+      "learning_rate": 1.8156589339333154e-05,
+      "loss": 0.992,
       "step": 2107
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.527699524267576e-06,
-      "loss": 0.6745,
+      "epoch": 0.2206174777603349,
+      "grad_norm": 1.995289142653823,
+      "learning_rate": 1.815462780557806e-05,
+      "loss": 0.833,
       "step": 2108
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.518777120634798e-06,
-      "loss": 0.7419,
+      "epoch": 0.2207221350078493,
+      "grad_norm": 2.528094256454689,
+      "learning_rate": 1.8152665334839704e-05,
+      "loss": 1.1359,
       "step": 2109
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.509860950210905e-06,
-      "loss": 0.6872,
+      "epoch": 0.2208267922553637,
+      "grad_norm": 2.8889436520004237,
+      "learning_rate": 1.815070192734357e-05,
+      "loss": 0.9671,
       "step": 2110
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.500951023135308e-06,
-      "loss": 0.6998,
+      "epoch": 0.22093144950287807,
+      "grad_norm": 2.566640937673796,
+      "learning_rate": 1.8148737583315266e-05,
+      "loss": 1.0503,
       "step": 2111
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.492047349540308e-06,
-      "loss": 0.6863,
+      "epoch": 0.22103610675039245,
+      "grad_norm": 2.1263834504381656,
+      "learning_rate": 1.8146772302980494e-05,
+      "loss": 0.7854,
       "step": 2112
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.483149939551105e-06,
-      "loss": 0.6969,
+      "epoch": 0.22114076399790686,
+      "grad_norm": 2.121646140657449,
+      "learning_rate": 1.814480608656507e-05,
+      "loss": 0.8763,
       "step": 2113
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.474258803285774e-06,
-      "loss": 0.7998,
+      "epoch": 0.22124542124542124,
+      "grad_norm": 2.4127305165775956,
+      "learning_rate": 1.814283893429491e-05,
+      "loss": 0.906,
       "step": 2114
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.465373950855249e-06,
-      "loss": 0.6863,
+      "epoch": 0.22135007849293564,
+      "grad_norm": 2.4552352187699493,
+      "learning_rate": 1.8140870846396055e-05,
+      "loss": 1.1067,
       "step": 2115
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.4564953923633295e-06,
-      "loss": 0.6583,
+      "epoch": 0.22145473574045002,
+      "grad_norm": 2.116477172570462,
+      "learning_rate": 1.8138901823094634e-05,
+      "loss": 0.9615,
       "step": 2116
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.4476231379066445e-06,
-      "loss": 0.6266,
+      "epoch": 0.22155939298796443,
+      "grad_norm": 2.1641002061702035,
+      "learning_rate": 1.8136931864616894e-05,
+      "loss": 1.0059,
       "step": 2117
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.4387571975746655e-06,
-      "loss": 0.7109,
+      "epoch": 0.2216640502354788,
+      "grad_norm": 2.418244872883079,
+      "learning_rate": 1.8134960971189186e-05,
+      "loss": 1.1253,
       "step": 2118
     },
     {
-      "epoch": 1.39,
-      "learning_rate": 4.429897581449677e-06,
-      "loss": 0.7029,
+      "epoch": 0.2217687074829932,
+      "grad_norm": 2.550192195359462,
+      "learning_rate": 1.8132989143037973e-05,
+      "loss": 0.9129,
       "step": 2119
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.4210442996067724e-06,
-      "loss": 0.7064,
+      "epoch": 0.2218733647305076,
+      "grad_norm": 2.1808748350442664,
+      "learning_rate": 1.8131016380389822e-05,
+      "loss": 0.9744,
       "step": 2120
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.412197362113848e-06,
-      "loss": 0.7868,
+      "epoch": 0.22197802197802197,
+      "grad_norm": 1.8551489114253599,
+      "learning_rate": 1.8129042683471404e-05,
+      "loss": 0.8682,
       "step": 2121
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.403356779031572e-06,
-      "loss": 0.7483,
+      "epoch": 0.22208267922553637,
+      "grad_norm": 2.08940663789057,
+      "learning_rate": 1.81270680525095e-05,
+      "loss": 0.9402,
       "step": 2122
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.394522560413399e-06,
-      "loss": 0.7122,
+      "epoch": 0.22218733647305075,
+      "grad_norm": 2.2964529328748506,
+      "learning_rate": 1.8125092487731005e-05,
+      "loss": 0.9279,
       "step": 2123
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.385694716305541e-06,
-      "loss": 0.6946,
+      "epoch": 0.22229199372056516,
+      "grad_norm": 2.215371699637496,
+      "learning_rate": 1.812311598936291e-05,
+      "loss": 1.0981,
       "step": 2124
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.376873256746962e-06,
-      "loss": 0.7781,
+      "epoch": 0.22239665096807953,
+      "grad_norm": 2.652740527835587,
+      "learning_rate": 1.8121138557632322e-05,
+      "loss": 1.083,
       "step": 2125
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.368058191769363e-06,
-      "loss": 0.7361,
+      "epoch": 0.22250130821559394,
+      "grad_norm": 2.5149433842670503,
+      "learning_rate": 1.811916019276645e-05,
+      "loss": 0.8659,
       "step": 2126
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.359249531397177e-06,
-      "loss": 0.6632,
+      "epoch": 0.22260596546310832,
+      "grad_norm": 1.9949643722746215,
+      "learning_rate": 1.811718089499261e-05,
+      "loss": 1.0477,
       "step": 2127
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.350447285647552e-06,
-      "loss": 0.6804,
+      "epoch": 0.22271062271062272,
+      "grad_norm": 2.6367101170753173,
+      "learning_rate": 1.8115200664538234e-05,
+      "loss": 0.9465,
       "step": 2128
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.341651464530341e-06,
-      "loss": 0.7049,
+      "epoch": 0.2228152799581371,
+      "grad_norm": 2.1662145017836645,
+      "learning_rate": 1.8113219501630848e-05,
+      "loss": 0.9172,
       "step": 2129
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.332862078048095e-06,
-      "loss": 0.7433,
+      "epoch": 0.22291993720565148,
+      "grad_norm": 2.139400049602698,
+      "learning_rate": 1.8111237406498096e-05,
+      "loss": 0.9777,
       "step": 2130
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.324079136196034e-06,
-      "loss": 0.7752,
+      "epoch": 0.2230245944531659,
+      "grad_norm": 2.0637795213950327,
+      "learning_rate": 1.810925437936772e-05,
+      "loss": 0.8885,
       "step": 2131
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.315302648962066e-06,
-      "loss": 0.7045,
+      "epoch": 0.22312925170068026,
+      "grad_norm": 2.4433179860559733,
+      "learning_rate": 1.810727042046758e-05,
+      "loss": 1.1147,
       "step": 2132
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.30653262632675e-06,
-      "loss": 0.7195,
+      "epoch": 0.22323390894819467,
+      "grad_norm": 2.507296458731943,
+      "learning_rate": 1.810528553002563e-05,
+      "loss": 1.0424,
       "step": 2133
     },
     {
-      "epoch": 1.4,
-      "learning_rate": 4.297769078263296e-06,
-      "loss": 0.8205,
+      "epoch": 0.22333856619570905,
+      "grad_norm": 2.476296110697989,
+      "learning_rate": 1.810329970826994e-05,
+      "loss": 1.0969,
       "step": 2134
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.2890120147375545e-06,
-      "loss": 0.7518,
+      "epoch": 0.22344322344322345,
+      "grad_norm": 2.2719919935385966,
+      "learning_rate": 1.810131295542869e-05,
+      "loss": 1.086,
       "step": 2135
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.280261445707985e-06,
-      "loss": 0.7425,
+      "epoch": 0.22354788069073783,
+      "grad_norm": 2.1254287766207867,
+      "learning_rate": 1.8099325271730158e-05,
+      "loss": 1.0048,
       "step": 2136
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.271517381125687e-06,
-      "loss": 0.73,
+      "epoch": 0.2236525379382522,
+      "grad_norm": 2.2899571405767913,
+      "learning_rate": 1.8097336657402733e-05,
+      "loss": 1.025,
       "step": 2137
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.262779830934346e-06,
-      "loss": 0.7006,
+      "epoch": 0.22375719518576662,
+      "grad_norm": 2.493384145123412,
+      "learning_rate": 1.809534711267491e-05,
+      "loss": 0.9832,
       "step": 2138
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.254048805070248e-06,
-      "loss": 0.6843,
+      "epoch": 0.223861852433281,
+      "grad_norm": 2.3792722351135867,
+      "learning_rate": 1.8093356637775296e-05,
+      "loss": 1.1609,
       "step": 2139
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.2453243134622465e-06,
-      "loss": 0.6841,
+      "epoch": 0.2239665096807954,
+      "grad_norm": 2.007926261369221,
+      "learning_rate": 1.8091365232932595e-05,
+      "loss": 0.9712,
       "step": 2140
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.2366063660317805e-06,
-      "loss": 0.6899,
+      "epoch": 0.22407116692830978,
+      "grad_norm": 2.483067842936785,
+      "learning_rate": 1.8089372898375622e-05,
+      "loss": 1.0444,
       "step": 2141
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.2278949726928354e-06,
-      "loss": 0.7306,
+      "epoch": 0.22417582417582418,
+      "grad_norm": 2.3009244591558815,
+      "learning_rate": 1.808737963433331e-05,
+      "loss": 1.13,
       "step": 2142
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.219190143351951e-06,
-      "loss": 0.6526,
+      "epoch": 0.22428048142333856,
+      "grad_norm": 2.3317608975873028,
+      "learning_rate": 1.808538544103468e-05,
+      "loss": 1.0947,
       "step": 2143
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.210491887908201e-06,
-      "loss": 0.7064,
+      "epoch": 0.22438513867085297,
+      "grad_norm": 2.058037185074867,
+      "learning_rate": 1.8083390318708875e-05,
+      "loss": 1.0688,
       "step": 2144
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.201800216253176e-06,
-      "loss": 0.7287,
+      "epoch": 0.22448979591836735,
+      "grad_norm": 2.1821994697350813,
+      "learning_rate": 1.8081394267585134e-05,
+      "loss": 1.0108,
       "step": 2145
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.193115138270986e-06,
-      "loss": 0.7231,
+      "epoch": 0.22459445316588172,
+      "grad_norm": 2.093805533273431,
+      "learning_rate": 1.8079397287892808e-05,
+      "loss": 1.0334,
       "step": 2146
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.184436663838245e-06,
-      "loss": 0.6422,
+      "epoch": 0.22469911041339613,
+      "grad_norm": 2.230362524509398,
+      "learning_rate": 1.807739937986136e-05,
+      "loss": 0.983,
       "step": 2147
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.175764802824049e-06,
-      "loss": 0.7402,
+      "epoch": 0.2248037676609105,
+      "grad_norm": 2.2669738777674517,
+      "learning_rate": 1.8075400543720342e-05,
+      "loss": 1.0644,
       "step": 2148
     },
     {
-      "epoch": 1.41,
-      "learning_rate": 4.167099565089989e-06,
-      "loss": 0.7965,
+      "epoch": 0.22490842490842491,
+      "grad_norm": 2.4539503469583237,
+      "learning_rate": 1.8073400779699434e-05,
+      "loss": 1.0708,
       "step": 2149
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.158440960490104e-06,
-      "loss": 0.6677,
+      "epoch": 0.2250130821559393,
+      "grad_norm": 2.366911153866586,
+      "learning_rate": 1.807140008802841e-05,
+      "loss": 1.0431,
       "step": 2150
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.149788998870902e-06,
-      "loss": 0.7098,
+      "epoch": 0.2251177394034537,
+      "grad_norm": 2.144584808902021,
+      "learning_rate": 1.8069398468937153e-05,
+      "loss": 0.9618,
       "step": 2151
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.141143690071335e-06,
-      "loss": 0.8215,
+      "epoch": 0.22522239665096808,
+      "grad_norm": 2.540674460417446,
+      "learning_rate": 1.806739592265565e-05,
+      "loss": 1.138,
       "step": 2152
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.132505043922792e-06,
-      "loss": 0.7118,
+      "epoch": 0.22532705389848248,
+      "grad_norm": 2.206689167590175,
+      "learning_rate": 1.8065392449414004e-05,
+      "loss": 1.0123,
       "step": 2153
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.123873070249074e-06,
-      "loss": 0.6459,
+      "epoch": 0.22543171114599686,
+      "grad_norm": 2.2813596687579474,
+      "learning_rate": 1.806338804944242e-05,
+      "loss": 1.07,
       "step": 2154
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.115247778866407e-06,
-      "loss": 0.7414,
+      "epoch": 0.22553636839351124,
+      "grad_norm": 2.4718690247538873,
+      "learning_rate": 1.8061382722971196e-05,
+      "loss": 0.9068,
       "step": 2155
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.106629179583412e-06,
-      "loss": 0.7304,
+      "epoch": 0.22564102564102564,
+      "grad_norm": 2.3583234330786507,
+      "learning_rate": 1.8059376470230757e-05,
+      "loss": 1.055,
       "step": 2156
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.0980172822010985e-06,
-      "loss": 0.6996,
+      "epoch": 0.22574568288854002,
+      "grad_norm": 2.4215574962079205,
+      "learning_rate": 1.8057369291451626e-05,
+      "loss": 1.0428,
       "step": 2157
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.0894120965128615e-06,
-      "loss": 0.7429,
+      "epoch": 0.22585034013605443,
+      "grad_norm": 1.9930502014891032,
+      "learning_rate": 1.805536118686443e-05,
+      "loss": 0.9008,
       "step": 2158
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.080813632304455e-06,
-      "loss": 0.6918,
+      "epoch": 0.2259549973835688,
+      "grad_norm": 2.3283037616599027,
+      "learning_rate": 1.80533521566999e-05,
+      "loss": 1.0235,
       "step": 2159
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.072221899353995e-06,
-      "loss": 0.7265,
+      "epoch": 0.2260596546310832,
+      "grad_norm": 2.3801695391927753,
+      "learning_rate": 1.805134220118888e-05,
+      "loss": 1.022,
       "step": 2160
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.063636907431942e-06,
-      "loss": 0.6741,
+      "epoch": 0.2261643118785976,
+      "grad_norm": 2.397543403584425,
+      "learning_rate": 1.8049331320562324e-05,
+      "loss": 0.915,
       "step": 2161
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.055058666301087e-06,
-      "loss": 0.7794,
+      "epoch": 0.226268969126112,
+      "grad_norm": 2.626509392011337,
+      "learning_rate": 1.804731951505128e-05,
+      "loss": 1.0158,
       "step": 2162
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.046487185716552e-06,
-      "loss": 0.7572,
+      "epoch": 0.22637362637362637,
+      "grad_norm": 2.6074284746843395,
+      "learning_rate": 1.804530678488691e-05,
+      "loss": 0.9578,
       "step": 2163
     },
     {
-      "epoch": 1.42,
-      "learning_rate": 4.037922475425759e-06,
-      "loss": 0.6487,
+      "epoch": 0.22647828362114075,
+      "grad_norm": 2.2563607196090665,
+      "learning_rate": 1.8043293130300482e-05,
+      "loss": 0.9736,
       "step": 2164
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 4.029364545168438e-06,
-      "loss": 0.6464,
+      "epoch": 0.22658294086865516,
+      "grad_norm": 2.5275115770936285,
+      "learning_rate": 1.8041278551523365e-05,
+      "loss": 1.1927,
       "step": 2165
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 4.020813404676612e-06,
-      "loss": 0.7904,
+      "epoch": 0.22668759811616954,
+      "grad_norm": 3.2122832502725074,
+      "learning_rate": 1.8039263048787042e-05,
+      "loss": 1.0338,
       "step": 2166
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 4.012269063674581e-06,
-      "loss": 0.663,
+      "epoch": 0.22679225536368394,
+      "grad_norm": 2.1422445536418078,
+      "learning_rate": 1.80372466223231e-05,
+      "loss": 0.8647,
       "step": 2167
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 4.0037315318789e-06,
-      "loss": 0.7922,
+      "epoch": 0.22689691261119832,
+      "grad_norm": 2.30508825920427,
+      "learning_rate": 1.8035229272363226e-05,
+      "loss": 0.9192,
       "step": 2168
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.9952008189984025e-06,
-      "loss": 0.6516,
+      "epoch": 0.22700156985871273,
+      "grad_norm": 2.3600467400757346,
+      "learning_rate": 1.803321099913922e-05,
+      "loss": 1.0098,
       "step": 2169
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.986676934734151e-06,
-      "loss": 0.7492,
+      "epoch": 0.2271062271062271,
+      "grad_norm": 2.239202924612382,
+      "learning_rate": 1.8031191802882984e-05,
+      "loss": 1.0374,
       "step": 2170
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.978159888779451e-06,
-      "loss": 0.6673,
+      "epoch": 0.2272108843537415,
+      "grad_norm": 2.007640041272985,
+      "learning_rate": 1.8029171683826528e-05,
+      "loss": 0.9717,
       "step": 2171
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.96964969081983e-06,
-      "loss": 0.6747,
+      "epoch": 0.2273155416012559,
+      "grad_norm": 2.0849186079241453,
+      "learning_rate": 1.802715064220197e-05,
+      "loss": 0.8889,
       "step": 2172
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.961146350533021e-06,
-      "loss": 0.5325,
+      "epoch": 0.22742019884877027,
+      "grad_norm": 2.441637760590174,
+      "learning_rate": 1.8025128678241532e-05,
+      "loss": 1.0504,
       "step": 2173
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.9526498775889645e-06,
-      "loss": 0.7031,
+      "epoch": 0.22752485609628467,
+      "grad_norm": 1.9626865084302276,
+      "learning_rate": 1.802310579217754e-05,
+      "loss": 0.9714,
       "step": 2174
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.944160281649793e-06,
-      "loss": 0.6815,
+      "epoch": 0.22762951334379905,
+      "grad_norm": 2.434914606469338,
+      "learning_rate": 1.8021081984242426e-05,
+      "loss": 1.0148,
       "step": 2175
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.935677572369817e-06,
-      "loss": 0.6953,
+      "epoch": 0.22773417059131346,
+      "grad_norm": 2.464927468743602,
+      "learning_rate": 1.801905725466873e-05,
+      "loss": 0.9263,
       "step": 2176
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.927201759395518e-06,
-      "loss": 0.7131,
+      "epoch": 0.22783882783882783,
+      "grad_norm": 2.1365896230444017,
+      "learning_rate": 1.8017031603689105e-05,
+      "loss": 1.0849,
       "step": 2177
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.918732852365524e-06,
-      "loss": 0.6739,
+      "epoch": 0.22794348508634224,
+      "grad_norm": 2.0108040763164894,
+      "learning_rate": 1.801500503153629e-05,
+      "loss": 0.8429,
       "step": 2178
     },
     {
-      "epoch": 1.43,
-      "learning_rate": 3.910270860910623e-06,
-      "loss": 0.6494,
+      "epoch": 0.22804814233385662,
+      "grad_norm": 2.5168079546538507,
+      "learning_rate": 1.8012977538443156e-05,
+      "loss": 1.0145,
       "step": 2179
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.901815794653729e-06,
-      "loss": 0.6589,
+      "epoch": 0.228152799581371,
+      "grad_norm": 2.2024177758411088,
+      "learning_rate": 1.801094912464265e-05,
+      "loss": 1.1185,
       "step": 2180
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.893367663209896e-06,
-      "loss": 0.6446,
+      "epoch": 0.2282574568288854,
+      "grad_norm": 1.8831830056093484,
+      "learning_rate": 1.8008919790367854e-05,
+      "loss": 0.9436,
       "step": 2181
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.884926476186269e-06,
-      "loss": 0.6824,
+      "epoch": 0.22836211407639978,
+      "grad_norm": 2.5755945919781063,
+      "learning_rate": 1.800688953585194e-05,
+      "loss": 1.0213,
       "step": 2182
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.876492243182113e-06,
-      "loss": 0.7825,
+      "epoch": 0.2284667713239142,
+      "grad_norm": 2.3639675977701535,
+      "learning_rate": 1.8004858361328185e-05,
+      "loss": 1.009,
       "step": 2183
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.8680649737887784e-06,
-      "loss": 0.6295,
+      "epoch": 0.22857142857142856,
+      "grad_norm": 2.4516774443504716,
+      "learning_rate": 1.8002826267029977e-05,
+      "loss": 1.0469,
       "step": 2184
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.859644677589699e-06,
-      "loss": 0.5602,
+      "epoch": 0.22867608581894297,
+      "grad_norm": 2.2277531735938414,
+      "learning_rate": 1.800079325319081e-05,
+      "loss": 1.1314,
       "step": 2185
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.851231364160379e-06,
-      "loss": 0.6399,
+      "epoch": 0.22878074306645735,
+      "grad_norm": 2.1946636768725116,
+      "learning_rate": 1.7998759320044276e-05,
+      "loss": 1.1728,
       "step": 2186
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.842825043068377e-06,
-      "loss": 0.7782,
+      "epoch": 0.22888540031397175,
+      "grad_norm": 1.8628296808775904,
+      "learning_rate": 1.799672446782408e-05,
+      "loss": 0.7964,
       "step": 2187
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.834425723873303e-06,
-      "loss": 0.6725,
+      "epoch": 0.22899005756148613,
+      "grad_norm": 2.2414870196187984,
+      "learning_rate": 1.7994688696764037e-05,
+      "loss": 1.0465,
       "step": 2188
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.826033416126807e-06,
-      "loss": 0.6962,
+      "epoch": 0.2290947148090005,
+      "grad_norm": 2.1280884247041056,
+      "learning_rate": 1.799265200709805e-05,
+      "loss": 0.934,
       "step": 2189
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.817648129372563e-06,
-      "loss": 0.7005,
+      "epoch": 0.22919937205651492,
+      "grad_norm": 2.4695395417950308,
+      "learning_rate": 1.7990614399060144e-05,
+      "loss": 1.078,
       "step": 2190
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.809269873146263e-06,
-      "loss": 0.7011,
+      "epoch": 0.2293040293040293,
+      "grad_norm": 2.240850930676284,
+      "learning_rate": 1.798857587288445e-05,
+      "loss": 1.0257,
       "step": 2191
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.800898656975599e-06,
-      "loss": 0.658,
+      "epoch": 0.2294086865515437,
+      "grad_norm": 2.4333038713844988,
+      "learning_rate": 1.798653642880519e-05,
+      "loss": 0.9771,
       "step": 2192
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.792534490380263e-06,
-      "loss": 0.7342,
+      "epoch": 0.22951334379905808,
+      "grad_norm": 2.3919820342346827,
+      "learning_rate": 1.7984496067056704e-05,
+      "loss": 1.0281,
       "step": 2193
     },
     {
-      "epoch": 1.44,
-      "learning_rate": 3.7841773828719277e-06,
-      "loss": 0.7488,
+      "epoch": 0.22961800104657248,
+      "grad_norm": 2.9237043226574633,
+      "learning_rate": 1.7982454787873436e-05,
+      "loss": 1.0971,
       "step": 2194
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.775827343954239e-06,
-      "loss": 0.6383,
+      "epoch": 0.22972265829408686,
+      "grad_norm": 2.499578413555501,
+      "learning_rate": 1.7980412591489932e-05,
+      "loss": 1.0503,
       "step": 2195
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.767484383122799e-06,
-      "loss": 0.6609,
+      "epoch": 0.22982731554160127,
+      "grad_norm": 2.1481213076928793,
+      "learning_rate": 1.7978369478140838e-05,
+      "loss": 1.0634,
       "step": 2196
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.759148509865168e-06,
-      "loss": 0.6105,
+      "epoch": 0.22993197278911565,
+      "grad_norm": 2.0306762217088568,
+      "learning_rate": 1.797632544806092e-05,
+      "loss": 0.9728,
       "step": 2197
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.750819733660844e-06,
-      "loss": 0.7654,
+      "epoch": 0.23003663003663002,
+      "grad_norm": 2.483176372385859,
+      "learning_rate": 1.7974280501485037e-05,
+      "loss": 0.9275,
       "step": 2198
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.7424980639812534e-06,
-      "loss": 0.756,
+      "epoch": 0.23014128728414443,
+      "grad_norm": 2.319447402532949,
+      "learning_rate": 1.797223463864816e-05,
+      "loss": 1.146,
       "step": 2199
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.734183510289742e-06,
-      "loss": 0.5883,
+      "epoch": 0.2302459445316588,
+      "grad_norm": 2.277766050423453,
+      "learning_rate": 1.797018785978536e-05,
+      "loss": 0.8616,
       "step": 2200
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.725876082041563e-06,
-      "loss": 0.8041,
+      "epoch": 0.23035060177917321,
+      "grad_norm": 2.3369318419186786,
+      "learning_rate": 1.796814016513182e-05,
+      "loss": 1.0268,
       "step": 2201
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.717575788683866e-06,
-      "loss": 0.5943,
+      "epoch": 0.2304552590266876,
+      "grad_norm": 2.597419543719222,
+      "learning_rate": 1.7966091554922823e-05,
+      "loss": 0.9834,
       "step": 2202
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.7092826396556882e-06,
-      "loss": 0.7253,
+      "epoch": 0.230559916274202,
+      "grad_norm": 2.011269693810685,
+      "learning_rate": 1.7964042029393755e-05,
+      "loss": 0.9039,
       "step": 2203
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.700996644387944e-06,
-      "loss": 0.7387,
+      "epoch": 0.23066457352171638,
+      "grad_norm": 2.2717414735854837,
+      "learning_rate": 1.7961991588780114e-05,
+      "loss": 1.0691,
       "step": 2204
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.6927178123034034e-06,
-      "loss": 0.7315,
+      "epoch": 0.23076923076923078,
+      "grad_norm": 2.2621982011763326,
+      "learning_rate": 1.79599402333175e-05,
+      "loss": 1.0414,
       "step": 2205
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.684446152816702e-06,
-      "loss": 0.801,
+      "epoch": 0.23087388801674516,
+      "grad_norm": 2.1871162127683323,
+      "learning_rate": 1.7957887963241613e-05,
+      "loss": 1.0806,
       "step": 2206
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.6761816753343114e-06,
-      "loss": 0.7474,
+      "epoch": 0.23097854526425954,
+      "grad_norm": 2.2467526764365604,
+      "learning_rate": 1.7955834778788266e-05,
+      "loss": 1.0368,
       "step": 2207
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.6679243892545414e-06,
-      "loss": 0.6163,
+      "epoch": 0.23108320251177394,
+      "grad_norm": 2.49703081611621,
+      "learning_rate": 1.795378068019338e-05,
+      "loss": 1.0251,
       "step": 2208
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.659674303967522e-06,
-      "loss": 0.6888,
+      "epoch": 0.23118785975928832,
+      "grad_norm": 2.2621831608178393,
+      "learning_rate": 1.7951725667692967e-05,
+      "loss": 0.9758,
       "step": 2209
     },
     {
-      "epoch": 1.45,
-      "learning_rate": 3.6514314288551877e-06,
-      "loss": 0.7455,
+      "epoch": 0.23129251700680273,
+      "grad_norm": 2.8459265279372192,
+      "learning_rate": 1.7949669741523156e-05,
+      "loss": 1.0373,
       "step": 2210
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.6431957732912793e-06,
-      "loss": 0.7855,
+      "epoch": 0.2313971742543171,
+      "grad_norm": 2.6273563201754957,
+      "learning_rate": 1.7947612901920174e-05,
+      "loss": 0.8752,
       "step": 2211
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.634967346641334e-06,
-      "loss": 0.7015,
+      "epoch": 0.2315018315018315,
+      "grad_norm": 2.5197273005659264,
+      "learning_rate": 1.794555514912036e-05,
+      "loss": 0.9882,
       "step": 2212
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.6267461582626584e-06,
-      "loss": 0.7168,
+      "epoch": 0.2316064887493459,
+      "grad_norm": 2.274914592268484,
+      "learning_rate": 1.7943496483360152e-05,
+      "loss": 0.9677,
       "step": 2213
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.6185322175043336e-06,
-      "loss": 0.7709,
+      "epoch": 0.2317111459968603,
+      "grad_norm": 2.428626518972456,
+      "learning_rate": 1.7941436904876095e-05,
+      "loss": 1.1441,
       "step": 2214
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.6103255337071918e-06,
-      "loss": 0.7098,
+      "epoch": 0.23181580324437467,
+      "grad_norm": 2.394836969811865,
+      "learning_rate": 1.7939376413904836e-05,
+      "loss": 0.9864,
       "step": 2215
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.602126116203819e-06,
-      "loss": 0.717,
+      "epoch": 0.23192046049188905,
+      "grad_norm": 2.6595728488415182,
+      "learning_rate": 1.7937315010683135e-05,
+      "loss": 1.0768,
       "step": 2216
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.5939339743185376e-06,
-      "loss": 0.6801,
+      "epoch": 0.23202511773940346,
+      "grad_norm": 2.1906514810889846,
+      "learning_rate": 1.7935252695447846e-05,
+      "loss": 0.9923,
       "step": 2217
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.5857491173673976e-06,
-      "loss": 0.6834,
+      "epoch": 0.23212977498691784,
+      "grad_norm": 2.3158045248201065,
+      "learning_rate": 1.793318946843594e-05,
+      "loss": 1.0749,
       "step": 2218
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.5775715546581546e-06,
-      "loss": 0.7986,
+      "epoch": 0.23223443223443224,
+      "grad_norm": 2.2698926380173563,
+      "learning_rate": 1.793112532988448e-05,
+      "loss": 1.0514,
       "step": 2219
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.5694012954902803e-06,
-      "loss": 0.8078,
+      "epoch": 0.23233908948194662,
+      "grad_norm": 2.3318940874120564,
+      "learning_rate": 1.7929060280030642e-05,
+      "loss": 1.0105,
       "step": 2220
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.5612383491549384e-06,
-      "loss": 0.7479,
+      "epoch": 0.23244374672946103,
+      "grad_norm": 2.3711702488900857,
+      "learning_rate": 1.7926994319111704e-05,
+      "loss": 1.0331,
       "step": 2221
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.553082724934973e-06,
-      "loss": 0.6753,
+      "epoch": 0.2325484039769754,
+      "grad_norm": 2.150975144233381,
+      "learning_rate": 1.7924927447365048e-05,
+      "loss": 1.0613,
       "step": 2222
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.544934432104905e-06,
-      "loss": 0.7859,
+      "epoch": 0.23265306122448978,
+      "grad_norm": 2.536648515717417,
+      "learning_rate": 1.7922859665028164e-05,
+      "loss": 1.0065,
       "step": 2223
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.5367934799309167e-06,
-      "loss": 0.7151,
+      "epoch": 0.2327577184720042,
+      "grad_norm": 1.9397689780705203,
+      "learning_rate": 1.792079097233864e-05,
+      "loss": 0.9451,
       "step": 2224
     },
     {
-      "epoch": 1.46,
-      "learning_rate": 3.5286598776708436e-06,
-      "loss": 0.7203,
+      "epoch": 0.23286237571951857,
+      "grad_norm": 2.2704190613976114,
+      "learning_rate": 1.791872136953418e-05,
+      "loss": 0.9858,
       "step": 2225
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.5205336345741603e-06,
-      "loss": 0.7241,
+      "epoch": 0.23296703296703297,
+      "grad_norm": 2.8347645497669656,
+      "learning_rate": 1.7916650856852577e-05,
+      "loss": 1.1249,
       "step": 2226
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.512414759881976e-06,
-      "loss": 0.719,
+      "epoch": 0.23307169021454735,
+      "grad_norm": 2.408645013203526,
+      "learning_rate": 1.7914579434531746e-05,
+      "loss": 0.9415,
       "step": 2227
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.5043032628270225e-06,
-      "loss": 0.6583,
+      "epoch": 0.23317634746206176,
+      "grad_norm": 2.5151198920475557,
+      "learning_rate": 1.7912507102809692e-05,
+      "loss": 1.0078,
       "step": 2228
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.49619915263363e-06,
-      "loss": 0.8169,
+      "epoch": 0.23328100470957613,
+      "grad_norm": 2.049430149571946,
+      "learning_rate": 1.7910433861924533e-05,
+      "loss": 0.9601,
       "step": 2229
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.488102438517742e-06,
-      "loss": 0.6847,
+      "epoch": 0.23338566195709054,
+      "grad_norm": 2.406663254302629,
+      "learning_rate": 1.7908359712114484e-05,
+      "loss": 0.9539,
       "step": 2230
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.4800131296868842e-06,
-      "loss": 0.5998,
+      "epoch": 0.23349031920460492,
+      "grad_norm": 2.32058310508671,
+      "learning_rate": 1.7906284653617874e-05,
+      "loss": 1.0463,
       "step": 2231
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.4719312353401647e-06,
-      "loss": 0.7476,
+      "epoch": 0.2335949764521193,
+      "grad_norm": 2.1243497033328076,
+      "learning_rate": 1.790420868667313e-05,
+      "loss": 0.8934,
       "step": 2232
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.4638567646682576e-06,
-      "loss": 0.753,
+      "epoch": 0.2336996336996337,
+      "grad_norm": 2.172322288650889,
+      "learning_rate": 1.7902131811518784e-05,
+      "loss": 0.8929,
       "step": 2233
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.4557897268533935e-06,
-      "loss": 0.6882,
+      "epoch": 0.23380429094714808,
+      "grad_norm": 2.626463122324986,
+      "learning_rate": 1.7900054028393475e-05,
+      "loss": 0.9942,
       "step": 2234
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.4477301310693544e-06,
-      "loss": 0.756,
+      "epoch": 0.23390894819466249,
+      "grad_norm": 2.277059459157924,
+      "learning_rate": 1.789797533753594e-05,
+      "loss": 1.0414,
       "step": 2235
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.4396779864814568e-06,
-      "loss": 0.5769,
+      "epoch": 0.23401360544217686,
+      "grad_norm": 2.676163205276392,
+      "learning_rate": 1.789589573918503e-05,
+      "loss": 0.8041,
       "step": 2236
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.4316333022465465e-06,
-      "loss": 0.7097,
+      "epoch": 0.23411826268969127,
+      "grad_norm": 2.1358940445833188,
+      "learning_rate": 1.7893815233579692e-05,
+      "loss": 1.0874,
       "step": 2237
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.423596087512976e-06,
-      "loss": 0.5898,
+      "epoch": 0.23422291993720565,
+      "grad_norm": 2.499667365100757,
+      "learning_rate": 1.7891733820958983e-05,
+      "loss": 0.9692,
       "step": 2238
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.415566351420615e-06,
-      "loss": 0.6184,
+      "epoch": 0.23432757718472005,
+      "grad_norm": 2.8983591306679104,
+      "learning_rate": 1.788965150156206e-05,
+      "loss": 0.8887,
       "step": 2239
     },
     {
-      "epoch": 1.47,
-      "learning_rate": 3.407544103100824e-06,
-      "loss": 0.6735,
+      "epoch": 0.23443223443223443,
+      "grad_norm": 2.4745131578348856,
+      "learning_rate": 1.788756827562818e-05,
+      "loss": 0.975,
       "step": 2240
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.399529351676447e-06,
-      "loss": 0.7342,
+      "epoch": 0.2345368916797488,
+      "grad_norm": 2.3648247665584434,
+      "learning_rate": 1.788548414339672e-05,
+      "loss": 0.924,
       "step": 2241
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.39152210626181e-06,
-      "loss": 0.5299,
+      "epoch": 0.23464154892726322,
+      "grad_norm": 2.3253167404589643,
+      "learning_rate": 1.7883399105107146e-05,
+      "loss": 1.0481,
       "step": 2242
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.3835223759626845e-06,
-      "loss": 0.7169,
+      "epoch": 0.2347462061747776,
+      "grad_norm": 2.2078932669131306,
+      "learning_rate": 1.7881313160999035e-05,
+      "loss": 1.0526,
       "step": 2243
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.3755301698763197e-06,
-      "loss": 0.7386,
+      "epoch": 0.234850863422292,
+      "grad_norm": 2.260715971363778,
+      "learning_rate": 1.7879226311312063e-05,
+      "loss": 0.9958,
       "step": 2244
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.367545497091396e-06,
-      "loss": 0.6406,
+      "epoch": 0.23495552066980638,
+      "grad_norm": 2.103577147400585,
+      "learning_rate": 1.7877138556286014e-05,
+      "loss": 1.0592,
       "step": 2245
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.3595683666880286e-06,
-      "loss": 0.7105,
+      "epoch": 0.23506017791732078,
+      "grad_norm": 2.7075505673940774,
+      "learning_rate": 1.787504989616078e-05,
+      "loss": 1.0385,
       "step": 2246
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.3515987877377498e-06,
-      "loss": 0.7014,
+      "epoch": 0.23516483516483516,
+      "grad_norm": 2.6763027546096807,
+      "learning_rate": 1.7872960331176347e-05,
+      "loss": 0.972,
       "step": 2247
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.343636769303512e-06,
-      "loss": 0.7209,
+      "epoch": 0.23526949241234957,
+      "grad_norm": 2.3328246332033227,
+      "learning_rate": 1.7870869861572807e-05,
+      "loss": 0.8997,
       "step": 2248
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.3356823204396706e-06,
-      "loss": 0.6988,
+      "epoch": 0.23537414965986395,
+      "grad_norm": 2.4266627069351108,
+      "learning_rate": 1.786877848759037e-05,
+      "loss": 0.9698,
       "step": 2249
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.3277354501919677e-06,
-      "loss": 0.7941,
+      "epoch": 0.23547880690737832,
+      "grad_norm": 2.093701504040857,
+      "learning_rate": 1.7866686209469328e-05,
+      "loss": 0.871,
       "step": 2250
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.319796167597533e-06,
-      "loss": 0.7455,
+      "epoch": 0.23558346415489273,
+      "grad_norm": 2.4270676083889087,
+      "learning_rate": 1.7864593027450097e-05,
+      "loss": 1.156,
       "step": 2251
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.3118644816848577e-06,
-      "loss": 0.7725,
+      "epoch": 0.2356881214024071,
+      "grad_norm": 2.6995988050304716,
+      "learning_rate": 1.7862498941773184e-05,
+      "loss": 1.0907,
       "step": 2252
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.303940401473804e-06,
-      "loss": 0.6013,
+      "epoch": 0.23579277864992151,
+      "grad_norm": 2.0633497910278256,
+      "learning_rate": 1.78604039526792e-05,
+      "loss": 0.9509,
       "step": 2253
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.2960239359755808e-06,
-      "loss": 0.759,
+      "epoch": 0.2358974358974359,
+      "grad_norm": 2.337404008068736,
+      "learning_rate": 1.785830806040887e-05,
+      "loss": 1.0717,
       "step": 2254
     },
     {
-      "epoch": 1.48,
-      "learning_rate": 3.288115094192734e-06,
-      "loss": 0.6683,
+      "epoch": 0.2360020931449503,
+      "grad_norm": 2.2055920312977295,
+      "learning_rate": 1.785621126520301e-05,
+      "loss": 0.8269,
       "step": 2255
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.2802138851191558e-06,
-      "loss": 0.8061,
+      "epoch": 0.23610675039246468,
+      "grad_norm": 2.5981676845560764,
+      "learning_rate": 1.7854113567302557e-05,
+      "loss": 0.9251,
       "step": 2256
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.2723203177400352e-06,
-      "loss": 0.7195,
+      "epoch": 0.23621140763997908,
+      "grad_norm": 2.183621120398911,
+      "learning_rate": 1.785201496694853e-05,
+      "loss": 1.0284,
       "step": 2257
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.264434401031887e-06,
-      "loss": 0.7414,
+      "epoch": 0.23631606488749346,
+      "grad_norm": 2.2329023327526,
+      "learning_rate": 1.784991546438206e-05,
+      "loss": 1.0342,
       "step": 2258
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.2565561439625205e-06,
-      "loss": 0.6787,
+      "epoch": 0.23642072213500784,
+      "grad_norm": 2.871738043676333,
+      "learning_rate": 1.7847815059844395e-05,
+      "loss": 1.1025,
       "step": 2259
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.248685555491039e-06,
-      "loss": 0.8007,
+      "epoch": 0.23652537938252224,
+      "grad_norm": 2.4479413257311378,
+      "learning_rate": 1.784571375357687e-05,
+      "loss": 1.0315,
       "step": 2260
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.240822644567815e-06,
-      "loss": 0.7207,
+      "epoch": 0.23663003663003662,
+      "grad_norm": 2.315149134540888,
+      "learning_rate": 1.7843611545820926e-05,
+      "loss": 1.1481,
       "step": 2261
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.2329674201344997e-06,
-      "loss": 0.6228,
+      "epoch": 0.23673469387755103,
+      "grad_norm": 2.4131417962503177,
+      "learning_rate": 1.784150843681812e-05,
+      "loss": 0.9819,
       "step": 2262
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.2251198911240002e-06,
-      "loss": 0.6543,
+      "epoch": 0.2368393511250654,
+      "grad_norm": 2.009874656026206,
+      "learning_rate": 1.7839404426810095e-05,
+      "loss": 0.8828,
       "step": 2263
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.217280066460472e-06,
-      "loss": 0.8337,
+      "epoch": 0.2369440083725798,
+      "grad_norm": 2.30932078574492,
+      "learning_rate": 1.7837299516038608e-05,
+      "loss": 1.0608,
       "step": 2264
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.2094479550593107e-06,
-      "loss": 0.6155,
+      "epoch": 0.2370486656200942,
+      "grad_norm": 2.056529909679889,
+      "learning_rate": 1.7835193704745523e-05,
+      "loss": 0.931,
       "step": 2265
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.201623565827139e-06,
-      "loss": 0.7811,
+      "epoch": 0.23715332286760857,
+      "grad_norm": 1.9976085370847383,
+      "learning_rate": 1.7833086993172797e-05,
+      "loss": 0.8794,
       "step": 2266
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.1938069076617983e-06,
-      "loss": 0.6261,
+      "epoch": 0.23725798011512297,
+      "grad_norm": 2.3451892003689854,
+      "learning_rate": 1.7830979381562493e-05,
+      "loss": 1.0935,
       "step": 2267
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.1859979894523375e-06,
-      "loss": 0.8209,
+      "epoch": 0.23736263736263735,
+      "grad_norm": 2.3332512825866756,
+      "learning_rate": 1.7828870870156783e-05,
+      "loss": 1.0952,
       "step": 2268
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.1781968200790094e-06,
-      "loss": 0.686,
+      "epoch": 0.23746729461015176,
+      "grad_norm": 2.4752809555669257,
+      "learning_rate": 1.7826761459197943e-05,
+      "loss": 0.9602,
       "step": 2269
     },
     {
-      "epoch": 1.49,
-      "learning_rate": 3.170403408413243e-06,
-      "loss": 0.788,
+      "epoch": 0.23757195185766614,
+      "grad_norm": 2.2358190924088017,
+      "learning_rate": 1.7824651148928343e-05,
+      "loss": 1.1246,
       "step": 2270
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.1626177633176557e-06,
-      "loss": 0.7175,
+      "epoch": 0.23767660910518054,
+      "grad_norm": 2.6059633471148915,
+      "learning_rate": 1.782253993959046e-05,
+      "loss": 1.0941,
       "step": 2271
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.154839893646028e-06,
-      "loss": 0.7192,
+      "epoch": 0.23778126635269492,
+      "grad_norm": 2.084559757590681,
+      "learning_rate": 1.7820427831426887e-05,
+      "loss": 0.9926,
       "step": 2272
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.1470698082433014e-06,
-      "loss": 0.7721,
+      "epoch": 0.23788592360020933,
+      "grad_norm": 2.2523944777498968,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.9949,
       "step": 2273
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.139307515945563e-06,
-      "loss": 0.7285,
+      "epoch": 0.2379905808477237,
+      "grad_norm": 2.8181662408595187,
+      "learning_rate": 1.781620091959349e-05,
+      "loss": 1.012,
       "step": 2274
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.1315530255800384e-06,
-      "loss": 0.6929,
+      "epoch": 0.23809523809523808,
+      "grad_norm": 2.328558412920668,
+      "learning_rate": 1.781408611640935e-05,
+      "loss": 1.0071,
       "step": 2275
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.1238063459650802e-06,
-      "loss": 0.6265,
+      "epoch": 0.2381998953427525,
+      "grad_norm": 2.183389318821081,
+      "learning_rate": 1.781197041537087e-05,
+      "loss": 1.0634,
       "step": 2276
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.116067485910159e-06,
-      "loss": 0.6691,
+      "epoch": 0.23830455259026687,
+      "grad_norm": 2.2754131017205417,
+      "learning_rate": 1.7809853816721157e-05,
+      "loss": 0.9493,
       "step": 2277
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.1083364542158535e-06,
-      "loss": 0.7067,
+      "epoch": 0.23840920983778127,
+      "grad_norm": 2.180076226224215,
+      "learning_rate": 1.7807736320703405e-05,
+      "loss": 0.98,
       "step": 2278
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.1006132596738426e-06,
-      "loss": 0.7177,
+      "epoch": 0.23851386708529565,
+      "grad_norm": 2.1011853330474772,
+      "learning_rate": 1.7805617927560926e-05,
+      "loss": 1.0353,
       "step": 2279
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.092897911066883e-06,
-      "loss": 0.661,
+      "epoch": 0.23861852433281006,
+      "grad_norm": 2.685310139913174,
+      "learning_rate": 1.780349863753712e-05,
+      "loss": 0.8339,
       "step": 2280
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.0851904171688196e-06,
-      "loss": 0.659,
+      "epoch": 0.23872318158032443,
+      "grad_norm": 2.2643234179394183,
+      "learning_rate": 1.7801378450875504e-05,
+      "loss": 0.9221,
       "step": 2281
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.0774907867445623e-06,
-      "loss": 0.6698,
+      "epoch": 0.23882783882783884,
+      "grad_norm": 2.0845511977759914,
+      "learning_rate": 1.7799257367819687e-05,
+      "loss": 1.0691,
       "step": 2282
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.0697990285500802e-06,
-      "loss": 0.698,
+      "epoch": 0.23893249607535322,
+      "grad_norm": 2.1996457975610286,
+      "learning_rate": 1.779713538861339e-05,
+      "loss": 1.0552,
       "step": 2283
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.06211515133238e-06,
-      "loss": 0.5568,
+      "epoch": 0.2390371533228676,
+      "grad_norm": 2.563382788930006,
+      "learning_rate": 1.7795012513500427e-05,
+      "loss": 1.0386,
       "step": 2284
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.054439163829519e-06,
-      "loss": 0.6433,
+      "epoch": 0.239141810570382,
+      "grad_norm": 2.482883361316951,
+      "learning_rate": 1.7792888742724727e-05,
+      "loss": 1.116,
       "step": 2285
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 3.0467710747705735e-06,
-      "loss": 0.678,
+      "epoch": 0.23924646781789638,
+      "grad_norm": 2.173594595079943,
+      "learning_rate": 1.7790764076530312e-05,
+      "loss": 1.0827,
       "step": 2286
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 3.0391108928756486e-06,
-      "loss": 0.6623,
+      "epoch": 0.23935112506541079,
+      "grad_norm": 1.973340347588087,
+      "learning_rate": 1.778863851516131e-05,
+      "loss": 1.068,
       "step": 2287
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 3.0314586268558486e-06,
-      "loss": 0.6125,
+      "epoch": 0.23945578231292516,
+      "grad_norm": 2.266375557620404,
+      "learning_rate": 1.7786512058861952e-05,
+      "loss": 0.8809,
       "step": 2288
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 3.0238142854132722e-06,
-      "loss": 0.7927,
+      "epoch": 0.23956043956043957,
+      "grad_norm": 1.952846674363708,
+      "learning_rate": 1.7784384707876576e-05,
+      "loss": 1.0294,
       "step": 2289
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 3.0161778772410153e-06,
-      "loss": 0.73,
+      "epoch": 0.23966509680795395,
+      "grad_norm": 2.2539440156526975,
+      "learning_rate": 1.7782256462449615e-05,
+      "loss": 1.0162,
       "step": 2290
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 3.0085494110231493e-06,
-      "loss": 0.7054,
+      "epoch": 0.23976975405546835,
+      "grad_norm": 2.278316834840628,
+      "learning_rate": 1.7780127322825615e-05,
+      "loss": 1.0605,
       "step": 2291
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 3.0009288954347115e-06,
-      "loss": 0.6187,
+      "epoch": 0.23987441130298273,
+      "grad_norm": 2.1166188092910674,
+      "learning_rate": 1.777799728924921e-05,
+      "loss": 0.9187,
       "step": 2292
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.9933163391417042e-06,
-      "loss": 0.7471,
+      "epoch": 0.2399790685504971,
+      "grad_norm": 2.2188937843904473,
+      "learning_rate": 1.7775866361965145e-05,
+      "loss": 0.9865,
       "step": 2293
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.985711750801068e-06,
-      "loss": 0.6812,
+      "epoch": 0.24008372579801152,
+      "grad_norm": 2.2834201572097936,
+      "learning_rate": 1.7773734541218277e-05,
+      "loss": 1.027,
       "step": 2294
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.97811513906069e-06,
-      "loss": 0.81,
+      "epoch": 0.2401883830455259,
+      "grad_norm": 2.3639841943193223,
+      "learning_rate": 1.777160182725355e-05,
+      "loss": 0.998,
       "step": 2295
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.9705265125593863e-06,
-      "loss": 0.7013,
+      "epoch": 0.2402930402930403,
+      "grad_norm": 2.6545219995276783,
+      "learning_rate": 1.7769468220316016e-05,
+      "loss": 1.0317,
       "step": 2296
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.9629458799268897e-06,
-      "loss": 0.6364,
+      "epoch": 0.24039769754055468,
+      "grad_norm": 2.806578222987241,
+      "learning_rate": 1.7767333720650836e-05,
+      "loss": 1.0528,
       "step": 2297
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.955373249783844e-06,
-      "loss": 0.7189,
+      "epoch": 0.24050235478806908,
+      "grad_norm": 2.450717164421872,
+      "learning_rate": 1.7765198328503262e-05,
+      "loss": 1.0429,
       "step": 2298
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.947808630741791e-06,
-      "loss": 0.6564,
+      "epoch": 0.24060701203558346,
+      "grad_norm": 2.3826191374301326,
+      "learning_rate": 1.7763062044118662e-05,
+      "loss": 0.9769,
       "step": 2299
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.9402520314031646e-06,
-      "loss": 0.6106,
+      "epoch": 0.24071166928309787,
+      "grad_norm": 2.5379042446021782,
+      "learning_rate": 1.7760924867742493e-05,
+      "loss": 1.0339,
       "step": 2300
     },
     {
-      "epoch": 1.51,
-      "learning_rate": 2.932703460361277e-06,
-      "loss": 0.7337,
+      "epoch": 0.24081632653061225,
+      "grad_norm": 2.3669619873635352,
+      "learning_rate": 1.7758786799620327e-05,
+      "loss": 1.0572,
       "step": 2301
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.9251629262003135e-06,
-      "loss": 0.5722,
+      "epoch": 0.24092098377812662,
+      "grad_norm": 2.2289468079215142,
+      "learning_rate": 1.7756647839997824e-05,
+      "loss": 1.0289,
       "step": 2302
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.9176304374953123e-06,
-      "loss": 0.6796,
+      "epoch": 0.24102564102564103,
+      "grad_norm": 2.2529359568142517,
+      "learning_rate": 1.7754507989120762e-05,
+      "loss": 0.9694,
       "step": 2303
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.910106002812172e-06,
-      "loss": 0.7378,
+      "epoch": 0.2411302982731554,
+      "grad_norm": 2.222490094675877,
+      "learning_rate": 1.7752367247235015e-05,
+      "loss": 0.9996,
       "step": 2304
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.902589630707624e-06,
-      "loss": 0.5941,
+      "epoch": 0.2412349555206698,
+      "grad_norm": 2.3032261423529254,
+      "learning_rate": 1.7750225614586557e-05,
+      "loss": 1.0049,
       "step": 2305
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.895081329729239e-06,
-      "loss": 0.6849,
+      "epoch": 0.2413396127681842,
+      "grad_norm": 2.158558091358129,
+      "learning_rate": 1.774808309142146e-05,
+      "loss": 1.013,
       "step": 2306
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.8875811084154025e-06,
-      "loss": 0.7581,
+      "epoch": 0.2414442700156986,
+      "grad_norm": 2.0156168291391134,
+      "learning_rate": 1.7745939677985912e-05,
+      "loss": 0.9208,
       "step": 2307
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.8800889752953156e-06,
-      "loss": 0.6943,
+      "epoch": 0.24154892726321298,
+      "grad_norm": 2.2427438503410353,
+      "learning_rate": 1.7743795374526186e-05,
+      "loss": 1.1672,
       "step": 2308
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.872604938888981e-06,
-      "loss": 0.655,
+      "epoch": 0.24165358451072738,
+      "grad_norm": 2.416172594533138,
+      "learning_rate": 1.774165018128868e-05,
+      "loss": 1.0452,
       "step": 2309
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.8651290077071914e-06,
-      "loss": 0.6559,
+      "epoch": 0.24175824175824176,
+      "grad_norm": 2.5723958354089254,
+      "learning_rate": 1.7739504098519872e-05,
+      "loss": 0.973,
       "step": 2310
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.8576611902515283e-06,
-      "loss": 0.6779,
+      "epoch": 0.24186289900575614,
+      "grad_norm": 2.9877649086017666,
+      "learning_rate": 1.773735712646635e-05,
+      "loss": 0.9828,
       "step": 2311
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.8502014950143376e-06,
-      "loss": 0.6959,
+      "epoch": 0.24196755625327054,
+      "grad_norm": 2.3078412731916367,
+      "learning_rate": 1.7735209265374816e-05,
+      "loss": 1.0852,
       "step": 2312
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.842749930478734e-06,
-      "loss": 0.6175,
+      "epoch": 0.24207221350078492,
+      "grad_norm": 2.1969741515349233,
+      "learning_rate": 1.773306051549205e-05,
+      "loss": 1.0456,
       "step": 2313
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.8353065051185867e-06,
-      "loss": 0.7628,
+      "epoch": 0.24217687074829933,
+      "grad_norm": 2.5003537240671885,
+      "learning_rate": 1.7730910877064955e-05,
+      "loss": 0.9049,
       "step": 2314
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.827871227398508e-06,
-      "loss": 0.7496,
+      "epoch": 0.2422815279958137,
+      "grad_norm": 2.519922133498177,
+      "learning_rate": 1.772876035034053e-05,
+      "loss": 1.0232,
       "step": 2315
     },
     {
-      "epoch": 1.52,
-      "learning_rate": 2.820444105773845e-06,
-      "loss": 0.6525,
+      "epoch": 0.2423861852433281,
+      "grad_norm": 2.477645819821275,
+      "learning_rate": 1.7726608935565874e-05,
+      "loss": 0.9229,
       "step": 2316
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.8130251486906666e-06,
-      "loss": 0.7134,
+      "epoch": 0.2424908424908425,
+      "grad_norm": 2.6081751548110685,
+      "learning_rate": 1.7724456632988188e-05,
+      "loss": 1.0518,
       "step": 2317
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.805614364585758e-06,
-      "loss": 0.7566,
+      "epoch": 0.24259549973835687,
+      "grad_norm": 2.5904924581138005,
+      "learning_rate": 1.7722303442854774e-05,
+      "loss": 0.8876,
       "step": 2318
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.7982117618866177e-06,
-      "loss": 0.7241,
+      "epoch": 0.24270015698587127,
+      "grad_norm": 2.2369477600349845,
+      "learning_rate": 1.7720149365413036e-05,
+      "loss": 0.9971,
       "step": 2319
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.790817349011432e-06,
-      "loss": 0.6949,
+      "epoch": 0.24280481423338565,
+      "grad_norm": 2.4995571839955786,
+      "learning_rate": 1.771799440091049e-05,
+      "loss": 0.8509,
       "step": 2320
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.783431134369079e-06,
-      "loss": 0.6533,
+      "epoch": 0.24290947148090006,
+      "grad_norm": 2.2952550447378774,
+      "learning_rate": 1.7715838549594744e-05,
+      "loss": 1.0066,
       "step": 2321
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.776053126359105e-06,
-      "loss": 0.6817,
+      "epoch": 0.24301412872841444,
+      "grad_norm": 2.190996661293464,
+      "learning_rate": 1.7713681811713504e-05,
+      "loss": 1.0152,
       "step": 2322
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.7686833333717334e-06,
-      "loss": 0.6128,
+      "epoch": 0.24311878597592884,
+      "grad_norm": 2.259684687855643,
+      "learning_rate": 1.7711524187514586e-05,
+      "loss": 0.8633,
       "step": 2323
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.761321763787841e-06,
-      "loss": 0.6683,
+      "epoch": 0.24322344322344322,
+      "grad_norm": 2.121917767970034,
+      "learning_rate": 1.7709365677245906e-05,
+      "loss": 0.9001,
       "step": 2324
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.753968425978958e-06,
-      "loss": 0.6115,
+      "epoch": 0.24332810047095763,
+      "grad_norm": 2.2233654020848475,
+      "learning_rate": 1.7707206281155482e-05,
+      "loss": 1.1153,
       "step": 2325
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.746623328307243e-06,
-      "loss": 0.6595,
+      "epoch": 0.243432757718472,
+      "grad_norm": 2.444459263315588,
+      "learning_rate": 1.770504599949143e-05,
+      "loss": 1.0226,
       "step": 2326
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.739286479125494e-06,
-      "loss": 0.7621,
+      "epoch": 0.24353741496598638,
+      "grad_norm": 2.349907556465352,
+      "learning_rate": 1.7702884832501978e-05,
+      "loss": 1.0579,
       "step": 2327
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.7319578867771237e-06,
-      "loss": 0.6654,
+      "epoch": 0.2436420722135008,
+      "grad_norm": 2.484238355922039,
+      "learning_rate": 1.7700722780435437e-05,
+      "loss": 0.9627,
       "step": 2328
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.7246375595961584e-06,
-      "loss": 0.6504,
+      "epoch": 0.24374672946101517,
+      "grad_norm": 2.4085127927382732,
+      "learning_rate": 1.7698559843540242e-05,
+      "loss": 1.0758,
       "step": 2329
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.7173255059072234e-06,
-      "loss": 0.6322,
+      "epoch": 0.24385138670852957,
+      "grad_norm": 2.4015208510522745,
+      "learning_rate": 1.769639602206491e-05,
+      "loss": 1.0412,
       "step": 2330
     },
     {
-      "epoch": 1.53,
-      "learning_rate": 2.7100217340255352e-06,
-      "loss": 0.6879,
+      "epoch": 0.24395604395604395,
+      "grad_norm": 2.2260639525963675,
+      "learning_rate": 1.769423131625808e-05,
+      "loss": 1.0326,
       "step": 2331
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.7027262522568944e-06,
-      "loss": 0.736,
+      "epoch": 0.24406070120355836,
+      "grad_norm": 2.2914496037327288,
+      "learning_rate": 1.769206572636847e-05,
+      "loss": 1.1783,
       "step": 2332
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.6954390688976706e-06,
-      "loss": 0.6613,
+      "epoch": 0.24416535845107273,
+      "grad_norm": 2.226935286535785,
+      "learning_rate": 1.7689899252644912e-05,
+      "loss": 0.9281,
       "step": 2333
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.6881601922348e-06,
-      "loss": 0.729,
+      "epoch": 0.24427001569858714,
+      "grad_norm": 2.835666597627508,
+      "learning_rate": 1.768773189533634e-05,
+      "loss": 1.016,
       "step": 2334
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.6808896305457733e-06,
-      "loss": 0.5926,
+      "epoch": 0.24437467294610152,
+      "grad_norm": 2.7851995018242417,
+      "learning_rate": 1.7685563654691797e-05,
+      "loss": 1.0492,
       "step": 2335
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.6736273920986166e-06,
-      "loss": 0.6028,
+      "epoch": 0.2444793301936159,
+      "grad_norm": 2.1483770409252445,
+      "learning_rate": 1.7683394530960402e-05,
+      "loss": 1.0236,
       "step": 2336
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.6663734851519017e-06,
-      "loss": 0.66,
+      "epoch": 0.2445839874411303,
+      "grad_norm": 2.170226481420682,
+      "learning_rate": 1.76812245243914e-05,
+      "loss": 1.0287,
       "step": 2337
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.65912791795472e-06,
-      "loss": 0.6421,
+      "epoch": 0.24468864468864468,
+      "grad_norm": 2.5889214185058895,
+      "learning_rate": 1.7679053635234134e-05,
+      "loss": 0.9133,
       "step": 2338
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.651890698746682e-06,
-      "loss": 0.7681,
+      "epoch": 0.24479330193615909,
+      "grad_norm": 2.526081416072903,
+      "learning_rate": 1.7676881863738034e-05,
+      "loss": 1.1081,
       "step": 2339
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.6446618357579035e-06,
-      "loss": 0.5866,
+      "epoch": 0.24489795918367346,
+      "grad_norm": 1.928523666124805,
+      "learning_rate": 1.767470921015265e-05,
+      "loss": 0.7997,
       "step": 2340
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.6374413372089968e-06,
-      "loss": 0.637,
+      "epoch": 0.24500261643118787,
+      "grad_norm": 2.20948369983928,
+      "learning_rate": 1.767253567472761e-05,
+      "loss": 0.9804,
       "step": 2341
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.630229211311064e-06,
-      "loss": 0.6654,
+      "epoch": 0.24510727367870225,
+      "grad_norm": 2.0194710314361632,
+      "learning_rate": 1.7670361257712674e-05,
+      "loss": 0.987,
       "step": 2342
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.623025466265684e-06,
-      "loss": 0.5687,
+      "epoch": 0.24521193092621665,
+      "grad_norm": 2.5584737884538935,
+      "learning_rate": 1.766818595935768e-05,
+      "loss": 1.0407,
       "step": 2343
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.6158301102649085e-06,
-      "loss": 0.6714,
+      "epoch": 0.24531658817373103,
+      "grad_norm": 2.236875174210803,
+      "learning_rate": 1.7666009779912574e-05,
+      "loss": 1.0802,
       "step": 2344
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.6086431514912438e-06,
-      "loss": 0.7063,
+      "epoch": 0.2454212454212454,
+      "grad_norm": 2.0282577109477686,
+      "learning_rate": 1.7663832719627404e-05,
+      "loss": 0.953,
       "step": 2345
     },
     {
-      "epoch": 1.54,
-      "learning_rate": 2.6014645981176488e-06,
-      "loss": 0.7165,
+      "epoch": 0.24552590266875982,
+      "grad_norm": 2.073084277889144,
+      "learning_rate": 1.7661654778752315e-05,
+      "loss": 1.0081,
       "step": 2346
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.594294458307528e-06,
-      "loss": 0.6265,
+      "epoch": 0.2456305599162742,
+      "grad_norm": 2.149914350909731,
+      "learning_rate": 1.765947595753756e-05,
+      "loss": 1.0456,
       "step": 2347
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.5871327402147173e-06,
-      "loss": 0.6936,
+      "epoch": 0.2457352171637886,
+      "grad_norm": 2.2340276298746535,
+      "learning_rate": 1.7657296256233494e-05,
+      "loss": 1.0553,
       "step": 2348
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.5799794519834632e-06,
-      "loss": 0.5665,
+      "epoch": 0.24583987441130298,
+      "grad_norm": 2.0128834946003784,
+      "learning_rate": 1.7655115675090566e-05,
+      "loss": 1.0138,
       "step": 2349
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.5728346017484453e-06,
-      "loss": 0.7499,
+      "epoch": 0.24594453165881738,
+      "grad_norm": 2.032422061077809,
+      "learning_rate": 1.7652934214359324e-05,
+      "loss": 1.005,
       "step": 2350
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.565698197634735e-06,
-      "loss": 0.6924,
+      "epoch": 0.24604918890633176,
+      "grad_norm": 2.2528606162122515,
+      "learning_rate": 1.7650751874290427e-05,
+      "loss": 1.1266,
       "step": 2351
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.5585702477578024e-06,
-      "loss": 0.7012,
+      "epoch": 0.24615384615384617,
+      "grad_norm": 2.1025231064248566,
+      "learning_rate": 1.7648568655134633e-05,
+      "loss": 1.1058,
       "step": 2352
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.5514507602235063e-06,
-      "loss": 0.7002,
+      "epoch": 0.24625850340136055,
+      "grad_norm": 2.1702704933156896,
+      "learning_rate": 1.7646384557142796e-05,
+      "loss": 1.0684,
       "step": 2353
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.5443397431280703e-06,
-      "loss": 0.6393,
+      "epoch": 0.24636316064887492,
+      "grad_norm": 2.343302825772316,
+      "learning_rate": 1.764419958056587e-05,
+      "loss": 1.0974,
       "step": 2354
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.5372372045581006e-06,
-      "loss": 0.6277,
+      "epoch": 0.24646781789638933,
+      "grad_norm": 2.3639946811692045,
+      "learning_rate": 1.7642013725654915e-05,
+      "loss": 0.9979,
       "step": 2355
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.530143152590553e-06,
-      "loss": 0.7114,
+      "epoch": 0.2465724751439037,
+      "grad_norm": 2.336282852242526,
+      "learning_rate": 1.763982699266109e-05,
+      "loss": 1.0527,
       "step": 2356
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.523057595292734e-06,
-      "loss": 0.6846,
+      "epoch": 0.2466771323914181,
+      "grad_norm": 2.2842090040435368,
+      "learning_rate": 1.7637639381835658e-05,
+      "loss": 0.959,
       "step": 2357
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.515980540722295e-06,
-      "loss": 0.6611,
+      "epoch": 0.2467817896389325,
+      "grad_norm": 1.9972596299075889,
+      "learning_rate": 1.7635450893429977e-05,
+      "loss": 0.9649,
       "step": 2358
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.508911996927207e-06,
-      "loss": 0.5545,
+      "epoch": 0.2468864468864469,
+      "grad_norm": 1.9988043374236892,
+      "learning_rate": 1.763326152769551e-05,
+      "loss": 1.1658,
       "step": 2359
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.5018519719457725e-06,
-      "loss": 0.6492,
+      "epoch": 0.24699110413396128,
+      "grad_norm": 2.0588021433267896,
+      "learning_rate": 1.763107128488382e-05,
+      "loss": 1.0258,
       "step": 2360
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.4948004738066013e-06,
-      "loss": 0.582,
+      "epoch": 0.24709576138147565,
+      "grad_norm": 1.8656858774757512,
+      "learning_rate": 1.7628880165246572e-05,
+      "loss": 0.9793,
       "step": 2361
     },
     {
-      "epoch": 1.55,
-      "learning_rate": 2.4877575105286146e-06,
-      "loss": 0.6536,
+      "epoch": 0.24720041862899006,
+      "grad_norm": 2.218035457125112,
+      "learning_rate": 1.7626688169035524e-05,
+      "loss": 0.9543,
       "step": 2362
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.4807230901210155e-06,
-      "loss": 0.5477,
+      "epoch": 0.24730507587650444,
+      "grad_norm": 1.9653508672118056,
+      "learning_rate": 1.7624495296502545e-05,
+      "loss": 0.8786,
       "step": 2363
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.4736972205832997e-06,
-      "loss": 0.7073,
+      "epoch": 0.24740973312401884,
+      "grad_norm": 1.9480601105860273,
+      "learning_rate": 1.7622301547899605e-05,
+      "loss": 0.8262,
       "step": 2364
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.4666799099052384e-06,
-      "loss": 0.6569,
+      "epoch": 0.24751439037153322,
+      "grad_norm": 2.333099449535231,
+      "learning_rate": 1.7620106923478756e-05,
+      "loss": 1.0377,
       "step": 2365
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.459671166066869e-06,
-      "loss": 0.6961,
+      "epoch": 0.24761904761904763,
+      "grad_norm": 2.0818968205046455,
+      "learning_rate": 1.761791142349218e-05,
+      "loss": 0.8524,
       "step": 2366
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.45267099703849e-06,
-      "loss": 0.7883,
+      "epoch": 0.247723704866562,
+      "grad_norm": 2.3374344327141303,
+      "learning_rate": 1.761571504819214e-05,
+      "loss": 1.1216,
       "step": 2367
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.445679410780638e-06,
-      "loss": 0.7122,
+      "epoch": 0.2478283621140764,
+      "grad_norm": 2.4268357445524424,
+      "learning_rate": 1.7613517797830995e-05,
+      "loss": 1.0762,
       "step": 2368
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.4386964152441016e-06,
-      "loss": 0.8225,
+      "epoch": 0.2479330193615908,
+      "grad_norm": 2.590696844481067,
+      "learning_rate": 1.7611319672661227e-05,
+      "loss": 0.9002,
       "step": 2369
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.431722018369894e-06,
-      "loss": 0.6341,
+      "epoch": 0.24803767660910517,
+      "grad_norm": 2.909652369590741,
+      "learning_rate": 1.7609120672935396e-05,
+      "loss": 1.0978,
       "step": 2370
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.424756228089251e-06,
-      "loss": 0.5397,
+      "epoch": 0.24814233385661957,
+      "grad_norm": 2.576806230412862,
+      "learning_rate": 1.7606920798906176e-05,
+      "loss": 1.0588,
       "step": 2371
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.4177990523236216e-06,
-      "loss": 0.6188,
+      "epoch": 0.24824699110413395,
+      "grad_norm": 2.0096476869887,
+      "learning_rate": 1.7604720050826333e-05,
+      "loss": 0.9611,
       "step": 2372
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.410850498984659e-06,
-      "loss": 0.7613,
+      "epoch": 0.24835164835164836,
+      "grad_norm": 2.2977012476580208,
+      "learning_rate": 1.7602518428948742e-05,
+      "loss": 1.0389,
       "step": 2373
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.4039105759742098e-06,
-      "loss": 0.6933,
+      "epoch": 0.24845630559916274,
+      "grad_norm": 2.945269543421569,
+      "learning_rate": 1.760031593352637e-05,
+      "loss": 1.0266,
       "step": 2374
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.396979291184306e-06,
-      "loss": 0.6117,
+      "epoch": 0.24856096284667714,
+      "grad_norm": 2.1640744849684155,
+      "learning_rate": 1.7598112564812292e-05,
+      "loss": 1.1166,
       "step": 2375
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.3900566524971593e-06,
-      "loss": 0.6098,
+      "epoch": 0.24866562009419152,
+      "grad_norm": 2.1568155494244285,
+      "learning_rate": 1.7595908323059675e-05,
+      "loss": 1.1188,
       "step": 2376
     },
     {
-      "epoch": 1.56,
-      "learning_rate": 2.3831426677851423e-06,
-      "loss": 0.7346,
+      "epoch": 0.24877027734170593,
+      "grad_norm": 2.584518671074471,
+      "learning_rate": 1.7593703208521794e-05,
+      "loss": 1.0749,
       "step": 2377
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3762373449107935e-06,
-      "loss": 0.6459,
+      "epoch": 0.2488749345892203,
+      "grad_norm": 2.6817350218871767,
+      "learning_rate": 1.7591497221452022e-05,
+      "loss": 1.0488,
       "step": 2378
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.369340691726798e-06,
-      "loss": 0.7263,
+      "epoch": 0.24897959183673468,
+      "grad_norm": 2.0586150260625797,
+      "learning_rate": 1.7589290362103827e-05,
+      "loss": 1.0097,
       "step": 2379
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3624527160759812e-06,
-      "loss": 0.6671,
+      "epoch": 0.2490842490842491,
+      "grad_norm": 2.459396989353071,
+      "learning_rate": 1.7587082630730786e-05,
+      "loss": 1.0839,
       "step": 2380
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3555734257913045e-06,
-      "loss": 0.6921,
+      "epoch": 0.24918890633176347,
+      "grad_norm": 2.4145810020805314,
+      "learning_rate": 1.758487402758657e-05,
+      "loss": 1.0401,
       "step": 2381
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3487028286958458e-06,
-      "loss": 0.603,
+      "epoch": 0.24929356357927787,
+      "grad_norm": 2.3872298010666952,
+      "learning_rate": 1.7582664552924953e-05,
+      "loss": 1.1245,
       "step": 2382
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3418409326028037e-06,
-      "loss": 0.6765,
+      "epoch": 0.24939822082679225,
+      "grad_norm": 2.466040223475429,
+      "learning_rate": 1.7580454206999805e-05,
+      "loss": 0.8704,
       "step": 2383
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.334987745315478e-06,
-      "loss": 0.7422,
+      "epoch": 0.24950287807430666,
+      "grad_norm": 2.327011051347902,
+      "learning_rate": 1.75782429900651e-05,
+      "loss": 1.0102,
       "step": 2384
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.328143274627266e-06,
-      "loss": 0.825,
+      "epoch": 0.24960753532182103,
+      "grad_norm": 2.222950496140978,
+      "learning_rate": 1.7576030902374917e-05,
+      "loss": 0.8952,
       "step": 2385
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3213075283216547e-06,
-      "loss": 0.646,
+      "epoch": 0.24971219256933544,
+      "grad_norm": 2.0002826498689266,
+      "learning_rate": 1.7573817944183426e-05,
+      "loss": 0.977,
       "step": 2386
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3144805141722038e-06,
-      "loss": 0.7573,
+      "epoch": 0.24981684981684982,
+      "grad_norm": 2.261276443674775,
+      "learning_rate": 1.7571604115744895e-05,
+      "loss": 1.0078,
       "step": 2387
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3076622399425485e-06,
-      "loss": 0.7257,
+      "epoch": 0.2499215070643642,
+      "grad_norm": 2.31943025867348,
+      "learning_rate": 1.7569389417313704e-05,
+      "loss": 1.0752,
       "step": 2388
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.3008527133863836e-06,
-      "loss": 0.7305,
+      "epoch": 0.2500261643118786,
+      "grad_norm": 2.2233824615996007,
+      "learning_rate": 1.7567173849144322e-05,
+      "loss": 1.0591,
       "step": 2389
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.294051942247457e-06,
-      "loss": 0.7919,
+      "epoch": 0.250130821559393,
+      "grad_norm": 2.335773242353471,
+      "learning_rate": 1.7564957411491324e-05,
+      "loss": 1.0174,
       "step": 2390
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.2872599342595557e-06,
-      "loss": 0.6848,
+      "epoch": 0.2502354788069074,
+      "grad_norm": 2.274951953356186,
+      "learning_rate": 1.756274010460939e-05,
+      "loss": 1.088,
       "step": 2391
     },
     {
-      "epoch": 1.57,
-      "learning_rate": 2.2804766971465054e-06,
-      "loss": 0.571,
+      "epoch": 0.2503401360544218,
+      "grad_norm": 1.928296571150385,
+      "learning_rate": 1.756052192875328e-05,
+      "loss": 0.8277,
       "step": 2392
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.2737022386221552e-06,
-      "loss": 0.6643,
+      "epoch": 0.25044479330193614,
+      "grad_norm": 2.2156454730583555,
+      "learning_rate": 1.7558302884177876e-05,
+      "loss": 1.0403,
       "step": 2393
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.2669365663903764e-06,
-      "loss": 0.6848,
+      "epoch": 0.25054945054945055,
+      "grad_norm": 2.1520512914884153,
+      "learning_rate": 1.755608297113815e-05,
+      "loss": 1.0643,
       "step": 2394
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.2601796881450468e-06,
-      "loss": 0.6377,
+      "epoch": 0.25065410779696495,
+      "grad_norm": 2.2740218868907927,
+      "learning_rate": 1.7553862189889172e-05,
+      "loss": 1.0674,
       "step": 2395
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.253431611570035e-06,
-      "loss": 0.6544,
+      "epoch": 0.2507587650444793,
+      "grad_norm": 2.1433388658904318,
+      "learning_rate": 1.7551640540686114e-05,
+      "loss": 0.9212,
       "step": 2396
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.24669234433921e-06,
-      "loss": 0.8275,
+      "epoch": 0.2508634222919937,
+      "grad_norm": 2.5749819609252764,
+      "learning_rate": 1.754941802378425e-05,
+      "loss": 0.9503,
       "step": 2397
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.2399618941164202e-06,
-      "loss": 0.7129,
+      "epoch": 0.2509680795395081,
+      "grad_norm": 2.4391582025577843,
+      "learning_rate": 1.754719463943895e-05,
+      "loss": 1.0504,
       "step": 2398
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.2332402685554876e-06,
-      "loss": 0.753,
+      "epoch": 0.2510727367870225,
+      "grad_norm": 2.517643489899688,
+      "learning_rate": 1.7544970387905687e-05,
+      "loss": 0.97,
       "step": 2399
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.2265274753002008e-06,
-      "loss": 0.5883,
+      "epoch": 0.25117739403453687,
+      "grad_norm": 2.202807983897776,
+      "learning_rate": 1.754274526944003e-05,
+      "loss": 0.8349,
       "step": 2400
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.2198235219842967e-06,
-      "loss": 0.6828,
+      "epoch": 0.2512820512820513,
+      "grad_norm": 2.9251756781669127,
+      "learning_rate": 1.7540519284297652e-05,
+      "loss": 1.1699,
       "step": 2401
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.213128416231468e-06,
-      "loss": 0.6688,
+      "epoch": 0.2513867085295657,
+      "grad_norm": 2.356785224256036,
+      "learning_rate": 1.753829243273432e-05,
+      "loss": 0.9605,
       "step": 2402
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.2064421656553415e-06,
-      "loss": 0.6618,
+      "epoch": 0.2514913657770801,
+      "grad_norm": 2.423973980202741,
+      "learning_rate": 1.7536064715005906e-05,
+      "loss": 0.8849,
       "step": 2403
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.199764777859477e-06,
-      "loss": 0.7577,
+      "epoch": 0.25159602302459444,
+      "grad_norm": 2.446059031088621,
+      "learning_rate": 1.7533836131368383e-05,
+      "loss": 1.093,
       "step": 2404
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.1930962604373517e-06,
-      "loss": 0.8434,
+      "epoch": 0.25170068027210885,
+      "grad_norm": 2.399076855736286,
+      "learning_rate": 1.7531606682077814e-05,
+      "loss": 1.0542,
       "step": 2405
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.1864366209723597e-06,
-      "loss": 0.6372,
+      "epoch": 0.25180533751962325,
+      "grad_norm": 2.2136011597833267,
+      "learning_rate": 1.7529376367390366e-05,
+      "loss": 0.915,
       "step": 2406
     },
     {
-      "epoch": 1.58,
-      "learning_rate": 2.179785867037797e-06,
-      "loss": 0.7029,
+      "epoch": 0.2519099947671376,
+      "grad_norm": 2.226461148880702,
+      "learning_rate": 1.7527145187562314e-05,
+      "loss": 0.9114,
       "step": 2407
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.1731440061968536e-06,
-      "loss": 0.6978,
+      "epoch": 0.252014652014652,
+      "grad_norm": 2.41593091900151,
+      "learning_rate": 1.7524913142850017e-05,
+      "loss": 1.0315,
       "step": 2408
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.166511046002612e-06,
-      "loss": 0.6433,
+      "epoch": 0.2521193092621664,
+      "grad_norm": 2.2319401902707146,
+      "learning_rate": 1.7522680233509947e-05,
+      "loss": 0.987,
       "step": 2409
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.1598869939980225e-06,
-      "loss": 0.6724,
+      "epoch": 0.2522239665096808,
+      "grad_norm": 2.7690125251281366,
+      "learning_rate": 1.752044645979867e-05,
+      "loss": 1.1371,
       "step": 2410
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.153271857715915e-06,
-      "loss": 0.6688,
+      "epoch": 0.25232862375719517,
+      "grad_norm": 2.286815126513628,
+      "learning_rate": 1.7518211821972845e-05,
+      "loss": 1.1027,
       "step": 2411
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.1466656446789757e-06,
-      "loss": 0.532,
+      "epoch": 0.2524332810047096,
+      "grad_norm": 2.0820445058724695,
+      "learning_rate": 1.7515976320289247e-05,
+      "loss": 1.0881,
       "step": 2412
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.140068362399745e-06,
-      "loss": 0.775,
+      "epoch": 0.252537938252224,
+      "grad_norm": 2.987553814644758,
+      "learning_rate": 1.751373995500473e-05,
+      "loss": 1.107,
       "step": 2413
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.133480018380608e-06,
-      "loss": 0.7806,
+      "epoch": 0.25264259549973833,
+      "grad_norm": 1.9581027443263326,
+      "learning_rate": 1.7511502726376257e-05,
+      "loss": 0.8612,
       "step": 2414
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.1269006201137833e-06,
-      "loss": 0.6316,
+      "epoch": 0.25274725274725274,
+      "grad_norm": 2.376542007505258,
+      "learning_rate": 1.7509264634660896e-05,
+      "loss": 1.006,
       "step": 2415
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.1203301750813166e-06,
-      "loss": 0.7329,
+      "epoch": 0.25285190999476714,
+      "grad_norm": 2.80707883970879,
+      "learning_rate": 1.7507025680115807e-05,
+      "loss": 1.1002,
       "step": 2416
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.1137686907550737e-06,
-      "loss": 0.6746,
+      "epoch": 0.25295656724228155,
+      "grad_norm": 1.6942186283185672,
+      "learning_rate": 1.7504785862998246e-05,
+      "loss": 0.8788,
       "step": 2417
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.1072161745967313e-06,
-      "loss": 0.6451,
+      "epoch": 0.2530612244897959,
+      "grad_norm": 2.5777349929769433,
+      "learning_rate": 1.750254518356558e-05,
+      "loss": 1.103,
       "step": 2418
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.1006726340577597e-06,
-      "loss": 0.6995,
+      "epoch": 0.2531658817373103,
+      "grad_norm": 2.4495279259341776,
+      "learning_rate": 1.7500303642075257e-05,
+      "loss": 1.0692,
       "step": 2419
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.094138076579433e-06,
-      "loss": 0.695,
+      "epoch": 0.2532705389848247,
+      "grad_norm": 1.8822814244979826,
+      "learning_rate": 1.7498061238784843e-05,
+      "loss": 0.8956,
       "step": 2420
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.0876125095928033e-06,
-      "loss": 0.7856,
+      "epoch": 0.2533751962323391,
+      "grad_norm": 2.223853110843607,
+      "learning_rate": 1.7495817973951997e-05,
+      "loss": 0.969,
       "step": 2421
     },
     {
-      "epoch": 1.59,
-      "learning_rate": 2.0810959405187005e-06,
-      "loss": 0.6353,
+      "epoch": 0.25347985347985347,
+      "grad_norm": 2.238360132707453,
+      "learning_rate": 1.7493573847834465e-05,
+      "loss": 1.0264,
       "step": 2422
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.0745883767677246e-06,
-      "loss": 0.7336,
+      "epoch": 0.2535845107273679,
+      "grad_norm": 2.0197198632761606,
+      "learning_rate": 1.749132886069011e-05,
+      "loss": 0.9527,
       "step": 2423
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.068089825740225e-06,
-      "loss": 0.7266,
+      "epoch": 0.2536891679748823,
+      "grad_norm": 2.22631274546597,
+      "learning_rate": 1.748908301277688e-05,
+      "loss": 0.9435,
       "step": 2424
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.0616002948263157e-06,
-      "loss": 0.7532,
+      "epoch": 0.25379382522239663,
+      "grad_norm": 2.299475782910513,
+      "learning_rate": 1.748683630435283e-05,
+      "loss": 1.1404,
       "step": 2425
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.0551197914058464e-06,
-      "loss": 0.6806,
+      "epoch": 0.25389848246991104,
+      "grad_norm": 2.5843227105301034,
+      "learning_rate": 1.7484588735676117e-05,
+      "loss": 0.8494,
       "step": 2426
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.0486483228484012e-06,
-      "loss": 0.7049,
+      "epoch": 0.25400313971742544,
+      "grad_norm": 1.9572204780029898,
+      "learning_rate": 1.748234030700498e-05,
+      "loss": 0.9776,
       "step": 2427
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.0421858965132856e-06,
-      "loss": 0.6314,
+      "epoch": 0.25410779696493985,
+      "grad_norm": 3.069066964069534,
+      "learning_rate": 1.7480091018597775e-05,
+      "loss": 0.968,
       "step": 2428
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.035732519749528e-06,
-      "loss": 0.6876,
+      "epoch": 0.2542124542124542,
+      "grad_norm": 2.5784423660832405,
+      "learning_rate": 1.7477840870712946e-05,
+      "loss": 1.0712,
       "step": 2429
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.0292881998958634e-06,
-      "loss": 0.6627,
+      "epoch": 0.2543171114599686,
+      "grad_norm": 2.400723658989671,
+      "learning_rate": 1.7475589863609042e-05,
+      "loss": 0.9463,
       "step": 2430
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.022852944280731e-06,
-      "loss": 0.7102,
+      "epoch": 0.254421768707483,
+      "grad_norm": 2.332512218507682,
+      "learning_rate": 1.747333799754471e-05,
+      "loss": 1.0427,
       "step": 2431
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.016426760222259e-06,
-      "loss": 0.6204,
+      "epoch": 0.25452642595499736,
+      "grad_norm": 2.069397809708061,
+      "learning_rate": 1.747108527277869e-05,
+      "loss": 0.9479,
       "step": 2432
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.0100096550282543e-06,
-      "loss": 0.5164,
+      "epoch": 0.25463108320251177,
+      "grad_norm": 2.284939347387363,
+      "learning_rate": 1.7468831689569827e-05,
+      "loss": 1.0377,
       "step": 2433
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 2.0036016359962084e-06,
-      "loss": 0.7561,
+      "epoch": 0.25473574045002617,
+      "grad_norm": 2.013249004544656,
+      "learning_rate": 1.7466577248177062e-05,
+      "loss": 1.0894,
       "step": 2434
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 1.9972027104132763e-06,
-      "loss": 0.7087,
+      "epoch": 0.2548403976975406,
+      "grad_norm": 2.095695638030201,
+      "learning_rate": 1.746432194885944e-05,
+      "loss": 1.0162,
       "step": 2435
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 1.990812885556269e-06,
-      "loss": 0.7155,
+      "epoch": 0.2549450549450549,
+      "grad_norm": 2.308895697628336,
+      "learning_rate": 1.7462065791876087e-05,
+      "loss": 0.9949,
       "step": 2436
     },
     {
-      "epoch": 1.6,
-      "learning_rate": 1.9844321686916602e-06,
-      "loss": 0.7704,
+      "epoch": 0.25504971219256933,
+      "grad_norm": 2.1787458199138183,
+      "learning_rate": 1.745980877748625e-05,
+      "loss": 1.0669,
       "step": 2437
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.9780605670755473e-06,
-      "loss": 0.6917,
+      "epoch": 0.25515436944008374,
+      "grad_norm": 2.27381391199627,
+      "learning_rate": 1.7457550905949263e-05,
+      "loss": 0.9188,
       "step": 2438
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.971698087953676e-06,
-      "loss": 0.6903,
+      "epoch": 0.2552590266875981,
+      "grad_norm": 2.0547178362816423,
+      "learning_rate": 1.745529217752456e-05,
+      "loss": 1.0003,
       "step": 2439
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.965344738561414e-06,
-      "loss": 0.6161,
+      "epoch": 0.2553636839351125,
+      "grad_norm": 2.003826247372063,
+      "learning_rate": 1.7453032592471667e-05,
+      "loss": 0.8219,
       "step": 2440
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.9590005261237475e-06,
-      "loss": 0.6375,
+      "epoch": 0.2554683411826269,
+      "grad_norm": 2.2165668382792676,
+      "learning_rate": 1.7450772151050226e-05,
+      "loss": 1.078,
       "step": 2441
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.9526654578552673e-06,
-      "loss": 0.7197,
+      "epoch": 0.2555729984301413,
+      "grad_norm": 2.406127059643738,
+      "learning_rate": 1.744851085351996e-05,
+      "loss": 1.0154,
       "step": 2442
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.9463395409601713e-06,
-      "loss": 0.6918,
+      "epoch": 0.25567765567765566,
+      "grad_norm": 1.9609166344880296,
+      "learning_rate": 1.7446248700140694e-05,
+      "loss": 1.0885,
       "step": 2443
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.940022782632248e-06,
-      "loss": 0.6887,
+      "epoch": 0.25578231292517006,
+      "grad_norm": 2.0711606782794947,
+      "learning_rate": 1.744398569117236e-05,
+      "loss": 1.0171,
       "step": 2444
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.9337151900548734e-06,
-      "loss": 0.7296,
+      "epoch": 0.25588697017268447,
+      "grad_norm": 2.0871258754382636,
+      "learning_rate": 1.7441721826874976e-05,
+      "loss": 0.9873,
       "step": 2445
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.9274167704009973e-06,
-      "loss": 0.6461,
+      "epoch": 0.2559916274201989,
+      "grad_norm": 2.112116868972381,
+      "learning_rate": 1.743945710750867e-05,
+      "loss": 0.9903,
       "step": 2446
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.921127530833138e-06,
-      "loss": 0.7237,
+      "epoch": 0.2560962846677132,
+      "grad_norm": 1.8482931898345583,
+      "learning_rate": 1.7437191533333663e-05,
+      "loss": 0.8978,
       "step": 2447
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.9148474785033768e-06,
-      "loss": 0.6848,
+      "epoch": 0.25620094191522763,
+      "grad_norm": 2.361301370582731,
+      "learning_rate": 1.7434925104610273e-05,
+      "loss": 1.0131,
       "step": 2448
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.908576620553345e-06,
-      "loss": 0.6442,
+      "epoch": 0.25630559916274204,
+      "grad_norm": 2.1470314899465275,
+      "learning_rate": 1.7432657821598912e-05,
+      "loss": 1.1222,
       "step": 2449
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.902314964114219e-06,
-      "loss": 0.7891,
+      "epoch": 0.2564102564102564,
+      "grad_norm": 2.243154931598059,
+      "learning_rate": 1.74303896845601e-05,
+      "loss": 0.9956,
       "step": 2450
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.8960625163067158e-06,
-      "loss": 0.6919,
+      "epoch": 0.2565149136577708,
+      "grad_norm": 2.090796158960878,
+      "learning_rate": 1.7428120693754453e-05,
+      "loss": 0.9429,
       "step": 2451
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.889819284241069e-06,
-      "loss": 0.7969,
+      "epoch": 0.2566195709052852,
+      "grad_norm": 2.129012668639158,
+      "learning_rate": 1.7425850849442677e-05,
+      "loss": 0.9842,
       "step": 2452
     },
     {
-      "epoch": 1.61,
-      "learning_rate": 1.8835852750170414e-06,
-      "loss": 0.7172,
+      "epoch": 0.2567242281527996,
+      "grad_norm": 2.587169226816751,
+      "learning_rate": 1.742358015188559e-05,
+      "loss": 0.9024,
       "step": 2453
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8773604957239067e-06,
-      "loss": 0.6268,
+      "epoch": 0.25682888540031396,
+      "grad_norm": 2.186311266652209,
+      "learning_rate": 1.742130860134409e-05,
+      "loss": 1.0747,
       "step": 2454
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8711449534404424e-06,
-      "loss": 0.6722,
+      "epoch": 0.25693354264782836,
+      "grad_norm": 2.2324519069996223,
+      "learning_rate": 1.7419036198079186e-05,
+      "loss": 1.0882,
       "step": 2455
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8649386552349136e-06,
-      "loss": 0.7591,
+      "epoch": 0.25703819989534277,
+      "grad_norm": 2.044693604835276,
+      "learning_rate": 1.7416762942351986e-05,
+      "loss": 1.0221,
       "step": 2456
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8587416081650866e-06,
-      "loss": 0.684,
+      "epoch": 0.2571428571428571,
+      "grad_norm": 2.4466601197617344,
+      "learning_rate": 1.7414488834423687e-05,
+      "loss": 1.1308,
       "step": 2457
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8525538192782e-06,
-      "loss": 0.554,
+      "epoch": 0.2572475143903715,
+      "grad_norm": 2.473780111943485,
+      "learning_rate": 1.7412213874555594e-05,
+      "loss": 1.1312,
       "step": 2458
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8463752956109616e-06,
-      "loss": 0.6304,
+      "epoch": 0.25735217163788593,
+      "grad_norm": 2.4776047017601948,
+      "learning_rate": 1.7409938063009092e-05,
+      "loss": 0.9445,
       "step": 2459
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8402060441895508e-06,
-      "loss": 0.675,
+      "epoch": 0.25745682888540034,
+      "grad_norm": 2.4309185606448636,
+      "learning_rate": 1.740766140004569e-05,
+      "loss": 1.1362,
       "step": 2460
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8340460720295917e-06,
-      "loss": 0.6858,
+      "epoch": 0.2575614861329147,
+      "grad_norm": 2.3084232636893853,
+      "learning_rate": 1.740538388592698e-05,
+      "loss": 0.9599,
       "step": 2461
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.827895386136166e-06,
-      "loss": 0.6739,
+      "epoch": 0.2576661433804291,
+      "grad_norm": 2.2338587055904693,
+      "learning_rate": 1.7403105520914644e-05,
+      "loss": 1.0872,
       "step": 2462
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8217539935037897e-06,
-      "loss": 0.7592,
+      "epoch": 0.2577708006279435,
+      "grad_norm": 2.32586858159479,
+      "learning_rate": 1.7400826305270477e-05,
+      "loss": 1.1381,
       "step": 2463
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8156219011164121e-06,
-      "loss": 0.6757,
+      "epoch": 0.2578754578754579,
+      "grad_norm": 1.9459202093623904,
+      "learning_rate": 1.7398546239256367e-05,
+      "loss": 0.888,
       "step": 2464
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8094991159474107e-06,
-      "loss": 0.6849,
+      "epoch": 0.25798011512297225,
+      "grad_norm": 2.1322352083234266,
+      "learning_rate": 1.7396265323134293e-05,
+      "loss": 1.0406,
       "step": 2465
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.8033856449595689e-06,
-      "loss": 0.608,
+      "epoch": 0.25808477237048666,
+      "grad_norm": 2.1704661597271118,
+      "learning_rate": 1.739398355716634e-05,
+      "loss": 0.9406,
       "step": 2466
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.7972814951050855e-06,
-      "loss": 0.8048,
+      "epoch": 0.25818942961800107,
+      "grad_norm": 2.5070809156506564,
+      "learning_rate": 1.7391700941614687e-05,
+      "loss": 0.9449,
       "step": 2467
     },
     {
-      "epoch": 1.62,
-      "learning_rate": 1.7911866733255556e-06,
-      "loss": 0.5433,
+      "epoch": 0.2582940868655154,
+      "grad_norm": 2.2868072880319286,
+      "learning_rate": 1.7389417476741615e-05,
+      "loss": 1.1516,
       "step": 2468
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.785101186551975e-06,
-      "loss": 0.698,
+      "epoch": 0.2583987441130298,
+      "grad_norm": 2.2741590192268526,
+      "learning_rate": 1.7387133162809492e-05,
+      "loss": 1.0228,
       "step": 2469
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.7790250417047116e-06,
-      "loss": 0.6862,
+      "epoch": 0.2585034013605442,
+      "grad_norm": 2.361486277697732,
+      "learning_rate": 1.738484800008079e-05,
+      "loss": 0.9587,
       "step": 2470
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.772958245693518e-06,
-      "loss": 0.668,
+      "epoch": 0.25860805860805863,
+      "grad_norm": 2.1605751628195358,
+      "learning_rate": 1.738256198881809e-05,
+      "loss": 1.087,
       "step": 2471
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.7669008054175108e-06,
-      "loss": 0.6286,
+      "epoch": 0.258712715855573,
+      "grad_norm": 2.0828368567159368,
+      "learning_rate": 1.7380275129284047e-05,
+      "loss": 1.0085,
       "step": 2472
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.7608527277651711e-06,
-      "loss": 0.6147,
+      "epoch": 0.2588173731030874,
+      "grad_norm": 2.180627297007591,
+      "learning_rate": 1.7377987421741428e-05,
+      "loss": 1.0592,
       "step": 2473
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.7548140196143338e-06,
-      "loss": 0.6577,
+      "epoch": 0.2589220303506018,
+      "grad_norm": 2.2690156377111035,
+      "learning_rate": 1.73756988664531e-05,
+      "loss": 1.1374,
       "step": 2474
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.7487846878321691e-06,
-      "loss": 0.7557,
+      "epoch": 0.25902668759811615,
+      "grad_norm": 2.16338779374175,
+      "learning_rate": 1.7373409463682017e-05,
+      "loss": 0.999,
       "step": 2475
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.7427647392751967e-06,
-      "loss": 0.5973,
+      "epoch": 0.25913134484563055,
+      "grad_norm": 2.1101239537203216,
+      "learning_rate": 1.7371119213691243e-05,
+      "loss": 0.9949,
       "step": 2476
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.7367541807892586e-06,
-      "loss": 0.5939,
+      "epoch": 0.25923600209314496,
+      "grad_norm": 2.0134589337565085,
+      "learning_rate": 1.7368828116743927e-05,
+      "loss": 0.9535,
       "step": 2477
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.7307530192095212e-06,
-      "loss": 0.6397,
+      "epoch": 0.25934065934065936,
+      "grad_norm": 2.0592210690399995,
+      "learning_rate": 1.7366536173103325e-05,
+      "loss": 1.0145,
       "step": 2478
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.7247612613604637e-06,
-      "loss": 0.7555,
+      "epoch": 0.2594453165881737,
+      "grad_norm": 2.418699264093393,
+      "learning_rate": 1.736424338303278e-05,
+      "loss": 0.9113,
       "step": 2479
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.718778914055873e-06,
-      "loss": 0.714,
+      "epoch": 0.2595499738356881,
+      "grad_norm": 2.6190682853440084,
+      "learning_rate": 1.7361949746795744e-05,
+      "loss": 1.1143,
       "step": 2480
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.712805984098832e-06,
-      "loss": 0.6922,
+      "epoch": 0.2596546310832025,
+      "grad_norm": 2.2236033474143975,
+      "learning_rate": 1.7359655264655755e-05,
+      "loss": 0.9108,
       "step": 2481
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.7068424782817162e-06,
-      "loss": 0.8039,
+      "epoch": 0.2597592883307169,
+      "grad_norm": 2.1342142709306207,
+      "learning_rate": 1.7357359936876457e-05,
+      "loss": 1.077,
       "step": 2482
     },
     {
-      "epoch": 1.63,
-      "learning_rate": 1.700888403386186e-06,
-      "loss": 0.6169,
+      "epoch": 0.2598639455782313,
+      "grad_norm": 2.0438697294062544,
+      "learning_rate": 1.735506376372159e-05,
+      "loss": 1.0832,
       "step": 2483
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.694943766183168e-06,
-      "loss": 0.6418,
+      "epoch": 0.2599686028257457,
+      "grad_norm": 2.2685453564325933,
+      "learning_rate": 1.7352766745454982e-05,
+      "loss": 1.0577,
       "step": 2484
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6890085734328666e-06,
-      "loss": 0.7807,
+      "epoch": 0.2600732600732601,
+      "grad_norm": 2.277948996427162,
+      "learning_rate": 1.7350468882340572e-05,
+      "loss": 1.1536,
       "step": 2485
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6830828318847415e-06,
-      "loss": 0.6801,
+      "epoch": 0.26017791732077444,
+      "grad_norm": 2.2839736714043117,
+      "learning_rate": 1.7348170174642387e-05,
+      "loss": 1.0408,
       "step": 2486
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6771665482775058e-06,
-      "loss": 0.7606,
+      "epoch": 0.26028257456828885,
+      "grad_norm": 2.1783014726910532,
+      "learning_rate": 1.7345870622624552e-05,
+      "loss": 1.0058,
       "step": 2487
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6712597293391153e-06,
-      "loss": 0.6118,
+      "epoch": 0.26038723181580326,
+      "grad_norm": 2.316943295044069,
+      "learning_rate": 1.734357022655129e-05,
+      "loss": 1.1339,
       "step": 2488
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6653623817867638e-06,
-      "loss": 0.6275,
+      "epoch": 0.26049188906331766,
+      "grad_norm": 2.2969718645888144,
+      "learning_rate": 1.7341268986686924e-05,
+      "loss": 1.0786,
       "step": 2489
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6594745123268751e-06,
-      "loss": 0.6102,
+      "epoch": 0.260596546310832,
+      "grad_norm": 2.1322028884382185,
+      "learning_rate": 1.7338966903295865e-05,
+      "loss": 0.8794,
       "step": 2490
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6535961276550948e-06,
-      "loss": 0.6771,
+      "epoch": 0.2607012035583464,
+      "grad_norm": 2.3892801496526994,
+      "learning_rate": 1.7336663976642634e-05,
+      "loss": 1.0003,
       "step": 2491
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6477272344562789e-06,
-      "loss": 0.5991,
+      "epoch": 0.2608058608058608,
+      "grad_norm": 2.3446023789762047,
+      "learning_rate": 1.7334360206991842e-05,
+      "loss": 0.9991,
       "step": 2492
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6418678394044962e-06,
-      "loss": 0.6772,
+      "epoch": 0.2609105180533752,
+      "grad_norm": 2.222641398320293,
+      "learning_rate": 1.7332055594608188e-05,
+      "loss": 1.0998,
       "step": 2493
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6360179491630056e-06,
-      "loss": 0.6505,
+      "epoch": 0.2610151753008896,
+      "grad_norm": 2.0463322929767784,
+      "learning_rate": 1.7329750139756484e-05,
+      "loss": 1.0081,
       "step": 2494
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6301775703842637e-06,
-      "loss": 0.7815,
+      "epoch": 0.261119832548404,
+      "grad_norm": 2.156653372785402,
+      "learning_rate": 1.732744384270163e-05,
+      "loss": 1.0108,
       "step": 2495
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6243467097099097e-06,
-      "loss": 0.614,
+      "epoch": 0.2612244897959184,
+      "grad_norm": 2.2798443834114432,
+      "learning_rate": 1.732513670370863e-05,
+      "loss": 0.9697,
       "step": 2496
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6185253737707607e-06,
-      "loss": 0.6995,
+      "epoch": 0.26132914704343274,
+      "grad_norm": 2.5215817238707827,
+      "learning_rate": 1.7322828723042566e-05,
+      "loss": 1.0863,
       "step": 2497
     },
     {
-      "epoch": 1.64,
-      "learning_rate": 1.6127135691867945e-06,
-      "loss": 0.5755,
+      "epoch": 0.26143380429094715,
+      "grad_norm": 2.1107145371174347,
+      "learning_rate": 1.732051990096864e-05,
+      "loss": 1.0245,
       "step": 2498
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.6069113025671546e-06,
-      "loss": 0.6053,
+      "epoch": 0.26153846153846155,
+      "grad_norm": 2.3976283399692537,
+      "learning_rate": 1.7318210237752137e-05,
+      "loss": 0.9136,
       "step": 2499
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.6011185805101447e-06,
-      "loss": 0.6471,
+      "epoch": 0.2616431187859759,
+      "grad_norm": 2.182358026003498,
+      "learning_rate": 1.7315899733658443e-05,
+      "loss": 0.9992,
       "step": 2500
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5953354096032048e-06,
-      "loss": 0.5996,
+      "epoch": 0.2617477760334903,
+      "grad_norm": 2.2560760531967947,
+      "learning_rate": 1.7313588388953037e-05,
+      "loss": 0.9428,
       "step": 2501
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5895617964229215e-06,
-      "loss": 0.6437,
+      "epoch": 0.2618524332810047,
+      "grad_norm": 2.409534450316176,
+      "learning_rate": 1.73112762039015e-05,
+      "loss": 1.1515,
       "step": 2502
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5837977475350008e-06,
-      "loss": 0.6752,
+      "epoch": 0.2619570905285191,
+      "grad_norm": 2.0440248637735365,
+      "learning_rate": 1.7308963178769507e-05,
+      "loss": 0.9662,
       "step": 2503
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5780432694942815e-06,
-      "loss": 0.8052,
+      "epoch": 0.26206174777603347,
+      "grad_norm": 2.2837513962147105,
+      "learning_rate": 1.7306649313822826e-05,
+      "loss": 1.0215,
       "step": 2504
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.572298368844719e-06,
-      "loss": 0.6576,
+      "epoch": 0.2621664050235479,
+      "grad_norm": 2.0651869724475986,
+      "learning_rate": 1.7304334609327326e-05,
+      "loss": 1.0749,
       "step": 2505
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5665630521193753e-06,
-      "loss": 0.7189,
+      "epoch": 0.2622710622710623,
+      "grad_norm": 2.2950964938180034,
+      "learning_rate": 1.7302019065548973e-05,
+      "loss": 1.0932,
       "step": 2506
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5608373258404087e-06,
-      "loss": 0.747,
+      "epoch": 0.2623757195185767,
+      "grad_norm": 2.6958405572322697,
+      "learning_rate": 1.7299702682753826e-05,
+      "loss": 0.8741,
       "step": 2507
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5551211965190783e-06,
-      "loss": 0.7151,
+      "epoch": 0.26248037676609104,
+      "grad_norm": 2.3931112849601615,
+      "learning_rate": 1.7297385461208044e-05,
+      "loss": 0.9674,
       "step": 2508
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5494146706557257e-06,
-      "loss": 0.6727,
+      "epoch": 0.26258503401360545,
+      "grad_norm": 2.0885169426285652,
+      "learning_rate": 1.7295067401177877e-05,
+      "loss": 1.0842,
       "step": 2509
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5437177547397742e-06,
-      "loss": 0.6923,
+      "epoch": 0.26268969126111985,
+      "grad_norm": 2.3250017262626503,
+      "learning_rate": 1.7292748502929678e-05,
+      "loss": 1.012,
       "step": 2510
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.538030455249716e-06,
-      "loss": 0.655,
+      "epoch": 0.2627943485086342,
+      "grad_norm": 2.379541018104846,
+      "learning_rate": 1.7290428766729893e-05,
+      "loss": 1.1306,
       "step": 2511
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5323527786531113e-06,
-      "loss": 0.5931,
+      "epoch": 0.2628990057561486,
+      "grad_norm": 2.1706941809048153,
+      "learning_rate": 1.728810819284506e-05,
+      "loss": 0.9991,
       "step": 2512
     },
     {
-      "epoch": 1.65,
-      "learning_rate": 1.5266847314065726e-06,
-      "loss": 0.6519,
+      "epoch": 0.263003663003663,
+      "grad_norm": 2.075305628554083,
+      "learning_rate": 1.7285786781541825e-05,
+      "loss": 1.0348,
       "step": 2513
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.521026319955764e-06,
-      "loss": 0.741,
+      "epoch": 0.2631083202511774,
+      "grad_norm": 2.0890810561710933,
+      "learning_rate": 1.7283464533086917e-05,
+      "loss": 0.8281,
       "step": 2514
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.515377550735394e-06,
-      "loss": 0.6867,
+      "epoch": 0.26321297749869177,
+      "grad_norm": 2.3435013116193035,
+      "learning_rate": 1.728114144774717e-05,
+      "loss": 1.0713,
       "step": 2515
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.5097384301692041e-06,
-      "loss": 0.6973,
+      "epoch": 0.2633176347462062,
+      "grad_norm": 2.1669632646702155,
+      "learning_rate": 1.7278817525789513e-05,
+      "loss": 1.1107,
       "step": 2516
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.5041089646699603e-06,
-      "loss": 0.5613,
+      "epoch": 0.2634222919937206,
+      "grad_norm": 2.171445878531362,
+      "learning_rate": 1.7276492767480962e-05,
+      "loss": 0.9642,
       "step": 2517
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4984891606394535e-06,
-      "loss": 0.6384,
+      "epoch": 0.26352694924123493,
+      "grad_norm": 2.417950887212835,
+      "learning_rate": 1.7274167173088643e-05,
+      "loss": 1.0002,
       "step": 2518
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4928790244684866e-06,
-      "loss": 0.7297,
+      "epoch": 0.26363160648874934,
+      "grad_norm": 2.0843255174378092,
+      "learning_rate": 1.727184074287977e-05,
+      "loss": 1.0267,
       "step": 2519
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.487278562536868e-06,
-      "loss": 0.663,
+      "epoch": 0.26373626373626374,
+      "grad_norm": 2.0538455916227134,
+      "learning_rate": 1.7269513477121652e-05,
+      "loss": 0.8743,
       "step": 2520
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4816877812134034e-06,
-      "loss": 0.7282,
+      "epoch": 0.26384092098377815,
+      "grad_norm": 2.1773574455703892,
+      "learning_rate": 1.7267185376081702e-05,
+      "loss": 0.9744,
       "step": 2521
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4761066868558915e-06,
-      "loss": 0.7148,
+      "epoch": 0.2639455782312925,
+      "grad_norm": 2.5975951297910167,
+      "learning_rate": 1.7264856440027417e-05,
+      "loss": 0.9104,
       "step": 2522
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4705352858111144e-06,
-      "loss": 0.7311,
+      "epoch": 0.2640502354788069,
+      "grad_norm": 2.03025817791244,
+      "learning_rate": 1.72625266692264e-05,
+      "loss": 0.987,
       "step": 2523
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4649735844148304e-06,
-      "loss": 0.7404,
+      "epoch": 0.2641548927263213,
+      "grad_norm": 2.549525974397687,
+      "learning_rate": 1.726019606394635e-05,
+      "loss": 1.0157,
       "step": 2524
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4594215889917707e-06,
-      "loss": 0.6626,
+      "epoch": 0.26425954997383566,
+      "grad_norm": 3.248084272655294,
+      "learning_rate": 1.7257864624455048e-05,
+      "loss": 1.0282,
       "step": 2525
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4538793058556222e-06,
-      "loss": 0.6253,
+      "epoch": 0.26436420722135007,
+      "grad_norm": 2.6468935665860807,
+      "learning_rate": 1.7255532351020393e-05,
+      "loss": 0.9803,
       "step": 2526
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4483467413090324e-06,
-      "loss": 0.6016,
+      "epoch": 0.2644688644688645,
+      "grad_norm": 2.088664398222349,
+      "learning_rate": 1.7253199243910357e-05,
+      "loss": 1.1068,
       "step": 2527
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4428239016435953e-06,
-      "loss": 0.6774,
+      "epoch": 0.2645735217163789,
+      "grad_norm": 2.181269418325987,
+      "learning_rate": 1.725086530339303e-05,
+      "loss": 0.9588,
       "step": 2528
     },
     {
-      "epoch": 1.66,
-      "learning_rate": 1.4373107931398477e-06,
-      "loss": 0.6316,
+      "epoch": 0.26467817896389323,
+      "grad_norm": 1.9358065140447658,
+      "learning_rate": 1.7248530529736575e-05,
+      "loss": 0.9188,
       "step": 2529
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.4318074220672595e-06,
-      "loss": 0.6527,
+      "epoch": 0.26478283621140764,
+      "grad_norm": 2.182930759674411,
+      "learning_rate": 1.724619492320927e-05,
+      "loss": 1.1289,
       "step": 2530
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.426313794684221e-06,
-      "loss": 0.659,
+      "epoch": 0.26488749345892204,
+      "grad_norm": 2.1868917273765227,
+      "learning_rate": 1.724385848407948e-05,
+      "loss": 0.9698,
       "step": 2531
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.4208299172380512e-06,
-      "loss": 0.6483,
+      "epoch": 0.26499215070643645,
+      "grad_norm": 2.0248128570915918,
+      "learning_rate": 1.7241521212615663e-05,
+      "loss": 0.9791,
       "step": 2532
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.4153557959649788e-06,
-      "loss": 0.6334,
+      "epoch": 0.2650968079539508,
+      "grad_norm": 2.4684093554379167,
+      "learning_rate": 1.7239183109086384e-05,
+      "loss": 1.0178,
       "step": 2533
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.4098914370901385e-06,
-      "loss": 0.6306,
+      "epoch": 0.2652014652014652,
+      "grad_norm": 2.3248588536485535,
+      "learning_rate": 1.7236844173760286e-05,
+      "loss": 0.9927,
       "step": 2534
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.4044368468275549e-06,
-      "loss": 0.7679,
+      "epoch": 0.2653061224489796,
+      "grad_norm": 2.3426825101423945,
+      "learning_rate": 1.7234504406906124e-05,
+      "loss": 1.0613,
       "step": 2535
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.3989920313801552e-06,
-      "loss": 0.6251,
+      "epoch": 0.26541077969649396,
+      "grad_norm": 2.1367495439409643,
+      "learning_rate": 1.7232163808792733e-05,
+      "loss": 1.0705,
       "step": 2536
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.3935569969397434e-06,
-      "loss": 0.5506,
+      "epoch": 0.26551543694400837,
+      "grad_norm": 2.3408082854159264,
+      "learning_rate": 1.7229822379689067e-05,
+      "loss": 1.1775,
       "step": 2537
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.3881317496870062e-06,
-      "loss": 0.672,
+      "epoch": 0.26562009419152277,
+      "grad_norm": 2.2416930097749646,
+      "learning_rate": 1.722748011986415e-05,
+      "loss": 1.0386,
       "step": 2538
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.3827162957914975e-06,
-      "loss": 0.5093,
+      "epoch": 0.2657247514390372,
+      "grad_norm": 2.1645588725679112,
+      "learning_rate": 1.7225137029587115e-05,
+      "loss": 0.9904,
       "step": 2539
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.37731064141163e-06,
-      "loss": 0.5369,
+      "epoch": 0.2658294086865515,
+      "grad_norm": 2.1423029600647925,
+      "learning_rate": 1.7222793109127193e-05,
+      "loss": 0.987,
       "step": 2540
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.3719147926946797e-06,
-      "loss": 0.6313,
+      "epoch": 0.26593406593406593,
+      "grad_norm": 2.360699966256053,
+      "learning_rate": 1.7220448358753693e-05,
+      "loss": 1.0804,
       "step": 2541
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.3665287557767682e-06,
-      "loss": 0.7615,
+      "epoch": 0.26603872318158034,
+      "grad_norm": 2.484820983087851,
+      "learning_rate": 1.7218102778736046e-05,
+      "loss": 1.1052,
       "step": 2542
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.361152536782856e-06,
-      "loss": 0.7162,
+      "epoch": 0.2661433804290947,
+      "grad_norm": 2.5559902418339298,
+      "learning_rate": 1.7215756369343755e-05,
+      "loss": 1.166,
       "step": 2543
     },
     {
-      "epoch": 1.67,
-      "learning_rate": 1.3557861418267514e-06,
-      "loss": 0.5893,
+      "epoch": 0.2662480376766091,
+      "grad_norm": 2.273062565811467,
+      "learning_rate": 1.721340913084643e-05,
+      "loss": 1.0314,
       "step": 2544
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.3504295770110731e-06,
-      "loss": 0.6642,
+      "epoch": 0.2663526949241235,
+      "grad_norm": 2.390606522743288,
+      "learning_rate": 1.721106106351377e-05,
+      "loss": 1.0458,
       "step": 2545
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.3450828484272727e-06,
-      "loss": 0.6711,
+      "epoch": 0.2664573521716379,
+      "grad_norm": 2.3461424836775313,
+      "learning_rate": 1.720871216761558e-05,
+      "loss": 0.9914,
       "step": 2546
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.339745962155613e-06,
-      "loss": 0.6091,
+      "epoch": 0.26656200941915226,
+      "grad_norm": 2.1093537321214333,
+      "learning_rate": 1.7206362443421742e-05,
+      "loss": 0.9635,
       "step": 2547
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.3344189242651672e-06,
-      "loss": 0.7028,
+      "epoch": 0.26666666666666666,
+      "grad_norm": 2.2076603089511333,
+      "learning_rate": 1.7204011891202256e-05,
+      "loss": 0.8798,
       "step": 2548
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.3291017408138007e-06,
-      "loss": 0.7104,
+      "epoch": 0.26677132391418107,
+      "grad_norm": 2.1198208472221376,
+      "learning_rate": 1.7201660511227195e-05,
+      "loss": 1.0799,
       "step": 2549
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.32379441784818e-06,
-      "loss": 0.6543,
+      "epoch": 0.2668759811616955,
+      "grad_norm": 2.257648787705932,
+      "learning_rate": 1.7199308303766745e-05,
+      "loss": 1.1216,
       "step": 2550
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.3184969614037567e-06,
-      "loss": 0.6472,
+      "epoch": 0.2669806384092098,
+      "grad_norm": 2.167568660205902,
+      "learning_rate": 1.7196955269091175e-05,
+      "loss": 0.9121,
       "step": 2551
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.3132093775047616e-06,
-      "loss": 0.8348,
+      "epoch": 0.26708529565672423,
+      "grad_norm": 2.070104233631567,
+      "learning_rate": 1.7194601407470857e-05,
+      "loss": 1.1141,
       "step": 2552
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.3079316721641987e-06,
-      "loss": 0.7482,
+      "epoch": 0.26718995290423864,
+      "grad_norm": 2.121285217113809,
+      "learning_rate": 1.719224671917625e-05,
+      "loss": 1.0316,
       "step": 2553
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.3026638513838386e-06,
-      "loss": 0.588,
+      "epoch": 0.267294610151753,
+      "grad_norm": 2.1816334225398997,
+      "learning_rate": 1.718989120447792e-05,
+      "loss": 0.8018,
       "step": 2554
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.29740592115421e-06,
-      "loss": 0.5869,
+      "epoch": 0.2673992673992674,
+      "grad_norm": 2.2372260823335557,
+      "learning_rate": 1.718753486364651e-05,
+      "loss": 0.936,
       "step": 2555
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.292157887454596e-06,
-      "loss": 0.6731,
+      "epoch": 0.2675039246467818,
+      "grad_norm": 2.1411788143946895,
+      "learning_rate": 1.7185177696952773e-05,
+      "loss": 1.0857,
       "step": 2556
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.2869197562530223e-06,
-      "loss": 0.7313,
+      "epoch": 0.2676085818942962,
+      "grad_norm": 2.06941219144131,
+      "learning_rate": 1.7182819704667552e-05,
+      "loss": 0.9053,
       "step": 2557
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.2816915335062596e-06,
-      "loss": 0.72,
+      "epoch": 0.26771323914181056,
+      "grad_norm": 2.152176539140967,
+      "learning_rate": 1.718046088706179e-05,
+      "loss": 0.9493,
       "step": 2558
     },
     {
-      "epoch": 1.68,
-      "learning_rate": 1.2764732251598022e-06,
-      "loss": 0.5977,
+      "epoch": 0.26781789638932496,
+      "grad_norm": 2.09104261234392,
+      "learning_rate": 1.7178101244406512e-05,
+      "loss": 1.0374,
       "step": 2559
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2712648371478763e-06,
-      "loss": 0.7444,
+      "epoch": 0.26792255363683937,
+      "grad_norm": 2.346305270831672,
+      "learning_rate": 1.7175740776972855e-05,
+      "loss": 0.9768,
       "step": 2560
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2660663753934256e-06,
-      "loss": 0.6558,
+      "epoch": 0.2680272108843537,
+      "grad_norm": 2.2927895598403847,
+      "learning_rate": 1.7173379485032028e-05,
+      "loss": 0.939,
       "step": 2561
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.260877845808105e-06,
-      "loss": 0.6358,
+      "epoch": 0.2681318681318681,
+      "grad_norm": 2.3628912312336148,
+      "learning_rate": 1.7171017368855363e-05,
+      "loss": 1.0344,
       "step": 2562
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2556992542922708e-06,
-      "loss": 0.7264,
+      "epoch": 0.26823652537938253,
+      "grad_norm": 2.210950954427031,
+      "learning_rate": 1.7168654428714262e-05,
+      "loss": 1.0587,
       "step": 2563
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2505306067349853e-06,
-      "loss": 0.7023,
+      "epoch": 0.26834118262689693,
+      "grad_norm": 2.4183417784541517,
+      "learning_rate": 1.7166290664880234e-05,
+      "loss": 0.9456,
       "step": 2564
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2453719090139994e-06,
-      "loss": 0.7608,
+      "epoch": 0.2684458398744113,
+      "grad_norm": 2.123364599494718,
+      "learning_rate": 1.7163926077624886e-05,
+      "loss": 0.9227,
       "step": 2565
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2402231669957465e-06,
-      "loss": 0.6659,
+      "epoch": 0.2685504971219257,
+      "grad_norm": 2.1478356534162586,
+      "learning_rate": 1.7161560667219907e-05,
+      "loss": 0.9954,
       "step": 2566
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2350843865353447e-06,
-      "loss": 0.5094,
+      "epoch": 0.2686551543694401,
+      "grad_norm": 2.2147876235241797,
+      "learning_rate": 1.7159194433937087e-05,
+      "loss": 0.9726,
       "step": 2567
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2299555734765733e-06,
-      "loss": 0.7642,
+      "epoch": 0.26875981161695445,
+      "grad_norm": 2.1406444243414526,
+      "learning_rate": 1.7156827378048313e-05,
+      "loss": 0.9457,
       "step": 2568
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2248367336518874e-06,
-      "loss": 0.6149,
+      "epoch": 0.26886446886446885,
+      "grad_norm": 1.979769022328967,
+      "learning_rate": 1.7154459499825564e-05,
+      "loss": 0.9121,
       "step": 2569
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2197278728823947e-06,
-      "loss": 0.7276,
+      "epoch": 0.26896912611198326,
+      "grad_norm": 2.0747397671929626,
+      "learning_rate": 1.715209079954092e-05,
+      "loss": 1.0564,
       "step": 2570
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.214628996977857e-06,
-      "loss": 0.6943,
+      "epoch": 0.26907378335949766,
+      "grad_norm": 2.2405940574201164,
+      "learning_rate": 1.7149721277466537e-05,
+      "loss": 1.0748,
       "step": 2571
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.2095401117366823e-06,
-      "loss": 0.7002,
+      "epoch": 0.269178440607012,
+      "grad_norm": 2.38814683134122,
+      "learning_rate": 1.7147350933874693e-05,
+      "loss": 0.9308,
       "step": 2572
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.204461222945913e-06,
-      "loss": 0.6412,
+      "epoch": 0.2692830978545264,
+      "grad_norm": 2.1247668633842394,
+      "learning_rate": 1.7144979769037732e-05,
+      "loss": 1.0777,
       "step": 2573
     },
     {
-      "epoch": 1.69,
-      "learning_rate": 1.1993923363812288e-06,
-      "loss": 0.6831,
+      "epoch": 0.2693877551020408,
+      "grad_norm": 2.262069730010127,
+      "learning_rate": 1.714260778322811e-05,
+      "loss": 1.0581,
       "step": 2574
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1943334578069277e-06,
-      "loss": 0.8331,
+      "epoch": 0.26949241234955523,
+      "grad_norm": 2.184376285021379,
+      "learning_rate": 1.7140234976718376e-05,
+      "loss": 1.0185,
       "step": 2575
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1892845929759412e-06,
-      "loss": 0.712,
+      "epoch": 0.2695970695970696,
+      "grad_norm": 2.267123524429682,
+      "learning_rate": 1.7137861349781172e-05,
+      "loss": 0.9554,
       "step": 2576
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1842457476297953e-06,
-      "loss": 0.7024,
+      "epoch": 0.269701726844584,
+      "grad_norm": 2.36727063416499,
+      "learning_rate": 1.7135486902689226e-05,
+      "loss": 1.0104,
       "step": 2577
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.179216927498632e-06,
-      "loss": 0.6374,
+      "epoch": 0.2698063840920984,
+      "grad_norm": 2.0589416272264223,
+      "learning_rate": 1.713311163571537e-05,
+      "loss": 1.0443,
       "step": 2578
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1741981383011924e-06,
-      "loss": 0.6785,
+      "epoch": 0.26991104133961275,
+      "grad_norm": 1.902080690208934,
+      "learning_rate": 1.7130735549132528e-05,
+      "loss": 1.0204,
       "step": 2579
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1691893857448089e-06,
-      "loss": 0.7947,
+      "epoch": 0.27001569858712715,
+      "grad_norm": 2.4326326349397838,
+      "learning_rate": 1.7128358643213715e-05,
+      "loss": 1.0687,
       "step": 2580
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1641906755254007e-06,
-      "loss": 0.8104,
+      "epoch": 0.27012035583464156,
+      "grad_norm": 2.068315109987545,
+      "learning_rate": 1.7125980918232043e-05,
+      "loss": 1.0579,
       "step": 2581
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.159202013327464e-06,
-      "loss": 0.671,
+      "epoch": 0.27022501308215596,
+      "grad_norm": 2.2810820908724567,
+      "learning_rate": 1.712360237446072e-05,
+      "loss": 1.0034,
       "step": 2582
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.154223404824073e-06,
-      "loss": 0.7137,
+      "epoch": 0.2703296703296703,
+      "grad_norm": 2.16614232204801,
+      "learning_rate": 1.7121223012173037e-05,
+      "loss": 1.0208,
       "step": 2583
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1492548556768668e-06,
-      "loss": 0.6039,
+      "epoch": 0.2704343275771847,
+      "grad_norm": 2.3587680319750266,
+      "learning_rate": 1.71188428316424e-05,
+      "loss": 0.9352,
       "step": 2584
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1442963715360444e-06,
-      "loss": 0.721,
+      "epoch": 0.2705389848246991,
+      "grad_norm": 2.1472264377221766,
+      "learning_rate": 1.711646183314229e-05,
+      "loss": 1.0184,
       "step": 2585
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.139347958040361e-06,
-      "loss": 0.6585,
+      "epoch": 0.2706436420722135,
+      "grad_norm": 2.114763044771736,
+      "learning_rate": 1.711408001694628e-05,
+      "loss": 0.9823,
       "step": 2586
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1344096208171186e-06,
-      "loss": 0.6241,
+      "epoch": 0.2707482993197279,
+      "grad_norm": 2.241676810783072,
+      "learning_rate": 1.7111697383328066e-05,
+      "loss": 1.0956,
       "step": 2587
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1294813654821591e-06,
-      "loss": 0.877,
+      "epoch": 0.2708529565672423,
+      "grad_norm": 2.1020077851868635,
+      "learning_rate": 1.71093139325614e-05,
+      "loss": 1.0291,
       "step": 2588
     },
     {
-      "epoch": 1.7,
-      "learning_rate": 1.1245631976398608e-06,
-      "loss": 0.6719,
+      "epoch": 0.2709576138147567,
+      "grad_norm": 2.022324455238823,
+      "learning_rate": 1.710692966492015e-05,
+      "loss": 0.8429,
       "step": 2589
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.1196551228831321e-06,
-      "loss": 0.723,
+      "epoch": 0.27106227106227104,
+      "grad_norm": 2.112019727111849,
+      "learning_rate": 1.7104544580678276e-05,
+      "loss": 1.0672,
       "step": 2590
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.1147571467933983e-06,
-      "loss": 0.6761,
+      "epoch": 0.27116692830978545,
+      "grad_norm": 2.2294841470996736,
+      "learning_rate": 1.7102158680109827e-05,
+      "loss": 1.0076,
       "step": 2591
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.1098692749406058e-06,
-      "loss": 0.7836,
+      "epoch": 0.27127158555729985,
+      "grad_norm": 2.0630709446164976,
+      "learning_rate": 1.7099771963488948e-05,
+      "loss": 0.9595,
       "step": 2592
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.10499151288321e-06,
-      "loss": 0.6073,
+      "epoch": 0.27137624280481426,
+      "grad_norm": 2.17182249171102,
+      "learning_rate": 1.709738443108988e-05,
+      "loss": 0.9664,
       "step": 2593
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.1001238661681657e-06,
-      "loss": 0.5906,
+      "epoch": 0.2714809000523286,
+      "grad_norm": 1.6976481509832229,
+      "learning_rate": 1.7094996083186947e-05,
+      "loss": 0.8269,
       "step": 2594
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.0952663403309294e-06,
-      "loss": 0.8204,
+      "epoch": 0.271585557299843,
+      "grad_norm": 2.297472828435568,
+      "learning_rate": 1.709260692005459e-05,
+      "loss": 1.0357,
       "step": 2595
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.0904189408954445e-06,
-      "loss": 0.7257,
+      "epoch": 0.2716902145473574,
+      "grad_norm": 2.215787382610391,
+      "learning_rate": 1.7090216941967314e-05,
+      "loss": 1.0665,
       "step": 2596
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.085581673374142e-06,
-      "loss": 0.6876,
+      "epoch": 0.2717948717948718,
+      "grad_norm": 1.9350048535413022,
+      "learning_rate": 1.7087826149199735e-05,
+      "loss": 0.9186,
       "step": 2597
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.0807545432679278e-06,
-      "loss": 0.6238,
+      "epoch": 0.2718995290423862,
+      "grad_norm": 2.449759441348814,
+      "learning_rate": 1.7085434542026568e-05,
+      "loss": 1.0463,
       "step": 2598
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.0759375560661834e-06,
-      "loss": 0.7294,
+      "epoch": 0.2720041862899006,
+      "grad_norm": 2.045258414864514,
+      "learning_rate": 1.7083042120722606e-05,
+      "loss": 0.999,
       "step": 2599
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.07113071724675e-06,
-      "loss": 0.6923,
+      "epoch": 0.272108843537415,
+      "grad_norm": 2.1280959060956333,
+      "learning_rate": 1.7080648885562746e-05,
+      "loss": 1.0704,
       "step": 2600
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.0663340322759318e-06,
-      "loss": 0.5949,
+      "epoch": 0.27221350078492934,
+      "grad_norm": 2.058863219285702,
+      "learning_rate": 1.7078254836821978e-05,
+      "loss": 0.9272,
       "step": 2601
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.061547506608488e-06,
-      "loss": 0.673,
+      "epoch": 0.27231815803244375,
+      "grad_norm": 2.139899360821639,
+      "learning_rate": 1.7075859974775377e-05,
+      "loss": 0.9393,
       "step": 2602
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.0567711456876218e-06,
-      "loss": 0.6742,
+      "epoch": 0.27242281527995815,
+      "grad_norm": 2.4744357953899474,
+      "learning_rate": 1.7073464299698122e-05,
+      "loss": 1.1305,
       "step": 2603
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.0520049549449796e-06,
-      "loss": 0.7142,
+      "epoch": 0.2725274725274725,
+      "grad_norm": 2.070308973029564,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.9733,
       "step": 2604
     },
     {
-      "epoch": 1.71,
-      "learning_rate": 1.047248939800637e-06,
-      "loss": 0.7914,
+      "epoch": 0.2726321297749869,
+      "grad_norm": 2.2526454703490364,
+      "learning_rate": 1.706867051155281e-05,
+      "loss": 0.9585,
       "step": 2605
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 1.0425031056631007e-06,
-      "loss": 0.6933,
+      "epoch": 0.2727367870225013,
+      "grad_norm": 2.212572397139927,
+      "learning_rate": 1.7066272399035568e-05,
+      "loss": 0.9461,
       "step": 2606
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 1.0377674579293062e-06,
-      "loss": 0.649,
+      "epoch": 0.2728414442700157,
+      "grad_norm": 2.263865126136771,
+      "learning_rate": 1.70638734745893e-05,
+      "loss": 1.091,
       "step": 2607
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 1.0330420019845965e-06,
-      "loss": 0.6101,
+      "epoch": 0.27294610151753007,
+      "grad_norm": 2.1151782209613854,
+      "learning_rate": 1.7061473738489655e-05,
+      "loss": 0.8887,
       "step": 2608
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 1.02832674320273e-06,
-      "loss": 0.7183,
+      "epoch": 0.2730507587650445,
+      "grad_norm": 2.246083764897805,
+      "learning_rate": 1.705907319101236e-05,
+      "loss": 1.1668,
       "step": 2609
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 1.0236216869458626e-06,
-      "loss": 0.7793,
+      "epoch": 0.2731554160125589,
+      "grad_norm": 2.0910504342650262,
+      "learning_rate": 1.7056671832433246e-05,
+      "loss": 0.9767,
       "step": 2610
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 1.0189268385645524e-06,
-      "loss": 0.6259,
+      "epoch": 0.27326007326007323,
+      "grad_norm": 1.844847039254374,
+      "learning_rate": 1.7054269663028232e-05,
+      "loss": 0.9653,
       "step": 2611
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 1.0142422033977506e-06,
-      "loss": 0.6424,
+      "epoch": 0.27336473050758764,
+      "grad_norm": 2.324658565103168,
+      "learning_rate": 1.7051866683073337e-05,
+      "loss": 1.0032,
       "step": 2612
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 1.0095677867727927e-06,
-      "loss": 0.6323,
+      "epoch": 0.27346938775510204,
+      "grad_norm": 2.3338728511235636,
+      "learning_rate": 1.7049462892844663e-05,
+      "loss": 1.0501,
       "step": 2613
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 1.0049035940053897e-06,
-      "loss": 0.6899,
+      "epoch": 0.27357404500261645,
+      "grad_norm": 2.3092244039265895,
+      "learning_rate": 1.704705829261841e-05,
+      "loss": 1.0663,
       "step": 2614
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 1.0002496303996312e-06,
-      "loss": 0.6545,
+      "epoch": 0.2736787022501308,
+      "grad_norm": 2.158121468687123,
+      "learning_rate": 1.7044652882670874e-05,
+      "loss": 0.9789,
       "step": 2615
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 9.956059012479725e-07,
-      "loss": 0.6734,
+      "epoch": 0.2737833594976452,
+      "grad_norm": 2.4459807471934174,
+      "learning_rate": 1.704224666327844e-05,
+      "loss": 0.9862,
       "step": 2616
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 9.909724118312304e-07,
-      "loss": 0.589,
+      "epoch": 0.2738880167451596,
+      "grad_norm": 1.97502784186903,
+      "learning_rate": 1.703983963471759e-05,
+      "loss": 0.9387,
       "step": 2617
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 9.863491674185776e-07,
-      "loss": 0.7629,
+      "epoch": 0.273992673992674,
+      "grad_norm": 2.2367807182106834,
+      "learning_rate": 1.703743179726489e-05,
+      "loss": 0.9994,
       "step": 2618
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 9.817361732675346e-07,
-      "loss": 0.6731,
+      "epoch": 0.27409733124018837,
+      "grad_norm": 1.8582838838815607,
+      "learning_rate": 1.703502315119702e-05,
+      "loss": 0.969,
       "step": 2619
     },
     {
-      "epoch": 1.72,
-      "learning_rate": 9.77133434623968e-07,
-      "loss": 0.7219,
+      "epoch": 0.2742019884877028,
+      "grad_norm": 2.021019921533219,
+      "learning_rate": 1.7032613696790718e-05,
+      "loss": 1.0694,
       "step": 2620
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.7254095672208e-07,
-      "loss": 0.6475,
+      "epoch": 0.2743066457352172,
+      "grad_norm": 2.1420561465239993,
+      "learning_rate": 1.7030203434322852e-05,
+      "loss": 1.1157,
       "step": 2621
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.679587447844052e-07,
-      "loss": 0.7196,
+      "epoch": 0.27441130298273153,
+      "grad_norm": 2.192997753489374,
+      "learning_rate": 1.7027792364070358e-05,
+      "loss": 1.1059,
       "step": 2622
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.633868040218043e-07,
-      "loss": 0.634,
+      "epoch": 0.27451596023024594,
+      "grad_norm": 2.363367853081556,
+      "learning_rate": 1.7025380486310275e-05,
+      "loss": 1.024,
       "step": 2623
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.588251396334524e-07,
-      "loss": 0.717,
+      "epoch": 0.27462061747776034,
+      "grad_norm": 2.0560880376217803,
+      "learning_rate": 1.702296780131973e-05,
+      "loss": 0.9622,
       "step": 2624
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.542737568068472e-07,
-      "loss": 0.715,
+      "epoch": 0.27472527472527475,
+      "grad_norm": 2.0836175068267004,
+      "learning_rate": 1.7020554309375947e-05,
+      "loss": 0.8943,
       "step": 2625
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.497326607177881e-07,
-      "loss": 0.7918,
+      "epoch": 0.2748299319727891,
+      "grad_norm": 2.1909340338926797,
+      "learning_rate": 1.7018140010756246e-05,
+      "loss": 1.0327,
       "step": 2626
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.452018565303789e-07,
-      "loss": 0.524,
+      "epoch": 0.2749345892203035,
+      "grad_norm": 2.5561649979411385,
+      "learning_rate": 1.7015724905738025e-05,
+      "loss": 0.9729,
       "step": 2627
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.40681349397019e-07,
-      "loss": 0.6232,
+      "epoch": 0.2750392464678179,
+      "grad_norm": 2.0793636772474717,
+      "learning_rate": 1.7013308994598796e-05,
+      "loss": 0.8657,
       "step": 2628
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.361711444583965e-07,
-      "loss": 0.6308,
+      "epoch": 0.27514390371533226,
+      "grad_norm": 2.092569812266147,
+      "learning_rate": 1.701089227761614e-05,
+      "loss": 0.9508,
       "step": 2629
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.316712468434874e-07,
-      "loss": 0.743,
+      "epoch": 0.27524856096284667,
+      "grad_norm": 2.0132533158168804,
+      "learning_rate": 1.7008474755067755e-05,
+      "loss": 0.938,
       "step": 2630
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.271816616695439e-07,
-      "loss": 0.7237,
+      "epoch": 0.2753532182103611,
+      "grad_norm": 1.926555041783822,
+      "learning_rate": 1.700605642723141e-05,
+      "loss": 1.0228,
       "step": 2631
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.22702394042092e-07,
-      "loss": 0.6994,
+      "epoch": 0.2754578754578755,
+      "grad_norm": 2.234682199384651,
+      "learning_rate": 1.700363729438498e-05,
+      "loss": 1.091,
       "step": 2632
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.182334490549205e-07,
-      "loss": 0.7045,
+      "epoch": 0.27556253270538983,
+      "grad_norm": 2.424151503612507,
+      "learning_rate": 1.7001217356806424e-05,
+      "loss": 1.1339,
       "step": 2633
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.137748317900852e-07,
-      "loss": 0.5457,
+      "epoch": 0.27566718995290423,
+      "grad_norm": 2.3376824460058794,
+      "learning_rate": 1.6998796614773802e-05,
+      "loss": 0.9105,
       "step": 2634
     },
     {
-      "epoch": 1.73,
-      "learning_rate": 9.093265473178958e-07,
-      "loss": 0.732,
+      "epoch": 0.27577184720041864,
+      "grad_norm": 2.057373304612056,
+      "learning_rate": 1.6996375068565264e-05,
+      "loss": 0.9211,
       "step": 2635
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 9.048886006969093e-07,
-      "loss": 0.6142,
+      "epoch": 0.27587650444793305,
+      "grad_norm": 2.187999947763851,
+      "learning_rate": 1.6993952718459044e-05,
+      "loss": 1.0002,
       "step": 2636
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 9.004609969739308e-07,
-      "loss": 0.6379,
+      "epoch": 0.2759811616954474,
+      "grad_norm": 3.049341917756041,
+      "learning_rate": 1.699152956473348e-05,
+      "loss": 1.0756,
       "step": 2637
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.960437411839939e-07,
-      "loss": 0.6108,
+      "epoch": 0.2760858189429618,
+      "grad_norm": 2.014925562032507,
+      "learning_rate": 1.6989105607666993e-05,
+      "loss": 0.9493,
       "step": 2638
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.916368383503782e-07,
-      "loss": 0.6138,
+      "epoch": 0.2761904761904762,
+      "grad_norm": 2.2153507871464293,
+      "learning_rate": 1.6986680847538107e-05,
+      "loss": 0.9276,
       "step": 2639
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.872402934845814e-07,
-      "loss": 0.714,
+      "epoch": 0.27629513343799056,
+      "grad_norm": 2.042229591409745,
+      "learning_rate": 1.6984255284625425e-05,
+      "loss": 0.9397,
       "step": 2640
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.828541115863253e-07,
-      "loss": 0.7185,
+      "epoch": 0.27639979068550496,
+      "grad_norm": 2.317973974796922,
+      "learning_rate": 1.6981828919207656e-05,
+      "loss": 0.9842,
       "step": 2641
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.784782976435425e-07,
-      "loss": 0.7271,
+      "epoch": 0.27650444793301937,
+      "grad_norm": 2.0908721172521103,
+      "learning_rate": 1.6979401751563584e-05,
+      "loss": 1.0226,
       "step": 2642
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.741128566323797e-07,
-      "loss": 0.7367,
+      "epoch": 0.2766091051805338,
+      "grad_norm": 2.3306273872258294,
+      "learning_rate": 1.697697378197211e-05,
+      "loss": 0.9604,
       "step": 2643
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.697577935171864e-07,
-      "loss": 0.6225,
+      "epoch": 0.2767137624280481,
+      "grad_norm": 2.037113746353718,
+      "learning_rate": 1.69745450107122e-05,
+      "loss": 1.0226,
       "step": 2644
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.654131132505106e-07,
-      "loss": 0.6469,
+      "epoch": 0.27681841967556253,
+      "grad_norm": 2.3954710797112524,
+      "learning_rate": 1.697211543806293e-05,
+      "loss": 0.9924,
       "step": 2645
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.610788207730958e-07,
-      "loss": 0.7463,
+      "epoch": 0.27692307692307694,
+      "grad_norm": 2.2168247323986083,
+      "learning_rate": 1.6969685064303462e-05,
+      "loss": 1.0382,
       "step": 2646
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.567549210138637e-07,
-      "loss": 0.5993,
+      "epoch": 0.2770277341705913,
+      "grad_norm": 2.497079920708726,
+      "learning_rate": 1.696725388971305e-05,
+      "loss": 0.9435,
       "step": 2647
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.524414188899266e-07,
-      "loss": 0.6791,
+      "epoch": 0.2771323914181057,
+      "grad_norm": 2.526133193397156,
+      "learning_rate": 1.6964821914571046e-05,
+      "loss": 1.1236,
       "step": 2648
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.481383193065684e-07,
-      "loss": 0.7428,
+      "epoch": 0.2772370486656201,
+      "grad_norm": 2.626567114858534,
+      "learning_rate": 1.6962389139156883e-05,
+      "loss": 1.1044,
       "step": 2649
     },
     {
-      "epoch": 1.74,
-      "learning_rate": 8.438456271572437e-07,
-      "loss": 0.7658,
+      "epoch": 0.2773417059131345,
+      "grad_norm": 2.2181232399143402,
+      "learning_rate": 1.6959955563750094e-05,
+      "loss": 1.0042,
       "step": 2650
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.395633473235764e-07,
-      "loss": 0.7767,
+      "epoch": 0.27744636316064886,
+      "grad_norm": 2.2945097994513053,
+      "learning_rate": 1.69575211886303e-05,
+      "loss": 0.8661,
       "step": 2651
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.352914846753413e-07,
-      "loss": 0.6969,
+      "epoch": 0.27755102040816326,
+      "grad_norm": 2.310307859900674,
+      "learning_rate": 1.6955086014077215e-05,
+      "loss": 1.0534,
       "step": 2652
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.310300440704732e-07,
-      "loss": 0.5453,
+      "epoch": 0.27765567765567767,
+      "grad_norm": 2.069211757222885,
+      "learning_rate": 1.6952650040370652e-05,
+      "loss": 1.034,
       "step": 2653
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.267790303550527e-07,
-      "loss": 0.8743,
+      "epoch": 0.277760334903192,
+      "grad_norm": 1.9076079670724744,
+      "learning_rate": 1.6950213267790504e-05,
+      "loss": 0.8245,
       "step": 2654
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.225384483633036e-07,
-      "loss": 0.6619,
+      "epoch": 0.2778649921507064,
+      "grad_norm": 3.6932632423547944,
+      "learning_rate": 1.694777569661676e-05,
+      "loss": 0.8772,
       "step": 2655
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.183083029175853e-07,
-      "loss": 0.541,
+      "epoch": 0.27796964939822083,
+      "grad_norm": 1.9051520913310178,
+      "learning_rate": 1.6945337327129504e-05,
+      "loss": 0.8986,
       "step": 2656
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.140885988283898e-07,
-      "loss": 0.8292,
+      "epoch": 0.27807430664573524,
+      "grad_norm": 2.305637947733258,
+      "learning_rate": 1.694289815960891e-05,
+      "loss": 1.0223,
       "step": 2657
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.098793408943361e-07,
-      "loss": 0.7288,
+      "epoch": 0.2781789638932496,
+      "grad_norm": 2.7028733357321615,
+      "learning_rate": 1.6940458194335243e-05,
+      "loss": 0.9702,
       "step": 2658
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.056805339021645e-07,
-      "loss": 0.6667,
+      "epoch": 0.278283621140764,
+      "grad_norm": 2.390388801259551,
+      "learning_rate": 1.693801743158886e-05,
+      "loss": 0.9889,
       "step": 2659
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 8.014921826267286e-07,
-      "loss": 0.7894,
+      "epoch": 0.2783882783882784,
+      "grad_norm": 2.2381841916149066,
+      "learning_rate": 1.693557587165021e-05,
+      "loss": 1.0082,
       "step": 2660
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 7.973142918309928e-07,
-      "loss": 0.7047,
+      "epoch": 0.2784929356357928,
+      "grad_norm": 2.071420896938347,
+      "learning_rate": 1.693313351479983e-05,
+      "loss": 1.0009,
       "step": 2661
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 7.931468662660246e-07,
-      "loss": 0.6641,
+      "epoch": 0.27859759288330715,
+      "grad_norm": 2.446957221763722,
+      "learning_rate": 1.693069036131836e-05,
+      "loss": 1.0082,
       "step": 2662
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 7.88989910670993e-07,
-      "loss": 0.6732,
+      "epoch": 0.27870225013082156,
+      "grad_norm": 2.105944137294003,
+      "learning_rate": 1.692824641148651e-05,
+      "loss": 0.8971,
       "step": 2663
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 7.8484342977316e-07,
-      "loss": 0.6306,
+      "epoch": 0.27880690737833597,
+      "grad_norm": 2.3614811300833884,
+      "learning_rate": 1.6925801665585103e-05,
+      "loss": 0.9576,
       "step": 2664
     },
     {
-      "epoch": 1.75,
-      "learning_rate": 7.807074282878723e-07,
-      "loss": 0.6054,
+      "epoch": 0.2789115646258503,
+      "grad_norm": 2.058948239976716,
+      "learning_rate": 1.692335612389505e-05,
+      "loss": 0.9072,
       "step": 2665
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.765819109185635e-07,
-      "loss": 0.6509,
+      "epoch": 0.2790162218733647,
+      "grad_norm": 2.2051843696736153,
+      "learning_rate": 1.692090978669734e-05,
+      "loss": 0.9603,
       "step": 2666
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.724668823567439e-07,
-      "loss": 0.6439,
+      "epoch": 0.27912087912087913,
+      "grad_norm": 2.2044668414956456,
+      "learning_rate": 1.6918462654273063e-05,
+      "loss": 1.0132,
       "step": 2667
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.683623472819956e-07,
-      "loss": 0.767,
+      "epoch": 0.27922553636839353,
+      "grad_norm": 2.388456279940295,
+      "learning_rate": 1.6916014726903408e-05,
+      "loss": 0.9757,
       "step": 2668
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.642683103619686e-07,
-      "loss": 0.7687,
+      "epoch": 0.2793301936159079,
+      "grad_norm": 2.1759554884870957,
+      "learning_rate": 1.6913566004869637e-05,
+      "loss": 1.0602,
       "step": 2669
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.601847762523718e-07,
-      "loss": 0.6927,
+      "epoch": 0.2794348508634223,
+      "grad_norm": 2.4236744464592825,
+      "learning_rate": 1.6911116488453118e-05,
+      "loss": 1.096,
       "step": 2670
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.561117495969717e-07,
-      "loss": 0.671,
+      "epoch": 0.2795395081109367,
+      "grad_norm": 2.354246032793071,
+      "learning_rate": 1.69086661779353e-05,
+      "loss": 0.8912,
       "step": 2671
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.520492350275876e-07,
-      "loss": 0.6598,
+      "epoch": 0.27964416535845105,
+      "grad_norm": 1.9463786670522631,
+      "learning_rate": 1.6906215073597736e-05,
+      "loss": 1.0506,
       "step": 2672
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.479972371640831e-07,
-      "loss": 0.6614,
+      "epoch": 0.27974882260596545,
+      "grad_norm": 2.3854108573613213,
+      "learning_rate": 1.690376317572206e-05,
+      "loss": 1.0731,
       "step": 2673
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.439557606143633e-07,
-      "loss": 0.716,
+      "epoch": 0.27985347985347986,
+      "grad_norm": 2.287680313241345,
+      "learning_rate": 1.690131048459e-05,
+      "loss": 1.0453,
       "step": 2674
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.39924809974365e-07,
-      "loss": 0.6041,
+      "epoch": 0.27995813710099426,
+      "grad_norm": 2.23259788225129,
+      "learning_rate": 1.6898857000483375e-05,
+      "loss": 1.039,
       "step": 2675
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.359043898280571e-07,
-      "loss": 0.6687,
+      "epoch": 0.2800627943485086,
+      "grad_norm": 2.4640669787362626,
+      "learning_rate": 1.6896402723684095e-05,
+      "loss": 1.035,
       "step": 2676
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.318945047474357e-07,
-      "loss": 0.7462,
+      "epoch": 0.280167451596023,
+      "grad_norm": 2.226817030735215,
+      "learning_rate": 1.689394765447416e-05,
+      "loss": 1.0474,
       "step": 2677
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.278951592925154e-07,
-      "loss": 0.6986,
+      "epoch": 0.2802721088435374,
+      "grad_norm": 1.9193868173180237,
+      "learning_rate": 1.6891491793135663e-05,
+      "loss": 1.0085,
       "step": 2678
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.239063580113203e-07,
-      "loss": 0.5393,
+      "epoch": 0.28037676609105183,
+      "grad_norm": 2.1659873373450838,
+      "learning_rate": 1.6889035139950795e-05,
+      "loss": 0.9146,
       "step": 2679
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.19928105439891e-07,
-      "loss": 0.6693,
+      "epoch": 0.2804814233385662,
+      "grad_norm": 2.57303138488225,
+      "learning_rate": 1.6886577695201816e-05,
+      "loss": 1.1251,
       "step": 2680
     },
     {
-      "epoch": 1.76,
-      "learning_rate": 7.159604061022663e-07,
-      "loss": 0.6553,
+      "epoch": 0.2805860805860806,
+      "grad_norm": 2.042240644425295,
+      "learning_rate": 1.6884119459171104e-05,
+      "loss": 0.9829,
       "step": 2681
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 7.120032645104891e-07,
-      "loss": 0.6441,
+      "epoch": 0.280690737833595,
+      "grad_norm": 1.931077914764023,
+      "learning_rate": 1.688166043214111e-05,
+      "loss": 1.0262,
       "step": 2682
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 7.080566851645954e-07,
-      "loss": 0.6279,
+      "epoch": 0.28079539508110934,
+      "grad_norm": 2.6511446782137083,
+      "learning_rate": 1.687920061439438e-05,
+      "loss": 1.0604,
       "step": 2683
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 7.041206725526029e-07,
-      "loss": 0.6176,
+      "epoch": 0.28090005232862375,
+      "grad_norm": 2.3009554948835698,
+      "learning_rate": 1.6876740006213556e-05,
+      "loss": 0.8862,
       "step": 2684
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 7.0019523115052e-07,
-      "loss": 0.7056,
+      "epoch": 0.28100470957613816,
+      "grad_norm": 2.4972612313324274,
+      "learning_rate": 1.6874278607881362e-05,
+      "loss": 0.9708,
       "step": 2685
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.962803654223327e-07,
-      "loss": 0.7062,
+      "epoch": 0.28110936682365256,
+      "grad_norm": 2.156106891119093,
+      "learning_rate": 1.687181641968062e-05,
+      "loss": 0.9762,
       "step": 2686
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.923760798199986e-07,
-      "loss": 0.6981,
+      "epoch": 0.2812140240711669,
+      "grad_norm": 2.211635752927475,
+      "learning_rate": 1.6869353441894245e-05,
+      "loss": 1.0014,
       "step": 2687
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.884823787834471e-07,
-      "loss": 0.7004,
+      "epoch": 0.2813186813186813,
+      "grad_norm": 2.4317542466014284,
+      "learning_rate": 1.6866889674805233e-05,
+      "loss": 0.9357,
       "step": 2688
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.845992667405621e-07,
-      "loss": 0.5681,
+      "epoch": 0.2814233385661957,
+      "grad_norm": 2.4931495962019987,
+      "learning_rate": 1.686442511869667e-05,
+      "loss": 0.9232,
       "step": 2689
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.807267481071966e-07,
-      "loss": 0.724,
+      "epoch": 0.2815279958137101,
+      "grad_norm": 2.1696324074929394,
+      "learning_rate": 1.6861959773851754e-05,
+      "loss": 0.8919,
       "step": 2690
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.768648272871503e-07,
-      "loss": 0.6296,
+      "epoch": 0.2816326530612245,
+      "grad_norm": 2.168218378578173,
+      "learning_rate": 1.685949364055375e-05,
+      "loss": 1.1053,
       "step": 2691
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.730135086721746e-07,
-      "loss": 0.7343,
+      "epoch": 0.2817373103087389,
+      "grad_norm": 2.334913438897909,
+      "learning_rate": 1.6857026719086014e-05,
+      "loss": 1.0497,
       "step": 2692
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.691727966419603e-07,
-      "loss": 0.564,
+      "epoch": 0.2818419675562533,
+      "grad_norm": 2.307342605366626,
+      "learning_rate": 1.6854559009732006e-05,
+      "loss": 1.0982,
       "step": 2693
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.653426955641396e-07,
-      "loss": 0.6441,
+      "epoch": 0.28194662480376764,
+      "grad_norm": 2.11751636176187,
+      "learning_rate": 1.685209051277528e-05,
+      "loss": 0.9288,
       "step": 2694
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.615232097942769e-07,
-      "loss": 0.6008,
+      "epoch": 0.28205128205128205,
+      "grad_norm": 1.702869333805771,
+      "learning_rate": 1.684962122849946e-05,
+      "loss": 0.8285,
       "step": 2695
     },
     {
-      "epoch": 1.77,
-      "learning_rate": 6.577143436758659e-07,
-      "loss": 0.5842,
+      "epoch": 0.28215593929879645,
+      "grad_norm": 2.5478588197388983,
+      "learning_rate": 1.6847151157188274e-05,
+      "loss": 1.1483,
       "step": 2696
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.53916101540324e-07,
-      "loss": 0.7402,
+      "epoch": 0.2822605965463108,
+      "grad_norm": 1.6423189951148351,
+      "learning_rate": 1.6844680299125542e-05,
+      "loss": 0.8226,
       "step": 2697
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.501284877069824e-07,
-      "loss": 0.6289,
+      "epoch": 0.2823652537938252,
+      "grad_norm": 2.7051398620041796,
+      "learning_rate": 1.6842208654595164e-05,
+      "loss": 1.1181,
       "step": 2698
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.463515064830916e-07,
-      "loss": 0.6598,
+      "epoch": 0.2824699110413396,
+      "grad_norm": 2.328813950042279,
+      "learning_rate": 1.6839736223881144e-05,
+      "loss": 1.0891,
       "step": 2699
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.42585162163808e-07,
-      "loss": 0.5905,
+      "epoch": 0.282574568288854,
+      "grad_norm": 2.2355207375652766,
+      "learning_rate": 1.6837263007267567e-05,
+      "loss": 0.8753,
       "step": 2700
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.38829459032193e-07,
-      "loss": 0.7595,
+      "epoch": 0.2826792255363684,
+      "grad_norm": 2.395946624547171,
+      "learning_rate": 1.683478900503861e-05,
+      "loss": 1.0681,
       "step": 2701
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.350844013592062e-07,
-      "loss": 0.715,
+      "epoch": 0.2827838827838828,
+      "grad_norm": 1.8895809027956487,
+      "learning_rate": 1.6832314217478538e-05,
+      "loss": 0.966,
       "step": 2702
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.313499934037015e-07,
-      "loss": 0.7258,
+      "epoch": 0.2828885400313972,
+      "grad_norm": 2.108842553403623,
+      "learning_rate": 1.6829838644871716e-05,
+      "loss": 0.9917,
       "step": 2703
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.276262394124222e-07,
-      "loss": 0.5514,
+      "epoch": 0.2829931972789116,
+      "grad_norm": 1.9546264051078577,
+      "learning_rate": 1.6827362287502583e-05,
+      "loss": 0.9654,
       "step": 2704
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.239131436199941e-07,
-      "loss": 0.6828,
+      "epoch": 0.28309785452642594,
+      "grad_norm": 2.1750684526478095,
+      "learning_rate": 1.6824885145655685e-05,
+      "loss": 0.9892,
       "step": 2705
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.202107102489274e-07,
-      "loss": 0.6106,
+      "epoch": 0.28320251177394035,
+      "grad_norm": 2.2396694828655472,
+      "learning_rate": 1.6822407219615646e-05,
+      "loss": 0.8008,
       "step": 2706
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.165189435095997e-07,
-      "loss": 0.6825,
+      "epoch": 0.28330716902145475,
+      "grad_norm": 2.0811467507416777,
+      "learning_rate": 1.6819928509667193e-05,
+      "loss": 1.0043,
       "step": 2707
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.12837847600265e-07,
-      "loss": 0.7687,
+      "epoch": 0.2834118262689691,
+      "grad_norm": 2.156062381671543,
+      "learning_rate": 1.6817449016095124e-05,
+      "loss": 0.8708,
       "step": 2708
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.091674267070391e-07,
-      "loss": 0.6662,
+      "epoch": 0.2835164835164835,
+      "grad_norm": 2.2971033102764524,
+      "learning_rate": 1.681496873918434e-05,
+      "loss": 0.9787,
       "step": 2709
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.05507685003901e-07,
-      "loss": 0.8031,
+      "epoch": 0.2836211407639979,
+      "grad_norm": 1.8028079138511994,
+      "learning_rate": 1.681248767921984e-05,
+      "loss": 0.7918,
       "step": 2710
     },
     {
-      "epoch": 1.78,
-      "learning_rate": 6.018586266526838e-07,
-      "loss": 0.6603,
+      "epoch": 0.2837257980115123,
+      "grad_norm": 1.9797359746272927,
+      "learning_rate": 1.6810005836486693e-05,
+      "loss": 0.9433,
       "step": 2711
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.982202558030692e-07,
-      "loss": 0.7243,
+      "epoch": 0.28383045525902667,
+      "grad_norm": 2.1233134417416246,
+      "learning_rate": 1.6807523211270065e-05,
+      "loss": 1.1024,
       "step": 2712
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.945925765925875e-07,
-      "loss": 0.5889,
+      "epoch": 0.2839351125065411,
+      "grad_norm": 2.0119377968605217,
+      "learning_rate": 1.6805039803855225e-05,
+      "loss": 1.0896,
       "step": 2713
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.909755931466121e-07,
-      "loss": 0.5792,
+      "epoch": 0.2840397697540555,
+      "grad_norm": 2.1762408296082083,
+      "learning_rate": 1.6802555614527513e-05,
+      "loss": 1.1555,
       "step": 2714
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.873693095783529e-07,
-      "loss": 0.706,
+      "epoch": 0.28414442700156983,
+      "grad_norm": 2.0014014481336195,
+      "learning_rate": 1.6800070643572374e-05,
+      "loss": 1.0324,
       "step": 2715
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.837737299888502e-07,
-      "loss": 0.6878,
+      "epoch": 0.28424908424908424,
+      "grad_norm": 2.2844238601393676,
+      "learning_rate": 1.679758489127533e-05,
+      "loss": 1.0019,
       "step": 2716
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.80188858466969e-07,
-      "loss": 0.761,
+      "epoch": 0.28435374149659864,
+      "grad_norm": 2.107100861748374,
+      "learning_rate": 1.6795098357922004e-05,
+      "loss": 0.8909,
       "step": 2717
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.766146990894039e-07,
-      "loss": 0.6688,
+      "epoch": 0.28445839874411305,
+      "grad_norm": 2.309820312590304,
+      "learning_rate": 1.67926110437981e-05,
+      "loss": 1.0038,
       "step": 2718
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.730512559206625e-07,
-      "loss": 0.6699,
+      "epoch": 0.2845630559916274,
+      "grad_norm": 2.9601375842932005,
+      "learning_rate": 1.679012294918942e-05,
+      "loss": 1.0204,
       "step": 2719
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.694985330130698e-07,
-      "loss": 0.6538,
+      "epoch": 0.2846677132391418,
+      "grad_norm": 2.301335233810261,
+      "learning_rate": 1.6787634074381844e-05,
+      "loss": 1.1004,
       "step": 2720
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.659565344067552e-07,
-      "loss": 0.7084,
+      "epoch": 0.2847723704866562,
+      "grad_norm": 2.2912609106687367,
+      "learning_rate": 1.6785144419661356e-05,
+      "loss": 0.8342,
       "step": 2721
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.624252641296568e-07,
-      "loss": 0.6618,
+      "epoch": 0.2848770277341706,
+      "grad_norm": 2.2352473794305903,
+      "learning_rate": 1.678265398531402e-05,
+      "loss": 0.9728,
       "step": 2722
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.5890472619751e-07,
-      "loss": 0.6645,
+      "epoch": 0.28498168498168497,
+      "grad_norm": 2.378174318133523,
+      "learning_rate": 1.6780162771625987e-05,
+      "loss": 1.0774,
       "step": 2723
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.553949246138468e-07,
-      "loss": 0.6506,
+      "epoch": 0.2850863422291994,
+      "grad_norm": 2.18094544554846,
+      "learning_rate": 1.6777670778883507e-05,
+      "loss": 1.049,
       "step": 2724
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.5189586336999e-07,
-      "loss": 0.6687,
+      "epoch": 0.2851909994767138,
+      "grad_norm": 2.4504749993683723,
+      "learning_rate": 1.6775178007372915e-05,
+      "loss": 0.8661,
       "step": 2725
     },
     {
-      "epoch": 1.79,
-      "learning_rate": 5.484075464450456e-07,
-      "loss": 0.6963,
+      "epoch": 0.28529565672422813,
+      "grad_norm": 2.282848626968931,
+      "learning_rate": 1.677268445738064e-05,
+      "loss": 1.0587,
       "step": 2726
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.449299778059069e-07,
-      "loss": 0.6859,
+      "epoch": 0.28540031397174254,
+      "grad_norm": 2.2046921458707653,
+      "learning_rate": 1.677019012919319e-05,
+      "loss": 0.8776,
       "step": 2727
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.414631614072385e-07,
-      "loss": 0.6964,
+      "epoch": 0.28550497121925694,
+      "grad_norm": 1.9336248225640682,
+      "learning_rate": 1.6767695023097164e-05,
+      "loss": 0.9612,
       "step": 2728
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.380071011914823e-07,
-      "loss": 0.6917,
+      "epoch": 0.28560962846677135,
+      "grad_norm": 2.1614112374322523,
+      "learning_rate": 1.6765199139379265e-05,
+      "loss": 1.0527,
       "step": 2729
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.345618010888431e-07,
-      "loss": 0.6154,
+      "epoch": 0.2857142857142857,
+      "grad_norm": 2.026874347225188,
+      "learning_rate": 1.676270247832627e-05,
+      "loss": 0.9901,
       "step": 2730
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.311272650172949e-07,
-      "loss": 0.7428,
+      "epoch": 0.2858189429618001,
+      "grad_norm": 2.5279801213743056,
+      "learning_rate": 1.6760205040225052e-05,
+      "loss": 1.0751,
       "step": 2731
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.277034968825667e-07,
-      "loss": 0.6999,
+      "epoch": 0.2859236002093145,
+      "grad_norm": 2.408066662111667,
+      "learning_rate": 1.6757706825362565e-05,
+      "loss": 1.0041,
       "step": 2732
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.242905005781451e-07,
-      "loss": 0.6909,
+      "epoch": 0.28602825745682886,
+      "grad_norm": 2.2561275528825693,
+      "learning_rate": 1.675520783402587e-05,
+      "loss": 1.0136,
       "step": 2733
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.208882799852666e-07,
-      "loss": 0.7756,
+      "epoch": 0.28613291470434327,
+      "grad_norm": 2.079863026171039,
+      "learning_rate": 1.6752708066502097e-05,
+      "loss": 1.0614,
       "step": 2734
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.174968389729118e-07,
-      "loss": 0.6534,
+      "epoch": 0.28623757195185767,
+      "grad_norm": 2.081290751507961,
+      "learning_rate": 1.675020752307848e-05,
+      "loss": 1.0573,
       "step": 2735
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.141161813978046e-07,
-      "loss": 0.654,
+      "epoch": 0.2863422291993721,
+      "grad_norm": 2.030844176643617,
+      "learning_rate": 1.6747706204042335e-05,
+      "loss": 1.1131,
       "step": 2736
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.107463111044053e-07,
-      "loss": 0.6402,
+      "epoch": 0.28644688644688643,
+      "grad_norm": 2.1050959828240936,
+      "learning_rate": 1.6745204109681064e-05,
+      "loss": 0.9127,
       "step": 2737
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.073872319249073e-07,
-      "loss": 0.6297,
+      "epoch": 0.28655154369440083,
+      "grad_norm": 2.055509983409726,
+      "learning_rate": 1.6742701240282174e-05,
+      "loss": 0.9553,
       "step": 2738
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.04038947679234e-07,
-      "loss": 0.7351,
+      "epoch": 0.28665620094191524,
+      "grad_norm": 2.137470369141178,
+      "learning_rate": 1.6740197596133238e-05,
+      "loss": 1.0597,
       "step": 2739
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 5.007014621750272e-07,
-      "loss": 0.6511,
+      "epoch": 0.2867608581894296,
+      "grad_norm": 2.048361552464316,
+      "learning_rate": 1.6737693177521936e-05,
+      "loss": 0.8446,
       "step": 2740
     },
     {
-      "epoch": 1.8,
-      "learning_rate": 4.973747792076555e-07,
-      "loss": 0.579,
+      "epoch": 0.286865515436944,
+      "grad_norm": 2.12112325524762,
+      "learning_rate": 1.673518798473603e-05,
+      "loss": 1.0618,
       "step": 2741
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.940589025601983e-07,
-      "loss": 0.7071,
+      "epoch": 0.2869701726844584,
+      "grad_norm": 2.6242506865312363,
+      "learning_rate": 1.6732682018063368e-05,
+      "loss": 0.8252,
       "step": 2742
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.907538360034502e-07,
-      "loss": 0.6111,
+      "epoch": 0.2870748299319728,
+      "grad_norm": 3.0996650803222674,
+      "learning_rate": 1.6730175277791895e-05,
+      "loss": 1.0003,
       "step": 2743
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.874595832959061e-07,
-      "loss": 0.6774,
+      "epoch": 0.28717948717948716,
+      "grad_norm": 2.350027250748929,
+      "learning_rate": 1.6727667764209638e-05,
+      "loss": 0.9309,
       "step": 2744
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.841761481837703e-07,
-      "loss": 0.7553,
+      "epoch": 0.28728414442700156,
+      "grad_norm": 2.0077156617023855,
+      "learning_rate": 1.6725159477604716e-05,
+      "loss": 1.0252,
       "step": 2745
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.809035344009416e-07,
-      "loss": 0.6495,
+      "epoch": 0.28738880167451597,
+      "grad_norm": 2.635741023964438,
+      "learning_rate": 1.672265041826534e-05,
+      "loss": 0.8513,
       "step": 2746
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.776417456690152e-07,
-      "loss": 0.7881,
+      "epoch": 0.2874934589220304,
+      "grad_norm": 2.2433336022783172,
+      "learning_rate": 1.67201405864798e-05,
+      "loss": 1.0492,
       "step": 2747
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.74390785697274e-07,
-      "loss": 0.6295,
+      "epoch": 0.2875981161695447,
+      "grad_norm": 2.0807771153561547,
+      "learning_rate": 1.6717629982536484e-05,
+      "loss": 0.9764,
       "step": 2748
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.7115065818268636e-07,
-      "loss": 0.7305,
+      "epoch": 0.28770277341705913,
+      "grad_norm": 2.3123250176131807,
+      "learning_rate": 1.6715118606723867e-05,
+      "loss": 0.9599,
       "step": 2749
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.6792136680990363e-07,
-      "loss": 0.7171,
+      "epoch": 0.28780743066457354,
+      "grad_norm": 2.0101181547077998,
+      "learning_rate": 1.6712606459330503e-05,
+      "loss": 0.8779,
       "step": 2750
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.6470291525125386e-07,
-      "loss": 0.6919,
+      "epoch": 0.2879120879120879,
+      "grad_norm": 2.0975900598381467,
+      "learning_rate": 1.6710093540645056e-05,
+      "loss": 0.9257,
       "step": 2751
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.614953071667383e-07,
-      "loss": 0.7825,
+      "epoch": 0.2880167451596023,
+      "grad_norm": 1.958047612574768,
+      "learning_rate": 1.6707579850956256e-05,
+      "loss": 0.8114,
       "step": 2752
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.582985462040268e-07,
-      "loss": 0.7472,
+      "epoch": 0.2881214024071167,
+      "grad_norm": 2.1295560514183425,
+      "learning_rate": 1.6705065390552934e-05,
+      "loss": 1.0102,
       "step": 2753
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.5511263599845145e-07,
-      "loss": 0.6995,
+      "epoch": 0.2882260596546311,
+      "grad_norm": 2.3322838337100795,
+      "learning_rate": 1.6702550159724005e-05,
+      "loss": 1.0251,
       "step": 2754
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.5193758017301084e-07,
-      "loss": 0.6419,
+      "epoch": 0.28833071690214546,
+      "grad_norm": 2.0269061447136276,
+      "learning_rate": 1.6700034158758476e-05,
+      "loss": 0.9773,
       "step": 2755
     },
     {
-      "epoch": 1.81,
-      "learning_rate": 4.4877338233835223e-07,
-      "loss": 0.6327,
+      "epoch": 0.28843537414965986,
+      "grad_norm": 2.2135086571849927,
+      "learning_rate": 1.6697517387945437e-05,
+      "loss": 1.0397,
       "step": 2756
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.456200460927862e-07,
-      "loss": 0.6755,
+      "epoch": 0.28854003139717427,
+      "grad_norm": 3.020623424201838,
+      "learning_rate": 1.669499984757408e-05,
+      "loss": 1.1779,
       "step": 2757
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.4247757502225983e-07,
-      "loss": 0.6664,
+      "epoch": 0.2886446886446886,
+      "grad_norm": 2.3913345388059675,
+      "learning_rate": 1.669248153793366e-05,
+      "loss": 0.9306,
       "step": 2758
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.3934597270037017e-07,
-      "loss": 0.6987,
+      "epoch": 0.288749345892203,
+      "grad_norm": 2.0642701710655196,
+      "learning_rate": 1.6689962459313547e-05,
+      "loss": 1.0335,
       "step": 2759
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.3622524268835507e-07,
-      "loss": 0.6946,
+      "epoch": 0.28885400313971743,
+      "grad_norm": 2.084776755214591,
+      "learning_rate": 1.6687442612003185e-05,
+      "loss": 0.8789,
       "step": 2760
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.3311538853508697e-07,
-      "loss": 0.7739,
+      "epoch": 0.28895866038723184,
+      "grad_norm": 2.1559657822579616,
+      "learning_rate": 1.668492199629211e-05,
+      "loss": 0.873,
       "step": 2761
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.3001641377707125e-07,
-      "loss": 0.6093,
+      "epoch": 0.2890633176347462,
+      "grad_norm": 2.1968258921987043,
+      "learning_rate": 1.668240061246995e-05,
+      "loss": 1.0585,
       "step": 2762
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.2692832193843794e-07,
-      "loss": 0.6907,
+      "epoch": 0.2891679748822606,
+      "grad_norm": 2.037768963343927,
+      "learning_rate": 1.6679878460826407e-05,
+      "loss": 1.0065,
       "step": 2763
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.2385111653094667e-07,
-      "loss": 0.7846,
+      "epoch": 0.289272632129775,
+      "grad_norm": 2.311144324625343,
+      "learning_rate": 1.6677355541651292e-05,
+      "loss": 1.0024,
       "step": 2764
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.207848010539717e-07,
-      "loss": 0.8014,
+      "epoch": 0.2893772893772894,
+      "grad_norm": 2.5667946832040536,
+      "learning_rate": 1.6674831855234486e-05,
+      "loss": 0.9484,
       "step": 2765
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.1772937899450804e-07,
-      "loss": 0.6396,
+      "epoch": 0.28948194662480375,
+      "grad_norm": 2.434219899597603,
+      "learning_rate": 1.667230740186597e-05,
+      "loss": 1.0179,
       "step": 2766
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.1468485382716085e-07,
-      "loss": 0.6481,
+      "epoch": 0.28958660387231816,
+      "grad_norm": 2.7601269464079676,
+      "learning_rate": 1.6669782181835807e-05,
+      "loss": 1.0228,
       "step": 2767
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.1165122901414055e-07,
-      "loss": 0.742,
+      "epoch": 0.28969126111983257,
+      "grad_norm": 2.5403438469183697,
+      "learning_rate": 1.666725619543415e-05,
+      "loss": 1.047,
       "step": 2768
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.0862850800526655e-07,
-      "loss": 0.7864,
+      "epoch": 0.2897959183673469,
+      "grad_norm": 2.051984141112977,
+      "learning_rate": 1.666472944295124e-05,
+      "loss": 0.9447,
       "step": 2769
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.056166942379558e-07,
-      "loss": 0.7122,
+      "epoch": 0.2899005756148613,
+      "grad_norm": 2.512139628224453,
+      "learning_rate": 1.666220192467741e-05,
+      "loss": 1.1367,
       "step": 2770
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 4.026157911372242e-07,
-      "loss": 0.7419,
+      "epoch": 0.29000523286237573,
+      "grad_norm": 2.718503682041811,
+      "learning_rate": 1.6659673640903067e-05,
+      "loss": 1.0922,
       "step": 2771
     },
     {
-      "epoch": 1.82,
-      "learning_rate": 3.9962580211567405e-07,
-      "loss": 0.7055,
+      "epoch": 0.29010989010989013,
+      "grad_norm": 2.5150069168743885,
+      "learning_rate": 1.6657144591918726e-05,
+      "loss": 1.1432,
       "step": 2772
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.9664673057350223e-07,
-      "loss": 0.7459,
+      "epoch": 0.2902145473574045,
+      "grad_norm": 2.3080940442950304,
+      "learning_rate": 1.665461477801497e-05,
+      "loss": 1.1565,
       "step": 2773
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.9367857989848765e-07,
-      "loss": 0.6831,
+      "epoch": 0.2903192046049189,
+      "grad_norm": 3.162428243155795,
+      "learning_rate": 1.6652084199482496e-05,
+      "loss": 1.1392,
       "step": 2774
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.907213534659926e-07,
-      "loss": 0.6474,
+      "epoch": 0.2904238618524333,
+      "grad_norm": 2.460964843259728,
+      "learning_rate": 1.6649552856612056e-05,
+      "loss": 1.0668,
       "step": 2775
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.877750546389536e-07,
-      "loss": 0.7284,
+      "epoch": 0.29052851909994765,
+      "grad_norm": 2.2837818852824747,
+      "learning_rate": 1.6647020749694513e-05,
+      "loss": 1.0405,
       "step": 2776
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.8483968676788055e-07,
-      "loss": 0.6127,
+      "epoch": 0.29063317634746205,
+      "grad_norm": 2.3757375317221348,
+      "learning_rate": 1.6644487879020812e-05,
+      "loss": 0.988,
       "step": 2777
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.819152531908543e-07,
-      "loss": 0.726,
+      "epoch": 0.29073783359497646,
+      "grad_norm": 2.0992714934923065,
+      "learning_rate": 1.6641954244881984e-05,
+      "loss": 1.0678,
       "step": 2778
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.790017572335214e-07,
-      "loss": 0.6844,
+      "epoch": 0.29084249084249086,
+      "grad_norm": 2.4400916221420963,
+      "learning_rate": 1.6639419847569147e-05,
+      "loss": 1.0578,
       "step": 2779
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.7609920220908813e-07,
-      "loss": 0.5946,
+      "epoch": 0.2909471480900052,
+      "grad_norm": 2.1469546303983043,
+      "learning_rate": 1.6636884687373508e-05,
+      "loss": 0.9257,
       "step": 2780
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.73207591418322e-07,
-      "loss": 0.6261,
+      "epoch": 0.2910518053375196,
+      "grad_norm": 2.334369619059665,
+      "learning_rate": 1.663434876458637e-05,
+      "loss": 0.9772,
       "step": 2781
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.703269281495392e-07,
-      "loss": 0.8017,
+      "epoch": 0.291156462585034,
+      "grad_norm": 2.3345181976583653,
+      "learning_rate": 1.6631812079499106e-05,
+      "loss": 1.0597,
       "step": 2782
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.674572156786127e-07,
-      "loss": 0.6984,
+      "epoch": 0.2912611198325484,
+      "grad_norm": 2.0700463369958815,
+      "learning_rate": 1.6629274632403193e-05,
+      "loss": 1.0176,
       "step": 2783
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.6459845726895984e-07,
-      "loss": 0.6563,
+      "epoch": 0.2913657770800628,
+      "grad_norm": 2.202087466914214,
+      "learning_rate": 1.6626736423590186e-05,
+      "loss": 1.0216,
       "step": 2784
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.6175065617154113e-07,
-      "loss": 0.6191,
+      "epoch": 0.2914704343275772,
+      "grad_norm": 2.0357122851821883,
+      "learning_rate": 1.6624197453351723e-05,
+      "loss": 1.0036,
       "step": 2785
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.5891381562485504e-07,
-      "loss": 0.7208,
+      "epoch": 0.2915750915750916,
+      "grad_norm": 2.6405418320254457,
+      "learning_rate": 1.662165772197955e-05,
+      "loss": 1.0112,
       "step": 2786
     },
     {
-      "epoch": 1.83,
-      "learning_rate": 3.5608793885493766e-07,
-      "loss": 0.6748,
+      "epoch": 0.29167974882260594,
+      "grad_norm": 2.1803315886051453,
+      "learning_rate": 1.661911722976548e-05,
+      "loss": 0.8626,
       "step": 2787
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.53273029075355e-07,
-      "loss": 0.5896,
+      "epoch": 0.29178440607012035,
+      "grad_norm": 1.9949267891582323,
+      "learning_rate": 1.6616575977001423e-05,
+      "loss": 0.8258,
       "step": 2788
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.504690894872065e-07,
-      "loss": 0.6115,
+      "epoch": 0.29188906331763476,
+      "grad_norm": 2.1296048269861965,
+      "learning_rate": 1.661403396397937e-05,
+      "loss": 0.9722,
       "step": 2789
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.476761232791115e-07,
-      "loss": 0.84,
+      "epoch": 0.29199372056514916,
+      "grad_norm": 1.9633811576783402,
+      "learning_rate": 1.661149119099141e-05,
+      "loss": 0.9531,
       "step": 2790
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.4489413362721156e-07,
-      "loss": 0.6417,
+      "epoch": 0.2920983778126635,
+      "grad_norm": 2.4559017455704146,
+      "learning_rate": 1.6608947658329705e-05,
+      "loss": 0.9185,
       "step": 2791
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.4212312369516496e-07,
-      "loss": 0.6939,
+      "epoch": 0.2922030350601779,
+      "grad_norm": 2.6490541995498593,
+      "learning_rate": 1.6606403366286522e-05,
+      "loss": 1.0208,
       "step": 2792
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.393630966341477e-07,
-      "loss": 0.6228,
+      "epoch": 0.2923076923076923,
+      "grad_norm": 2.230218351877687,
+      "learning_rate": 1.6603858315154194e-05,
+      "loss": 0.9397,
       "step": 2793
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.3661405558284124e-07,
-      "loss": 0.6105,
+      "epoch": 0.2924123495552067,
+      "grad_norm": 2.386055723778366,
+      "learning_rate": 1.660131250522516e-05,
+      "loss": 0.8659,
       "step": 2794
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.338760036674382e-07,
-      "loss": 0.6057,
+      "epoch": 0.2925170068027211,
+      "grad_norm": 2.223350214103197,
+      "learning_rate": 1.659876593679194e-05,
+      "loss": 1.0085,
       "step": 2795
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.311489440016302e-07,
-      "loss": 0.541,
+      "epoch": 0.2926216640502355,
+      "grad_norm": 2.362704726315233,
+      "learning_rate": 1.6596218610147134e-05,
+      "loss": 1.0483,
       "step": 2796
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.284328796866121e-07,
-      "loss": 0.7407,
+      "epoch": 0.2927263212977499,
+      "grad_norm": 2.554013585062049,
+      "learning_rate": 1.6593670525583437e-05,
+      "loss": 0.9678,
       "step": 2797
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.25727813811072e-07,
-      "loss": 0.6918,
+      "epoch": 0.29283097854526424,
+      "grad_norm": 2.9600036988498477,
+      "learning_rate": 1.659112168339363e-05,
+      "loss": 0.855,
       "step": 2798
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.230337494511937e-07,
-      "loss": 0.6731,
+      "epoch": 0.29293563579277865,
+      "grad_norm": 2.1825672871247135,
+      "learning_rate": 1.6588572083870583e-05,
+      "loss": 1.0531,
       "step": 2799
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.2035068967064876e-07,
-      "loss": 0.6961,
+      "epoch": 0.29304029304029305,
+      "grad_norm": 2.0933057115237603,
+      "learning_rate": 1.6586021727307247e-05,
+      "loss": 1.0515,
       "step": 2800
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.176786375205931e-07,
-      "loss": 0.7355,
+      "epoch": 0.2931449502878074,
+      "grad_norm": 2.400496066375719,
+      "learning_rate": 1.658347061399666e-05,
+      "loss": 1.0419,
       "step": 2801
     },
     {
-      "epoch": 1.84,
-      "learning_rate": 3.150175960396673e-07,
-      "loss": 0.6905,
+      "epoch": 0.2932496075353218,
+      "grad_norm": 2.016597627986569,
+      "learning_rate": 1.6580918744231955e-05,
+      "loss": 0.9104,
       "step": 2802
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 3.123675682539884e-07,
-      "loss": 0.6961,
+      "epoch": 0.2933542647828362,
+      "grad_norm": 2.3757951069404326,
+      "learning_rate": 1.6578366118306343e-05,
+      "loss": 1.1151,
       "step": 2803
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 3.0972855717715134e-07,
-      "loss": 0.5558,
+      "epoch": 0.2934589220303506,
+      "grad_norm": 2.0100169684030114,
+      "learning_rate": 1.6575812736513133e-05,
+      "loss": 0.9798,
       "step": 2804
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 3.0710056581021886e-07,
-      "loss": 0.5556,
+      "epoch": 0.29356357927786497,
+      "grad_norm": 2.1122429080682945,
+      "learning_rate": 1.6573258599145704e-05,
+      "loss": 0.9442,
       "step": 2805
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 3.0448359714172593e-07,
-      "loss": 0.6529,
+      "epoch": 0.2936682365253794,
+      "grad_norm": 2.154489488864629,
+      "learning_rate": 1.657070370649754e-05,
+      "loss": 1.0333,
       "step": 2806
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 3.018776541476709e-07,
-      "loss": 0.6608,
+      "epoch": 0.2937728937728938,
+      "grad_norm": 2.275200742495966,
+      "learning_rate": 1.65681480588622e-05,
+      "loss": 1.083,
       "step": 2807
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.9928273979151433e-07,
-      "loss": 0.779,
+      "epoch": 0.2938775510204082,
+      "grad_norm": 2.1710490075798514,
+      "learning_rate": 1.656559165653333e-05,
+      "loss": 1.0605,
       "step": 2808
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.966988570241747e-07,
-      "loss": 0.6715,
+      "epoch": 0.29398220826792254,
+      "grad_norm": 2.598428034733466,
+      "learning_rate": 1.6563034499804672e-05,
+      "loss": 1.0464,
       "step": 2809
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.9412600878402696e-07,
-      "loss": 0.56,
+      "epoch": 0.29408686551543695,
+      "grad_norm": 2.151540858385915,
+      "learning_rate": 1.6560476588970044e-05,
+      "loss": 1.0399,
       "step": 2810
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.915641979968953e-07,
-      "loss": 0.8199,
+      "epoch": 0.29419152276295135,
+      "grad_norm": 2.1278728592287868,
+      "learning_rate": 1.6557917924323358e-05,
+      "loss": 1.0988,
       "step": 2811
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.8901342757605256e-07,
-      "loss": 0.7911,
+      "epoch": 0.2942961800104657,
+      "grad_norm": 2.1705417888870944,
+      "learning_rate": 1.6555358506158604e-05,
+      "loss": 0.9765,
       "step": 2812
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.8647370042222064e-07,
-      "loss": 0.6145,
+      "epoch": 0.2944008372579801,
+      "grad_norm": 2.206832470090881,
+      "learning_rate": 1.6552798334769874e-05,
+      "loss": 0.9632,
       "step": 2813
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.839450194235549e-07,
-      "loss": 0.745,
+      "epoch": 0.2945054945054945,
+      "grad_norm": 2.1115011840786586,
+      "learning_rate": 1.655023741045133e-05,
+      "loss": 1.1335,
       "step": 2814
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.8142738745565724e-07,
-      "loss": 0.7071,
+      "epoch": 0.2946101517530089,
+      "grad_norm": 1.90252480707429,
+      "learning_rate": 1.6547675733497226e-05,
+      "loss": 1.0063,
       "step": 2815
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.789208073815608e-07,
-      "loss": 0.7781,
+      "epoch": 0.29471480900052327,
+      "grad_norm": 2.349864203175167,
+      "learning_rate": 1.6545113304201906e-05,
+      "loss": 1.018,
       "step": 2816
     },
     {
-      "epoch": 1.85,
-      "learning_rate": 2.764252820517299e-07,
-      "loss": 0.6849,
+      "epoch": 0.2948194662480377,
+      "grad_norm": 1.9420251810407316,
+      "learning_rate": 1.6542550122859804e-05,
+      "loss": 0.88,
       "step": 2817
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.739408143040623e-07,
-      "loss": 0.7383,
+      "epoch": 0.2949241234955521,
+      "grad_norm": 2.2700830883100767,
+      "learning_rate": 1.6539986189765425e-05,
+      "loss": 1.0005,
       "step": 2818
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.714674069638712e-07,
-      "loss": 0.6071,
+      "epoch": 0.29502878074306643,
+      "grad_norm": 2.214309833369823,
+      "learning_rate": 1.6537421505213377e-05,
+      "loss": 0.9971,
       "step": 2819
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.6900506284390446e-07,
-      "loss": 0.725,
+      "epoch": 0.29513343799058084,
+      "grad_norm": 2.0402371350258415,
+      "learning_rate": 1.6534856069498345e-05,
+      "loss": 1.0959,
       "step": 2820
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.6655378474432e-07,
-      "loss": 0.6599,
+      "epoch": 0.29523809523809524,
+      "grad_norm": 2.0958531308968196,
+      "learning_rate": 1.6532289882915104e-05,
+      "loss": 1.0072,
       "step": 2821
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.641135754526958e-07,
-      "loss": 0.6231,
+      "epoch": 0.29534275248560965,
+      "grad_norm": 1.858230244811686,
+      "learning_rate": 1.6529722945758512e-05,
+      "loss": 0.8212,
       "step": 2822
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.6168443774401976e-07,
-      "loss": 0.6586,
+      "epoch": 0.295447409733124,
+      "grad_norm": 1.817144737007705,
+      "learning_rate": 1.6527155258323517e-05,
+      "loss": 1.0134,
       "step": 2823
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.592663743806911e-07,
-      "loss": 0.8006,
+      "epoch": 0.2955520669806384,
+      "grad_norm": 2.4276651505529454,
+      "learning_rate": 1.652458682090515e-05,
+      "loss": 1.0547,
       "step": 2824
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.5685938811251476e-07,
-      "loss": 0.6761,
+      "epoch": 0.2956567242281528,
+      "grad_norm": 2.963051911787662,
+      "learning_rate": 1.6522017633798534e-05,
+      "loss": 1.0579,
       "step": 2825
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.544634816766989e-07,
-      "loss": 0.6951,
+      "epoch": 0.2957613814756672,
+      "grad_norm": 2.2165478606094355,
+      "learning_rate": 1.6519447697298866e-05,
+      "loss": 0.9951,
       "step": 2826
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.520786577978518e-07,
-      "loss": 0.6571,
+      "epoch": 0.29586603872318157,
+      "grad_norm": 1.9820489456783845,
+      "learning_rate": 1.6516877011701443e-05,
+      "loss": 0.7579,
       "step": 2827
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.4970491918797855e-07,
-      "loss": 0.6903,
+      "epoch": 0.295970695970696,
+      "grad_norm": 2.211338495085747,
+      "learning_rate": 1.651430557730164e-05,
+      "loss": 1.0503,
       "step": 2828
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.4734226854647883e-07,
-      "loss": 0.7043,
+      "epoch": 0.2960753532182104,
+      "grad_norm": 2.0412669316228964,
+      "learning_rate": 1.6511733394394922e-05,
+      "loss": 0.8973,
       "step": 2829
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.449907085601422e-07,
-      "loss": 0.6137,
+      "epoch": 0.29618001046572473,
+      "grad_norm": 1.930076486135168,
+      "learning_rate": 1.6509160463276835e-05,
+      "loss": 0.945,
       "step": 2830
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.426502419031451e-07,
-      "loss": 0.8411,
+      "epoch": 0.29628466771323914,
+      "grad_norm": 2.2334950597171677,
+      "learning_rate": 1.6506586784243015e-05,
+      "loss": 0.9924,
       "step": 2831
     },
     {
-      "epoch": 1.86,
-      "learning_rate": 2.4032087123705414e-07,
-      "loss": 0.7115,
+      "epoch": 0.29638932496075354,
+      "grad_norm": 2.3824449949926727,
+      "learning_rate": 1.650401235758918e-05,
+      "loss": 0.8262,
       "step": 2832
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.3800259921081037e-07,
-      "loss": 0.5807,
+      "epoch": 0.29649398220826795,
+      "grad_norm": 2.2784589120562795,
+      "learning_rate": 1.6501437183611148e-05,
+      "loss": 1.1117,
       "step": 2833
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.35695428460736e-07,
-      "loss": 0.7759,
+      "epoch": 0.2965986394557823,
+      "grad_norm": 2.250028215509174,
+      "learning_rate": 1.64988612626048e-05,
+      "loss": 1.0606,
       "step": 2834
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.3339936161053235e-07,
-      "loss": 0.6828,
+      "epoch": 0.2967032967032967,
+      "grad_norm": 2.3794072494274827,
+      "learning_rate": 1.6496284594866115e-05,
+      "loss": 0.8615,
       "step": 2835
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.311144012712707e-07,
-      "loss": 0.6282,
+      "epoch": 0.2968079539508111,
+      "grad_norm": 2.470955073403848,
+      "learning_rate": 1.6493707180691166e-05,
+      "loss": 0.9446,
       "step": 2836
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.2884055004139038e-07,
-      "loss": 0.7145,
+      "epoch": 0.29691261119832546,
+      "grad_norm": 2.1185633078780004,
+      "learning_rate": 1.6491129020376094e-05,
+      "loss": 1.1212,
       "step": 2837
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.265778105066996e-07,
-      "loss": 0.6468,
+      "epoch": 0.29701726844583987,
+      "grad_norm": 2.40021678899163,
+      "learning_rate": 1.6488550114217138e-05,
+      "loss": 1.1674,
       "step": 2838
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.243261852403711e-07,
-      "loss": 0.6459,
+      "epoch": 0.29712192569335427,
+      "grad_norm": 2.8426932715804276,
+      "learning_rate": 1.6485970462510624e-05,
+      "loss": 0.9707,
       "step": 2839
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.220856768029367e-07,
-      "loss": 0.6344,
+      "epoch": 0.2972265829408687,
+      "grad_norm": 1.9555538468974023,
+      "learning_rate": 1.6483390065552952e-05,
+      "loss": 0.9935,
       "step": 2840
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.1985628774228717e-07,
-      "loss": 0.5841,
+      "epoch": 0.29733124018838303,
+      "grad_norm": 2.001436241338665,
+      "learning_rate": 1.648080892364062e-05,
+      "loss": 0.8675,
       "step": 2841
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.1763802059367012e-07,
-      "loss": 0.6143,
+      "epoch": 0.29743589743589743,
+      "grad_norm": 2.3809732845318456,
+      "learning_rate": 1.6478227037070205e-05,
+      "loss": 1.0029,
       "step": 2842
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.154308778796821e-07,
-      "loss": 0.5321,
+      "epoch": 0.29754055468341184,
+      "grad_norm": 2.052147061815293,
+      "learning_rate": 1.6475644406138372e-05,
+      "loss": 0.9261,
       "step": 2843
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.1323486211026979e-07,
-      "loss": 0.7149,
+      "epoch": 0.2976452119309262,
+      "grad_norm": 1.8579343591461566,
+      "learning_rate": 1.6473061031141868e-05,
+      "loss": 0.9575,
       "step": 2844
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.1104997578273e-07,
-      "loss": 0.683,
+      "epoch": 0.2977498691784406,
+      "grad_norm": 2.0536350820054743,
+      "learning_rate": 1.6470476912377526e-05,
+      "loss": 0.8525,
       "step": 2845
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.088762213816986e-07,
-      "loss": 0.7292,
+      "epoch": 0.297854526425955,
+      "grad_norm": 2.7957230117068432,
+      "learning_rate": 1.646789205014227e-05,
+      "loss": 1.0007,
       "step": 2846
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.0671360137915375e-07,
-      "loss": 0.7957,
+      "epoch": 0.2979591836734694,
+      "grad_norm": 1.8368027105734126,
+      "learning_rate": 1.646530644473311e-05,
+      "loss": 0.8189,
       "step": 2847
     },
     {
-      "epoch": 1.87,
-      "learning_rate": 2.0456211823441153e-07,
-      "loss": 0.6577,
+      "epoch": 0.29806384092098376,
+      "grad_norm": 1.9674169831122257,
+      "learning_rate": 1.6462720096447134e-05,
+      "loss": 0.8723,
       "step": 2848
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 2.0242177439412592e-07,
-      "loss": 0.684,
+      "epoch": 0.29816849816849816,
+      "grad_norm": 2.1875273033291136,
+      "learning_rate": 1.6460133005581512e-05,
+      "loss": 0.953,
       "step": 2849
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 2.0029257229227993e-07,
-      "loss": 0.7029,
+      "epoch": 0.29827315541601257,
+      "grad_norm": 1.9643878581413023,
+      "learning_rate": 1.6457545172433515e-05,
+      "loss": 0.8681,
       "step": 2850
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.9817451435018676e-07,
-      "loss": 0.7119,
+      "epoch": 0.298377812663527,
+      "grad_norm": 2.071057314759131,
+      "learning_rate": 1.6454956597300486e-05,
+      "loss": 1.0712,
       "step": 2851
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.9606760297648742e-07,
-      "loss": 0.6767,
+      "epoch": 0.2984824699110413,
+      "grad_norm": 3.0502621658394604,
+      "learning_rate": 1.645236728047986e-05,
+      "loss": 0.9732,
       "step": 2852
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.9397184056714868e-07,
-      "loss": 0.6691,
+      "epoch": 0.29858712715855573,
+      "grad_norm": 2.208086371273379,
+      "learning_rate": 1.6449777222269153e-05,
+      "loss": 1.1401,
       "step": 2853
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.9188722950545636e-07,
-      "loss": 0.6608,
+      "epoch": 0.29869178440607014,
+      "grad_norm": 2.374468906716569,
+      "learning_rate": 1.6447186422965962e-05,
+      "loss": 0.9075,
       "step": 2854
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.8981377216201636e-07,
-      "loss": 0.6504,
+      "epoch": 0.2987964416535845,
+      "grad_norm": 2.075865806365837,
+      "learning_rate": 1.644459488286799e-05,
+      "loss": 0.9824,
       "step": 2855
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.877514708947481e-07,
-      "loss": 0.6051,
+      "epoch": 0.2989010989010989,
+      "grad_norm": 2.1721807151694814,
+      "learning_rate": 1.6442002602272994e-05,
+      "loss": 1.0686,
       "step": 2856
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.8570032804888783e-07,
-      "loss": 0.6595,
+      "epoch": 0.2990057561486133,
+      "grad_norm": 2.1427069344984546,
+      "learning_rate": 1.6439409581478843e-05,
+      "loss": 1.0767,
       "step": 2857
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.836603459569808e-07,
-      "loss": 0.7658,
+      "epoch": 0.2991104133961277,
+      "grad_norm": 2.3195474794616504,
+      "learning_rate": 1.6436815820783477e-05,
+      "loss": 1.0688,
       "step": 2858
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.816315269388802e-07,
-      "loss": 0.7564,
+      "epoch": 0.29921507064364206,
+      "grad_norm": 2.078009990693235,
+      "learning_rate": 1.6434221320484928e-05,
+      "loss": 1.0676,
       "step": 2859
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.7961387330174606e-07,
-      "loss": 0.5965,
+      "epoch": 0.29931972789115646,
+      "grad_norm": 2.652882959489745,
+      "learning_rate": 1.64316260808813e-05,
+      "loss": 1.1277,
       "step": 2860
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.7760738734003858e-07,
-      "loss": 0.631,
+      "epoch": 0.29942438513867087,
+      "grad_norm": 1.9324618218876926,
+      "learning_rate": 1.6429030102270802e-05,
+      "loss": 0.9057,
       "step": 2861
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.756120713355214e-07,
-      "loss": 0.6346,
+      "epoch": 0.2995290423861852,
+      "grad_norm": 2.1542787245014763,
+      "learning_rate": 1.6426433384951707e-05,
+      "loss": 1.0036,
       "step": 2862
     },
     {
-      "epoch": 1.88,
-      "learning_rate": 1.7362792755725278e-07,
-      "loss": 0.6879,
+      "epoch": 0.2996336996336996,
+      "grad_norm": 2.316215732282425,
+      "learning_rate": 1.6423835929222393e-05,
+      "loss": 1.0189,
       "step": 2863
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.7165495826158896e-07,
-      "loss": 0.6362,
+      "epoch": 0.29973835688121403,
+      "grad_norm": 2.0832717770915785,
+      "learning_rate": 1.6421237735381305e-05,
+      "loss": 0.9312,
       "step": 2864
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.696931656921752e-07,
-      "loss": 0.5911,
+      "epoch": 0.29984301412872844,
+      "grad_norm": 2.0990095053120794,
+      "learning_rate": 1.6418638803726988e-05,
+      "loss": 1.0452,
       "step": 2865
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.6774255207994915e-07,
-      "loss": 0.7731,
+      "epoch": 0.2999476713762428,
+      "grad_norm": 2.3858579539152416,
+      "learning_rate": 1.6416039134558058e-05,
+      "loss": 0.8431,
       "step": 2866
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.6580311964313645e-07,
-      "loss": 0.6895,
+      "epoch": 0.3000523286237572,
+      "grad_norm": 2.0445888077165084,
+      "learning_rate": 1.641343872817322e-05,
+      "loss": 0.921,
       "step": 2867
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.6387487058724504e-07,
-      "loss": 0.7009,
+      "epoch": 0.3001569858712716,
+      "grad_norm": 2.3463310835524225,
+      "learning_rate": 1.6410837584871276e-05,
+      "loss": 1.0862,
       "step": 2868
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.6195780710506648e-07,
-      "loss": 0.7321,
+      "epoch": 0.300261643118786,
+      "grad_norm": 2.5259232781216316,
+      "learning_rate": 1.6408235704951098e-05,
+      "loss": 0.9665,
       "step": 2869
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.600519313766724e-07,
-      "loss": 0.6184,
+      "epoch": 0.30036630036630035,
+      "grad_norm": 2.4845918292536244,
+      "learning_rate": 1.6405633088711646e-05,
+      "loss": 1.0941,
       "step": 2870
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.5815724556941135e-07,
-      "loss": 0.6653,
+      "epoch": 0.30047095761381476,
+      "grad_norm": 2.250546185914107,
+      "learning_rate": 1.6403029736451964e-05,
+      "loss": 1.0009,
       "step": 2871
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.562737518379065e-07,
-      "loss": 0.6153,
+      "epoch": 0.30057561486132917,
+      "grad_norm": 2.2255990588605004,
+      "learning_rate": 1.6400425648471187e-05,
+      "loss": 0.9602,
       "step": 2872
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.5440145232405557e-07,
-      "loss": 0.7641,
+      "epoch": 0.3006802721088435,
+      "grad_norm": 2.17264095954443,
+      "learning_rate": 1.6397820825068528e-05,
+      "loss": 0.9746,
       "step": 2873
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.525403491570221e-07,
-      "loss": 0.662,
+      "epoch": 0.3007849293563579,
+      "grad_norm": 2.425953579485797,
+      "learning_rate": 1.6395215266543284e-05,
+      "loss": 0.8,
       "step": 2874
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.506904444532431e-07,
-      "loss": 0.7431,
+      "epoch": 0.3008895866038723,
+      "grad_norm": 2.4373639606890998,
+      "learning_rate": 1.6392608973194847e-05,
+      "loss": 0.9525,
       "step": 2875
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.488517403164147e-07,
-      "loss": 0.7146,
+      "epoch": 0.30099424385138673,
+      "grad_norm": 2.6105194549403894,
+      "learning_rate": 1.6390001945322674e-05,
+      "loss": 1.0145,
       "step": 2876
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.4702423883749983e-07,
-      "loss": 0.6645,
+      "epoch": 0.3010989010989011,
+      "grad_norm": 2.2916573263775644,
+      "learning_rate": 1.6387394183226327e-05,
+      "loss": 0.9484,
       "step": 2877
     },
     {
-      "epoch": 1.89,
-      "learning_rate": 1.4520794209472056e-07,
-      "loss": 0.672,
+      "epoch": 0.3012035583464155,
+      "grad_norm": 2.385274366893819,
+      "learning_rate": 1.6384785687205438e-05,
+      "loss": 1.1407,
       "step": 2878
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.4340285215355577e-07,
-      "loss": 0.6396,
+      "epoch": 0.3013082155939299,
+      "grad_norm": 2.3566579851084395,
+      "learning_rate": 1.6382176457559732e-05,
+      "loss": 1.0001,
       "step": 2879
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.4160897106674343e-07,
-      "loss": 0.6892,
+      "epoch": 0.30141287284144425,
+      "grad_norm": 2.395710605917217,
+      "learning_rate": 1.637956649458901e-05,
+      "loss": 1.0079,
       "step": 2880
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.3982630087427062e-07,
-      "loss": 0.651,
+      "epoch": 0.30151753008895865,
+      "grad_norm": 2.0298439688834753,
+      "learning_rate": 1.637695579859317e-05,
+      "loss": 1.0266,
       "step": 2881
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.3805484360337906e-07,
-      "loss": 0.6363,
+      "epoch": 0.30162218733647306,
+      "grad_norm": 2.3803391461053116,
+      "learning_rate": 1.637434436987218e-05,
+      "loss": 0.953,
       "step": 2882
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.362946012685573e-07,
-      "loss": 0.5743,
+      "epoch": 0.30172684458398746,
+      "grad_norm": 2.339978618501876,
+      "learning_rate": 1.6371732208726098e-05,
+      "loss": 1.0875,
       "step": 2883
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.3454557587154083e-07,
-      "loss": 0.6203,
+      "epoch": 0.3018315018315018,
+      "grad_norm": 2.0575949988075752,
+      "learning_rate": 1.6369119315455067e-05,
+      "loss": 1.1127,
       "step": 2884
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.3280776940130858e-07,
-      "loss": 0.7716,
+      "epoch": 0.3019361590790162,
+      "grad_norm": 2.553619190127974,
+      "learning_rate": 1.6366505690359316e-05,
+      "loss": 0.8558,
       "step": 2885
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.3108118383408418e-07,
-      "loss": 0.7167,
+      "epoch": 0.3020408163265306,
+      "grad_norm": 2.2250332246678606,
+      "learning_rate": 1.6363891333739153e-05,
+      "loss": 0.9865,
       "step": 2886
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.2936582113332707e-07,
-      "loss": 0.6412,
+      "epoch": 0.302145473574045,
+      "grad_norm": 2.263025646780511,
+      "learning_rate": 1.6361276245894982e-05,
+      "loss": 0.94,
       "step": 2887
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.276616832497346e-07,
-      "loss": 0.6639,
+      "epoch": 0.3022501308215594,
+      "grad_norm": 3.2202185282441733,
+      "learning_rate": 1.6358660427127267e-05,
+      "loss": 0.9522,
       "step": 2888
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.2596877212124327e-07,
-      "loss": 0.7092,
+      "epoch": 0.3023547880690738,
+      "grad_norm": 2.6014220655520686,
+      "learning_rate": 1.6356043877736582e-05,
+      "loss": 1.0032,
       "step": 2889
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.2428708967301973e-07,
-      "loss": 0.6886,
+      "epoch": 0.3024594453165882,
+      "grad_norm": 2.5626669505776145,
+      "learning_rate": 1.635342659802357e-05,
+      "loss": 1.0856,
       "step": 2890
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.226166378174598e-07,
-      "loss": 0.7492,
+      "epoch": 0.30256410256410254,
+      "grad_norm": 2.343833892601788,
+      "learning_rate": 1.6350808588288964e-05,
+      "loss": 1.0161,
       "step": 2891
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.2095741845419174e-07,
-      "loss": 0.5752,
+      "epoch": 0.30266875981161695,
+      "grad_norm": 2.586511158457927,
+      "learning_rate": 1.6348189848833574e-05,
+      "loss": 1.0082,
       "step": 2892
     },
     {
-      "epoch": 1.9,
-      "learning_rate": 1.1930943347006618e-07,
-      "loss": 0.6927,
+      "epoch": 0.30277341705913136,
+      "grad_norm": 1.7944614432349806,
+      "learning_rate": 1.6345570379958302e-05,
+      "loss": 0.8952,
       "step": 2893
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.1767268473916182e-07,
-      "loss": 0.7242,
+      "epoch": 0.30287807430664576,
+      "grad_norm": 2.2117394132793047,
+      "learning_rate": 1.634295018196413e-05,
+      "loss": 0.9989,
       "step": 2894
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.1604717412277866e-07,
-      "loss": 0.6736,
+      "epoch": 0.3029827315541601,
+      "grad_norm": 2.12160379753247,
+      "learning_rate": 1.6340329255152124e-05,
+      "loss": 1.0493,
       "step": 2895
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.144329034694347e-07,
-      "loss": 0.727,
+      "epoch": 0.3030873888016745,
+      "grad_norm": 2.5588546522643707,
+      "learning_rate": 1.633770759982343e-05,
+      "loss": 0.9269,
       "step": 2896
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.1282987461486927e-07,
-      "loss": 0.5983,
+      "epoch": 0.3031920460491889,
+      "grad_norm": 2.4173578432974008,
+      "learning_rate": 1.633508521627929e-05,
+      "loss": 0.915,
       "step": 2897
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.1123808938203306e-07,
-      "loss": 0.7715,
+      "epoch": 0.3032967032967033,
+      "grad_norm": 1.9462185216594179,
+      "learning_rate": 1.633246210482101e-05,
+      "loss": 1.0375,
       "step": 2898
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.0965754958109587e-07,
-      "loss": 0.6424,
+      "epoch": 0.3034013605442177,
+      "grad_norm": 3.1222882590726995,
+      "learning_rate": 1.6329838265750005e-05,
+      "loss": 1.0446,
       "step": 2899
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.0808825700943437e-07,
-      "loss": 0.7114,
+      "epoch": 0.3035060177917321,
+      "grad_norm": 2.2579929742271796,
+      "learning_rate": 1.6327213699367746e-05,
+      "loss": 1.0918,
       "step": 2900
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.0653021345163995e-07,
-      "loss": 0.668,
+      "epoch": 0.3036106750392465,
+      "grad_norm": 2.130384120094451,
+      "learning_rate": 1.6324588405975807e-05,
+      "loss": 0.8925,
       "step": 2901
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.049834206795064e-07,
-      "loss": 0.6584,
+      "epoch": 0.30371533228676084,
+      "grad_norm": 2.2025752462669486,
+      "learning_rate": 1.6321962385875837e-05,
+      "loss": 0.9963,
       "step": 2902
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.0344788045203668e-07,
-      "loss": 0.5823,
+      "epoch": 0.30381998953427525,
+      "grad_norm": 2.385277002826419,
+      "learning_rate": 1.6319335639369577e-05,
+      "loss": 1.0002,
       "step": 2903
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.0192359451543731e-07,
-      "loss": 0.6654,
+      "epoch": 0.30392464678178965,
+      "grad_norm": 1.9086149634590928,
+      "learning_rate": 1.631670816675884e-05,
+      "loss": 0.8627,
       "step": 2904
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 1.0041056460311393e-07,
-      "loss": 0.5543,
+      "epoch": 0.304029304029304,
+      "grad_norm": 2.172309116560836,
+      "learning_rate": 1.6314079968345527e-05,
+      "loss": 0.9886,
       "step": 2905
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 9.890879243567686e-08,
-      "loss": 0.6764,
+      "epoch": 0.3041339612768184,
+      "grad_norm": 2.286003474124238,
+      "learning_rate": 1.631145104443163e-05,
+      "loss": 0.8937,
       "step": 2906
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 9.741827972092777e-08,
-      "loss": 0.6297,
+      "epoch": 0.3042386185243328,
+      "grad_norm": 2.1054654506329804,
+      "learning_rate": 1.630882139531921e-05,
+      "loss": 0.9913,
       "step": 2907
     },
     {
-      "epoch": 1.91,
-      "learning_rate": 9.593902815386968e-08,
-      "loss": 0.6081,
+      "epoch": 0.3043432757718472,
+      "grad_norm": 2.188221004194734,
+      "learning_rate": 1.6306191021310423e-05,
+      "loss": 1.0191,
       "step": 2908
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 9.447103941669588e-08,
-      "loss": 0.6845,
+      "epoch": 0.30444793301936157,
+      "grad_norm": 2.1904821257327374,
+      "learning_rate": 1.6303559922707503e-05,
+      "loss": 0.8218,
       "step": 2909
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 9.301431517879323e-08,
-      "loss": 0.6775,
+      "epoch": 0.304552590266876,
+      "grad_norm": 2.5842095448369844,
+      "learning_rate": 1.6300928099812772e-05,
+      "loss": 0.9329,
       "step": 2910
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 9.156885709674101e-08,
-      "loss": 0.6371,
+      "epoch": 0.3046572475143904,
+      "grad_norm": 2.132246343055798,
+      "learning_rate": 1.6298295552928626e-05,
+      "loss": 0.8609,
       "step": 2911
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 9.013466681429994e-08,
-      "loss": 0.6572,
+      "epoch": 0.3047619047619048,
+      "grad_norm": 2.2252491798180514,
+      "learning_rate": 1.6295662282357555e-05,
+      "loss": 1.118,
       "step": 2912
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.871174596242427e-08,
-      "loss": 0.6672,
+      "epoch": 0.30486656200941914,
+      "grad_norm": 2.098601354897258,
+      "learning_rate": 1.6293028288402123e-05,
+      "loss": 0.9224,
       "step": 2913
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.73000961592485e-08,
-      "loss": 0.6575,
+      "epoch": 0.30497121925693355,
+      "grad_norm": 2.012587732978616,
+      "learning_rate": 1.6290393571364987e-05,
+      "loss": 1.0842,
       "step": 2914
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.589971901009186e-08,
-      "loss": 0.6667,
+      "epoch": 0.30507587650444795,
+      "grad_norm": 2.3614383102118555,
+      "learning_rate": 1.628775813154887e-05,
+      "loss": 1.0265,
       "step": 2915
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.451061610745381e-08,
-      "loss": 0.6693,
+      "epoch": 0.3051805337519623,
+      "grad_norm": 2.7878156230807134,
+      "learning_rate": 1.6285121969256605e-05,
+      "loss": 1.1703,
       "step": 2916
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.313278903101075e-08,
-      "loss": 0.6127,
+      "epoch": 0.3052851909994767,
+      "grad_norm": 2.1636226722521856,
+      "learning_rate": 1.6282485084791086e-05,
+      "loss": 1.072,
       "step": 2917
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.17662393476204e-08,
-      "loss": 0.6235,
+      "epoch": 0.3053898482469911,
+      "grad_norm": 1.914547100044751,
+      "learning_rate": 1.627984747845529e-05,
+      "loss": 0.8617,
       "step": 2918
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 8.041096861131303e-08,
-      "loss": 0.5349,
+      "epoch": 0.3054945054945055,
+      "grad_norm": 2.4366729944284753,
+      "learning_rate": 1.6277209150552285e-05,
+      "loss": 1.042,
       "step": 2919
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 7.906697836329247e-08,
-      "loss": 0.8139,
+      "epoch": 0.30559916274201987,
+      "grad_norm": 1.8211650521764042,
+      "learning_rate": 1.627457010138523e-05,
+      "loss": 1.0024,
       "step": 2920
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 7.773427013193502e-08,
-      "loss": 0.7365,
+      "epoch": 0.3057038199895343,
+      "grad_norm": 2.4531522351641417,
+      "learning_rate": 1.6271930331257345e-05,
+      "loss": 0.9587,
       "step": 2921
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 7.641284543278838e-08,
-      "loss": 0.7617,
+      "epoch": 0.3058084772370487,
+      "grad_norm": 2.5130079196870208,
+      "learning_rate": 1.6269289840471952e-05,
+      "loss": 1.0224,
       "step": 2922
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 7.510270576856605e-08,
-      "loss": 0.6038,
+      "epoch": 0.30591313448456303,
+      "grad_norm": 1.9747172743888348,
+      "learning_rate": 1.6266648629332448e-05,
+      "loss": 0.9663,
       "step": 2923
     },
     {
-      "epoch": 1.92,
-      "learning_rate": 7.38038526291518e-08,
-      "loss": 0.7505,
+      "epoch": 0.30601779173207744,
+      "grad_norm": 2.3944082981680705,
+      "learning_rate": 1.6264006698142318e-05,
+      "loss": 0.9934,
       "step": 2924
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 7.25162874915919e-08,
-      "loss": 0.7365,
+      "epoch": 0.30612244897959184,
+      "grad_norm": 2.508078569407357,
+      "learning_rate": 1.6261364047205114e-05,
+      "loss": 1.017,
       "step": 2925
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 7.12400118200951e-08,
-      "loss": 0.6976,
+      "epoch": 0.30622710622710625,
+      "grad_norm": 2.33208975658998,
+      "learning_rate": 1.6258720676824487e-05,
+      "loss": 0.8922,
       "step": 2926
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 6.997502706603487e-08,
-      "loss": 0.6077,
+      "epoch": 0.3063317634746206,
+      "grad_norm": 1.9714230710953324,
+      "learning_rate": 1.625607658730417e-05,
+      "loss": 0.9413,
       "step": 2927
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 6.872133466794163e-08,
-      "loss": 0.6821,
+      "epoch": 0.306436420722135,
+      "grad_norm": 2.006875902861234,
+      "learning_rate": 1.625343177894797e-05,
+      "loss": 0.9562,
       "step": 2928
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 6.747893605150824e-08,
-      "loss": 0.6679,
+      "epoch": 0.3065410779696494,
+      "grad_norm": 1.9744248707683965,
+      "learning_rate": 1.625078625205978e-05,
+      "loss": 1.0386,
       "step": 2929
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 6.624783262958012e-08,
-      "loss": 0.6501,
+      "epoch": 0.30664573521716376,
+      "grad_norm": 2.3082227558275377,
+      "learning_rate": 1.6248140006943577e-05,
+      "loss": 1.1613,
       "step": 2930
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 6.50280258021596e-08,
-      "loss": 0.7814,
+      "epoch": 0.30675039246467817,
+      "grad_norm": 2.1702890355942337,
+      "learning_rate": 1.624549304390342e-05,
+      "loss": 0.9356,
       "step": 2931
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 6.381951695640154e-08,
-      "loss": 0.6804,
+      "epoch": 0.3068550497121926,
+      "grad_norm": 2.424739431310441,
+      "learning_rate": 1.6242845363243455e-05,
+      "loss": 1.1796,
       "step": 2932
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 6.262230746661657e-08,
-      "loss": 0.5597,
+      "epoch": 0.306959706959707,
+      "grad_norm": 2.171158515959578,
+      "learning_rate": 1.62401969652679e-05,
+      "loss": 1.109,
       "step": 2933
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 6.143639869426232e-08,
-      "loss": 0.6584,
+      "epoch": 0.30706436420722133,
+      "grad_norm": 2.186889110464127,
+      "learning_rate": 1.6237547850281065e-05,
+      "loss": 0.9602,
       "step": 2934
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 6.026179198794335e-08,
-      "loss": 0.838,
+      "epoch": 0.30716902145473574,
+      "grad_norm": 2.404453450737933,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 1.0674,
       "step": 2935
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 5.9098488683417834e-08,
-      "loss": 0.5844,
+      "epoch": 0.30727367870225014,
+      "grad_norm": 2.2426489372917224,
+      "learning_rate": 1.6232247470491188e-05,
+      "loss": 1.0203,
       "step": 2936
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 5.794649010358422e-08,
-      "loss": 0.836,
+      "epoch": 0.30737833594976455,
+      "grad_norm": 2.1305208739832135,
+      "learning_rate": 1.622959620629717e-05,
+      "loss": 0.8933,
       "step": 2937
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 5.680579755848681e-08,
-      "loss": 0.7431,
+      "epoch": 0.3074829931972789,
+      "grad_norm": 2.475213166949153,
+      "learning_rate": 1.6226944226309916e-05,
+      "loss": 0.9715,
       "step": 2938
     },
     {
-      "epoch": 1.93,
-      "learning_rate": 5.5676412345316844e-08,
-      "loss": 0.647,
+      "epoch": 0.3075876504447933,
+      "grad_norm": 2.229771800563122,
+      "learning_rate": 1.6224291530834147e-05,
+      "loss": 1.0384,
       "step": 2939
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 5.455833574839919e-08,
-      "loss": 0.5488,
+      "epoch": 0.3076923076923077,
+      "grad_norm": 1.9303541858289048,
+      "learning_rate": 1.6221638120174668e-05,
+      "loss": 1.0791,
       "step": 2940
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 5.3451569039205674e-08,
-      "loss": 0.6635,
+      "epoch": 0.30779696493982206,
+      "grad_norm": 2.3662003014744437,
+      "learning_rate": 1.6218983994636355e-05,
+      "loss": 1.0525,
       "step": 2941
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 5.235611347634173e-08,
-      "loss": 0.6223,
+      "epoch": 0.30790162218733647,
+      "grad_norm": 2.1079944888721127,
+      "learning_rate": 1.621632915452417e-05,
+      "loss": 0.9803,
       "step": 2942
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 5.12719703055542e-08,
-      "loss": 0.6356,
+      "epoch": 0.30800627943485087,
+      "grad_norm": 2.092870856439816,
+      "learning_rate": 1.6213673600143168e-05,
+      "loss": 0.9583,
       "step": 2943
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 5.0199140759721324e-08,
-      "loss": 0.7753,
+      "epoch": 0.3081109366823653,
+      "grad_norm": 2.243256700536703,
+      "learning_rate": 1.6211017331798473e-05,
+      "loss": 0.9972,
       "step": 2944
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.913762605885719e-08,
-      "loss": 0.6892,
+      "epoch": 0.3082155939298796,
+      "grad_norm": 2.205695168816815,
+      "learning_rate": 1.6208360349795293e-05,
+      "loss": 1.0897,
       "step": 2945
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.8087427410110635e-08,
-      "loss": 0.6796,
+      "epoch": 0.30832025117739403,
+      "grad_norm": 2.4776470468527103,
+      "learning_rate": 1.620570265443892e-05,
+      "loss": 1.0388,
       "step": 2946
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.7048546007759656e-08,
-      "loss": 0.7213,
+      "epoch": 0.30842490842490844,
+      "grad_norm": 2.344647528820799,
+      "learning_rate": 1.620304424603474e-05,
+      "loss": 1.0009,
       "step": 2947
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.602098303321256e-08,
-      "loss": 0.6885,
+      "epoch": 0.3085295656724228,
+      "grad_norm": 2.108670142100681,
+      "learning_rate": 1.6200385124888195e-05,
+      "loss": 0.9776,
       "step": 2948
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.5004739655006844e-08,
-      "loss": 0.6352,
+      "epoch": 0.3086342229199372,
+      "grad_norm": 2.3408563934308897,
+      "learning_rate": 1.6197725291304833e-05,
+      "loss": 0.9592,
       "step": 2949
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.399981702880918e-08,
-      "loss": 0.6829,
+      "epoch": 0.3087388801674516,
+      "grad_norm": 2.18058930736239,
+      "learning_rate": 1.6195064745590267e-05,
+      "loss": 1.049,
       "step": 2950
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.3006216297407645e-08,
-      "loss": 0.666,
+      "epoch": 0.308843537414966,
+      "grad_norm": 2.1381863651278854,
+      "learning_rate": 1.619240348805021e-05,
+      "loss": 0.962,
       "step": 2951
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.2023938590720625e-08,
-      "loss": 0.7365,
+      "epoch": 0.30894819466248036,
+      "grad_norm": 1.8703885232830255,
+      "learning_rate": 1.6189741518990433e-05,
+      "loss": 1.0408,
       "step": 2952
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.1052985025787914e-08,
-      "loss": 0.6313,
+      "epoch": 0.30905285190999476,
+      "grad_norm": 2.170717558554135,
+      "learning_rate": 1.618707883871681e-05,
+      "loss": 1.136,
       "step": 2953
     },
     {
-      "epoch": 1.94,
-      "learning_rate": 4.009335670676851e-08,
-      "loss": 0.7312,
+      "epoch": 0.30915750915750917,
+      "grad_norm": 2.365760154975676,
+      "learning_rate": 1.6184415447535284e-05,
+      "loss": 0.9957,
       "step": 2954
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.914505472494723e-08,
-      "loss": 0.7549,
+      "epoch": 0.3092621664050236,
+      "grad_norm": 2.2575554436817606,
+      "learning_rate": 1.6181751345751887e-05,
+      "loss": 1.0823,
       "step": 2955
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.820808015872701e-08,
-      "loss": 0.6622,
+      "epoch": 0.3093668236525379,
+      "grad_norm": 2.00512183891559,
+      "learning_rate": 1.6179086533672726e-05,
+      "loss": 1.064,
       "step": 2956
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.728243407362664e-08,
-      "loss": 0.6694,
+      "epoch": 0.30947148090005233,
+      "grad_norm": 2.0900482163912932,
+      "learning_rate": 1.6176421011604e-05,
+      "loss": 1.0277,
       "step": 2957
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.636811752228519e-08,
-      "loss": 0.6484,
+      "epoch": 0.30957613814756674,
+      "grad_norm": 2.028018177800534,
+      "learning_rate": 1.6173754779851978e-05,
+      "loss": 0.9109,
       "step": 2958
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.546513154445874e-08,
-      "loss": 0.7049,
+      "epoch": 0.3096807953950811,
+      "grad_norm": 1.911257078717531,
+      "learning_rate": 1.6171087838723015e-05,
+      "loss": 1.0144,
       "step": 2959
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.457347716701587e-08,
-      "loss": 0.6954,
+      "epoch": 0.3097854526425955,
+      "grad_norm": 2.2217963538032284,
+      "learning_rate": 1.616842018852355e-05,
+      "loss": 0.9425,
       "step": 2960
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.3693155403938805e-08,
-      "loss": 0.7506,
+      "epoch": 0.3098901098901099,
+      "grad_norm": 2.59813827847896,
+      "learning_rate": 1.61657518295601e-05,
+      "loss": 1.0,
       "step": 2961
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.282416725632454e-08,
-      "loss": 0.7205,
+      "epoch": 0.3099947671376243,
+      "grad_norm": 2.16386391537696,
+      "learning_rate": 1.6163082762139265e-05,
+      "loss": 0.836,
       "step": 2962
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.1966513712380356e-08,
-      "loss": 0.6839,
+      "epoch": 0.31009942438513866,
+      "grad_norm": 2.8465111690795064,
+      "learning_rate": 1.616041298656773e-05,
+      "loss": 1.008,
       "step": 2963
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.1120195747422755e-08,
-      "loss": 0.7295,
+      "epoch": 0.31020408163265306,
+      "grad_norm": 2.58047061290731,
+      "learning_rate": 1.6157742503152253e-05,
+      "loss": 0.9143,
       "step": 2964
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 3.0285214323879655e-08,
-      "loss": 0.6969,
+      "epoch": 0.31030873888016747,
+      "grad_norm": 1.9097639693984674,
+      "learning_rate": 1.6155071312199676e-05,
+      "loss": 0.8861,
       "step": 2965
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 2.9461570391287053e-08,
-      "loss": 0.5346,
+      "epoch": 0.3104133961276818,
+      "grad_norm": 2.124026315060564,
+      "learning_rate": 1.615239941401693e-05,
+      "loss": 0.965,
       "step": 2966
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 2.864926488628572e-08,
-      "loss": 0.6258,
+      "epoch": 0.3105180533751962,
+      "grad_norm": 2.39401063448098,
+      "learning_rate": 1.614972680891102e-05,
+      "loss": 1.0503,
       "step": 2967
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 2.78482987326234e-08,
-      "loss": 0.6871,
+      "epoch": 0.31062271062271063,
+      "grad_norm": 1.8182089689462924,
+      "learning_rate": 1.6147053497189032e-05,
+      "loss": 0.8512,
       "step": 2968
     },
     {
-      "epoch": 1.95,
-      "learning_rate": 2.7058672841154822e-08,
-      "loss": 0.7879,
+      "epoch": 0.31072736787022504,
+      "grad_norm": 2.0677360921080115,
+      "learning_rate": 1.6144379479158136e-05,
+      "loss": 0.9903,
       "step": 2969
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.628038810983613e-08,
-      "loss": 0.6921,
+      "epoch": 0.3108320251177394,
+      "grad_norm": 2.428228198980805,
+      "learning_rate": 1.614170475512558e-05,
+      "loss": 0.8599,
       "step": 2970
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.5513445423728246e-08,
-      "loss": 0.6992,
+      "epoch": 0.3109366823652538,
+      "grad_norm": 2.0105272211059027,
+      "learning_rate": 1.61390293253987e-05,
+      "loss": 0.9592,
       "step": 2971
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.4757845654992396e-08,
-      "loss": 0.6271,
+      "epoch": 0.3110413396127682,
+      "grad_norm": 2.323659569472944,
+      "learning_rate": 1.6136353190284903e-05,
+      "loss": 0.9728,
       "step": 2972
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.4013589662892356e-08,
-      "loss": 0.5533,
+      "epoch": 0.31114599686028255,
+      "grad_norm": 1.9635717769932388,
+      "learning_rate": 1.6133676350091686e-05,
+      "loss": 0.9582,
       "step": 2973
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.3280678293788883e-08,
-      "loss": 0.6439,
+      "epoch": 0.31125065410779695,
+      "grad_norm": 1.9068151782035103,
+      "learning_rate": 1.6130998805126622e-05,
+      "loss": 0.8325,
       "step": 2974
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.2559112381146386e-08,
-      "loss": 0.6682,
+      "epoch": 0.31135531135531136,
+      "grad_norm": 2.0598995448173314,
+      "learning_rate": 1.6128320555697365e-05,
+      "loss": 1.0608,
       "step": 2975
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.1848892745522932e-08,
-      "loss": 0.6897,
+      "epoch": 0.31145996860282577,
+      "grad_norm": 2.2373806491125463,
+      "learning_rate": 1.6125641602111655e-05,
+      "loss": 0.9931,
       "step": 2976
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.1150020194574682e-08,
-      "loss": 0.555,
+      "epoch": 0.3115646258503401,
+      "grad_norm": 2.0920136465295918,
+      "learning_rate": 1.6122961944677308e-05,
+      "loss": 0.9315,
       "step": 2977
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 2.0462495523057013e-08,
-      "loss": 0.6069,
+      "epoch": 0.3116692830978545,
+      "grad_norm": 2.082366198297534,
+      "learning_rate": 1.612028158370222e-05,
+      "loss": 0.9886,
       "step": 2978
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 1.978631951281673e-08,
-      "loss": 0.6374,
+      "epoch": 0.3117739403453689,
+      "grad_norm": 2.2060623386259635,
+      "learning_rate": 1.6117600519494373e-05,
+      "loss": 0.9608,
       "step": 2979
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 1.9121492932796526e-08,
-      "loss": 0.6606,
+      "epoch": 0.31187859759288333,
+      "grad_norm": 2.552233556749464,
+      "learning_rate": 1.611491875236182e-05,
+      "loss": 1.018,
       "step": 2980
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 1.846801653903274e-08,
-      "loss": 0.7836,
+      "epoch": 0.3119832548403977,
+      "grad_norm": 2.2999204774617157,
+      "learning_rate": 1.6112236282612705e-05,
+      "loss": 0.9881,
       "step": 2981
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 1.7825891074654267e-08,
-      "loss": 0.5954,
+      "epoch": 0.3120879120879121,
+      "grad_norm": 2.084664368943567,
+      "learning_rate": 1.6109553110555256e-05,
+      "loss": 0.9935,
       "step": 2982
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 1.7195117269881433e-08,
-      "loss": 0.7067,
+      "epoch": 0.3121925693354265,
+      "grad_norm": 2.077448020127175,
+      "learning_rate": 1.6106869236497767e-05,
+      "loss": 0.9272,
       "step": 2983
     },
     {
-      "epoch": 1.96,
-      "learning_rate": 1.6575695842027116e-08,
-      "loss": 0.6521,
+      "epoch": 0.31229722658294085,
+      "grad_norm": 1.9328287512609774,
+      "learning_rate": 1.610418466074862e-05,
+      "loss": 1.037,
       "step": 2984
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.59676274954923e-08,
-      "loss": 0.6676,
+      "epoch": 0.31240188383045525,
+      "grad_norm": 2.3308588787662674,
+      "learning_rate": 1.6101499383616284e-05,
+      "loss": 1.0365,
       "step": 2985
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.537091292176829e-08,
-      "loss": 0.6783,
+      "epoch": 0.31250654107796966,
+      "grad_norm": 2.1473404366037023,
+      "learning_rate": 1.60988134054093e-05,
+      "loss": 0.9637,
       "step": 2986
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.4785552799433388e-08,
-      "loss": 0.6663,
+      "epoch": 0.31261119832548406,
+      "grad_norm": 2.0439604173751627,
+      "learning_rate": 1.609612672643629e-05,
+      "loss": 0.8799,
       "step": 2987
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.4211547794157344e-08,
-      "loss": 0.6096,
+      "epoch": 0.3127158555729984,
+      "grad_norm": 2.285023732772858,
+      "learning_rate": 1.609343934700596e-05,
+      "loss": 1.0245,
       "step": 2988
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.3648898558693558e-08,
-      "loss": 0.6837,
+      "epoch": 0.3128205128205128,
+      "grad_norm": 2.47915241747156,
+      "learning_rate": 1.60907512674271e-05,
+      "loss": 1.05,
       "step": 2989
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.3097605732882435e-08,
-      "loss": 0.5759,
+      "epoch": 0.3129251700680272,
+      "grad_norm": 2.0809584106263057,
+      "learning_rate": 1.6088062488008575e-05,
+      "loss": 1.0695,
       "step": 2990
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.2557669943650264e-08,
-      "loss": 0.6818,
+      "epoch": 0.3130298273155416,
+      "grad_norm": 2.1591951982467927,
+      "learning_rate": 1.6085373009059322e-05,
+      "loss": 0.9742,
       "step": 2991
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.2029091805010328e-08,
-      "loss": 0.7021,
+      "epoch": 0.313134484563056,
+      "grad_norm": 1.8386586833176444,
+      "learning_rate": 1.6082682830888376e-05,
+      "loss": 0.8092,
       "step": 2992
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.151187191805625e-08,
-      "loss": 0.6455,
+      "epoch": 0.3132391418105704,
+      "grad_norm": 2.598989002728532,
+      "learning_rate": 1.607999195380484e-05,
+      "loss": 0.9899,
       "step": 2993
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.1006010870966422e-08,
-      "loss": 0.587,
+      "epoch": 0.3133437990580848,
+      "grad_norm": 2.7275001106943297,
+      "learning_rate": 1.6077300378117906e-05,
+      "loss": 0.9487,
       "step": 2994
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.0511509239004013e-08,
-      "loss": 0.7055,
+      "epoch": 0.31344845630559914,
+      "grad_norm": 2.2062421090046342,
+      "learning_rate": 1.6074608104136838e-05,
+      "loss": 0.9532,
       "step": 2995
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 1.0028367584512532e-08,
-      "loss": 0.7165,
+      "epoch": 0.31355311355311355,
+      "grad_norm": 2.0978845722449995,
+      "learning_rate": 1.6071915132170986e-05,
+      "loss": 0.8883,
       "step": 2996
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 9.556586456916928e-09,
-      "loss": 0.7311,
+      "epoch": 0.31365777080062796,
+      "grad_norm": 2.1704288073650653,
+      "learning_rate": 1.606922146252977e-05,
+      "loss": 1.0379,
       "step": 2997
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 9.096166392723593e-09,
-      "loss": 0.572,
+      "epoch": 0.31376242804814236,
+      "grad_norm": 2.0169657476089755,
+      "learning_rate": 1.6066527095522707e-05,
+      "loss": 1.0009,
       "step": 2998
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 8.647107915519259e-09,
-      "loss": 0.6315,
+      "epoch": 0.3138670852956567,
+      "grad_norm": 2.256441777616629,
+      "learning_rate": 1.6063832031459384e-05,
+      "loss": 0.972,
       "step": 2999
     },
     {
-      "epoch": 1.97,
-      "learning_rate": 8.209411535969881e-09,
-      "loss": 0.5715,
+      "epoch": 0.3139717425431711,
+      "grad_norm": 1.8938052813988708,
+      "learning_rate": 1.6061136270649473e-05,
+      "loss": 0.9554,
       "step": 3000
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 7.783077751820633e-09,
-      "loss": 0.5621,
+      "epoch": 0.3140763997906855,
+      "grad_norm": 2.5447202789705554,
+      "learning_rate": 1.6058439813402712e-05,
+      "loss": 1.0768,
       "step": 3001
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 7.3681070478948125e-09,
-      "loss": 0.6666,
+      "epoch": 0.3141810570381999,
+      "grad_norm": 2.0871529984817965,
+      "learning_rate": 1.605574266002893e-05,
+      "loss": 0.9473,
       "step": 3002
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 6.964499896096044e-09,
-      "loss": 0.672,
+      "epoch": 0.3142857142857143,
+      "grad_norm": 2.1668427542881097,
+      "learning_rate": 1.6053044810838048e-05,
+      "loss": 0.9402,
       "step": 3003
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 6.572256755401629e-09,
-      "loss": 0.7563,
+      "epoch": 0.3143903715332287,
+      "grad_norm": 2.3159990985111936,
+      "learning_rate": 1.6050346266140046e-05,
+      "loss": 0.9726,
       "step": 3004
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 6.191378071869203e-09,
-      "loss": 0.7738,
+      "epoch": 0.3144950287807431,
+      "grad_norm": 1.9154842831401457,
+      "learning_rate": 1.6047647026244988e-05,
+      "loss": 0.8467,
       "step": 3005
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 5.821864278631184e-09,
-      "loss": 0.7616,
+      "epoch": 0.31459968602825744,
+      "grad_norm": 2.25244360448727,
+      "learning_rate": 1.604494709146303e-05,
+      "loss": 1.1072,
       "step": 3006
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 5.463715795895885e-09,
-      "loss": 0.74,
+      "epoch": 0.31470434327577185,
+      "grad_norm": 2.207816514269716,
+      "learning_rate": 1.6042246462104394e-05,
+      "loss": 1.011,
       "step": 3007
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 5.116933030946403e-09,
-      "loss": 0.6748,
+      "epoch": 0.31480900052328625,
+      "grad_norm": 2.5547164500222044,
+      "learning_rate": 1.6039545138479395e-05,
+      "loss": 0.968,
       "step": 3008
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 4.781516378142837e-09,
-      "loss": 0.7536,
+      "epoch": 0.3149136577708006,
+      "grad_norm": 2.677040611649757,
+      "learning_rate": 1.603684312089841e-05,
+      "loss": 1.1204,
       "step": 3009
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 4.45746621891785e-09,
-      "loss": 0.6959,
+      "epoch": 0.315018315018315,
+      "grad_norm": 1.9755495487026258,
+      "learning_rate": 1.6034140409671916e-05,
+      "loss": 0.97,
       "step": 3010
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 4.144782921779999e-09,
-      "loss": 0.6405,
+      "epoch": 0.3151229722658294,
+      "grad_norm": 2.1587533164996566,
+      "learning_rate": 1.6031437005110456e-05,
+      "loss": 1.0208,
       "step": 3011
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 3.843466842310406e-09,
-      "loss": 0.6195,
+      "epoch": 0.3152276295133438,
+      "grad_norm": 1.9806892809879053,
+      "learning_rate": 1.602873290752466e-05,
+      "loss": 1.1013,
       "step": 3012
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 3.5535183231627523e-09,
-      "loss": 0.5632,
+      "epoch": 0.31533228676085817,
+      "grad_norm": 2.4483979214827745,
+      "learning_rate": 1.602602811722523e-05,
+      "loss": 1.1004,
       "step": 3013
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 3.274937694065505e-09,
-      "loss": 0.7038,
+      "epoch": 0.3154369440083726,
+      "grad_norm": 2.5101769689868436,
+      "learning_rate": 1.602332263452295e-05,
+      "loss": 0.9379,
       "step": 3014
     },
     {
-      "epoch": 1.98,
-      "learning_rate": 3.0077252718185844e-09,
-      "loss": 0.5748,
+      "epoch": 0.315541601255887,
+      "grad_norm": 2.538507265258334,
+      "learning_rate": 1.602061645972869e-05,
+      "loss": 1.0772,
       "step": 3015
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 2.751881360292252e-09,
-      "loss": 0.6874,
+      "epoch": 0.31564625850340133,
+      "grad_norm": 2.1327347713309415,
+      "learning_rate": 1.6017909593153394e-05,
+      "loss": 0.9862,
       "step": 3016
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 2.507406250433775e-09,
-      "loss": 0.584,
+      "epoch": 0.31575091575091574,
+      "grad_norm": 1.8079317995730482,
+      "learning_rate": 1.601520203510809e-05,
+      "loss": 0.9613,
       "step": 3017
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 2.27430022025521e-09,
-      "loss": 0.6519,
+      "epoch": 0.31585557299843015,
+      "grad_norm": 2.3477974076051424,
+      "learning_rate": 1.6012493785903874e-05,
+      "loss": 1.0938,
       "step": 3018
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 2.0525635348456195e-09,
-      "loss": 0.6832,
+      "epoch": 0.31596023024594455,
+      "grad_norm": 2.3995917756166953,
+      "learning_rate": 1.6009784845851936e-05,
+      "loss": 0.9591,
       "step": 3019
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 1.8421964463610776e-09,
-      "loss": 0.7397,
+      "epoch": 0.3160648874934589,
+      "grad_norm": 2.171025021213507,
+      "learning_rate": 1.6007075215263533e-05,
+      "loss": 1.0401,
       "step": 3020
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 1.6431991940302206e-09,
-      "loss": 0.728,
+      "epoch": 0.3161695447409733,
+      "grad_norm": 2.453631613652852,
+      "learning_rate": 1.6004364894450015e-05,
+      "loss": 1.0872,
       "step": 3021
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 1.455572004150918e-09,
-      "loss": 0.7637,
+      "epoch": 0.3162742019884877,
+      "grad_norm": 2.0649905270363003,
+      "learning_rate": 1.6001653883722794e-05,
+      "loss": 0.947,
       "step": 3022
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 1.279315090091382e-09,
-      "loss": 0.6522,
+      "epoch": 0.3163788592360021,
+      "grad_norm": 1.9845232246558884,
+      "learning_rate": 1.5998942183393372e-05,
+      "loss": 1.0143,
       "step": 3023
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 1.114428652290167e-09,
-      "loss": 0.7245,
+      "epoch": 0.31648351648351647,
+      "grad_norm": 2.027999737482337,
+      "learning_rate": 1.599622979377334e-05,
+      "loss": 1.058,
       "step": 3024
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 9.609128782550604e-10,
-      "loss": 0.7767,
+      "epoch": 0.3165881737310309,
+      "grad_norm": 2.386479104325292,
+      "learning_rate": 1.5993516715174345e-05,
+      "loss": 1.0363,
       "step": 3025
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 8.187679425630813e-10,
-      "loss": 0.6937,
+      "epoch": 0.3166928309785453,
+      "grad_norm": 2.026243723400283,
+      "learning_rate": 1.5990802947908132e-05,
+      "loss": 0.8436,
       "step": 3026
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 6.879940068604818e-10,
-      "loss": 0.6176,
+      "epoch": 0.31679748822605963,
+      "grad_norm": 2.4461769884069486,
+      "learning_rate": 1.598808849228651e-05,
+      "loss": 1.0391,
       "step": 3027
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 5.685912198627464e-10,
-      "loss": 0.7871,
+      "epoch": 0.31690214547357404,
+      "grad_norm": 2.735072516013726,
+      "learning_rate": 1.5985373348621384e-05,
+      "loss": 1.1361,
       "step": 3028
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 4.605597173534815e-10,
-      "loss": 0.6383,
+      "epoch": 0.31700680272108844,
+      "grad_norm": 2.0367817571925677,
+      "learning_rate": 1.598265751722473e-05,
+      "loss": 1.1262,
       "step": 3029
     },
     {
-      "epoch": 1.99,
-      "learning_rate": 3.63899622184416e-10,
-      "loss": 0.6543,
+      "epoch": 0.31711145996860285,
+      "grad_norm": 2.389592094981779,
+      "learning_rate": 1.5979940998408593e-05,
+      "loss": 1.0193,
       "step": 3030
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 2.786110442787315e-10,
-      "loss": 0.6971,
+      "epoch": 0.3172161172161172,
+      "grad_norm": 2.051597409307426,
+      "learning_rate": 1.597722379248512e-05,
+      "loss": 0.8997,
       "step": 3031
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 2.0469408062440131e-10,
-      "loss": 0.7434,
+      "epoch": 0.3173207744636316,
+      "grad_norm": 1.8589251291767148,
+      "learning_rate": 1.597450589976651e-05,
+      "loss": 0.957,
       "step": 3032
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 1.4214881527974123e-10,
-      "loss": 0.7192,
+      "epoch": 0.317425431711146,
+      "grad_norm": 2.2975676929699937,
+      "learning_rate": 1.5971787320565063e-05,
+      "loss": 0.9536,
       "step": 3033
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 9.097531937118931e-11,
-      "loss": 0.6138,
+      "epoch": 0.31753008895866036,
+      "grad_norm": 2.2772719111714297,
+      "learning_rate": 1.5969068055193146e-05,
+      "loss": 1.008,
       "step": 3034
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 5.1173651093305854e-11,
-      "loss": 0.6704,
+      "epoch": 0.31763474620617477,
+      "grad_norm": 2.348844355294729,
+      "learning_rate": 1.596634810396321e-05,
+      "loss": 0.9407,
       "step": 3035
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 2.274385570655291e-11,
-      "loss": 0.7346,
+      "epoch": 0.3177394034536892,
+      "grad_norm": 2.1144033212390494,
+      "learning_rate": 1.5963627467187783e-05,
+      "loss": 0.913,
       "step": 3036
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 5.685965542845395e-12,
-      "loss": 0.6479,
+      "epoch": 0.3178440607012036,
+      "grad_norm": 2.078038440024001,
+      "learning_rate": 1.5960906145179466e-05,
+      "loss": 0.9828,
       "step": 3037
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 0.0,
-      "loss": 0.5033,
+      "epoch": 0.31794871794871793,
+      "grad_norm": 2.1282457724139667,
+      "learning_rate": 1.595818413825095e-05,
+      "loss": 0.9836,
       "step": 3038
     },
     {
-      "epoch": 2.0,
-      "step": 3038,
-      "total_flos": 148653022003200.0,
-      "train_loss": 1.016740475225166,
-      "train_runtime": 19333.4248,
-      "train_samples_per_second": 5.027,
-      "train_steps_per_second": 0.157
-    }
-  ],
-  "max_steps": 3038,
-  "num_train_epochs": 2,
-  "total_flos": 148653022003200.0,
+      "epoch": 0.31805337519623234,
+      "grad_norm": 1.9799340852762877,
+      "learning_rate": 1.5955461446715002e-05,
+      "loss": 1.044,
+      "step": 3039
+    },
+    {
+      "epoch": 0.31815803244374674,
+      "grad_norm": 2.1983306415646693,
+      "learning_rate": 1.5952738070884458e-05,
+      "loss": 0.9124,
+      "step": 3040
+    },
+    {
+      "epoch": 0.31826268969126115,
+      "grad_norm": 1.974260011569216,
+      "learning_rate": 1.5950014011072238e-05,
+      "loss": 0.9891,
+      "step": 3041
+    },
+    {
+      "epoch": 0.3183673469387755,
+      "grad_norm": 2.310426349353939,
+      "learning_rate": 1.5947289267591354e-05,
+      "loss": 0.9872,
+      "step": 3042
+    },
+    {
+      "epoch": 0.3184720041862899,
+      "grad_norm": 1.9450362823631278,
+      "learning_rate": 1.5944563840754873e-05,
+      "loss": 1.0055,
+      "step": 3043
+    },
+    {
+      "epoch": 0.3185766614338043,
+      "grad_norm": 2.036067128143145,
+      "learning_rate": 1.5941837730875956e-05,
+      "loss": 0.9706,
+      "step": 3044
+    },
+    {
+      "epoch": 0.31868131868131866,
+      "grad_norm": 2.05039801626101,
+      "learning_rate": 1.593911093826784e-05,
+      "loss": 0.9528,
+      "step": 3045
+    },
+    {
+      "epoch": 0.31878597592883307,
+      "grad_norm": 2.2876059678714857,
+      "learning_rate": 1.5936383463243836e-05,
+      "loss": 0.9563,
+      "step": 3046
+    },
+    {
+      "epoch": 0.31889063317634747,
+      "grad_norm": 2.697037738761109,
+      "learning_rate": 1.5933655306117338e-05,
+      "loss": 1.0216,
+      "step": 3047
+    },
+    {
+      "epoch": 0.3189952904238619,
+      "grad_norm": 2.5022369955164385,
+      "learning_rate": 1.5930926467201816e-05,
+      "loss": 1.033,
+      "step": 3048
+    },
+    {
+      "epoch": 0.3190999476713762,
+      "grad_norm": 2.113886043265922,
+      "learning_rate": 1.592819694681082e-05,
+      "loss": 0.9792,
+      "step": 3049
+    },
+    {
+      "epoch": 0.31920460491889063,
+      "grad_norm": 2.237260875974751,
+      "learning_rate": 1.5925466745257977e-05,
+      "loss": 1.0271,
+      "step": 3050
+    },
+    {
+      "epoch": 0.31930926216640504,
+      "grad_norm": 2.1237623188008037,
+      "learning_rate": 1.592273586285699e-05,
+      "loss": 0.9983,
+      "step": 3051
+    },
+    {
+      "epoch": 0.3194139194139194,
+      "grad_norm": 2.2392068397320344,
+      "learning_rate": 1.5920004299921652e-05,
+      "loss": 1.0828,
+      "step": 3052
+    },
+    {
+      "epoch": 0.3195185766614338,
+      "grad_norm": 3.338921058059234,
+      "learning_rate": 1.5917272056765815e-05,
+      "loss": 1.0108,
+      "step": 3053
+    },
+    {
+      "epoch": 0.3196232339089482,
+      "grad_norm": 2.1076301049943647,
+      "learning_rate": 1.591453913370342e-05,
+      "loss": 1.0223,
+      "step": 3054
+    },
+    {
+      "epoch": 0.3197278911564626,
+      "grad_norm": 2.0503145112616825,
+      "learning_rate": 1.5911805531048495e-05,
+      "loss": 1.0396,
+      "step": 3055
+    },
+    {
+      "epoch": 0.31983254840397696,
+      "grad_norm": 2.4375226451961827,
+      "learning_rate": 1.5909071249115128e-05,
+      "loss": 1.07,
+      "step": 3056
+    },
+    {
+      "epoch": 0.31993720565149136,
+      "grad_norm": 1.9288776837663408,
+      "learning_rate": 1.5906336288217494e-05,
+      "loss": 0.9863,
+      "step": 3057
+    },
+    {
+      "epoch": 0.32004186289900577,
+      "grad_norm": 2.0943331121912605,
+      "learning_rate": 1.5903600648669847e-05,
+      "loss": 0.8811,
+      "step": 3058
+    },
+    {
+      "epoch": 0.3201465201465201,
+      "grad_norm": 2.0975483698938917,
+      "learning_rate": 1.590086433078652e-05,
+      "loss": 1.1083,
+      "step": 3059
+    },
+    {
+      "epoch": 0.3202511773940345,
+      "grad_norm": 1.991398727953297,
+      "learning_rate": 1.5898127334881916e-05,
+      "loss": 0.8839,
+      "step": 3060
+    },
+    {
+      "epoch": 0.32035583464154893,
+      "grad_norm": 2.2279462939438655,
+      "learning_rate": 1.5895389661270528e-05,
+      "loss": 0.9735,
+      "step": 3061
+    },
+    {
+      "epoch": 0.32046049188906334,
+      "grad_norm": 1.8857564494547183,
+      "learning_rate": 1.5892651310266912e-05,
+      "loss": 1.0815,
+      "step": 3062
+    },
+    {
+      "epoch": 0.3205651491365777,
+      "grad_norm": 2.1252881559677492,
+      "learning_rate": 1.5889912282185717e-05,
+      "loss": 0.956,
+      "step": 3063
+    },
+    {
+      "epoch": 0.3206698063840921,
+      "grad_norm": 2.2558822549332236,
+      "learning_rate": 1.588717257734166e-05,
+      "loss": 1.0757,
+      "step": 3064
+    },
+    {
+      "epoch": 0.3207744636316065,
+      "grad_norm": 2.3702017920147567,
+      "learning_rate": 1.588443219604954e-05,
+      "loss": 0.9591,
+      "step": 3065
+    },
+    {
+      "epoch": 0.3208791208791209,
+      "grad_norm": 1.9235192765898275,
+      "learning_rate": 1.5881691138624236e-05,
+      "loss": 1.0212,
+      "step": 3066
+    },
+    {
+      "epoch": 0.32098377812663526,
+      "grad_norm": 2.664761728680135,
+      "learning_rate": 1.5878949405380694e-05,
+      "loss": 1.1368,
+      "step": 3067
+    },
+    {
+      "epoch": 0.32108843537414966,
+      "grad_norm": 2.3349986003807683,
+      "learning_rate": 1.5876206996633943e-05,
+      "loss": 1.0861,
+      "step": 3068
+    },
+    {
+      "epoch": 0.32119309262166407,
+      "grad_norm": 2.0350639615647435,
+      "learning_rate": 1.58734639126991e-05,
+      "loss": 1.0298,
+      "step": 3069
+    },
+    {
+      "epoch": 0.3212977498691784,
+      "grad_norm": 2.1379630016372473,
+      "learning_rate": 1.5870720153891346e-05,
+      "loss": 1.0008,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3214024071166928,
+      "grad_norm": 2.423558596633754,
+      "learning_rate": 1.5867975720525947e-05,
+      "loss": 0.9308,
+      "step": 3071
+    },
+    {
+      "epoch": 0.32150706436420723,
+      "grad_norm": 2.0816034578512728,
+      "learning_rate": 1.586523061291824e-05,
+      "loss": 0.9646,
+      "step": 3072
+    },
+    {
+      "epoch": 0.32161172161172163,
+      "grad_norm": 2.560250524049416,
+      "learning_rate": 1.5862484831383643e-05,
+      "loss": 1.0258,
+      "step": 3073
+    },
+    {
+      "epoch": 0.321716378859236,
+      "grad_norm": 2.239124234673666,
+      "learning_rate": 1.5859738376237657e-05,
+      "loss": 1.0865,
+      "step": 3074
+    },
+    {
+      "epoch": 0.3218210361067504,
+      "grad_norm": 2.3699409843476666,
+      "learning_rate": 1.5856991247795853e-05,
+      "loss": 0.9975,
+      "step": 3075
+    },
+    {
+      "epoch": 0.3219256933542648,
+      "grad_norm": 1.9916759495681915,
+      "learning_rate": 1.5854243446373877e-05,
+      "loss": 0.9162,
+      "step": 3076
+    },
+    {
+      "epoch": 0.32203035060177915,
+      "grad_norm": 1.9898290475403624,
+      "learning_rate": 1.5851494972287466e-05,
+      "loss": 1.0024,
+      "step": 3077
+    },
+    {
+      "epoch": 0.32213500784929355,
+      "grad_norm": 2.0511214319935456,
+      "learning_rate": 1.584874582585242e-05,
+      "loss": 1.0025,
+      "step": 3078
+    },
+    {
+      "epoch": 0.32223966509680796,
+      "grad_norm": 2.180082289189908,
+      "learning_rate": 1.5845996007384626e-05,
+      "loss": 0.7952,
+      "step": 3079
+    },
+    {
+      "epoch": 0.32234432234432236,
+      "grad_norm": 2.0373190957074314,
+      "learning_rate": 1.5843245517200033e-05,
+      "loss": 0.9029,
+      "step": 3080
+    },
+    {
+      "epoch": 0.3224489795918367,
+      "grad_norm": 2.2294041825628117,
+      "learning_rate": 1.5840494355614686e-05,
+      "loss": 1.1581,
+      "step": 3081
+    },
+    {
+      "epoch": 0.3225536368393511,
+      "grad_norm": 2.3966321362004512,
+      "learning_rate": 1.5837742522944698e-05,
+      "loss": 0.9631,
+      "step": 3082
+    },
+    {
+      "epoch": 0.3226582940868655,
+      "grad_norm": 1.924447534905052,
+      "learning_rate": 1.5834990019506263e-05,
+      "loss": 1.0266,
+      "step": 3083
+    },
+    {
+      "epoch": 0.32276295133437993,
+      "grad_norm": 2.3585671923673908,
+      "learning_rate": 1.583223684561564e-05,
+      "loss": 1.0525,
+      "step": 3084
+    },
+    {
+      "epoch": 0.3228676085818943,
+      "grad_norm": 2.3735070616891334,
+      "learning_rate": 1.5829483001589185e-05,
+      "loss": 1.0239,
+      "step": 3085
+    },
+    {
+      "epoch": 0.3229722658294087,
+      "grad_norm": 2.107789442162236,
+      "learning_rate": 1.582672848774331e-05,
+      "loss": 0.9487,
+      "step": 3086
+    },
+    {
+      "epoch": 0.3230769230769231,
+      "grad_norm": 1.9939952977167719,
+      "learning_rate": 1.5823973304394526e-05,
+      "loss": 0.9027,
+      "step": 3087
+    },
+    {
+      "epoch": 0.32318158032443745,
+      "grad_norm": 1.9853892321221516,
+      "learning_rate": 1.5821217451859402e-05,
+      "loss": 0.9402,
+      "step": 3088
+    },
+    {
+      "epoch": 0.32328623757195185,
+      "grad_norm": 2.294801256285806,
+      "learning_rate": 1.5818460930454588e-05,
+      "loss": 1.082,
+      "step": 3089
+    },
+    {
+      "epoch": 0.32339089481946626,
+      "grad_norm": 2.0584151362636267,
+      "learning_rate": 1.5815703740496823e-05,
+      "loss": 0.9276,
+      "step": 3090
+    },
+    {
+      "epoch": 0.32349555206698066,
+      "grad_norm": 1.9598606086672974,
+      "learning_rate": 1.5812945882302907e-05,
+      "loss": 0.9895,
+      "step": 3091
+    },
+    {
+      "epoch": 0.323600209314495,
+      "grad_norm": 2.0072911970323135,
+      "learning_rate": 1.5810187356189722e-05,
+      "loss": 1.0237,
+      "step": 3092
+    },
+    {
+      "epoch": 0.3237048665620094,
+      "grad_norm": 1.8808583035296527,
+      "learning_rate": 1.580742816247423e-05,
+      "loss": 0.9557,
+      "step": 3093
+    },
+    {
+      "epoch": 0.3238095238095238,
+      "grad_norm": 2.241856615594079,
+      "learning_rate": 1.5804668301473473e-05,
+      "loss": 1.0337,
+      "step": 3094
+    },
+    {
+      "epoch": 0.3239141810570382,
+      "grad_norm": 2.238604082081988,
+      "learning_rate": 1.5801907773504563e-05,
+      "loss": 1.0139,
+      "step": 3095
+    },
+    {
+      "epoch": 0.3240188383045526,
+      "grad_norm": 2.086041350119793,
+      "learning_rate": 1.5799146578884686e-05,
+      "loss": 1.0195,
+      "step": 3096
+    },
+    {
+      "epoch": 0.324123495552067,
+      "grad_norm": 2.177520822966867,
+      "learning_rate": 1.579638471793111e-05,
+      "loss": 0.9967,
+      "step": 3097
+    },
+    {
+      "epoch": 0.3242281527995814,
+      "grad_norm": 2.590324178614816,
+      "learning_rate": 1.579362219096118e-05,
+      "loss": 0.9962,
+      "step": 3098
+    },
+    {
+      "epoch": 0.32433281004709574,
+      "grad_norm": 2.0699609992470407,
+      "learning_rate": 1.5790858998292313e-05,
+      "loss": 1.0322,
+      "step": 3099
+    },
+    {
+      "epoch": 0.32443746729461015,
+      "grad_norm": 2.206119437570607,
+      "learning_rate": 1.5788095140242014e-05,
+      "loss": 1.0231,
+      "step": 3100
+    },
+    {
+      "epoch": 0.32454212454212455,
+      "grad_norm": 1.8212021688643432,
+      "learning_rate": 1.5785330617127844e-05,
+      "loss": 0.8309,
+      "step": 3101
+    },
+    {
+      "epoch": 0.3246467817896389,
+      "grad_norm": 2.2727201676845583,
+      "learning_rate": 1.578256542926746e-05,
+      "loss": 0.986,
+      "step": 3102
+    },
+    {
+      "epoch": 0.3247514390371533,
+      "grad_norm": 2.1449608431036684,
+      "learning_rate": 1.5779799576978584e-05,
+      "loss": 1.099,
+      "step": 3103
+    },
+    {
+      "epoch": 0.3248560962846677,
+      "grad_norm": 2.480440020807763,
+      "learning_rate": 1.5777033060579022e-05,
+      "loss": 0.9017,
+      "step": 3104
+    },
+    {
+      "epoch": 0.3249607535321821,
+      "grad_norm": 2.0367956477342166,
+      "learning_rate": 1.577426588038665e-05,
+      "loss": 0.8987,
+      "step": 3105
+    },
+    {
+      "epoch": 0.3250654107796965,
+      "grad_norm": 1.9815075457418991,
+      "learning_rate": 1.5771498036719418e-05,
+      "loss": 0.8948,
+      "step": 3106
+    },
+    {
+      "epoch": 0.3251700680272109,
+      "grad_norm": 1.9726747058162075,
+      "learning_rate": 1.5768729529895365e-05,
+      "loss": 0.9563,
+      "step": 3107
+    },
+    {
+      "epoch": 0.3252747252747253,
+      "grad_norm": 2.1307459003781295,
+      "learning_rate": 1.57659603602326e-05,
+      "loss": 0.9732,
+      "step": 3108
+    },
+    {
+      "epoch": 0.3253793825222397,
+      "grad_norm": 4.287887282310838,
+      "learning_rate": 1.5763190528049295e-05,
+      "loss": 1.1042,
+      "step": 3109
+    },
+    {
+      "epoch": 0.32548403976975404,
+      "grad_norm": 2.1285826245261483,
+      "learning_rate": 1.5760420033663716e-05,
+      "loss": 1.0633,
+      "step": 3110
+    },
+    {
+      "epoch": 0.32558869701726845,
+      "grad_norm": 1.8011470361015922,
+      "learning_rate": 1.5757648877394197e-05,
+      "loss": 0.8628,
+      "step": 3111
+    },
+    {
+      "epoch": 0.32569335426478285,
+      "grad_norm": 2.151338719221326,
+      "learning_rate": 1.575487705955915e-05,
+      "loss": 1.0653,
+      "step": 3112
+    },
+    {
+      "epoch": 0.3257980115122972,
+      "grad_norm": 2.1328861874541563,
+      "learning_rate": 1.5752104580477065e-05,
+      "loss": 1.11,
+      "step": 3113
+    },
+    {
+      "epoch": 0.3259026687598116,
+      "grad_norm": 2.088911019800362,
+      "learning_rate": 1.5749331440466497e-05,
+      "loss": 1.0583,
+      "step": 3114
+    },
+    {
+      "epoch": 0.326007326007326,
+      "grad_norm": 1.7497905672760596,
+      "learning_rate": 1.5746557639846095e-05,
+      "loss": 0.8168,
+      "step": 3115
+    },
+    {
+      "epoch": 0.3261119832548404,
+      "grad_norm": 2.0703455005124765,
+      "learning_rate": 1.5743783178934572e-05,
+      "loss": 0.9295,
+      "step": 3116
+    },
+    {
+      "epoch": 0.32621664050235477,
+      "grad_norm": 2.051050023116992,
+      "learning_rate": 1.5741008058050718e-05,
+      "loss": 0.9884,
+      "step": 3117
+    },
+    {
+      "epoch": 0.3263212977498692,
+      "grad_norm": 1.8426576087061717,
+      "learning_rate": 1.57382322775134e-05,
+      "loss": 0.9219,
+      "step": 3118
+    },
+    {
+      "epoch": 0.3264259549973836,
+      "grad_norm": 2.1973030476954123,
+      "learning_rate": 1.5735455837641556e-05,
+      "loss": 0.9935,
+      "step": 3119
+    },
+    {
+      "epoch": 0.32653061224489793,
+      "grad_norm": 2.390653054556527,
+      "learning_rate": 1.5732678738754214e-05,
+      "loss": 1.1316,
+      "step": 3120
+    },
+    {
+      "epoch": 0.32663526949241234,
+      "grad_norm": 2.099033446066326,
+      "learning_rate": 1.5729900981170464e-05,
+      "loss": 1.036,
+      "step": 3121
+    },
+    {
+      "epoch": 0.32673992673992674,
+      "grad_norm": 1.7457495629091635,
+      "learning_rate": 1.5727122565209474e-05,
+      "loss": 0.8337,
+      "step": 3122
+    },
+    {
+      "epoch": 0.32684458398744115,
+      "grad_norm": 2.5405638057373365,
+      "learning_rate": 1.5724343491190488e-05,
+      "loss": 1.019,
+      "step": 3123
+    },
+    {
+      "epoch": 0.3269492412349555,
+      "grad_norm": 2.2469815540311826,
+      "learning_rate": 1.5721563759432835e-05,
+      "loss": 1.0026,
+      "step": 3124
+    },
+    {
+      "epoch": 0.3270538984824699,
+      "grad_norm": 2.397277989964288,
+      "learning_rate": 1.5718783370255905e-05,
+      "loss": 0.9063,
+      "step": 3125
+    },
+    {
+      "epoch": 0.3271585557299843,
+      "grad_norm": 2.353318814929306,
+      "learning_rate": 1.5716002323979174e-05,
+      "loss": 1.194,
+      "step": 3126
+    },
+    {
+      "epoch": 0.3272632129774987,
+      "grad_norm": 2.028797944646975,
+      "learning_rate": 1.571322062092219e-05,
+      "loss": 0.7781,
+      "step": 3127
+    },
+    {
+      "epoch": 0.32736787022501307,
+      "grad_norm": 2.1763538986167346,
+      "learning_rate": 1.571043826140457e-05,
+      "loss": 1.0173,
+      "step": 3128
+    },
+    {
+      "epoch": 0.3274725274725275,
+      "grad_norm": 2.2559481805621866,
+      "learning_rate": 1.5707655245746022e-05,
+      "loss": 1.0209,
+      "step": 3129
+    },
+    {
+      "epoch": 0.3275771847200419,
+      "grad_norm": 2.266294866644528,
+      "learning_rate": 1.5704871574266315e-05,
+      "loss": 1.1683,
+      "step": 3130
+    },
+    {
+      "epoch": 0.32768184196755623,
+      "grad_norm": 2.1728502972455774,
+      "learning_rate": 1.5702087247285297e-05,
+      "loss": 1.0714,
+      "step": 3131
+    },
+    {
+      "epoch": 0.32778649921507064,
+      "grad_norm": 2.412631577860173,
+      "learning_rate": 1.5699302265122894e-05,
+      "loss": 1.0777,
+      "step": 3132
+    },
+    {
+      "epoch": 0.32789115646258504,
+      "grad_norm": 2.4781137882095403,
+      "learning_rate": 1.569651662809911e-05,
+      "loss": 1.086,
+      "step": 3133
+    },
+    {
+      "epoch": 0.32799581371009945,
+      "grad_norm": 2.2300712669478036,
+      "learning_rate": 1.5693730336534017e-05,
+      "loss": 1.0021,
+      "step": 3134
+    },
+    {
+      "epoch": 0.3281004709576138,
+      "grad_norm": 2.2052539997942655,
+      "learning_rate": 1.5690943390747764e-05,
+      "loss": 1.0284,
+      "step": 3135
+    },
+    {
+      "epoch": 0.3282051282051282,
+      "grad_norm": 2.4413624909490497,
+      "learning_rate": 1.5688155791060577e-05,
+      "loss": 1.0287,
+      "step": 3136
+    },
+    {
+      "epoch": 0.3283097854526426,
+      "grad_norm": 2.556415622005147,
+      "learning_rate": 1.5685367537792765e-05,
+      "loss": 1.075,
+      "step": 3137
+    },
+    {
+      "epoch": 0.32841444270015696,
+      "grad_norm": 2.1283762756864677,
+      "learning_rate": 1.568257863126469e-05,
+      "loss": 0.8205,
+      "step": 3138
+    },
+    {
+      "epoch": 0.32851909994767137,
+      "grad_norm": 2.232790862732562,
+      "learning_rate": 1.567978907179682e-05,
+      "loss": 1.0162,
+      "step": 3139
+    },
+    {
+      "epoch": 0.3286237571951858,
+      "grad_norm": 2.3136474249151537,
+      "learning_rate": 1.5676998859709663e-05,
+      "loss": 1.0353,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3287284144427002,
+      "grad_norm": 2.638925386745023,
+      "learning_rate": 1.5674207995323828e-05,
+      "loss": 0.9493,
+      "step": 3141
+    },
+    {
+      "epoch": 0.32883307169021453,
+      "grad_norm": 2.506618787870084,
+      "learning_rate": 1.5671416478959995e-05,
+      "loss": 1.0422,
+      "step": 3142
+    },
+    {
+      "epoch": 0.32893772893772893,
+      "grad_norm": 2.0975015700370125,
+      "learning_rate": 1.5668624310938914e-05,
+      "loss": 0.9647,
+      "step": 3143
+    },
+    {
+      "epoch": 0.32904238618524334,
+      "grad_norm": 2.1397256829041456,
+      "learning_rate": 1.5665831491581405e-05,
+      "loss": 0.8599,
+      "step": 3144
+    },
+    {
+      "epoch": 0.3291470434327577,
+      "grad_norm": 2.63504903803654,
+      "learning_rate": 1.566303802120837e-05,
+      "loss": 1.0202,
+      "step": 3145
+    },
+    {
+      "epoch": 0.3292517006802721,
+      "grad_norm": 2.026871373608857,
+      "learning_rate": 1.566024390014079e-05,
+      "loss": 1.0174,
+      "step": 3146
+    },
+    {
+      "epoch": 0.3293563579277865,
+      "grad_norm": 2.222517598112396,
+      "learning_rate": 1.5657449128699708e-05,
+      "loss": 0.8657,
+      "step": 3147
+    },
+    {
+      "epoch": 0.3294610151753009,
+      "grad_norm": 1.8704721739634205,
+      "learning_rate": 1.5654653707206255e-05,
+      "loss": 1.0354,
+      "step": 3148
+    },
+    {
+      "epoch": 0.32956567242281526,
+      "grad_norm": 2.215919635887023,
+      "learning_rate": 1.5651857635981628e-05,
+      "loss": 1.0849,
+      "step": 3149
+    },
+    {
+      "epoch": 0.32967032967032966,
+      "grad_norm": 2.564246658121425,
+      "learning_rate": 1.56490609153471e-05,
+      "loss": 0.9606,
+      "step": 3150
+    },
+    {
+      "epoch": 0.32977498691784407,
+      "grad_norm": 2.1280666629677696,
+      "learning_rate": 1.564626354562402e-05,
+      "loss": 1.0398,
+      "step": 3151
+    },
+    {
+      "epoch": 0.3298796441653585,
+      "grad_norm": 2.007176609705347,
+      "learning_rate": 1.5643465527133815e-05,
+      "loss": 0.9364,
+      "step": 3152
+    },
+    {
+      "epoch": 0.3299843014128728,
+      "grad_norm": 2.374920213374514,
+      "learning_rate": 1.5640666860197975e-05,
+      "loss": 1.0571,
+      "step": 3153
+    },
+    {
+      "epoch": 0.33008895866038723,
+      "grad_norm": 1.9142213744766876,
+      "learning_rate": 1.5637867545138083e-05,
+      "loss": 0.96,
+      "step": 3154
+    },
+    {
+      "epoch": 0.33019361590790164,
+      "grad_norm": 2.2782769847496223,
+      "learning_rate": 1.563506758227578e-05,
+      "loss": 1.0653,
+      "step": 3155
+    },
+    {
+      "epoch": 0.330298273155416,
+      "grad_norm": 1.884020880571543,
+      "learning_rate": 1.563226697193279e-05,
+      "loss": 1.0129,
+      "step": 3156
+    },
+    {
+      "epoch": 0.3304029304029304,
+      "grad_norm": 2.2062503790267862,
+      "learning_rate": 1.5629465714430906e-05,
+      "loss": 0.9958,
+      "step": 3157
+    },
+    {
+      "epoch": 0.3305075876504448,
+      "grad_norm": 2.3997612816521086,
+      "learning_rate": 1.5626663810091997e-05,
+      "loss": 1.0093,
+      "step": 3158
+    },
+    {
+      "epoch": 0.3306122448979592,
+      "grad_norm": 2.103715083415718,
+      "learning_rate": 1.562386125923801e-05,
+      "loss": 0.9297,
+      "step": 3159
+    },
+    {
+      "epoch": 0.33071690214547356,
+      "grad_norm": 2.215995087737775,
+      "learning_rate": 1.5621058062190965e-05,
+      "loss": 0.9648,
+      "step": 3160
+    },
+    {
+      "epoch": 0.33082155939298796,
+      "grad_norm": 1.8692320023257354,
+      "learning_rate": 1.5618254219272954e-05,
+      "loss": 1.0103,
+      "step": 3161
+    },
+    {
+      "epoch": 0.33092621664050237,
+      "grad_norm": 2.0395101583342305,
+      "learning_rate": 1.561544973080614e-05,
+      "loss": 0.936,
+      "step": 3162
+    },
+    {
+      "epoch": 0.3310308738880167,
+      "grad_norm": 1.9092428105317705,
+      "learning_rate": 1.5612644597112773e-05,
+      "loss": 0.8791,
+      "step": 3163
+    },
+    {
+      "epoch": 0.3311355311355311,
+      "grad_norm": 2.340830105583403,
+      "learning_rate": 1.5609838818515162e-05,
+      "loss": 1.0334,
+      "step": 3164
+    },
+    {
+      "epoch": 0.33124018838304553,
+      "grad_norm": 2.500046118391191,
+      "learning_rate": 1.56070323953357e-05,
+      "loss": 0.8887,
+      "step": 3165
+    },
+    {
+      "epoch": 0.33134484563055994,
+      "grad_norm": 2.3218591928110794,
+      "learning_rate": 1.5604225327896846e-05,
+      "loss": 1.0837,
+      "step": 3166
+    },
+    {
+      "epoch": 0.3314495028780743,
+      "grad_norm": 2.119253835435931,
+      "learning_rate": 1.5601417616521143e-05,
+      "loss": 0.9716,
+      "step": 3167
+    },
+    {
+      "epoch": 0.3315541601255887,
+      "grad_norm": 2.038070808087544,
+      "learning_rate": 1.5598609261531202e-05,
+      "loss": 1.0377,
+      "step": 3168
+    },
+    {
+      "epoch": 0.3316588173731031,
+      "grad_norm": 2.4134269571581215,
+      "learning_rate": 1.5595800263249702e-05,
+      "loss": 0.8931,
+      "step": 3169
+    },
+    {
+      "epoch": 0.3317634746206175,
+      "grad_norm": 1.7317120142862439,
+      "learning_rate": 1.5592990621999414e-05,
+      "loss": 0.8002,
+      "step": 3170
+    },
+    {
+      "epoch": 0.33186813186813185,
+      "grad_norm": 2.150415859427035,
+      "learning_rate": 1.559018033810316e-05,
+      "loss": 0.8975,
+      "step": 3171
+    },
+    {
+      "epoch": 0.33197278911564626,
+      "grad_norm": 2.90864630369287,
+      "learning_rate": 1.5587369411883856e-05,
+      "loss": 0.998,
+      "step": 3172
+    },
+    {
+      "epoch": 0.33207744636316067,
+      "grad_norm": 2.271286673783965,
+      "learning_rate": 1.558455784366448e-05,
+      "loss": 1.021,
+      "step": 3173
+    },
+    {
+      "epoch": 0.332182103610675,
+      "grad_norm": 2.3150664350547885,
+      "learning_rate": 1.5581745633768086e-05,
+      "loss": 1.0001,
+      "step": 3174
+    },
+    {
+      "epoch": 0.3322867608581894,
+      "grad_norm": 3.363470350927157,
+      "learning_rate": 1.5578932782517804e-05,
+      "loss": 0.9165,
+      "step": 3175
+    },
+    {
+      "epoch": 0.33239141810570383,
+      "grad_norm": 2.08188408813949,
+      "learning_rate": 1.5576119290236836e-05,
+      "loss": 0.9887,
+      "step": 3176
+    },
+    {
+      "epoch": 0.33249607535321823,
+      "grad_norm": 1.930989000386139,
+      "learning_rate": 1.5573305157248454e-05,
+      "loss": 0.9224,
+      "step": 3177
+    },
+    {
+      "epoch": 0.3326007326007326,
+      "grad_norm": 1.9621642580428715,
+      "learning_rate": 1.5570490383876015e-05,
+      "loss": 0.9062,
+      "step": 3178
+    },
+    {
+      "epoch": 0.332705389848247,
+      "grad_norm": 1.9661202771421844,
+      "learning_rate": 1.5567674970442936e-05,
+      "loss": 0.8901,
+      "step": 3179
+    },
+    {
+      "epoch": 0.3328100470957614,
+      "grad_norm": 2.1572459606930625,
+      "learning_rate": 1.5564858917272716e-05,
+      "loss": 0.9869,
+      "step": 3180
+    },
+    {
+      "epoch": 0.33291470434327575,
+      "grad_norm": 2.474124091599375,
+      "learning_rate": 1.5562042224688926e-05,
+      "loss": 1.1371,
+      "step": 3181
+    },
+    {
+      "epoch": 0.33301936159079015,
+      "grad_norm": 1.8653639765364336,
+      "learning_rate": 1.555922489301521e-05,
+      "loss": 0.8485,
+      "step": 3182
+    },
+    {
+      "epoch": 0.33312401883830456,
+      "grad_norm": 2.1078524268043783,
+      "learning_rate": 1.555640692257528e-05,
+      "loss": 0.9803,
+      "step": 3183
+    },
+    {
+      "epoch": 0.33322867608581896,
+      "grad_norm": 2.139476652742253,
+      "learning_rate": 1.5553588313692936e-05,
+      "loss": 0.9596,
+      "step": 3184
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 2.141936726484649,
+      "learning_rate": 1.5550769066692036e-05,
+      "loss": 1.0432,
+      "step": 3185
+    },
+    {
+      "epoch": 0.3334379905808477,
+      "grad_norm": 2.12152468260362,
+      "learning_rate": 1.5547949181896515e-05,
+      "loss": 1.01,
+      "step": 3186
+    },
+    {
+      "epoch": 0.3335426478283621,
+      "grad_norm": 2.264842415629573,
+      "learning_rate": 1.5545128659630385e-05,
+      "loss": 1.0728,
+      "step": 3187
+    },
+    {
+      "epoch": 0.3336473050758765,
+      "grad_norm": 2.1304146804426924,
+      "learning_rate": 1.5542307500217734e-05,
+      "loss": 0.967,
+      "step": 3188
+    },
+    {
+      "epoch": 0.3337519623233909,
+      "grad_norm": 1.9276499028871452,
+      "learning_rate": 1.5539485703982714e-05,
+      "loss": 0.9121,
+      "step": 3189
+    },
+    {
+      "epoch": 0.3338566195709053,
+      "grad_norm": 1.9765479283756435,
+      "learning_rate": 1.553666327124956e-05,
+      "loss": 1.0808,
+      "step": 3190
+    },
+    {
+      "epoch": 0.3339612768184197,
+      "grad_norm": 1.7361909209283715,
+      "learning_rate": 1.553384020234257e-05,
+      "loss": 0.8643,
+      "step": 3191
+    },
+    {
+      "epoch": 0.33406593406593404,
+      "grad_norm": 2.1761684193729174,
+      "learning_rate": 1.553101649758612e-05,
+      "loss": 1.0361,
+      "step": 3192
+    },
+    {
+      "epoch": 0.33417059131344845,
+      "grad_norm": 2.523913526660812,
+      "learning_rate": 1.5528192157304663e-05,
+      "loss": 1.0898,
+      "step": 3193
+    },
+    {
+      "epoch": 0.33427524856096286,
+      "grad_norm": 2.118661622161171,
+      "learning_rate": 1.5525367181822722e-05,
+      "loss": 0.956,
+      "step": 3194
+    },
+    {
+      "epoch": 0.33437990580847726,
+      "grad_norm": 2.1545428502001123,
+      "learning_rate": 1.5522541571464894e-05,
+      "loss": 0.9939,
+      "step": 3195
+    },
+    {
+      "epoch": 0.3344845630559916,
+      "grad_norm": 2.2676647398119107,
+      "learning_rate": 1.5519715326555837e-05,
+      "loss": 1.0386,
+      "step": 3196
+    },
+    {
+      "epoch": 0.334589220303506,
+      "grad_norm": 2.06105181377489,
+      "learning_rate": 1.5516888447420303e-05,
+      "loss": 1.0757,
+      "step": 3197
+    },
+    {
+      "epoch": 0.3346938775510204,
+      "grad_norm": 2.3715768654167775,
+      "learning_rate": 1.5514060934383103e-05,
+      "loss": 0.997,
+      "step": 3198
+    },
+    {
+      "epoch": 0.3347985347985348,
+      "grad_norm": 2.027803037575869,
+      "learning_rate": 1.5511232787769124e-05,
+      "loss": 0.8679,
+      "step": 3199
+    },
+    {
+      "epoch": 0.3349031920460492,
+      "grad_norm": 2.1181232730913964,
+      "learning_rate": 1.5508404007903322e-05,
+      "loss": 0.9015,
+      "step": 3200
+    },
+    {
+      "epoch": 0.3350078492935636,
+      "grad_norm": 2.3226842222737365,
+      "learning_rate": 1.5505574595110734e-05,
+      "loss": 0.9824,
+      "step": 3201
+    },
+    {
+      "epoch": 0.335112506541078,
+      "grad_norm": 2.47655647815724,
+      "learning_rate": 1.5502744549716465e-05,
+      "loss": 0.9697,
+      "step": 3202
+    },
+    {
+      "epoch": 0.33521716378859234,
+      "grad_norm": 2.0566095932796986,
+      "learning_rate": 1.5499913872045694e-05,
+      "loss": 0.8571,
+      "step": 3203
+    },
+    {
+      "epoch": 0.33532182103610675,
+      "grad_norm": 2.28076401752358,
+      "learning_rate": 1.5497082562423665e-05,
+      "loss": 1.0462,
+      "step": 3204
+    },
+    {
+      "epoch": 0.33542647828362115,
+      "grad_norm": 2.3423053756381784,
+      "learning_rate": 1.5494250621175706e-05,
+      "loss": 1.01,
+      "step": 3205
+    },
+    {
+      "epoch": 0.3355311355311355,
+      "grad_norm": 2.5072831733976653,
+      "learning_rate": 1.5491418048627214e-05,
+      "loss": 1.0932,
+      "step": 3206
+    },
+    {
+      "epoch": 0.3356357927786499,
+      "grad_norm": 2.1432608031534723,
+      "learning_rate": 1.548858484510366e-05,
+      "loss": 0.8312,
+      "step": 3207
+    },
+    {
+      "epoch": 0.3357404500261643,
+      "grad_norm": 2.437167122394929,
+      "learning_rate": 1.5485751010930577e-05,
+      "loss": 0.927,
+      "step": 3208
+    },
+    {
+      "epoch": 0.3358451072736787,
+      "grad_norm": 2.0576201295605925,
+      "learning_rate": 1.5482916546433583e-05,
+      "loss": 1.0367,
+      "step": 3209
+    },
+    {
+      "epoch": 0.3359497645211931,
+      "grad_norm": 2.1847007438604362,
+      "learning_rate": 1.5480081451938362e-05,
+      "loss": 1.0442,
+      "step": 3210
+    },
+    {
+      "epoch": 0.3360544217687075,
+      "grad_norm": 2.068195651770208,
+      "learning_rate": 1.5477245727770672e-05,
+      "loss": 0.9245,
+      "step": 3211
+    },
+    {
+      "epoch": 0.3361590790162219,
+      "grad_norm": 2.1072128887111967,
+      "learning_rate": 1.5474409374256344e-05,
+      "loss": 0.9496,
+      "step": 3212
+    },
+    {
+      "epoch": 0.3362637362637363,
+      "grad_norm": 2.3804536046173914,
+      "learning_rate": 1.5471572391721282e-05,
+      "loss": 0.977,
+      "step": 3213
+    },
+    {
+      "epoch": 0.33636839351125064,
+      "grad_norm": 2.0895509232276406,
+      "learning_rate": 1.5468734780491465e-05,
+      "loss": 1.0336,
+      "step": 3214
+    },
+    {
+      "epoch": 0.33647305075876505,
+      "grad_norm": 2.168752460124443,
+      "learning_rate": 1.5465896540892932e-05,
+      "loss": 0.9361,
+      "step": 3215
+    },
+    {
+      "epoch": 0.33657770800627945,
+      "grad_norm": 2.1004970170203197,
+      "learning_rate": 1.5463057673251802e-05,
+      "loss": 0.9727,
+      "step": 3216
+    },
+    {
+      "epoch": 0.3366823652537938,
+      "grad_norm": 2.050190223227202,
+      "learning_rate": 1.5460218177894274e-05,
+      "loss": 0.9999,
+      "step": 3217
+    },
+    {
+      "epoch": 0.3367870225013082,
+      "grad_norm": 2.038917201160083,
+      "learning_rate": 1.5457378055146607e-05,
+      "loss": 0.962,
+      "step": 3218
+    },
+    {
+      "epoch": 0.3368916797488226,
+      "grad_norm": 2.0493286691810395,
+      "learning_rate": 1.545453730533514e-05,
+      "loss": 0.918,
+      "step": 3219
+    },
+    {
+      "epoch": 0.336996336996337,
+      "grad_norm": 2.009804376163448,
+      "learning_rate": 1.545169592878628e-05,
+      "loss": 0.7884,
+      "step": 3220
+    },
+    {
+      "epoch": 0.33710099424385137,
+      "grad_norm": 1.9418895146815451,
+      "learning_rate": 1.5448853925826505e-05,
+      "loss": 0.9334,
+      "step": 3221
+    },
+    {
+      "epoch": 0.3372056514913658,
+      "grad_norm": 2.0920719859447363,
+      "learning_rate": 1.5446011296782368e-05,
+      "loss": 0.9992,
+      "step": 3222
+    },
+    {
+      "epoch": 0.3373103087388802,
+      "grad_norm": 2.7317450031003876,
+      "learning_rate": 1.5443168041980488e-05,
+      "loss": 0.9853,
+      "step": 3223
+    },
+    {
+      "epoch": 0.33741496598639453,
+      "grad_norm": 2.5941883234023564,
+      "learning_rate": 1.544032416174757e-05,
+      "loss": 1.0523,
+      "step": 3224
+    },
+    {
+      "epoch": 0.33751962323390894,
+      "grad_norm": 2.2629384674782442,
+      "learning_rate": 1.5437479656410376e-05,
+      "loss": 1.1233,
+      "step": 3225
+    },
+    {
+      "epoch": 0.33762428048142334,
+      "grad_norm": 2.2335346692765126,
+      "learning_rate": 1.543463452629575e-05,
+      "loss": 0.9641,
+      "step": 3226
+    },
+    {
+      "epoch": 0.33772893772893775,
+      "grad_norm": 2.1221463470407427,
+      "learning_rate": 1.5431788771730597e-05,
+      "loss": 1.0066,
+      "step": 3227
+    },
+    {
+      "epoch": 0.3378335949764521,
+      "grad_norm": 2.2293540541218504,
+      "learning_rate": 1.5428942393041904e-05,
+      "loss": 1.0154,
+      "step": 3228
+    },
+    {
+      "epoch": 0.3379382522239665,
+      "grad_norm": 2.3900893182152845,
+      "learning_rate": 1.5426095390556725e-05,
+      "loss": 1.053,
+      "step": 3229
+    },
+    {
+      "epoch": 0.3380429094714809,
+      "grad_norm": 2.048307102090152,
+      "learning_rate": 1.542324776460218e-05,
+      "loss": 1.0966,
+      "step": 3230
+    },
+    {
+      "epoch": 0.33814756671899526,
+      "grad_norm": 2.141298333153114,
+      "learning_rate": 1.5420399515505478e-05,
+      "loss": 0.9943,
+      "step": 3231
+    },
+    {
+      "epoch": 0.33825222396650967,
+      "grad_norm": 2.1539784614556936,
+      "learning_rate": 1.5417550643593885e-05,
+      "loss": 1.0074,
+      "step": 3232
+    },
+    {
+      "epoch": 0.3383568812140241,
+      "grad_norm": 2.244060531894042,
+      "learning_rate": 1.5414701149194736e-05,
+      "loss": 1.0491,
+      "step": 3233
+    },
+    {
+      "epoch": 0.3384615384615385,
+      "grad_norm": 1.995913218973165,
+      "learning_rate": 1.541185103263545e-05,
+      "loss": 0.9373,
+      "step": 3234
+    },
+    {
+      "epoch": 0.33856619570905283,
+      "grad_norm": 2.172151133964098,
+      "learning_rate": 1.540900029424351e-05,
+      "loss": 0.9444,
+      "step": 3235
+    },
+    {
+      "epoch": 0.33867085295656724,
+      "grad_norm": 1.89676124574755,
+      "learning_rate": 1.5406148934346468e-05,
+      "loss": 0.8628,
+      "step": 3236
+    },
+    {
+      "epoch": 0.33877551020408164,
+      "grad_norm": 1.8828681953222985,
+      "learning_rate": 1.5403296953271958e-05,
+      "loss": 0.9545,
+      "step": 3237
+    },
+    {
+      "epoch": 0.33888016745159605,
+      "grad_norm": 1.8741424977674062,
+      "learning_rate": 1.5400444351347674e-05,
+      "loss": 0.966,
+      "step": 3238
+    },
+    {
+      "epoch": 0.3389848246991104,
+      "grad_norm": 2.0333171496050304,
+      "learning_rate": 1.5397591128901384e-05,
+      "loss": 0.9455,
+      "step": 3239
+    },
+    {
+      "epoch": 0.3390894819466248,
+      "grad_norm": 1.9816357166577598,
+      "learning_rate": 1.539473728626093e-05,
+      "loss": 1.0213,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3391941391941392,
+      "grad_norm": 2.363598537983322,
+      "learning_rate": 1.539188282375423e-05,
+      "loss": 1.0232,
+      "step": 3241
+    },
+    {
+      "epoch": 0.33929879644165356,
+      "grad_norm": 2.0397959543670314,
+      "learning_rate": 1.538902774170926e-05,
+      "loss": 1.0472,
+      "step": 3242
+    },
+    {
+      "epoch": 0.33940345368916797,
+      "grad_norm": 2.070254879911,
+      "learning_rate": 1.538617204045408e-05,
+      "loss": 1.0323,
+      "step": 3243
+    },
+    {
+      "epoch": 0.3395081109366824,
+      "grad_norm": 2.0695439666681814,
+      "learning_rate": 1.538331572031681e-05,
+      "loss": 1.063,
+      "step": 3244
+    },
+    {
+      "epoch": 0.3396127681841968,
+      "grad_norm": 2.1305043728996567,
+      "learning_rate": 1.5380458781625654e-05,
+      "loss": 1.0096,
+      "step": 3245
+    },
+    {
+      "epoch": 0.33971742543171113,
+      "grad_norm": 1.967788896520279,
+      "learning_rate": 1.5377601224708875e-05,
+      "loss": 1.002,
+      "step": 3246
+    },
+    {
+      "epoch": 0.33982208267922553,
+      "grad_norm": 2.1322066005520592,
+      "learning_rate": 1.5374743049894813e-05,
+      "loss": 1.0679,
+      "step": 3247
+    },
+    {
+      "epoch": 0.33992673992673994,
+      "grad_norm": 1.8579885877502837,
+      "learning_rate": 1.537188425751188e-05,
+      "loss": 0.8455,
+      "step": 3248
+    },
+    {
+      "epoch": 0.3400313971742543,
+      "grad_norm": 2.15408723090576,
+      "learning_rate": 1.536902484788856e-05,
+      "loss": 1.0029,
+      "step": 3249
+    },
+    {
+      "epoch": 0.3401360544217687,
+      "grad_norm": 2.4599439352791537,
+      "learning_rate": 1.5366164821353392e-05,
+      "loss": 1.0669,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3402407116692831,
+      "grad_norm": 2.2542203659980142,
+      "learning_rate": 1.5363304178235017e-05,
+      "loss": 1.0087,
+      "step": 3251
+    },
+    {
+      "epoch": 0.3403453689167975,
+      "grad_norm": 2.0949948404828187,
+      "learning_rate": 1.536044291886211e-05,
+      "loss": 0.9923,
+      "step": 3252
+    },
+    {
+      "epoch": 0.34045002616431186,
+      "grad_norm": 1.9315131178475988,
+      "learning_rate": 1.535758104356345e-05,
+      "loss": 1.0347,
+      "step": 3253
+    },
+    {
+      "epoch": 0.34055468341182626,
+      "grad_norm": 2.172260985606811,
+      "learning_rate": 1.535471855266787e-05,
+      "loss": 1.0161,
+      "step": 3254
+    },
+    {
+      "epoch": 0.34065934065934067,
+      "grad_norm": 2.03347702907374,
+      "learning_rate": 1.5351855446504268e-05,
+      "loss": 0.9756,
+      "step": 3255
+    },
+    {
+      "epoch": 0.3407639979068551,
+      "grad_norm": 2.1511761068591344,
+      "learning_rate": 1.534899172540163e-05,
+      "loss": 0.9812,
+      "step": 3256
+    },
+    {
+      "epoch": 0.3408686551543694,
+      "grad_norm": 2.014625870073615,
+      "learning_rate": 1.5346127389688996e-05,
+      "loss": 1.0583,
+      "step": 3257
+    },
+    {
+      "epoch": 0.34097331240188383,
+      "grad_norm": 2.424338507190857,
+      "learning_rate": 1.534326243969549e-05,
+      "loss": 1.1823,
+      "step": 3258
+    },
+    {
+      "epoch": 0.34107796964939824,
+      "grad_norm": 2.381135018727616,
+      "learning_rate": 1.5340396875750294e-05,
+      "loss": 0.9861,
+      "step": 3259
+    },
+    {
+      "epoch": 0.3411826268969126,
+      "grad_norm": 2.4725877348005585,
+      "learning_rate": 1.5337530698182674e-05,
+      "loss": 1.0594,
+      "step": 3260
+    },
+    {
+      "epoch": 0.341287284144427,
+      "grad_norm": 1.9709311757797843,
+      "learning_rate": 1.5334663907321957e-05,
+      "loss": 1.0043,
+      "step": 3261
+    },
+    {
+      "epoch": 0.3413919413919414,
+      "grad_norm": 2.0576277902991693,
+      "learning_rate": 1.533179650349754e-05,
+      "loss": 0.9401,
+      "step": 3262
+    },
+    {
+      "epoch": 0.3414965986394558,
+      "grad_norm": 1.9172132204679706,
+      "learning_rate": 1.5328928487038898e-05,
+      "loss": 1.0187,
+      "step": 3263
+    },
+    {
+      "epoch": 0.34160125588697016,
+      "grad_norm": 2.2358901429303026,
+      "learning_rate": 1.5326059858275568e-05,
+      "loss": 1.0068,
+      "step": 3264
+    },
+    {
+      "epoch": 0.34170591313448456,
+      "grad_norm": 2.130154051130339,
+      "learning_rate": 1.5323190617537165e-05,
+      "loss": 0.9645,
+      "step": 3265
+    },
+    {
+      "epoch": 0.34181057038199897,
+      "grad_norm": 1.9736640857121206,
+      "learning_rate": 1.5320320765153367e-05,
+      "loss": 0.9449,
+      "step": 3266
+    },
+    {
+      "epoch": 0.3419152276295133,
+      "grad_norm": 2.2610379556501226,
+      "learning_rate": 1.531745030145393e-05,
+      "loss": 0.9335,
+      "step": 3267
+    },
+    {
+      "epoch": 0.3420198848770277,
+      "grad_norm": 1.9304145684884284,
+      "learning_rate": 1.531457922676867e-05,
+      "loss": 0.9527,
+      "step": 3268
+    },
+    {
+      "epoch": 0.34212454212454213,
+      "grad_norm": 2.23937829886418,
+      "learning_rate": 1.531170754142749e-05,
+      "loss": 1.0275,
+      "step": 3269
+    },
+    {
+      "epoch": 0.34222919937205654,
+      "grad_norm": 1.952354365893699,
+      "learning_rate": 1.5308835245760342e-05,
+      "loss": 0.9735,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3423338566195709,
+      "grad_norm": 1.7116086172760445,
+      "learning_rate": 1.530596234009726e-05,
+      "loss": 0.9834,
+      "step": 3271
+    },
+    {
+      "epoch": 0.3424385138670853,
+      "grad_norm": 2.209172716848948,
+      "learning_rate": 1.5303088824768347e-05,
+      "loss": 0.9872,
+      "step": 3272
+    },
+    {
+      "epoch": 0.3425431711145997,
+      "grad_norm": 2.2644113537174504,
+      "learning_rate": 1.5300214700103784e-05,
+      "loss": 1.0005,
+      "step": 3273
+    },
+    {
+      "epoch": 0.3426478283621141,
+      "grad_norm": 2.2147137441366063,
+      "learning_rate": 1.52973399664338e-05,
+      "loss": 1.0471,
+      "step": 3274
+    },
+    {
+      "epoch": 0.34275248560962845,
+      "grad_norm": 2.202945184782102,
+      "learning_rate": 1.5294464624088718e-05,
+      "loss": 1.0856,
+      "step": 3275
+    },
+    {
+      "epoch": 0.34285714285714286,
+      "grad_norm": 2.036998214551154,
+      "learning_rate": 1.529158867339892e-05,
+      "loss": 1.0183,
+      "step": 3276
+    },
+    {
+      "epoch": 0.34296180010465727,
+      "grad_norm": 2.3880333648717778,
+      "learning_rate": 1.5288712114694855e-05,
+      "loss": 0.9732,
+      "step": 3277
+    },
+    {
+      "epoch": 0.3430664573521716,
+      "grad_norm": 2.293141355535519,
+      "learning_rate": 1.5285834948307047e-05,
+      "loss": 1.0911,
+      "step": 3278
+    },
+    {
+      "epoch": 0.343171114599686,
+      "grad_norm": 2.3667237873101103,
+      "learning_rate": 1.5282957174566086e-05,
+      "loss": 1.0271,
+      "step": 3279
+    },
+    {
+      "epoch": 0.34327577184720043,
+      "grad_norm": 2.5802091802539118,
+      "learning_rate": 1.528007879380264e-05,
+      "loss": 1.0383,
+      "step": 3280
+    },
+    {
+      "epoch": 0.34338042909471483,
+      "grad_norm": 2.2984418484959037,
+      "learning_rate": 1.527719980634743e-05,
+      "loss": 1.0983,
+      "step": 3281
+    },
+    {
+      "epoch": 0.3434850863422292,
+      "grad_norm": 1.9988451780658998,
+      "learning_rate": 1.527432021253127e-05,
+      "loss": 0.9391,
+      "step": 3282
+    },
+    {
+      "epoch": 0.3435897435897436,
+      "grad_norm": 1.9979482563308306,
+      "learning_rate": 1.5271440012685027e-05,
+      "loss": 0.9803,
+      "step": 3283
+    },
+    {
+      "epoch": 0.343694400837258,
+      "grad_norm": 2.0462333311095144,
+      "learning_rate": 1.5268559207139637e-05,
+      "loss": 0.953,
+      "step": 3284
+    },
+    {
+      "epoch": 0.34379905808477235,
+      "grad_norm": 2.018679553660942,
+      "learning_rate": 1.5265677796226114e-05,
+      "loss": 1.0009,
+      "step": 3285
+    },
+    {
+      "epoch": 0.34390371533228675,
+      "grad_norm": 1.8547261935286408,
+      "learning_rate": 1.5262795780275538e-05,
+      "loss": 1.0313,
+      "step": 3286
+    },
+    {
+      "epoch": 0.34400837257980116,
+      "grad_norm": 2.0846595576576843,
+      "learning_rate": 1.5259913159619054e-05,
+      "loss": 0.9638,
+      "step": 3287
+    },
+    {
+      "epoch": 0.34411302982731556,
+      "grad_norm": 2.2092617207892458,
+      "learning_rate": 1.5257029934587888e-05,
+      "loss": 1.048,
+      "step": 3288
+    },
+    {
+      "epoch": 0.3442176870748299,
+      "grad_norm": 2.115010279212093,
+      "learning_rate": 1.5254146105513328e-05,
+      "loss": 1.139,
+      "step": 3289
+    },
+    {
+      "epoch": 0.3443223443223443,
+      "grad_norm": 2.0072225659373903,
+      "learning_rate": 1.5251261672726727e-05,
+      "loss": 1.0207,
+      "step": 3290
+    },
+    {
+      "epoch": 0.3444270015698587,
+      "grad_norm": 1.9343608998324062,
+      "learning_rate": 1.5248376636559514e-05,
+      "loss": 0.8169,
+      "step": 3291
+    },
+    {
+      "epoch": 0.3445316588173731,
+      "grad_norm": 2.2114081845502795,
+      "learning_rate": 1.5245490997343187e-05,
+      "loss": 1.0064,
+      "step": 3292
+    },
+    {
+      "epoch": 0.3446363160648875,
+      "grad_norm": 2.19450830303441,
+      "learning_rate": 1.5242604755409307e-05,
+      "loss": 1.0446,
+      "step": 3293
+    },
+    {
+      "epoch": 0.3447409733124019,
+      "grad_norm": 2.9828919291241727,
+      "learning_rate": 1.5239717911089518e-05,
+      "loss": 0.9943,
+      "step": 3294
+    },
+    {
+      "epoch": 0.3448456305599163,
+      "grad_norm": 1.8836484748965538,
+      "learning_rate": 1.5236830464715517e-05,
+      "loss": 0.8834,
+      "step": 3295
+    },
+    {
+      "epoch": 0.34495028780743064,
+      "grad_norm": 2.484034519489334,
+      "learning_rate": 1.523394241661908e-05,
+      "loss": 0.8373,
+      "step": 3296
+    },
+    {
+      "epoch": 0.34505494505494505,
+      "grad_norm": 2.3959300576508262,
+      "learning_rate": 1.5231053767132047e-05,
+      "loss": 0.9524,
+      "step": 3297
+    },
+    {
+      "epoch": 0.34515960230245946,
+      "grad_norm": 2.0058874784680447,
+      "learning_rate": 1.5228164516586332e-05,
+      "loss": 0.9646,
+      "step": 3298
+    },
+    {
+      "epoch": 0.34526425954997386,
+      "grad_norm": 2.1806341190742944,
+      "learning_rate": 1.5225274665313916e-05,
+      "loss": 0.9534,
+      "step": 3299
+    },
+    {
+      "epoch": 0.3453689167974882,
+      "grad_norm": 2.3809715607129274,
+      "learning_rate": 1.5222384213646847e-05,
+      "loss": 1.0433,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3454735740450026,
+      "grad_norm": 2.243507765788451,
+      "learning_rate": 1.5219493161917248e-05,
+      "loss": 0.9947,
+      "step": 3301
+    },
+    {
+      "epoch": 0.345578231292517,
+      "grad_norm": 1.9486940004775435,
+      "learning_rate": 1.5216601510457305e-05,
+      "loss": 1.0537,
+      "step": 3302
+    },
+    {
+      "epoch": 0.3456828885400314,
+      "grad_norm": 2.1151878794379058,
+      "learning_rate": 1.5213709259599267e-05,
+      "loss": 0.9945,
+      "step": 3303
+    },
+    {
+      "epoch": 0.3457875457875458,
+      "grad_norm": 2.4085393603562073,
+      "learning_rate": 1.5210816409675473e-05,
+      "loss": 1.0041,
+      "step": 3304
+    },
+    {
+      "epoch": 0.3458922030350602,
+      "grad_norm": 2.0934987255320405,
+      "learning_rate": 1.5207922961018304e-05,
+      "loss": 1.0212,
+      "step": 3305
+    },
+    {
+      "epoch": 0.3459968602825746,
+      "grad_norm": 2.040087227874492,
+      "learning_rate": 1.5205028913960232e-05,
+      "loss": 1.045,
+      "step": 3306
+    },
+    {
+      "epoch": 0.34610151753008894,
+      "grad_norm": 2.0986542779184183,
+      "learning_rate": 1.5202134268833786e-05,
+      "loss": 0.9866,
+      "step": 3307
+    },
+    {
+      "epoch": 0.34620617477760335,
+      "grad_norm": 2.2898782266773425,
+      "learning_rate": 1.5199239025971567e-05,
+      "loss": 0.8739,
+      "step": 3308
+    },
+    {
+      "epoch": 0.34631083202511775,
+      "grad_norm": 2.7507453481175292,
+      "learning_rate": 1.5196343185706244e-05,
+      "loss": 0.9127,
+      "step": 3309
+    },
+    {
+      "epoch": 0.3464154892726321,
+      "grad_norm": 2.1750979583796366,
+      "learning_rate": 1.5193446748370551e-05,
+      "loss": 1.0279,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3465201465201465,
+      "grad_norm": 2.0867343208501365,
+      "learning_rate": 1.5190549714297303e-05,
+      "loss": 1.0373,
+      "step": 3311
+    },
+    {
+      "epoch": 0.3466248037676609,
+      "grad_norm": 2.5327508046734626,
+      "learning_rate": 1.5187652083819369e-05,
+      "loss": 1.0135,
+      "step": 3312
+    },
+    {
+      "epoch": 0.3467294610151753,
+      "grad_norm": 2.1859927718890555,
+      "learning_rate": 1.5184753857269697e-05,
+      "loss": 0.9699,
+      "step": 3313
+    },
+    {
+      "epoch": 0.3468341182626897,
+      "grad_norm": 2.009327763025113,
+      "learning_rate": 1.518185503498129e-05,
+      "loss": 0.907,
+      "step": 3314
+    },
+    {
+      "epoch": 0.3469387755102041,
+      "grad_norm": 2.2224539969275643,
+      "learning_rate": 1.5178955617287239e-05,
+      "loss": 1.083,
+      "step": 3315
+    },
+    {
+      "epoch": 0.3470434327577185,
+      "grad_norm": 2.4096293598821177,
+      "learning_rate": 1.517605560452069e-05,
+      "loss": 0.8995,
+      "step": 3316
+    },
+    {
+      "epoch": 0.3471480900052329,
+      "grad_norm": 2.146816047792693,
+      "learning_rate": 1.5173154997014857e-05,
+      "loss": 0.8222,
+      "step": 3317
+    },
+    {
+      "epoch": 0.34725274725274724,
+      "grad_norm": 2.3069897827667623,
+      "learning_rate": 1.5170253795103025e-05,
+      "loss": 0.9993,
+      "step": 3318
+    },
+    {
+      "epoch": 0.34735740450026165,
+      "grad_norm": 1.74531938947694,
+      "learning_rate": 1.5167351999118554e-05,
+      "loss": 0.9175,
+      "step": 3319
+    },
+    {
+      "epoch": 0.34746206174777605,
+      "grad_norm": 2.1288546754174993,
+      "learning_rate": 1.5164449609394863e-05,
+      "loss": 1.0347,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3475667189952904,
+      "grad_norm": 1.8501462739239363,
+      "learning_rate": 1.5161546626265442e-05,
+      "loss": 0.9374,
+      "step": 3321
+    },
+    {
+      "epoch": 0.3476713762428048,
+      "grad_norm": 2.416666658774087,
+      "learning_rate": 1.515864305006385e-05,
+      "loss": 1.1008,
+      "step": 3322
+    },
+    {
+      "epoch": 0.3477760334903192,
+      "grad_norm": 2.1683860693468913,
+      "learning_rate": 1.5155738881123711e-05,
+      "loss": 0.8261,
+      "step": 3323
+    },
+    {
+      "epoch": 0.3478806907378336,
+      "grad_norm": 1.9981929383306056,
+      "learning_rate": 1.5152834119778727e-05,
+      "loss": 0.8493,
+      "step": 3324
+    },
+    {
+      "epoch": 0.34798534798534797,
+      "grad_norm": 2.149137266048455,
+      "learning_rate": 1.5149928766362658e-05,
+      "loss": 1.0977,
+      "step": 3325
+    },
+    {
+      "epoch": 0.3480900052328624,
+      "grad_norm": 2.078026164203414,
+      "learning_rate": 1.5147022821209331e-05,
+      "loss": 1.107,
+      "step": 3326
+    },
+    {
+      "epoch": 0.3481946624803768,
+      "grad_norm": 2.3425230247962987,
+      "learning_rate": 1.5144116284652645e-05,
+      "loss": 0.9754,
+      "step": 3327
+    },
+    {
+      "epoch": 0.34829931972789113,
+      "grad_norm": 1.9091246891467695,
+      "learning_rate": 1.5141209157026576e-05,
+      "loss": 0.8426,
+      "step": 3328
+    },
+    {
+      "epoch": 0.34840397697540554,
+      "grad_norm": 1.9479723011505157,
+      "learning_rate": 1.5138301438665148e-05,
+      "loss": 1.0919,
+      "step": 3329
+    },
+    {
+      "epoch": 0.34850863422291994,
+      "grad_norm": 2.2310776562980412,
+      "learning_rate": 1.5135393129902469e-05,
+      "loss": 1.1552,
+      "step": 3330
+    },
+    {
+      "epoch": 0.34861329147043435,
+      "grad_norm": 2.0947428083855786,
+      "learning_rate": 1.513248423107271e-05,
+      "loss": 1.0404,
+      "step": 3331
+    },
+    {
+      "epoch": 0.3487179487179487,
+      "grad_norm": 2.066377865083883,
+      "learning_rate": 1.5129574742510107e-05,
+      "loss": 1.0278,
+      "step": 3332
+    },
+    {
+      "epoch": 0.3488226059654631,
+      "grad_norm": 2.099307559790499,
+      "learning_rate": 1.5126664664548969e-05,
+      "loss": 1.1413,
+      "step": 3333
+    },
+    {
+      "epoch": 0.3489272632129775,
+      "grad_norm": 2.154455920018568,
+      "learning_rate": 1.5123753997523665e-05,
+      "loss": 1.0198,
+      "step": 3334
+    },
+    {
+      "epoch": 0.34903192046049186,
+      "grad_norm": 2.2352534977710152,
+      "learning_rate": 1.5120842741768644e-05,
+      "loss": 1.0821,
+      "step": 3335
+    },
+    {
+      "epoch": 0.34913657770800627,
+      "grad_norm": 1.883735141874073,
+      "learning_rate": 1.5117930897618407e-05,
+      "loss": 1.0526,
+      "step": 3336
+    },
+    {
+      "epoch": 0.3492412349555207,
+      "grad_norm": 2.085232249057728,
+      "learning_rate": 1.5115018465407536e-05,
+      "loss": 0.9961,
+      "step": 3337
+    },
+    {
+      "epoch": 0.3493458922030351,
+      "grad_norm": 2.2405865672055567,
+      "learning_rate": 1.5112105445470677e-05,
+      "loss": 0.9606,
+      "step": 3338
+    },
+    {
+      "epoch": 0.34945054945054943,
+      "grad_norm": 2.210569750343716,
+      "learning_rate": 1.5109191838142537e-05,
+      "loss": 0.9826,
+      "step": 3339
+    },
+    {
+      "epoch": 0.34955520669806384,
+      "grad_norm": 1.8900569961244176,
+      "learning_rate": 1.5106277643757895e-05,
+      "loss": 0.9309,
+      "step": 3340
+    },
+    {
+      "epoch": 0.34965986394557824,
+      "grad_norm": 2.119499479604053,
+      "learning_rate": 1.5103362862651602e-05,
+      "loss": 0.8946,
+      "step": 3341
+    },
+    {
+      "epoch": 0.34976452119309265,
+      "grad_norm": 2.14866784367062,
+      "learning_rate": 1.5100447495158572e-05,
+      "loss": 1.0256,
+      "step": 3342
+    },
+    {
+      "epoch": 0.349869178440607,
+      "grad_norm": 2.21964636178506,
+      "learning_rate": 1.5097531541613784e-05,
+      "loss": 0.8199,
+      "step": 3343
+    },
+    {
+      "epoch": 0.3499738356881214,
+      "grad_norm": 2.0824336161678474,
+      "learning_rate": 1.5094615002352284e-05,
+      "loss": 1.0465,
+      "step": 3344
+    },
+    {
+      "epoch": 0.3500784929356358,
+      "grad_norm": 2.2906700002330354,
+      "learning_rate": 1.50916978777092e-05,
+      "loss": 1.0289,
+      "step": 3345
+    },
+    {
+      "epoch": 0.35018315018315016,
+      "grad_norm": 1.7870706054978838,
+      "learning_rate": 1.5088780168019704e-05,
+      "loss": 0.8907,
+      "step": 3346
+    },
+    {
+      "epoch": 0.35028780743066457,
+      "grad_norm": 2.350201821325944,
+      "learning_rate": 1.508586187361905e-05,
+      "loss": 0.9207,
+      "step": 3347
+    },
+    {
+      "epoch": 0.35039246467817897,
+      "grad_norm": 1.9237037149190483,
+      "learning_rate": 1.5082942994842557e-05,
+      "loss": 0.9421,
+      "step": 3348
+    },
+    {
+      "epoch": 0.3504971219256934,
+      "grad_norm": 2.6474938275921,
+      "learning_rate": 1.508002353202561e-05,
+      "loss": 1.0688,
+      "step": 3349
+    },
+    {
+      "epoch": 0.35060177917320773,
+      "grad_norm": 2.266954434535588,
+      "learning_rate": 1.5077103485503664e-05,
+      "loss": 0.938,
+      "step": 3350
+    },
+    {
+      "epoch": 0.35070643642072213,
+      "grad_norm": 1.838992898597147,
+      "learning_rate": 1.507418285561223e-05,
+      "loss": 1.0003,
+      "step": 3351
+    },
+    {
+      "epoch": 0.35081109366823654,
+      "grad_norm": 2.1043607397062702,
+      "learning_rate": 1.5071261642686903e-05,
+      "loss": 0.8642,
+      "step": 3352
+    },
+    {
+      "epoch": 0.3509157509157509,
+      "grad_norm": 2.092736361302439,
+      "learning_rate": 1.506833984706333e-05,
+      "loss": 1.0074,
+      "step": 3353
+    },
+    {
+      "epoch": 0.3510204081632653,
+      "grad_norm": 2.540716324430332,
+      "learning_rate": 1.5065417469077235e-05,
+      "loss": 1.112,
+      "step": 3354
+    },
+    {
+      "epoch": 0.3511250654107797,
+      "grad_norm": 2.2462673291428548,
+      "learning_rate": 1.5062494509064406e-05,
+      "loss": 1.053,
+      "step": 3355
+    },
+    {
+      "epoch": 0.3512297226582941,
+      "grad_norm": 1.9698347906672438,
+      "learning_rate": 1.5059570967360693e-05,
+      "loss": 0.9501,
+      "step": 3356
+    },
+    {
+      "epoch": 0.35133437990580846,
+      "grad_norm": 2.8732667059564707,
+      "learning_rate": 1.5056646844302018e-05,
+      "loss": 0.9699,
+      "step": 3357
+    },
+    {
+      "epoch": 0.35143903715332286,
+      "grad_norm": 2.300905830511158,
+      "learning_rate": 1.5053722140224371e-05,
+      "loss": 1.0558,
+      "step": 3358
+    },
+    {
+      "epoch": 0.35154369440083727,
+      "grad_norm": 2.1008549188317573,
+      "learning_rate": 1.5050796855463807e-05,
+      "loss": 1.0461,
+      "step": 3359
+    },
+    {
+      "epoch": 0.3516483516483517,
+      "grad_norm": 2.193971701711426,
+      "learning_rate": 1.5047870990356443e-05,
+      "loss": 0.8999,
+      "step": 3360
+    },
+    {
+      "epoch": 0.351753008895866,
+      "grad_norm": 1.9237206891325724,
+      "learning_rate": 1.5044944545238472e-05,
+      "loss": 0.919,
+      "step": 3361
+    },
+    {
+      "epoch": 0.35185766614338043,
+      "grad_norm": 2.3398493395367828,
+      "learning_rate": 1.5042017520446144e-05,
+      "loss": 0.9966,
+      "step": 3362
+    },
+    {
+      "epoch": 0.35196232339089484,
+      "grad_norm": 2.0245486796099414,
+      "learning_rate": 1.5039089916315778e-05,
+      "loss": 0.9552,
+      "step": 3363
+    },
+    {
+      "epoch": 0.3520669806384092,
+      "grad_norm": 2.146653040509672,
+      "learning_rate": 1.5036161733183772e-05,
+      "loss": 0.9682,
+      "step": 3364
+    },
+    {
+      "epoch": 0.3521716378859236,
+      "grad_norm": 2.3003979859770167,
+      "learning_rate": 1.5033232971386569e-05,
+      "loss": 0.9486,
+      "step": 3365
+    },
+    {
+      "epoch": 0.352276295133438,
+      "grad_norm": 2.080353917018897,
+      "learning_rate": 1.5030303631260691e-05,
+      "loss": 1.008,
+      "step": 3366
+    },
+    {
+      "epoch": 0.3523809523809524,
+      "grad_norm": 2.145495285206271,
+      "learning_rate": 1.5027373713142735e-05,
+      "loss": 0.8679,
+      "step": 3367
+    },
+    {
+      "epoch": 0.35248560962846676,
+      "grad_norm": 2.219416644891993,
+      "learning_rate": 1.5024443217369346e-05,
+      "loss": 0.8524,
+      "step": 3368
+    },
+    {
+      "epoch": 0.35259026687598116,
+      "grad_norm": 2.5932343785171432,
+      "learning_rate": 1.5021512144277246e-05,
+      "loss": 0.8825,
+      "step": 3369
+    },
+    {
+      "epoch": 0.35269492412349557,
+      "grad_norm": 2.2140426590236575,
+      "learning_rate": 1.5018580494203217e-05,
+      "loss": 1.0054,
+      "step": 3370
+    },
+    {
+      "epoch": 0.3527995813710099,
+      "grad_norm": 2.640222446891625,
+      "learning_rate": 1.501564826748412e-05,
+      "loss": 1.0553,
+      "step": 3371
+    },
+    {
+      "epoch": 0.3529042386185243,
+      "grad_norm": 2.6088669036618453,
+      "learning_rate": 1.5012715464456866e-05,
+      "loss": 0.9632,
+      "step": 3372
+    },
+    {
+      "epoch": 0.35300889586603873,
+      "grad_norm": 2.3393856800568513,
+      "learning_rate": 1.5009782085458447e-05,
+      "loss": 0.9854,
+      "step": 3373
+    },
+    {
+      "epoch": 0.35311355311355314,
+      "grad_norm": 2.13358684905417,
+      "learning_rate": 1.5006848130825909e-05,
+      "loss": 0.861,
+      "step": 3374
+    },
+    {
+      "epoch": 0.3532182103610675,
+      "grad_norm": 1.948963946557993,
+      "learning_rate": 1.500391360089637e-05,
+      "loss": 0.93,
+      "step": 3375
+    },
+    {
+      "epoch": 0.3533228676085819,
+      "grad_norm": 2.2621209126353374,
+      "learning_rate": 1.5000978496007013e-05,
+      "loss": 1.0111,
+      "step": 3376
+    },
+    {
+      "epoch": 0.3534275248560963,
+      "grad_norm": 2.175682145812267,
+      "learning_rate": 1.4998042816495091e-05,
+      "loss": 1.0821,
+      "step": 3377
+    },
+    {
+      "epoch": 0.35353218210361065,
+      "grad_norm": 2.118572813572351,
+      "learning_rate": 1.4995106562697917e-05,
+      "loss": 1.0052,
+      "step": 3378
+    },
+    {
+      "epoch": 0.35363683935112505,
+      "grad_norm": 2.2925720756150514,
+      "learning_rate": 1.4992169734952872e-05,
+      "loss": 0.8211,
+      "step": 3379
+    },
+    {
+      "epoch": 0.35374149659863946,
+      "grad_norm": 1.9505334994916146,
+      "learning_rate": 1.4989232333597404e-05,
+      "loss": 0.9554,
+      "step": 3380
+    },
+    {
+      "epoch": 0.35384615384615387,
+      "grad_norm": 2.0479847216050135,
+      "learning_rate": 1.4986294358969029e-05,
+      "loss": 1.0622,
+      "step": 3381
+    },
+    {
+      "epoch": 0.3539508110936682,
+      "grad_norm": 1.940644043048965,
+      "learning_rate": 1.498335581140532e-05,
+      "loss": 0.9721,
+      "step": 3382
+    },
+    {
+      "epoch": 0.3540554683411826,
+      "grad_norm": 2.357644955495196,
+      "learning_rate": 1.4980416691243929e-05,
+      "loss": 0.9845,
+      "step": 3383
+    },
+    {
+      "epoch": 0.35416012558869703,
+      "grad_norm": 1.9728875562735373,
+      "learning_rate": 1.4977476998822564e-05,
+      "loss": 0.9456,
+      "step": 3384
+    },
+    {
+      "epoch": 0.35426478283621143,
+      "grad_norm": 2.5759061498541773,
+      "learning_rate": 1.4974536734479e-05,
+      "loss": 1.0961,
+      "step": 3385
+    },
+    {
+      "epoch": 0.3543694400837258,
+      "grad_norm": 2.0650501070386436,
+      "learning_rate": 1.497159589855108e-05,
+      "loss": 0.9497,
+      "step": 3386
+    },
+    {
+      "epoch": 0.3544740973312402,
+      "grad_norm": 2.049511205390927,
+      "learning_rate": 1.4968654491376715e-05,
+      "loss": 0.9196,
+      "step": 3387
+    },
+    {
+      "epoch": 0.3545787545787546,
+      "grad_norm": 2.119804481767199,
+      "learning_rate": 1.4965712513293874e-05,
+      "loss": 1.1051,
+      "step": 3388
+    },
+    {
+      "epoch": 0.35468341182626895,
+      "grad_norm": 2.0198789773840153,
+      "learning_rate": 1.49627699646406e-05,
+      "loss": 0.9479,
+      "step": 3389
+    },
+    {
+      "epoch": 0.35478806907378335,
+      "grad_norm": 2.2514166159676376,
+      "learning_rate": 1.4959826845754996e-05,
+      "loss": 1.1349,
+      "step": 3390
+    },
+    {
+      "epoch": 0.35489272632129776,
+      "grad_norm": 1.9991992740675673,
+      "learning_rate": 1.4956883156975235e-05,
+      "loss": 0.9907,
+      "step": 3391
+    },
+    {
+      "epoch": 0.35499738356881216,
+      "grad_norm": 2.165858304353837,
+      "learning_rate": 1.495393889863955e-05,
+      "loss": 1.0128,
+      "step": 3392
+    },
+    {
+      "epoch": 0.3551020408163265,
+      "grad_norm": 2.046874393542093,
+      "learning_rate": 1.495099407108624e-05,
+      "loss": 0.9728,
+      "step": 3393
+    },
+    {
+      "epoch": 0.3552066980638409,
+      "grad_norm": 1.841861405528776,
+      "learning_rate": 1.4948048674653677e-05,
+      "loss": 0.9024,
+      "step": 3394
+    },
+    {
+      "epoch": 0.3553113553113553,
+      "grad_norm": 1.963788808953176,
+      "learning_rate": 1.4945102709680291e-05,
+      "loss": 1.047,
+      "step": 3395
+    },
+    {
+      "epoch": 0.3554160125588697,
+      "grad_norm": 1.8854311633911396,
+      "learning_rate": 1.4942156176504577e-05,
+      "loss": 0.9661,
+      "step": 3396
+    },
+    {
+      "epoch": 0.3555206698063841,
+      "grad_norm": 2.0001437303254312,
+      "learning_rate": 1.49392090754651e-05,
+      "loss": 1.0731,
+      "step": 3397
+    },
+    {
+      "epoch": 0.3556253270538985,
+      "grad_norm": 2.2429282145699645,
+      "learning_rate": 1.493626140690049e-05,
+      "loss": 1.0166,
+      "step": 3398
+    },
+    {
+      "epoch": 0.3557299843014129,
+      "grad_norm": 2.230032061722695,
+      "learning_rate": 1.4933313171149433e-05,
+      "loss": 0.9602,
+      "step": 3399
+    },
+    {
+      "epoch": 0.35583464154892724,
+      "grad_norm": 2.3611671076976517,
+      "learning_rate": 1.4930364368550697e-05,
+      "loss": 1.0514,
+      "step": 3400
+    },
+    {
+      "epoch": 0.35593929879644165,
+      "grad_norm": 1.9997981736421566,
+      "learning_rate": 1.4927414999443093e-05,
+      "loss": 0.8934,
+      "step": 3401
+    },
+    {
+      "epoch": 0.35604395604395606,
+      "grad_norm": 2.2260425699224378,
+      "learning_rate": 1.492446506416552e-05,
+      "loss": 1.105,
+      "step": 3402
+    },
+    {
+      "epoch": 0.35614861329147046,
+      "grad_norm": 2.6007886380070064,
+      "learning_rate": 1.4921514563056931e-05,
+      "loss": 1.0282,
+      "step": 3403
+    },
+    {
+      "epoch": 0.3562532705389848,
+      "grad_norm": 2.2911564149754664,
+      "learning_rate": 1.491856349645634e-05,
+      "loss": 0.9593,
+      "step": 3404
+    },
+    {
+      "epoch": 0.3563579277864992,
+      "grad_norm": 2.144042496081372,
+      "learning_rate": 1.4915611864702832e-05,
+      "loss": 1.0542,
+      "step": 3405
+    },
+    {
+      "epoch": 0.3564625850340136,
+      "grad_norm": 2.314520779479154,
+      "learning_rate": 1.4912659668135553e-05,
+      "loss": 1.0739,
+      "step": 3406
+    },
+    {
+      "epoch": 0.356567242281528,
+      "grad_norm": 2.0119651946546755,
+      "learning_rate": 1.4909706907093725e-05,
+      "loss": 1.0338,
+      "step": 3407
+    },
+    {
+      "epoch": 0.3566718995290424,
+      "grad_norm": 2.0594144199775517,
+      "learning_rate": 1.4906753581916618e-05,
+      "loss": 1.0535,
+      "step": 3408
+    },
+    {
+      "epoch": 0.3567765567765568,
+      "grad_norm": 2.628859441235232,
+      "learning_rate": 1.4903799692943575e-05,
+      "loss": 1.1152,
+      "step": 3409
+    },
+    {
+      "epoch": 0.3568812140240712,
+      "grad_norm": 2.3483961540859224,
+      "learning_rate": 1.4900845240514009e-05,
+      "loss": 1.0825,
+      "step": 3410
+    },
+    {
+      "epoch": 0.35698587127158554,
+      "grad_norm": 2.199102694351282,
+      "learning_rate": 1.4897890224967388e-05,
+      "loss": 1.0981,
+      "step": 3411
+    },
+    {
+      "epoch": 0.35709052851909995,
+      "grad_norm": 2.437840230574091,
+      "learning_rate": 1.4894934646643253e-05,
+      "loss": 1.0649,
+      "step": 3412
+    },
+    {
+      "epoch": 0.35719518576661435,
+      "grad_norm": 2.35758514512899,
+      "learning_rate": 1.4891978505881202e-05,
+      "loss": 1.0239,
+      "step": 3413
+    },
+    {
+      "epoch": 0.3572998430141287,
+      "grad_norm": 2.2981006350213304,
+      "learning_rate": 1.4889021803020904e-05,
+      "loss": 1.031,
+      "step": 3414
+    },
+    {
+      "epoch": 0.3574045002616431,
+      "grad_norm": 2.0708809030772484,
+      "learning_rate": 1.4886064538402091e-05,
+      "loss": 0.9587,
+      "step": 3415
+    },
+    {
+      "epoch": 0.3575091575091575,
+      "grad_norm": 2.454255588394301,
+      "learning_rate": 1.4883106712364558e-05,
+      "loss": 0.9636,
+      "step": 3416
+    },
+    {
+      "epoch": 0.3576138147566719,
+      "grad_norm": 1.870007612357461,
+      "learning_rate": 1.4880148325248163e-05,
+      "loss": 1.0022,
+      "step": 3417
+    },
+    {
+      "epoch": 0.35771847200418627,
+      "grad_norm": 2.1025070655550904,
+      "learning_rate": 1.4877189377392831e-05,
+      "loss": 1.0294,
+      "step": 3418
+    },
+    {
+      "epoch": 0.3578231292517007,
+      "grad_norm": 2.192703285083379,
+      "learning_rate": 1.4874229869138554e-05,
+      "loss": 1.1935,
+      "step": 3419
+    },
+    {
+      "epoch": 0.3579277864992151,
+      "grad_norm": 2.1921332205785555,
+      "learning_rate": 1.4871269800825384e-05,
+      "loss": 0.9556,
+      "step": 3420
+    },
+    {
+      "epoch": 0.35803244374672943,
+      "grad_norm": 2.3389520267003547,
+      "learning_rate": 1.486830917279344e-05,
+      "loss": 0.959,
+      "step": 3421
+    },
+    {
+      "epoch": 0.35813710099424384,
+      "grad_norm": 1.887005791860528,
+      "learning_rate": 1.4865347985382901e-05,
+      "loss": 0.9283,
+      "step": 3422
+    },
+    {
+      "epoch": 0.35824175824175825,
+      "grad_norm": 3.738025461337002,
+      "learning_rate": 1.4862386238934017e-05,
+      "loss": 1.0093,
+      "step": 3423
+    },
+    {
+      "epoch": 0.35834641548927265,
+      "grad_norm": 2.0931366465972268,
+      "learning_rate": 1.4859423933787097e-05,
+      "loss": 0.9058,
+      "step": 3424
+    },
+    {
+      "epoch": 0.358451072736787,
+      "grad_norm": 2.068043177767639,
+      "learning_rate": 1.4856461070282512e-05,
+      "loss": 1.044,
+      "step": 3425
+    },
+    {
+      "epoch": 0.3585557299843014,
+      "grad_norm": 2.2134904451583206,
+      "learning_rate": 1.4853497648760711e-05,
+      "loss": 0.8614,
+      "step": 3426
+    },
+    {
+      "epoch": 0.3586603872318158,
+      "grad_norm": 2.1848502762270368,
+      "learning_rate": 1.4850533669562189e-05,
+      "loss": 1.0403,
+      "step": 3427
+    },
+    {
+      "epoch": 0.3587650444793302,
+      "grad_norm": 2.190313212576901,
+      "learning_rate": 1.4847569133027512e-05,
+      "loss": 1.0629,
+      "step": 3428
+    },
+    {
+      "epoch": 0.35886970172684457,
+      "grad_norm": 2.2018800818289765,
+      "learning_rate": 1.4844604039497319e-05,
+      "loss": 1.0115,
+      "step": 3429
+    },
+    {
+      "epoch": 0.358974358974359,
+      "grad_norm": 1.8297165257801538,
+      "learning_rate": 1.4841638389312298e-05,
+      "loss": 0.9768,
+      "step": 3430
+    },
+    {
+      "epoch": 0.3590790162218734,
+      "grad_norm": 2.0502371253182177,
+      "learning_rate": 1.4838672182813213e-05,
+      "loss": 0.8625,
+      "step": 3431
+    },
+    {
+      "epoch": 0.35918367346938773,
+      "grad_norm": 2.0678658262449994,
+      "learning_rate": 1.4835705420340884e-05,
+      "loss": 0.9285,
+      "step": 3432
+    },
+    {
+      "epoch": 0.35928833071690214,
+      "grad_norm": 2.097477714048872,
+      "learning_rate": 1.48327381022362e-05,
+      "loss": 0.9737,
+      "step": 3433
+    },
+    {
+      "epoch": 0.35939298796441654,
+      "grad_norm": 2.216587571402107,
+      "learning_rate": 1.4829770228840111e-05,
+      "loss": 0.9205,
+      "step": 3434
+    },
+    {
+      "epoch": 0.35949764521193095,
+      "grad_norm": 2.242990585095391,
+      "learning_rate": 1.482680180049363e-05,
+      "loss": 1.0677,
+      "step": 3435
+    },
+    {
+      "epoch": 0.3596023024594453,
+      "grad_norm": 2.1956250689462595,
+      "learning_rate": 1.4823832817537838e-05,
+      "loss": 0.9873,
+      "step": 3436
+    },
+    {
+      "epoch": 0.3597069597069597,
+      "grad_norm": 2.250186078331522,
+      "learning_rate": 1.4820863280313874e-05,
+      "loss": 0.9965,
+      "step": 3437
+    },
+    {
+      "epoch": 0.3598116169544741,
+      "grad_norm": 2.080456079650869,
+      "learning_rate": 1.4817893189162954e-05,
+      "loss": 1.0024,
+      "step": 3438
+    },
+    {
+      "epoch": 0.35991627420198846,
+      "grad_norm": 2.0093127130290576,
+      "learning_rate": 1.4814922544426332e-05,
+      "loss": 1.0008,
+      "step": 3439
+    },
+    {
+      "epoch": 0.36002093144950287,
+      "grad_norm": 2.2251021302294083,
+      "learning_rate": 1.481195134644535e-05,
+      "loss": 1.0177,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3601255886970173,
+      "grad_norm": 1.9221853447634059,
+      "learning_rate": 1.4808979595561406e-05,
+      "loss": 0.8362,
+      "step": 3441
+    },
+    {
+      "epoch": 0.3602302459445317,
+      "grad_norm": 1.9001182786985462,
+      "learning_rate": 1.4806007292115955e-05,
+      "loss": 1.077,
+      "step": 3442
+    },
+    {
+      "epoch": 0.36033490319204603,
+      "grad_norm": 2.24243203183523,
+      "learning_rate": 1.4803034436450528e-05,
+      "loss": 0.9866,
+      "step": 3443
+    },
+    {
+      "epoch": 0.36043956043956044,
+      "grad_norm": 2.1248905323421217,
+      "learning_rate": 1.4800061028906703e-05,
+      "loss": 0.9718,
+      "step": 3444
+    },
+    {
+      "epoch": 0.36054421768707484,
+      "grad_norm": 2.1426511270144966,
+      "learning_rate": 1.4797087069826136e-05,
+      "loss": 1.0464,
+      "step": 3445
+    },
+    {
+      "epoch": 0.36064887493458925,
+      "grad_norm": 1.7983654364896233,
+      "learning_rate": 1.4794112559550542e-05,
+      "loss": 0.8639,
+      "step": 3446
+    },
+    {
+      "epoch": 0.3607535321821036,
+      "grad_norm": 2.053627748221558,
+      "learning_rate": 1.4791137498421692e-05,
+      "loss": 0.903,
+      "step": 3447
+    },
+    {
+      "epoch": 0.360858189429618,
+      "grad_norm": 1.8683934406849398,
+      "learning_rate": 1.4788161886781434e-05,
+      "loss": 0.867,
+      "step": 3448
+    },
+    {
+      "epoch": 0.3609628466771324,
+      "grad_norm": 2.101452076878475,
+      "learning_rate": 1.4785185724971667e-05,
+      "loss": 0.9869,
+      "step": 3449
+    },
+    {
+      "epoch": 0.36106750392464676,
+      "grad_norm": 2.496970829177601,
+      "learning_rate": 1.4782209013334364e-05,
+      "loss": 0.8874,
+      "step": 3450
+    },
+    {
+      "epoch": 0.36117216117216117,
+      "grad_norm": 1.9906643960387973,
+      "learning_rate": 1.4779231752211546e-05,
+      "loss": 1.0344,
+      "step": 3451
+    },
+    {
+      "epoch": 0.36127681841967557,
+      "grad_norm": 2.134773764897517,
+      "learning_rate": 1.477625394194531e-05,
+      "loss": 1.0026,
+      "step": 3452
+    },
+    {
+      "epoch": 0.36138147566719,
+      "grad_norm": 1.9360626015222449,
+      "learning_rate": 1.4773275582877818e-05,
+      "loss": 1.0467,
+      "step": 3453
+    },
+    {
+      "epoch": 0.36148613291470433,
+      "grad_norm": 2.3545030642754585,
+      "learning_rate": 1.477029667535128e-05,
+      "loss": 1.0638,
+      "step": 3454
+    },
+    {
+      "epoch": 0.36159079016221873,
+      "grad_norm": 2.178802837894888,
+      "learning_rate": 1.476731721970799e-05,
+      "loss": 0.9444,
+      "step": 3455
+    },
+    {
+      "epoch": 0.36169544740973314,
+      "grad_norm": 2.4728900427384155,
+      "learning_rate": 1.4764337216290283e-05,
+      "loss": 1.0585,
+      "step": 3456
+    },
+    {
+      "epoch": 0.3618001046572475,
+      "grad_norm": 2.203666741614066,
+      "learning_rate": 1.4761356665440571e-05,
+      "loss": 1.0357,
+      "step": 3457
+    },
+    {
+      "epoch": 0.3619047619047619,
+      "grad_norm": 2.1800575119431986,
+      "learning_rate": 1.475837556750133e-05,
+      "loss": 0.9192,
+      "step": 3458
+    },
+    {
+      "epoch": 0.3620094191522763,
+      "grad_norm": 2.237067611748822,
+      "learning_rate": 1.4755393922815086e-05,
+      "loss": 0.9418,
+      "step": 3459
+    },
+    {
+      "epoch": 0.3621140763997907,
+      "grad_norm": 2.0662472125409352,
+      "learning_rate": 1.4752411731724443e-05,
+      "loss": 1.0651,
+      "step": 3460
+    },
+    {
+      "epoch": 0.36221873364730506,
+      "grad_norm": 1.9800019356904621,
+      "learning_rate": 1.4749428994572058e-05,
+      "loss": 1.0229,
+      "step": 3461
+    },
+    {
+      "epoch": 0.36232339089481946,
+      "grad_norm": 2.193928271349309,
+      "learning_rate": 1.4746445711700648e-05,
+      "loss": 0.9627,
+      "step": 3462
+    },
+    {
+      "epoch": 0.36242804814233387,
+      "grad_norm": 2.1284831634544017,
+      "learning_rate": 1.474346188345301e-05,
+      "loss": 1.0074,
+      "step": 3463
+    },
+    {
+      "epoch": 0.3625327053898482,
+      "grad_norm": 1.9794414505430156,
+      "learning_rate": 1.4740477510171982e-05,
+      "loss": 0.9674,
+      "step": 3464
+    },
+    {
+      "epoch": 0.3626373626373626,
+      "grad_norm": 1.9781425949045437,
+      "learning_rate": 1.473749259220048e-05,
+      "loss": 0.9827,
+      "step": 3465
+    },
+    {
+      "epoch": 0.36274201988487703,
+      "grad_norm": 2.1680974375346174,
+      "learning_rate": 1.4734507129881473e-05,
+      "loss": 0.9404,
+      "step": 3466
+    },
+    {
+      "epoch": 0.36284667713239144,
+      "grad_norm": 2.104960303718753,
+      "learning_rate": 1.4731521123558e-05,
+      "loss": 0.9687,
+      "step": 3467
+    },
+    {
+      "epoch": 0.3629513343799058,
+      "grad_norm": 1.950557453989959,
+      "learning_rate": 1.4728534573573159e-05,
+      "loss": 0.766,
+      "step": 3468
+    },
+    {
+      "epoch": 0.3630559916274202,
+      "grad_norm": 2.2931497243072374,
+      "learning_rate": 1.472554748027011e-05,
+      "loss": 0.9558,
+      "step": 3469
+    },
+    {
+      "epoch": 0.3631606488749346,
+      "grad_norm": 2.114730292398457,
+      "learning_rate": 1.4722559843992075e-05,
+      "loss": 1.0599,
+      "step": 3470
+    },
+    {
+      "epoch": 0.363265306122449,
+      "grad_norm": 1.9387223713176414,
+      "learning_rate": 1.471957166508234e-05,
+      "loss": 0.9173,
+      "step": 3471
+    },
+    {
+      "epoch": 0.36336996336996336,
+      "grad_norm": 2.0196303025643934,
+      "learning_rate": 1.4716582943884254e-05,
+      "loss": 0.8972,
+      "step": 3472
+    },
+    {
+      "epoch": 0.36347462061747776,
+      "grad_norm": 2.3779507104223176,
+      "learning_rate": 1.4713593680741225e-05,
+      "loss": 1.0191,
+      "step": 3473
+    },
+    {
+      "epoch": 0.36357927786499217,
+      "grad_norm": 2.171151514960678,
+      "learning_rate": 1.4710603875996725e-05,
+      "loss": 0.9061,
+      "step": 3474
+    },
+    {
+      "epoch": 0.3636839351125065,
+      "grad_norm": 2.3376877060855974,
+      "learning_rate": 1.4707613529994294e-05,
+      "loss": 1.0131,
+      "step": 3475
+    },
+    {
+      "epoch": 0.3637885923600209,
+      "grad_norm": 2.240363414874548,
+      "learning_rate": 1.4704622643077524e-05,
+      "loss": 1.0021,
+      "step": 3476
+    },
+    {
+      "epoch": 0.36389324960753533,
+      "grad_norm": 2.2172322785021126,
+      "learning_rate": 1.4701631215590075e-05,
+      "loss": 1.0355,
+      "step": 3477
+    },
+    {
+      "epoch": 0.36399790685504974,
+      "grad_norm": 2.072175096832046,
+      "learning_rate": 1.4698639247875668e-05,
+      "loss": 1.0331,
+      "step": 3478
+    },
+    {
+      "epoch": 0.3641025641025641,
+      "grad_norm": 2.0743508713728844,
+      "learning_rate": 1.4695646740278084e-05,
+      "loss": 0.9647,
+      "step": 3479
+    },
+    {
+      "epoch": 0.3642072213500785,
+      "grad_norm": 1.8532767415518483,
+      "learning_rate": 1.4692653693141175e-05,
+      "loss": 0.8776,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3643118785975929,
+      "grad_norm": 2.302001815845674,
+      "learning_rate": 1.468966010680884e-05,
+      "loss": 0.9075,
+      "step": 3481
+    },
+    {
+      "epoch": 0.36441653584510725,
+      "grad_norm": 2.394789643374174,
+      "learning_rate": 1.4686665981625055e-05,
+      "loss": 1.0686,
+      "step": 3482
+    },
+    {
+      "epoch": 0.36452119309262165,
+      "grad_norm": 2.2837151238361697,
+      "learning_rate": 1.4683671317933843e-05,
+      "loss": 1.0853,
+      "step": 3483
+    },
+    {
+      "epoch": 0.36462585034013606,
+      "grad_norm": 2.025193817081867,
+      "learning_rate": 1.4680676116079307e-05,
+      "loss": 0.9073,
+      "step": 3484
+    },
+    {
+      "epoch": 0.36473050758765047,
+      "grad_norm": 2.4029403299228274,
+      "learning_rate": 1.4677680376405592e-05,
+      "loss": 0.9786,
+      "step": 3485
+    },
+    {
+      "epoch": 0.3648351648351648,
+      "grad_norm": 2.275681739422216,
+      "learning_rate": 1.4674684099256923e-05,
+      "loss": 1.1143,
+      "step": 3486
+    },
+    {
+      "epoch": 0.3649398220826792,
+      "grad_norm": 2.2365915095598576,
+      "learning_rate": 1.4671687284977572e-05,
+      "loss": 0.9557,
+      "step": 3487
+    },
+    {
+      "epoch": 0.3650444793301936,
+      "grad_norm": 2.2324428792489495,
+      "learning_rate": 1.4668689933911882e-05,
+      "loss": 0.961,
+      "step": 3488
+    },
+    {
+      "epoch": 0.36514913657770803,
+      "grad_norm": 2.187984861710933,
+      "learning_rate": 1.4665692046404253e-05,
+      "loss": 1.063,
+      "step": 3489
+    },
+    {
+      "epoch": 0.3652537938252224,
+      "grad_norm": 2.2899556426594243,
+      "learning_rate": 1.4662693622799152e-05,
+      "loss": 1.0662,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3653584510727368,
+      "grad_norm": 2.4136159700742716,
+      "learning_rate": 1.4659694663441102e-05,
+      "loss": 0.9803,
+      "step": 3491
+    },
+    {
+      "epoch": 0.3654631083202512,
+      "grad_norm": 1.7662751611483918,
+      "learning_rate": 1.4656695168674686e-05,
+      "loss": 0.8149,
+      "step": 3492
+    },
+    {
+      "epoch": 0.36556776556776555,
+      "grad_norm": 2.2806309857676226,
+      "learning_rate": 1.4653695138844556e-05,
+      "loss": 1.0818,
+      "step": 3493
+    },
+    {
+      "epoch": 0.36567242281527995,
+      "grad_norm": 1.7491999471330408,
+      "learning_rate": 1.4650694574295423e-05,
+      "loss": 0.8499,
+      "step": 3494
+    },
+    {
+      "epoch": 0.36577708006279436,
+      "grad_norm": 2.1237260413546526,
+      "learning_rate": 1.4647693475372054e-05,
+      "loss": 0.9252,
+      "step": 3495
+    },
+    {
+      "epoch": 0.36588173731030876,
+      "grad_norm": 2.0122406066004506,
+      "learning_rate": 1.4644691842419278e-05,
+      "loss": 0.9047,
+      "step": 3496
+    },
+    {
+      "epoch": 0.3659863945578231,
+      "grad_norm": 1.7997357542002177,
+      "learning_rate": 1.4641689675782e-05,
+      "loss": 0.9075,
+      "step": 3497
+    },
+    {
+      "epoch": 0.3660910518053375,
+      "grad_norm": 2.0095662635854126,
+      "learning_rate": 1.4638686975805168e-05,
+      "loss": 0.99,
+      "step": 3498
+    },
+    {
+      "epoch": 0.3661957090528519,
+      "grad_norm": 2.305964189521301,
+      "learning_rate": 1.4635683742833798e-05,
+      "loss": 0.9384,
+      "step": 3499
+    },
+    {
+      "epoch": 0.3663003663003663,
+      "grad_norm": 1.9289338620569885,
+      "learning_rate": 1.4632679977212965e-05,
+      "loss": 0.9879,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3664050235478807,
+      "grad_norm": 2.134260629081426,
+      "learning_rate": 1.4629675679287814e-05,
+      "loss": 0.9906,
+      "step": 3501
+    },
+    {
+      "epoch": 0.3665096807953951,
+      "grad_norm": 2.125024277582387,
+      "learning_rate": 1.4626670849403541e-05,
+      "loss": 0.8551,
+      "step": 3502
+    },
+    {
+      "epoch": 0.3666143380429095,
+      "grad_norm": 2.014078167441046,
+      "learning_rate": 1.4623665487905409e-05,
+      "loss": 0.9331,
+      "step": 3503
+    },
+    {
+      "epoch": 0.36671899529042384,
+      "grad_norm": 2.3595548560106674,
+      "learning_rate": 1.462065959513874e-05,
+      "loss": 1.0379,
+      "step": 3504
+    },
+    {
+      "epoch": 0.36682365253793825,
+      "grad_norm": 2.9757237588936847,
+      "learning_rate": 1.4617653171448915e-05,
+      "loss": 1.1066,
+      "step": 3505
+    },
+    {
+      "epoch": 0.36692830978545266,
+      "grad_norm": 2.246060097728677,
+      "learning_rate": 1.4614646217181384e-05,
+      "loss": 1.0206,
+      "step": 3506
+    },
+    {
+      "epoch": 0.367032967032967,
+      "grad_norm": 2.1279065491723577,
+      "learning_rate": 1.461163873268164e-05,
+      "loss": 0.9343,
+      "step": 3507
+    },
+    {
+      "epoch": 0.3671376242804814,
+      "grad_norm": 2.0617313415147582,
+      "learning_rate": 1.4608630718295262e-05,
+      "loss": 1.0097,
+      "step": 3508
+    },
+    {
+      "epoch": 0.3672422815279958,
+      "grad_norm": 1.8166947694326565,
+      "learning_rate": 1.4605622174367873e-05,
+      "loss": 1.0153,
+      "step": 3509
+    },
+    {
+      "epoch": 0.3673469387755102,
+      "grad_norm": 2.1056303565410115,
+      "learning_rate": 1.4602613101245156e-05,
+      "loss": 1.0769,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3674515960230246,
+      "grad_norm": 2.229739943150304,
+      "learning_rate": 1.4599603499272866e-05,
+      "loss": 1.0299,
+      "step": 3511
+    },
+    {
+      "epoch": 0.367556253270539,
+      "grad_norm": 2.1505713190979674,
+      "learning_rate": 1.4596593368796808e-05,
+      "loss": 1.0079,
+      "step": 3512
+    },
+    {
+      "epoch": 0.3676609105180534,
+      "grad_norm": 2.0137320566247445,
+      "learning_rate": 1.4593582710162854e-05,
+      "loss": 0.9401,
+      "step": 3513
+    },
+    {
+      "epoch": 0.3677655677655678,
+      "grad_norm": 1.97738685057115,
+      "learning_rate": 1.4590571523716932e-05,
+      "loss": 0.9479,
+      "step": 3514
+    },
+    {
+      "epoch": 0.36787022501308214,
+      "grad_norm": 2.1775568508800305,
+      "learning_rate": 1.4587559809805041e-05,
+      "loss": 0.8715,
+      "step": 3515
+    },
+    {
+      "epoch": 0.36797488226059655,
+      "grad_norm": 2.4652646105763663,
+      "learning_rate": 1.4584547568773225e-05,
+      "loss": 1.0586,
+      "step": 3516
+    },
+    {
+      "epoch": 0.36807953950811095,
+      "grad_norm": 2.2880185364374355,
+      "learning_rate": 1.4581534800967598e-05,
+      "loss": 0.9568,
+      "step": 3517
+    },
+    {
+      "epoch": 0.3681841967556253,
+      "grad_norm": 2.0626244211775986,
+      "learning_rate": 1.4578521506734337e-05,
+      "loss": 0.8189,
+      "step": 3518
+    },
+    {
+      "epoch": 0.3682888540031397,
+      "grad_norm": 2.0228867528318966,
+      "learning_rate": 1.4575507686419672e-05,
+      "loss": 0.9727,
+      "step": 3519
+    },
+    {
+      "epoch": 0.3683935112506541,
+      "grad_norm": 1.9220254744007452,
+      "learning_rate": 1.4572493340369899e-05,
+      "loss": 1.0084,
+      "step": 3520
+    },
+    {
+      "epoch": 0.3684981684981685,
+      "grad_norm": 2.2566198672292392,
+      "learning_rate": 1.4569478468931371e-05,
+      "loss": 1.0256,
+      "step": 3521
+    },
+    {
+      "epoch": 0.36860282574568287,
+      "grad_norm": 2.180361824104086,
+      "learning_rate": 1.4566463072450503e-05,
+      "loss": 1.044,
+      "step": 3522
+    },
+    {
+      "epoch": 0.3687074829931973,
+      "grad_norm": 2.623792129539169,
+      "learning_rate": 1.4563447151273774e-05,
+      "loss": 0.9307,
+      "step": 3523
+    },
+    {
+      "epoch": 0.3688121402407117,
+      "grad_norm": 2.4668276543972047,
+      "learning_rate": 1.4560430705747712e-05,
+      "loss": 0.9877,
+      "step": 3524
+    },
+    {
+      "epoch": 0.36891679748822603,
+      "grad_norm": 2.454886424734209,
+      "learning_rate": 1.4557413736218921e-05,
+      "loss": 0.8058,
+      "step": 3525
+    },
+    {
+      "epoch": 0.36902145473574044,
+      "grad_norm": 2.1383002692744557,
+      "learning_rate": 1.4554396243034052e-05,
+      "loss": 0.9637,
+      "step": 3526
+    },
+    {
+      "epoch": 0.36912611198325485,
+      "grad_norm": 2.187789113145796,
+      "learning_rate": 1.4551378226539821e-05,
+      "loss": 1.011,
+      "step": 3527
+    },
+    {
+      "epoch": 0.36923076923076925,
+      "grad_norm": 1.7592035312864471,
+      "learning_rate": 1.4548359687083005e-05,
+      "loss": 0.8919,
+      "step": 3528
+    },
+    {
+      "epoch": 0.3693354264782836,
+      "grad_norm": 2.1794650803355373,
+      "learning_rate": 1.454534062501044e-05,
+      "loss": 0.9867,
+      "step": 3529
+    },
+    {
+      "epoch": 0.369440083725798,
+      "grad_norm": 1.9320597793347067,
+      "learning_rate": 1.4542321040669025e-05,
+      "loss": 0.9989,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3695447409733124,
+      "grad_norm": 1.9447012593805288,
+      "learning_rate": 1.4539300934405712e-05,
+      "loss": 0.9174,
+      "step": 3531
+    },
+    {
+      "epoch": 0.3696493982208268,
+      "grad_norm": 2.161309649226642,
+      "learning_rate": 1.4536280306567521e-05,
+      "loss": 1.0729,
+      "step": 3532
+    },
+    {
+      "epoch": 0.36975405546834117,
+      "grad_norm": 2.199055254223592,
+      "learning_rate": 1.4533259157501523e-05,
+      "loss": 0.9474,
+      "step": 3533
+    },
+    {
+      "epoch": 0.3698587127158556,
+      "grad_norm": 1.9653608066864128,
+      "learning_rate": 1.4530237487554863e-05,
+      "loss": 0.8146,
+      "step": 3534
+    },
+    {
+      "epoch": 0.36996336996337,
+      "grad_norm": 2.0938517524245803,
+      "learning_rate": 1.4527215297074729e-05,
+      "loss": 1.0321,
+      "step": 3535
+    },
+    {
+      "epoch": 0.37006802721088433,
+      "grad_norm": 2.2264094228758307,
+      "learning_rate": 1.452419258640838e-05,
+      "loss": 1.068,
+      "step": 3536
+    },
+    {
+      "epoch": 0.37017268445839874,
+      "grad_norm": 2.003519287512863,
+      "learning_rate": 1.4521169355903134e-05,
+      "loss": 1.0125,
+      "step": 3537
+    },
+    {
+      "epoch": 0.37027734170591314,
+      "grad_norm": 2.06288913711494,
+      "learning_rate": 1.451814560590636e-05,
+      "loss": 1.0217,
+      "step": 3538
+    },
+    {
+      "epoch": 0.37038199895342755,
+      "grad_norm": 2.32869949729458,
+      "learning_rate": 1.4515121336765499e-05,
+      "loss": 0.9788,
+      "step": 3539
+    },
+    {
+      "epoch": 0.3704866562009419,
+      "grad_norm": 2.2629636033796503,
+      "learning_rate": 1.4512096548828041e-05,
+      "loss": 1.0602,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3705913134484563,
+      "grad_norm": 1.9185846266270503,
+      "learning_rate": 1.4509071242441543e-05,
+      "loss": 0.9762,
+      "step": 3541
+    },
+    {
+      "epoch": 0.3706959706959707,
+      "grad_norm": 2.2625877749763172,
+      "learning_rate": 1.4506045417953623e-05,
+      "loss": 1.0291,
+      "step": 3542
+    },
+    {
+      "epoch": 0.37080062794348506,
+      "grad_norm": 1.809617016801832,
+      "learning_rate": 1.4503019075711944e-05,
+      "loss": 0.9528,
+      "step": 3543
+    },
+    {
+      "epoch": 0.37090528519099947,
+      "grad_norm": 2.002346966236511,
+      "learning_rate": 1.4499992216064247e-05,
+      "loss": 0.8509,
+      "step": 3544
+    },
+    {
+      "epoch": 0.3710099424385139,
+      "grad_norm": 2.0335315203538697,
+      "learning_rate": 1.4496964839358322e-05,
+      "loss": 0.976,
+      "step": 3545
+    },
+    {
+      "epoch": 0.3711145996860283,
+      "grad_norm": 2.438768221681832,
+      "learning_rate": 1.449393694594202e-05,
+      "loss": 0.9592,
+      "step": 3546
+    },
+    {
+      "epoch": 0.37121925693354263,
+      "grad_norm": 2.188091743315195,
+      "learning_rate": 1.4490908536163256e-05,
+      "loss": 1.0567,
+      "step": 3547
+    },
+    {
+      "epoch": 0.37132391418105704,
+      "grad_norm": 2.0144776071078168,
+      "learning_rate": 1.448787961036999e-05,
+      "loss": 1.0472,
+      "step": 3548
+    },
+    {
+      "epoch": 0.37142857142857144,
+      "grad_norm": 1.9614612256190855,
+      "learning_rate": 1.4484850168910264e-05,
+      "loss": 0.8886,
+      "step": 3549
+    },
+    {
+      "epoch": 0.3715332286760858,
+      "grad_norm": 1.8138847376496081,
+      "learning_rate": 1.4481820212132163e-05,
+      "loss": 0.9632,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3716378859236002,
+      "grad_norm": 2.2730125431428863,
+      "learning_rate": 1.4478789740383831e-05,
+      "loss": 0.9063,
+      "step": 3551
+    },
+    {
+      "epoch": 0.3717425431711146,
+      "grad_norm": 2.4175931682424947,
+      "learning_rate": 1.4475758754013481e-05,
+      "loss": 1.0576,
+      "step": 3552
+    },
+    {
+      "epoch": 0.371847200418629,
+      "grad_norm": 1.9765613042579417,
+      "learning_rate": 1.4472727253369374e-05,
+      "loss": 0.9173,
+      "step": 3553
+    },
+    {
+      "epoch": 0.37195185766614336,
+      "grad_norm": 2.105898251066037,
+      "learning_rate": 1.4469695238799841e-05,
+      "loss": 1.0151,
+      "step": 3554
+    },
+    {
+      "epoch": 0.37205651491365777,
+      "grad_norm": 2.063365637058352,
+      "learning_rate": 1.4466662710653261e-05,
+      "loss": 0.9934,
+      "step": 3555
+    },
+    {
+      "epoch": 0.37216117216117217,
+      "grad_norm": 2.16238034321692,
+      "learning_rate": 1.4463629669278083e-05,
+      "loss": 1.095,
+      "step": 3556
+    },
+    {
+      "epoch": 0.3722658294086866,
+      "grad_norm": 2.064593186035993,
+      "learning_rate": 1.4460596115022808e-05,
+      "loss": 0.93,
+      "step": 3557
+    },
+    {
+      "epoch": 0.3723704866562009,
+      "grad_norm": 2.1031716067377406,
+      "learning_rate": 1.4457562048235993e-05,
+      "loss": 1.1012,
+      "step": 3558
+    },
+    {
+      "epoch": 0.37247514390371533,
+      "grad_norm": 2.040768861352751,
+      "learning_rate": 1.4454527469266269e-05,
+      "loss": 1.0625,
+      "step": 3559
+    },
+    {
+      "epoch": 0.37257980115122974,
+      "grad_norm": 2.1485703086705508,
+      "learning_rate": 1.4451492378462302e-05,
+      "loss": 1.119,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3726844583987441,
+      "grad_norm": 1.8403984678229408,
+      "learning_rate": 1.444845677617284e-05,
+      "loss": 0.9352,
+      "step": 3561
+    },
+    {
+      "epoch": 0.3727891156462585,
+      "grad_norm": 1.922757309168295,
+      "learning_rate": 1.4445420662746674e-05,
+      "loss": 1.0356,
+      "step": 3562
+    },
+    {
+      "epoch": 0.3728937728937729,
+      "grad_norm": 2.1213514343363564,
+      "learning_rate": 1.4442384038532664e-05,
+      "loss": 1.0686,
+      "step": 3563
+    },
+    {
+      "epoch": 0.3729984301412873,
+      "grad_norm": 2.112512593103979,
+      "learning_rate": 1.4439346903879724e-05,
+      "loss": 0.9844,
+      "step": 3564
+    },
+    {
+      "epoch": 0.37310308738880166,
+      "grad_norm": 2.1935515618029577,
+      "learning_rate": 1.443630925913682e-05,
+      "loss": 0.9552,
+      "step": 3565
+    },
+    {
+      "epoch": 0.37320774463631606,
+      "grad_norm": 1.963564429493465,
+      "learning_rate": 1.4433271104652993e-05,
+      "loss": 1.0212,
+      "step": 3566
+    },
+    {
+      "epoch": 0.37331240188383047,
+      "grad_norm": 2.1958400025455855,
+      "learning_rate": 1.4430232440777325e-05,
+      "loss": 1.0752,
+      "step": 3567
+    },
+    {
+      "epoch": 0.3734170591313448,
+      "grad_norm": 2.321242717364229,
+      "learning_rate": 1.4427193267858974e-05,
+      "loss": 0.9584,
+      "step": 3568
+    },
+    {
+      "epoch": 0.3735217163788592,
+      "grad_norm": 2.4134787950833174,
+      "learning_rate": 1.4424153586247139e-05,
+      "loss": 1.075,
+      "step": 3569
+    },
+    {
+      "epoch": 0.37362637362637363,
+      "grad_norm": 2.204301556211592,
+      "learning_rate": 1.4421113396291087e-05,
+      "loss": 1.0884,
+      "step": 3570
+    },
+    {
+      "epoch": 0.37373103087388804,
+      "grad_norm": 2.4179458922764874,
+      "learning_rate": 1.4418072698340147e-05,
+      "loss": 0.9831,
+      "step": 3571
+    },
+    {
+      "epoch": 0.3738356881214024,
+      "grad_norm": 2.07945845423551,
+      "learning_rate": 1.4415031492743692e-05,
+      "loss": 1.1162,
+      "step": 3572
+    },
+    {
+      "epoch": 0.3739403453689168,
+      "grad_norm": 2.383328131104694,
+      "learning_rate": 1.4411989779851175e-05,
+      "loss": 0.9214,
+      "step": 3573
+    },
+    {
+      "epoch": 0.3740450026164312,
+      "grad_norm": 2.811893175065896,
+      "learning_rate": 1.4408947560012081e-05,
+      "loss": 0.9125,
+      "step": 3574
+    },
+    {
+      "epoch": 0.3741496598639456,
+      "grad_norm": 2.177298784002885,
+      "learning_rate": 1.4405904833575979e-05,
+      "loss": 0.99,
+      "step": 3575
+    },
+    {
+      "epoch": 0.37425431711145996,
+      "grad_norm": 2.1489316661311437,
+      "learning_rate": 1.440286160089248e-05,
+      "loss": 1.0198,
+      "step": 3576
+    },
+    {
+      "epoch": 0.37435897435897436,
+      "grad_norm": 2.2742667087048436,
+      "learning_rate": 1.4399817862311255e-05,
+      "loss": 1.0415,
+      "step": 3577
+    },
+    {
+      "epoch": 0.37446363160648877,
+      "grad_norm": 2.2904104266630103,
+      "learning_rate": 1.439677361818204e-05,
+      "loss": 0.9554,
+      "step": 3578
+    },
+    {
+      "epoch": 0.3745682888540031,
+      "grad_norm": 2.0120333801365846,
+      "learning_rate": 1.4393728868854622e-05,
+      "loss": 0.9557,
+      "step": 3579
+    },
+    {
+      "epoch": 0.3746729461015175,
+      "grad_norm": 1.97838166233458,
+      "learning_rate": 1.4390683614678851e-05,
+      "loss": 0.9937,
+      "step": 3580
+    },
+    {
+      "epoch": 0.37477760334903193,
+      "grad_norm": 2.1059022035617803,
+      "learning_rate": 1.4387637856004633e-05,
+      "loss": 1.0074,
+      "step": 3581
+    },
+    {
+      "epoch": 0.37488226059654634,
+      "grad_norm": 2.0354065733173403,
+      "learning_rate": 1.438459159318193e-05,
+      "loss": 0.9762,
+      "step": 3582
+    },
+    {
+      "epoch": 0.3749869178440607,
+      "grad_norm": 2.2000318232269924,
+      "learning_rate": 1.4381544826560762e-05,
+      "loss": 0.9475,
+      "step": 3583
+    },
+    {
+      "epoch": 0.3750915750915751,
+      "grad_norm": 1.928430649502382,
+      "learning_rate": 1.4378497556491212e-05,
+      "loss": 0.8959,
+      "step": 3584
+    },
+    {
+      "epoch": 0.3751962323390895,
+      "grad_norm": 2.0181021559315897,
+      "learning_rate": 1.4375449783323419e-05,
+      "loss": 0.8834,
+      "step": 3585
+    },
+    {
+      "epoch": 0.37530088958660385,
+      "grad_norm": 2.179626276000838,
+      "learning_rate": 1.4372401507407575e-05,
+      "loss": 0.8578,
+      "step": 3586
+    },
+    {
+      "epoch": 0.37540554683411825,
+      "grad_norm": 1.9333891804031262,
+      "learning_rate": 1.436935272909393e-05,
+      "loss": 1.0699,
+      "step": 3587
+    },
+    {
+      "epoch": 0.37551020408163266,
+      "grad_norm": 1.9551604770253188,
+      "learning_rate": 1.4366303448732804e-05,
+      "loss": 0.9671,
+      "step": 3588
+    },
+    {
+      "epoch": 0.37561486132914707,
+      "grad_norm": 1.9683480948127803,
+      "learning_rate": 1.4363253666674559e-05,
+      "loss": 0.8561,
+      "step": 3589
+    },
+    {
+      "epoch": 0.3757195185766614,
+      "grad_norm": 2.2367501994279526,
+      "learning_rate": 1.4360203383269622e-05,
+      "loss": 0.8811,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3758241758241758,
+      "grad_norm": 2.108577595208127,
+      "learning_rate": 1.4357152598868478e-05,
+      "loss": 1.0762,
+      "step": 3591
+    },
+    {
+      "epoch": 0.3759288330716902,
+      "grad_norm": 2.0801774600295357,
+      "learning_rate": 1.4354101313821666e-05,
+      "loss": 1.0351,
+      "step": 3592
+    },
+    {
+      "epoch": 0.3760334903192046,
+      "grad_norm": 1.9951115073973749,
+      "learning_rate": 1.4351049528479788e-05,
+      "loss": 0.9729,
+      "step": 3593
+    },
+    {
+      "epoch": 0.376138147566719,
+      "grad_norm": 2.254132484905743,
+      "learning_rate": 1.4347997243193497e-05,
+      "loss": 0.9553,
+      "step": 3594
+    },
+    {
+      "epoch": 0.3762428048142334,
+      "grad_norm": 2.2575494533039846,
+      "learning_rate": 1.434494445831351e-05,
+      "loss": 0.9477,
+      "step": 3595
+    },
+    {
+      "epoch": 0.3763474620617478,
+      "grad_norm": 2.411830867087975,
+      "learning_rate": 1.4341891174190594e-05,
+      "loss": 0.8515,
+      "step": 3596
+    },
+    {
+      "epoch": 0.37645211930926215,
+      "grad_norm": 2.0691126406810563,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.9937,
+      "step": 3597
+    },
+    {
+      "epoch": 0.37655677655677655,
+      "grad_norm": 1.8560655060436033,
+      "learning_rate": 1.4335783109619356e-05,
+      "loss": 0.9589,
+      "step": 3598
+    },
+    {
+      "epoch": 0.37666143380429096,
+      "grad_norm": 1.9064645496752881,
+      "learning_rate": 1.4332728329872867e-05,
+      "loss": 1.0487,
+      "step": 3599
+    },
+    {
+      "epoch": 0.37676609105180536,
+      "grad_norm": 2.4222992304653923,
+      "learning_rate": 1.4329673052287105e-05,
+      "loss": 0.9782,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3768707482993197,
+      "grad_norm": 2.0131408426702126,
+      "learning_rate": 1.432661727721313e-05,
+      "loss": 1.0806,
+      "step": 3601
+    },
+    {
+      "epoch": 0.3769754055468341,
+      "grad_norm": 2.0235190034405948,
+      "learning_rate": 1.4323561005002064e-05,
+      "loss": 0.8947,
+      "step": 3602
+    },
+    {
+      "epoch": 0.3770800627943485,
+      "grad_norm": 2.1458509322659545,
+      "learning_rate": 1.432050423600507e-05,
+      "loss": 1.0682,
+      "step": 3603
+    },
+    {
+      "epoch": 0.3771847200418629,
+      "grad_norm": 2.266595492288038,
+      "learning_rate": 1.4317446970573386e-05,
+      "loss": 0.9339,
+      "step": 3604
+    },
+    {
+      "epoch": 0.3772893772893773,
+      "grad_norm": 1.7601596273055828,
+      "learning_rate": 1.4314389209058287e-05,
+      "loss": 0.7858,
+      "step": 3605
+    },
+    {
+      "epoch": 0.3773940345368917,
+      "grad_norm": 1.9387994291788024,
+      "learning_rate": 1.4311330951811125e-05,
+      "loss": 0.9226,
+      "step": 3606
+    },
+    {
+      "epoch": 0.3774986917844061,
+      "grad_norm": 1.8920940023031685,
+      "learning_rate": 1.4308272199183296e-05,
+      "loss": 0.9885,
+      "step": 3607
+    },
+    {
+      "epoch": 0.37760334903192044,
+      "grad_norm": 2.058831017637809,
+      "learning_rate": 1.4305212951526255e-05,
+      "loss": 0.8887,
+      "step": 3608
+    },
+    {
+      "epoch": 0.37770800627943485,
+      "grad_norm": 2.025314142101822,
+      "learning_rate": 1.4302153209191518e-05,
+      "loss": 0.9362,
+      "step": 3609
+    },
+    {
+      "epoch": 0.37781266352694926,
+      "grad_norm": 2.202887178692797,
+      "learning_rate": 1.4299092972530656e-05,
+      "loss": 0.9994,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3779173207744636,
+      "grad_norm": 2.1307137717760694,
+      "learning_rate": 1.42960322418953e-05,
+      "loss": 1.0086,
+      "step": 3611
+    },
+    {
+      "epoch": 0.378021978021978,
+      "grad_norm": 2.23112028651418,
+      "learning_rate": 1.4292971017637131e-05,
+      "loss": 1.0339,
+      "step": 3612
+    },
+    {
+      "epoch": 0.3781266352694924,
+      "grad_norm": 1.826253404688636,
+      "learning_rate": 1.4289909300107885e-05,
+      "loss": 0.8473,
+      "step": 3613
+    },
+    {
+      "epoch": 0.3782312925170068,
+      "grad_norm": 2.428885841856068,
+      "learning_rate": 1.4286847089659368e-05,
+      "loss": 1.001,
+      "step": 3614
+    },
+    {
+      "epoch": 0.3783359497645212,
+      "grad_norm": 1.8607496884224413,
+      "learning_rate": 1.4283784386643427e-05,
+      "loss": 1.0102,
+      "step": 3615
+    },
+    {
+      "epoch": 0.3784406070120356,
+      "grad_norm": 2.101434244434794,
+      "learning_rate": 1.4280721191411976e-05,
+      "loss": 1.0508,
+      "step": 3616
+    },
+    {
+      "epoch": 0.37854526425955,
+      "grad_norm": 1.9702696695086308,
+      "learning_rate": 1.4277657504316984e-05,
+      "loss": 0.9761,
+      "step": 3617
+    },
+    {
+      "epoch": 0.3786499215070644,
+      "grad_norm": 2.189847545110545,
+      "learning_rate": 1.427459332571047e-05,
+      "loss": 0.9961,
+      "step": 3618
+    },
+    {
+      "epoch": 0.37875457875457874,
+      "grad_norm": 2.30294474388609,
+      "learning_rate": 1.4271528655944522e-05,
+      "loss": 0.9599,
+      "step": 3619
+    },
+    {
+      "epoch": 0.37885923600209315,
+      "grad_norm": 1.8696799148064482,
+      "learning_rate": 1.4268463495371267e-05,
+      "loss": 1.0181,
+      "step": 3620
+    },
+    {
+      "epoch": 0.37896389324960755,
+      "grad_norm": 2.09128819824349,
+      "learning_rate": 1.426539784434291e-05,
+      "loss": 1.0629,
+      "step": 3621
+    },
+    {
+      "epoch": 0.3790685504971219,
+      "grad_norm": 2.0599948614085815,
+      "learning_rate": 1.4262331703211686e-05,
+      "loss": 0.9998,
+      "step": 3622
+    },
+    {
+      "epoch": 0.3791732077446363,
+      "grad_norm": 2.05462296115126,
+      "learning_rate": 1.4259265072329914e-05,
+      "loss": 0.8353,
+      "step": 3623
+    },
+    {
+      "epoch": 0.3792778649921507,
+      "grad_norm": 2.4818761789564814,
+      "learning_rate": 1.425619795204995e-05,
+      "loss": 0.885,
+      "step": 3624
+    },
+    {
+      "epoch": 0.3793825222396651,
+      "grad_norm": 1.9669887610252532,
+      "learning_rate": 1.425313034272421e-05,
+      "loss": 1.0076,
+      "step": 3625
+    },
+    {
+      "epoch": 0.37948717948717947,
+      "grad_norm": 2.4144006548804295,
+      "learning_rate": 1.4250062244705174e-05,
+      "loss": 1.0021,
+      "step": 3626
+    },
+    {
+      "epoch": 0.3795918367346939,
+      "grad_norm": 2.298079044624755,
+      "learning_rate": 1.4246993658345369e-05,
+      "loss": 0.8731,
+      "step": 3627
+    },
+    {
+      "epoch": 0.3796964939822083,
+      "grad_norm": 1.8240722916104797,
+      "learning_rate": 1.4243924583997386e-05,
+      "loss": 0.9051,
+      "step": 3628
+    },
+    {
+      "epoch": 0.37980115122972263,
+      "grad_norm": 2.5476300949512387,
+      "learning_rate": 1.4240855022013863e-05,
+      "loss": 1.1031,
+      "step": 3629
+    },
+    {
+      "epoch": 0.37990580847723704,
+      "grad_norm": 1.7527472576385803,
+      "learning_rate": 1.4237784972747501e-05,
+      "loss": 0.8347,
+      "step": 3630
+    },
+    {
+      "epoch": 0.38001046572475145,
+      "grad_norm": 2.0657909815224618,
+      "learning_rate": 1.4234714436551053e-05,
+      "loss": 0.853,
+      "step": 3631
+    },
+    {
+      "epoch": 0.38011512297226585,
+      "grad_norm": 2.0203292973651283,
+      "learning_rate": 1.4231643413777333e-05,
+      "loss": 0.9679,
+      "step": 3632
+    },
+    {
+      "epoch": 0.3802197802197802,
+      "grad_norm": 2.0882649368291553,
+      "learning_rate": 1.4228571904779209e-05,
+      "loss": 1.0813,
+      "step": 3633
+    },
+    {
+      "epoch": 0.3803244374672946,
+      "grad_norm": 2.1962421392589837,
+      "learning_rate": 1.4225499909909597e-05,
+      "loss": 1.0255,
+      "step": 3634
+    },
+    {
+      "epoch": 0.380429094714809,
+      "grad_norm": 2.0096595971812405,
+      "learning_rate": 1.4222427429521482e-05,
+      "loss": 0.9536,
+      "step": 3635
+    },
+    {
+      "epoch": 0.38053375196232336,
+      "grad_norm": 2.349899372814253,
+      "learning_rate": 1.4219354463967893e-05,
+      "loss": 0.9828,
+      "step": 3636
+    },
+    {
+      "epoch": 0.38063840920983777,
+      "grad_norm": 2.15612266191841,
+      "learning_rate": 1.4216281013601926e-05,
+      "loss": 0.8592,
+      "step": 3637
+    },
+    {
+      "epoch": 0.3807430664573522,
+      "grad_norm": 2.1616922957234164,
+      "learning_rate": 1.4213207078776723e-05,
+      "loss": 1.0161,
+      "step": 3638
+    },
+    {
+      "epoch": 0.3808477237048666,
+      "grad_norm": 2.0271980981355813,
+      "learning_rate": 1.4210132659845482e-05,
+      "loss": 0.8895,
+      "step": 3639
+    },
+    {
+      "epoch": 0.38095238095238093,
+      "grad_norm": 1.8793729122503329,
+      "learning_rate": 1.4207057757161465e-05,
+      "loss": 0.9433,
+      "step": 3640
+    },
+    {
+      "epoch": 0.38105703819989534,
+      "grad_norm": 2.2584716002700422,
+      "learning_rate": 1.4203982371077984e-05,
+      "loss": 0.9715,
+      "step": 3641
+    },
+    {
+      "epoch": 0.38116169544740974,
+      "grad_norm": 1.9923565504390046,
+      "learning_rate": 1.4200906501948405e-05,
+      "loss": 0.8118,
+      "step": 3642
+    },
+    {
+      "epoch": 0.38126635269492415,
+      "grad_norm": 2.3266543948061065,
+      "learning_rate": 1.4197830150126155e-05,
+      "loss": 1.0415,
+      "step": 3643
+    },
+    {
+      "epoch": 0.3813710099424385,
+      "grad_norm": 1.9354684638961517,
+      "learning_rate": 1.4194753315964707e-05,
+      "loss": 0.9977,
+      "step": 3644
+    },
+    {
+      "epoch": 0.3814756671899529,
+      "grad_norm": 2.0521405009628513,
+      "learning_rate": 1.4191675999817603e-05,
+      "loss": 0.9696,
+      "step": 3645
+    },
+    {
+      "epoch": 0.3815803244374673,
+      "grad_norm": 2.2085561140812775,
+      "learning_rate": 1.4188598202038428e-05,
+      "loss": 1.0361,
+      "step": 3646
+    },
+    {
+      "epoch": 0.38168498168498166,
+      "grad_norm": 2.447907112557791,
+      "learning_rate": 1.418551992298083e-05,
+      "loss": 1.0262,
+      "step": 3647
+    },
+    {
+      "epoch": 0.38178963893249607,
+      "grad_norm": 2.083006037799446,
+      "learning_rate": 1.4182441162998506e-05,
+      "loss": 0.9471,
+      "step": 3648
+    },
+    {
+      "epoch": 0.3818942961800105,
+      "grad_norm": 2.2012013601209466,
+      "learning_rate": 1.4179361922445214e-05,
+      "loss": 1.0669,
+      "step": 3649
+    },
+    {
+      "epoch": 0.3819989534275249,
+      "grad_norm": 2.356165760065059,
+      "learning_rate": 1.4176282201674764e-05,
+      "loss": 1.0236,
+      "step": 3650
+    },
+    {
+      "epoch": 0.38210361067503923,
+      "grad_norm": 2.1910213266631553,
+      "learning_rate": 1.4173202001041024e-05,
+      "loss": 0.9729,
+      "step": 3651
+    },
+    {
+      "epoch": 0.38220826792255364,
+      "grad_norm": 1.9619249365232363,
+      "learning_rate": 1.4170121320897911e-05,
+      "loss": 0.9726,
+      "step": 3652
+    },
+    {
+      "epoch": 0.38231292517006804,
+      "grad_norm": 2.7128459699808483,
+      "learning_rate": 1.416704016159941e-05,
+      "loss": 1.0244,
+      "step": 3653
+    },
+    {
+      "epoch": 0.3824175824175824,
+      "grad_norm": 1.9243530614458784,
+      "learning_rate": 1.4163958523499543e-05,
+      "loss": 0.8521,
+      "step": 3654
+    },
+    {
+      "epoch": 0.3825222396650968,
+      "grad_norm": 2.139320327183821,
+      "learning_rate": 1.4160876406952402e-05,
+      "loss": 0.9358,
+      "step": 3655
+    },
+    {
+      "epoch": 0.3826268969126112,
+      "grad_norm": 1.8395056038544089,
+      "learning_rate": 1.4157793812312126e-05,
+      "loss": 0.9061,
+      "step": 3656
+    },
+    {
+      "epoch": 0.3827315541601256,
+      "grad_norm": 2.094018606545468,
+      "learning_rate": 1.4154710739932913e-05,
+      "loss": 0.9147,
+      "step": 3657
+    },
+    {
+      "epoch": 0.38283621140763996,
+      "grad_norm": 2.0930445989438553,
+      "learning_rate": 1.4151627190169013e-05,
+      "loss": 0.9932,
+      "step": 3658
+    },
+    {
+      "epoch": 0.38294086865515437,
+      "grad_norm": 2.143172865705711,
+      "learning_rate": 1.4148543163374733e-05,
+      "loss": 0.9458,
+      "step": 3659
+    },
+    {
+      "epoch": 0.38304552590266877,
+      "grad_norm": 2.2091982738545033,
+      "learning_rate": 1.4145458659904433e-05,
+      "loss": 0.9759,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3831501831501832,
+      "grad_norm": 1.9567813439206125,
+      "learning_rate": 1.4142373680112528e-05,
+      "loss": 0.9427,
+      "step": 3661
+    },
+    {
+      "epoch": 0.3832548403976975,
+      "grad_norm": 2.041638652930433,
+      "learning_rate": 1.4139288224353494e-05,
+      "loss": 0.9669,
+      "step": 3662
+    },
+    {
+      "epoch": 0.38335949764521193,
+      "grad_norm": 2.194450474916531,
+      "learning_rate": 1.4136202292981848e-05,
+      "loss": 0.8689,
+      "step": 3663
+    },
+    {
+      "epoch": 0.38346415489272634,
+      "grad_norm": 2.048372715974301,
+      "learning_rate": 1.4133115886352176e-05,
+      "loss": 0.8711,
+      "step": 3664
+    },
+    {
+      "epoch": 0.3835688121402407,
+      "grad_norm": 2.081252667466177,
+      "learning_rate": 1.4130029004819109e-05,
+      "loss": 0.9163,
+      "step": 3665
+    },
+    {
+      "epoch": 0.3836734693877551,
+      "grad_norm": 2.063327599869014,
+      "learning_rate": 1.4126941648737334e-05,
+      "loss": 0.9385,
+      "step": 3666
+    },
+    {
+      "epoch": 0.3837781266352695,
+      "grad_norm": 2.2359367310037133,
+      "learning_rate": 1.4123853818461601e-05,
+      "loss": 1.1035,
+      "step": 3667
+    },
+    {
+      "epoch": 0.3838827838827839,
+      "grad_norm": 1.7634923657138273,
+      "learning_rate": 1.4120765514346703e-05,
+      "loss": 0.8968,
+      "step": 3668
+    },
+    {
+      "epoch": 0.38398744113029826,
+      "grad_norm": 2.141693015056442,
+      "learning_rate": 1.4117676736747494e-05,
+      "loss": 0.921,
+      "step": 3669
+    },
+    {
+      "epoch": 0.38409209837781266,
+      "grad_norm": 2.1476056016940035,
+      "learning_rate": 1.411458748601888e-05,
+      "loss": 1.0029,
+      "step": 3670
+    },
+    {
+      "epoch": 0.38419675562532707,
+      "grad_norm": 1.9647392959113228,
+      "learning_rate": 1.4111497762515822e-05,
+      "loss": 0.9642,
+      "step": 3671
+    },
+    {
+      "epoch": 0.3843014128728414,
+      "grad_norm": 2.5121210210965006,
+      "learning_rate": 1.410840756659334e-05,
+      "loss": 1.1309,
+      "step": 3672
+    },
+    {
+      "epoch": 0.3844060701203558,
+      "grad_norm": 1.9954842133931197,
+      "learning_rate": 1.4105316898606499e-05,
+      "loss": 0.9308,
+      "step": 3673
+    },
+    {
+      "epoch": 0.38451072736787023,
+      "grad_norm": 2.1044462151613135,
+      "learning_rate": 1.410222575891042e-05,
+      "loss": 0.8752,
+      "step": 3674
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 2.0953131458808962,
+      "learning_rate": 1.4099134147860287e-05,
+      "loss": 0.6864,
+      "step": 3675
+    },
+    {
+      "epoch": 0.384720041862899,
+      "grad_norm": 2.0201216369124335,
+      "learning_rate": 1.4096042065811333e-05,
+      "loss": 1.0058,
+      "step": 3676
+    },
+    {
+      "epoch": 0.3848246991104134,
+      "grad_norm": 2.1104902948779034,
+      "learning_rate": 1.409294951311884e-05,
+      "loss": 0.9358,
+      "step": 3677
+    },
+    {
+      "epoch": 0.3849293563579278,
+      "grad_norm": 2.0782997552273783,
+      "learning_rate": 1.408985649013815e-05,
+      "loss": 1.0495,
+      "step": 3678
+    },
+    {
+      "epoch": 0.38503401360544215,
+      "grad_norm": 2.2354543394222226,
+      "learning_rate": 1.4086762997224658e-05,
+      "loss": 1.028,
+      "step": 3679
+    },
+    {
+      "epoch": 0.38513867085295656,
+      "grad_norm": 2.344608781601967,
+      "learning_rate": 1.4083669034733816e-05,
+      "loss": 1.1165,
+      "step": 3680
+    },
+    {
+      "epoch": 0.38524332810047096,
+      "grad_norm": 2.093693999681073,
+      "learning_rate": 1.4080574603021121e-05,
+      "loss": 0.9505,
+      "step": 3681
+    },
+    {
+      "epoch": 0.38534798534798537,
+      "grad_norm": 1.9846470070251416,
+      "learning_rate": 1.4077479702442132e-05,
+      "loss": 0.9975,
+      "step": 3682
+    },
+    {
+      "epoch": 0.3854526425954997,
+      "grad_norm": 1.9333444382455947,
+      "learning_rate": 1.407438433335246e-05,
+      "loss": 0.8526,
+      "step": 3683
+    },
+    {
+      "epoch": 0.3855572998430141,
+      "grad_norm": 2.020798413551147,
+      "learning_rate": 1.4071288496107769e-05,
+      "loss": 0.8515,
+      "step": 3684
+    },
+    {
+      "epoch": 0.38566195709052853,
+      "grad_norm": 2.008663061169858,
+      "learning_rate": 1.4068192191063777e-05,
+      "loss": 0.9006,
+      "step": 3685
+    },
+    {
+      "epoch": 0.38576661433804293,
+      "grad_norm": 2.117370458308192,
+      "learning_rate": 1.4065095418576254e-05,
+      "loss": 1.0621,
+      "step": 3686
+    },
+    {
+      "epoch": 0.3858712715855573,
+      "grad_norm": 2.4013394801880836,
+      "learning_rate": 1.4061998179001025e-05,
+      "loss": 1.0932,
+      "step": 3687
+    },
+    {
+      "epoch": 0.3859759288330717,
+      "grad_norm": 2.1143080149970603,
+      "learning_rate": 1.4058900472693973e-05,
+      "loss": 0.8957,
+      "step": 3688
+    },
+    {
+      "epoch": 0.3860805860805861,
+      "grad_norm": 2.075156804250162,
+      "learning_rate": 1.4055802300011027e-05,
+      "loss": 1.0239,
+      "step": 3689
+    },
+    {
+      "epoch": 0.38618524332810045,
+      "grad_norm": 2.8275247517173216,
+      "learning_rate": 1.4052703661308175e-05,
+      "loss": 1.0158,
+      "step": 3690
+    },
+    {
+      "epoch": 0.38628990057561485,
+      "grad_norm": 1.7962929732068007,
+      "learning_rate": 1.4049604556941457e-05,
+      "loss": 0.924,
+      "step": 3691
+    },
+    {
+      "epoch": 0.38639455782312926,
+      "grad_norm": 2.01783824353475,
+      "learning_rate": 1.4046504987266964e-05,
+      "loss": 0.9401,
+      "step": 3692
+    },
+    {
+      "epoch": 0.38649921507064366,
+      "grad_norm": 2.2187966441047435,
+      "learning_rate": 1.4043404952640848e-05,
+      "loss": 1.0663,
+      "step": 3693
+    },
+    {
+      "epoch": 0.386603872318158,
+      "grad_norm": 1.999279925714374,
+      "learning_rate": 1.4040304453419306e-05,
+      "loss": 0.9407,
+      "step": 3694
+    },
+    {
+      "epoch": 0.3867085295656724,
+      "grad_norm": 2.130348742018532,
+      "learning_rate": 1.4037203489958593e-05,
+      "loss": 0.9873,
+      "step": 3695
+    },
+    {
+      "epoch": 0.3868131868131868,
+      "grad_norm": 2.325652550191666,
+      "learning_rate": 1.4034102062615009e-05,
+      "loss": 0.8994,
+      "step": 3696
+    },
+    {
+      "epoch": 0.3869178440607012,
+      "grad_norm": 2.3197905613136345,
+      "learning_rate": 1.4031000171744924e-05,
+      "loss": 0.9815,
+      "step": 3697
+    },
+    {
+      "epoch": 0.3870225013082156,
+      "grad_norm": 2.2641945333562385,
+      "learning_rate": 1.402789781770475e-05,
+      "loss": 1.0339,
+      "step": 3698
+    },
+    {
+      "epoch": 0.38712715855573,
+      "grad_norm": 2.3574842229562614,
+      "learning_rate": 1.4024795000850946e-05,
+      "loss": 0.9534,
+      "step": 3699
+    },
+    {
+      "epoch": 0.3872318158032444,
+      "grad_norm": 2.0918637615880566,
+      "learning_rate": 1.4021691721540038e-05,
+      "loss": 0.9808,
+      "step": 3700
+    },
+    {
+      "epoch": 0.38733647305075875,
+      "grad_norm": 2.075992424556234,
+      "learning_rate": 1.4018587980128602e-05,
+      "loss": 0.8702,
+      "step": 3701
+    },
+    {
+      "epoch": 0.38744113029827315,
+      "grad_norm": 1.8782927333898265,
+      "learning_rate": 1.401548377697326e-05,
+      "loss": 1.0075,
+      "step": 3702
+    },
+    {
+      "epoch": 0.38754578754578756,
+      "grad_norm": 2.083887164652835,
+      "learning_rate": 1.4012379112430692e-05,
+      "loss": 1.0557,
+      "step": 3703
+    },
+    {
+      "epoch": 0.38765044479330196,
+      "grad_norm": 2.389475853756546,
+      "learning_rate": 1.4009273986857625e-05,
+      "loss": 0.9441,
+      "step": 3704
+    },
+    {
+      "epoch": 0.3877551020408163,
+      "grad_norm": 2.1864008731163795,
+      "learning_rate": 1.4006168400610853e-05,
+      "loss": 0.9095,
+      "step": 3705
+    },
+    {
+      "epoch": 0.3878597592883307,
+      "grad_norm": 2.1399945231613624,
+      "learning_rate": 1.4003062354047211e-05,
+      "loss": 1.0264,
+      "step": 3706
+    },
+    {
+      "epoch": 0.3879644165358451,
+      "grad_norm": 2.039186746148691,
+      "learning_rate": 1.3999955847523588e-05,
+      "loss": 1.0078,
+      "step": 3707
+    },
+    {
+      "epoch": 0.3880690737833595,
+      "grad_norm": 1.7528376320399341,
+      "learning_rate": 1.3996848881396932e-05,
+      "loss": 0.7449,
+      "step": 3708
+    },
+    {
+      "epoch": 0.3881737310308739,
+      "grad_norm": 2.4937767704582265,
+      "learning_rate": 1.3993741456024233e-05,
+      "loss": 0.9109,
+      "step": 3709
+    },
+    {
+      "epoch": 0.3882783882783883,
+      "grad_norm": 2.129798206261498,
+      "learning_rate": 1.3990633571762547e-05,
+      "loss": 1.1029,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3883830455259027,
+      "grad_norm": 1.9957488140578195,
+      "learning_rate": 1.3987525228968972e-05,
+      "loss": 1.018,
+      "step": 3711
+    },
+    {
+      "epoch": 0.38848770277341704,
+      "grad_norm": 2.1354640527218947,
+      "learning_rate": 1.398441642800067e-05,
+      "loss": 0.8997,
+      "step": 3712
+    },
+    {
+      "epoch": 0.38859236002093145,
+      "grad_norm": 1.964424076142884,
+      "learning_rate": 1.3981307169214837e-05,
+      "loss": 0.9455,
+      "step": 3713
+    },
+    {
+      "epoch": 0.38869701726844585,
+      "grad_norm": 2.221101115356642,
+      "learning_rate": 1.397819745296874e-05,
+      "loss": 1.0268,
+      "step": 3714
+    },
+    {
+      "epoch": 0.3888016745159602,
+      "grad_norm": 2.3833418030736553,
+      "learning_rate": 1.3975087279619695e-05,
+      "loss": 1.04,
+      "step": 3715
+    },
+    {
+      "epoch": 0.3889063317634746,
+      "grad_norm": 1.9463675227198336,
+      "learning_rate": 1.397197664952506e-05,
+      "loss": 0.9108,
+      "step": 3716
+    },
+    {
+      "epoch": 0.389010989010989,
+      "grad_norm": 2.104522166089339,
+      "learning_rate": 1.3968865563042255e-05,
+      "loss": 1.0029,
+      "step": 3717
+    },
+    {
+      "epoch": 0.3891156462585034,
+      "grad_norm": 2.2293325863509623,
+      "learning_rate": 1.3965754020528755e-05,
+      "loss": 1.0006,
+      "step": 3718
+    },
+    {
+      "epoch": 0.3892203035060178,
+      "grad_norm": 2.225977481709003,
+      "learning_rate": 1.3962642022342075e-05,
+      "loss": 0.9976,
+      "step": 3719
+    },
+    {
+      "epoch": 0.3893249607535322,
+      "grad_norm": 2.1371606373643983,
+      "learning_rate": 1.3959529568839796e-05,
+      "loss": 0.9584,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3894296180010466,
+      "grad_norm": 2.1984571676095217,
+      "learning_rate": 1.3956416660379542e-05,
+      "loss": 0.9966,
+      "step": 3721
+    },
+    {
+      "epoch": 0.389534275248561,
+      "grad_norm": 2.3612095316369994,
+      "learning_rate": 1.3953303297318992e-05,
+      "loss": 1.0549,
+      "step": 3722
+    },
+    {
+      "epoch": 0.38963893249607534,
+      "grad_norm": 2.002770628808304,
+      "learning_rate": 1.3950189480015879e-05,
+      "loss": 0.9726,
+      "step": 3723
+    },
+    {
+      "epoch": 0.38974358974358975,
+      "grad_norm": 2.2409687498155817,
+      "learning_rate": 1.3947075208827986e-05,
+      "loss": 1.051,
+      "step": 3724
+    },
+    {
+      "epoch": 0.38984824699110415,
+      "grad_norm": 2.4369275610874417,
+      "learning_rate": 1.3943960484113154e-05,
+      "loss": 1.0222,
+      "step": 3725
+    },
+    {
+      "epoch": 0.3899529042386185,
+      "grad_norm": 2.001381883429874,
+      "learning_rate": 1.3940845306229263e-05,
+      "loss": 0.8523,
+      "step": 3726
+    },
+    {
+      "epoch": 0.3900575614861329,
+      "grad_norm": 2.379959624189031,
+      "learning_rate": 1.3937729675534259e-05,
+      "loss": 1.0068,
+      "step": 3727
+    },
+    {
+      "epoch": 0.3901622187336473,
+      "grad_norm": 2.065993568806048,
+      "learning_rate": 1.3934613592386133e-05,
+      "loss": 0.9928,
+      "step": 3728
+    },
+    {
+      "epoch": 0.3902668759811617,
+      "grad_norm": 2.14687057918789,
+      "learning_rate": 1.393149705714293e-05,
+      "loss": 0.9085,
+      "step": 3729
+    },
+    {
+      "epoch": 0.39037153322867607,
+      "grad_norm": 2.056457087459079,
+      "learning_rate": 1.3928380070162743e-05,
+      "loss": 1.031,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3904761904761905,
+      "grad_norm": 2.3376645905376447,
+      "learning_rate": 1.3925262631803722e-05,
+      "loss": 1.0316,
+      "step": 3731
+    },
+    {
+      "epoch": 0.3905808477237049,
+      "grad_norm": 2.345112786800701,
+      "learning_rate": 1.392214474242407e-05,
+      "loss": 1.0802,
+      "step": 3732
+    },
+    {
+      "epoch": 0.39068550497121923,
+      "grad_norm": 2.0352455265227443,
+      "learning_rate": 1.3919026402382034e-05,
+      "loss": 0.9909,
+      "step": 3733
+    },
+    {
+      "epoch": 0.39079016221873364,
+      "grad_norm": 2.0459097619442863,
+      "learning_rate": 1.3915907612035922e-05,
+      "loss": 0.9397,
+      "step": 3734
+    },
+    {
+      "epoch": 0.39089481946624804,
+      "grad_norm": 1.954972542902261,
+      "learning_rate": 1.3912788371744084e-05,
+      "loss": 0.825,
+      "step": 3735
+    },
+    {
+      "epoch": 0.39099947671376245,
+      "grad_norm": 2.1145543797342983,
+      "learning_rate": 1.3909668681864932e-05,
+      "loss": 1.0745,
+      "step": 3736
+    },
+    {
+      "epoch": 0.3911041339612768,
+      "grad_norm": 1.9636359750174326,
+      "learning_rate": 1.3906548542756921e-05,
+      "loss": 0.9003,
+      "step": 3737
+    },
+    {
+      "epoch": 0.3912087912087912,
+      "grad_norm": 2.405810011459009,
+      "learning_rate": 1.3903427954778566e-05,
+      "loss": 1.0238,
+      "step": 3738
+    },
+    {
+      "epoch": 0.3913134484563056,
+      "grad_norm": 2.0934967739240813,
+      "learning_rate": 1.3900306918288423e-05,
+      "loss": 1.004,
+      "step": 3739
+    },
+    {
+      "epoch": 0.39141810570381996,
+      "grad_norm": 2.017817533341966,
+      "learning_rate": 1.3897185433645106e-05,
+      "loss": 0.9562,
+      "step": 3740
+    },
+    {
+      "epoch": 0.39152276295133437,
+      "grad_norm": 2.0048073269684865,
+      "learning_rate": 1.389406350120729e-05,
+      "loss": 0.9614,
+      "step": 3741
+    },
+    {
+      "epoch": 0.3916274201988488,
+      "grad_norm": 2.1093352540533985,
+      "learning_rate": 1.3890941121333677e-05,
+      "loss": 0.9876,
+      "step": 3742
+    },
+    {
+      "epoch": 0.3917320774463632,
+      "grad_norm": 2.221682557807781,
+      "learning_rate": 1.3887818294383042e-05,
+      "loss": 0.9904,
+      "step": 3743
+    },
+    {
+      "epoch": 0.39183673469387753,
+      "grad_norm": 1.9422817754452815,
+      "learning_rate": 1.3884695020714207e-05,
+      "loss": 0.8561,
+      "step": 3744
+    },
+    {
+      "epoch": 0.39194139194139194,
+      "grad_norm": 2.080159971701426,
+      "learning_rate": 1.3881571300686037e-05,
+      "loss": 0.9193,
+      "step": 3745
+    },
+    {
+      "epoch": 0.39204604918890634,
+      "grad_norm": 2.1895432862916078,
+      "learning_rate": 1.3878447134657455e-05,
+      "loss": 0.9276,
+      "step": 3746
+    },
+    {
+      "epoch": 0.39215070643642075,
+      "grad_norm": 2.154912897139779,
+      "learning_rate": 1.3875322522987435e-05,
+      "loss": 0.9417,
+      "step": 3747
+    },
+    {
+      "epoch": 0.3922553636839351,
+      "grad_norm": 2.375072450587214,
+      "learning_rate": 1.3872197466035003e-05,
+      "loss": 1.0635,
+      "step": 3748
+    },
+    {
+      "epoch": 0.3923600209314495,
+      "grad_norm": 2.029749571414169,
+      "learning_rate": 1.3869071964159231e-05,
+      "loss": 0.9101,
+      "step": 3749
+    },
+    {
+      "epoch": 0.3924646781789639,
+      "grad_norm": 2.298693925354746,
+      "learning_rate": 1.3865946017719248e-05,
+      "loss": 1.0388,
+      "step": 3750
+    },
+    {
+      "epoch": 0.39256933542647826,
+      "grad_norm": 2.32567539400059,
+      "learning_rate": 1.3862819627074231e-05,
+      "loss": 0.9207,
+      "step": 3751
+    },
+    {
+      "epoch": 0.39267399267399267,
+      "grad_norm": 1.8738585362137903,
+      "learning_rate": 1.3859692792583403e-05,
+      "loss": 0.9748,
+      "step": 3752
+    },
+    {
+      "epoch": 0.3927786499215071,
+      "grad_norm": 1.903263299759873,
+      "learning_rate": 1.3856565514606053e-05,
+      "loss": 1.0709,
+      "step": 3753
+    },
+    {
+      "epoch": 0.3928833071690215,
+      "grad_norm": 2.1106708790964377,
+      "learning_rate": 1.3853437793501507e-05,
+      "loss": 1.0031,
+      "step": 3754
+    },
+    {
+      "epoch": 0.39298796441653583,
+      "grad_norm": 2.0628055794677493,
+      "learning_rate": 1.3850309629629146e-05,
+      "loss": 0.998,
+      "step": 3755
+    },
+    {
+      "epoch": 0.39309262166405023,
+      "grad_norm": 1.8840425280715498,
+      "learning_rate": 1.3847181023348404e-05,
+      "loss": 0.9089,
+      "step": 3756
+    },
+    {
+      "epoch": 0.39319727891156464,
+      "grad_norm": 2.2547660697868332,
+      "learning_rate": 1.3844051975018761e-05,
+      "loss": 0.9776,
+      "step": 3757
+    },
+    {
+      "epoch": 0.393301936159079,
+      "grad_norm": 1.9372101963139208,
+      "learning_rate": 1.3840922484999758e-05,
+      "loss": 0.9781,
+      "step": 3758
+    },
+    {
+      "epoch": 0.3934065934065934,
+      "grad_norm": 2.3442712849451657,
+      "learning_rate": 1.383779255365097e-05,
+      "loss": 0.9706,
+      "step": 3759
+    },
+    {
+      "epoch": 0.3935112506541078,
+      "grad_norm": 1.9728314083799356,
+      "learning_rate": 1.3834662181332043e-05,
+      "loss": 0.8493,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3936159079016222,
+      "grad_norm": 1.8671773883426177,
+      "learning_rate": 1.3831531368402653e-05,
+      "loss": 0.8621,
+      "step": 3761
+    },
+    {
+      "epoch": 0.39372056514913656,
+      "grad_norm": 2.0851610975093,
+      "learning_rate": 1.3828400115222542e-05,
+      "loss": 0.9945,
+      "step": 3762
+    },
+    {
+      "epoch": 0.39382522239665096,
+      "grad_norm": 2.0950765767067296,
+      "learning_rate": 1.38252684221515e-05,
+      "loss": 0.9923,
+      "step": 3763
+    },
+    {
+      "epoch": 0.39392987964416537,
+      "grad_norm": 2.0426883514777616,
+      "learning_rate": 1.382213628954936e-05,
+      "loss": 1.0127,
+      "step": 3764
+    },
+    {
+      "epoch": 0.3940345368916798,
+      "grad_norm": 2.1796168252887713,
+      "learning_rate": 1.381900371777601e-05,
+      "loss": 1.0253,
+      "step": 3765
+    },
+    {
+      "epoch": 0.3941391941391941,
+      "grad_norm": 2.056669859459506,
+      "learning_rate": 1.3815870707191393e-05,
+      "loss": 1.0314,
+      "step": 3766
+    },
+    {
+      "epoch": 0.39424385138670853,
+      "grad_norm": 2.058977648719983,
+      "learning_rate": 1.3812737258155495e-05,
+      "loss": 0.8932,
+      "step": 3767
+    },
+    {
+      "epoch": 0.39434850863422294,
+      "grad_norm": 2.3166421400138177,
+      "learning_rate": 1.380960337102836e-05,
+      "loss": 0.964,
+      "step": 3768
+    },
+    {
+      "epoch": 0.3944531658817373,
+      "grad_norm": 1.8919186197649047,
+      "learning_rate": 1.380646904617007e-05,
+      "loss": 0.7797,
+      "step": 3769
+    },
+    {
+      "epoch": 0.3945578231292517,
+      "grad_norm": 2.0859503146939256,
+      "learning_rate": 1.3803334283940772e-05,
+      "loss": 0.9242,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3946624803767661,
+      "grad_norm": 1.9003991213190858,
+      "learning_rate": 1.3800199084700655e-05,
+      "loss": 0.9279,
+      "step": 3771
+    },
+    {
+      "epoch": 0.3947671376242805,
+      "grad_norm": 1.9199813086679065,
+      "learning_rate": 1.3797063448809959e-05,
+      "loss": 0.8576,
+      "step": 3772
+    },
+    {
+      "epoch": 0.39487179487179486,
+      "grad_norm": 1.984278812039716,
+      "learning_rate": 1.3793927376628977e-05,
+      "loss": 0.9722,
+      "step": 3773
+    },
+    {
+      "epoch": 0.39497645211930926,
+      "grad_norm": 2.1654990774470426,
+      "learning_rate": 1.3790790868518044e-05,
+      "loss": 0.8915,
+      "step": 3774
+    },
+    {
+      "epoch": 0.39508110936682367,
+      "grad_norm": 2.0693477910040152,
+      "learning_rate": 1.378765392483756e-05,
+      "loss": 1.0609,
+      "step": 3775
+    },
+    {
+      "epoch": 0.395185766614338,
+      "grad_norm": 2.5214519932464263,
+      "learning_rate": 1.378451654594796e-05,
+      "loss": 0.8346,
+      "step": 3776
+    },
+    {
+      "epoch": 0.3952904238618524,
+      "grad_norm": 2.273499618027654,
+      "learning_rate": 1.3781378732209738e-05,
+      "loss": 0.971,
+      "step": 3777
+    },
+    {
+      "epoch": 0.39539508110936683,
+      "grad_norm": 2.0346748171851594,
+      "learning_rate": 1.3778240483983432e-05,
+      "loss": 0.921,
+      "step": 3778
+    },
+    {
+      "epoch": 0.39549973835688124,
+      "grad_norm": 2.4181856266842705,
+      "learning_rate": 1.3775101801629636e-05,
+      "loss": 0.9929,
+      "step": 3779
+    },
+    {
+      "epoch": 0.3956043956043956,
+      "grad_norm": 1.8889123693724994,
+      "learning_rate": 1.3771962685508991e-05,
+      "loss": 0.9226,
+      "step": 3780
+    },
+    {
+      "epoch": 0.39570905285191,
+      "grad_norm": 1.8892857747949212,
+      "learning_rate": 1.3768823135982185e-05,
+      "loss": 0.9675,
+      "step": 3781
+    },
+    {
+      "epoch": 0.3958137100994244,
+      "grad_norm": 2.421740842709229,
+      "learning_rate": 1.3765683153409963e-05,
+      "loss": 0.9961,
+      "step": 3782
+    },
+    {
+      "epoch": 0.39591836734693875,
+      "grad_norm": 2.19282905248247,
+      "learning_rate": 1.376254273815311e-05,
+      "loss": 1.045,
+      "step": 3783
+    },
+    {
+      "epoch": 0.39602302459445315,
+      "grad_norm": 2.6404415016056695,
+      "learning_rate": 1.375940189057247e-05,
+      "loss": 1.0346,
+      "step": 3784
+    },
+    {
+      "epoch": 0.39612768184196756,
+      "grad_norm": 2.087172492105292,
+      "learning_rate": 1.3756260611028932e-05,
+      "loss": 1.0005,
+      "step": 3785
+    },
+    {
+      "epoch": 0.39623233908948197,
+      "grad_norm": 1.8576562979717508,
+      "learning_rate": 1.3753118899883436e-05,
+      "loss": 0.8769,
+      "step": 3786
+    },
+    {
+      "epoch": 0.3963369963369963,
+      "grad_norm": 1.7251243282419155,
+      "learning_rate": 1.3749976757496969e-05,
+      "loss": 0.9917,
+      "step": 3787
+    },
+    {
+      "epoch": 0.3964416535845107,
+      "grad_norm": 2.3431283462971244,
+      "learning_rate": 1.3746834184230569e-05,
+      "loss": 0.9063,
+      "step": 3788
+    },
+    {
+      "epoch": 0.39654631083202513,
+      "grad_norm": 2.100375176753719,
+      "learning_rate": 1.3743691180445328e-05,
+      "loss": 1.056,
+      "step": 3789
+    },
+    {
+      "epoch": 0.39665096807953953,
+      "grad_norm": 1.883781822130877,
+      "learning_rate": 1.374054774650238e-05,
+      "loss": 0.9573,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3967556253270539,
+      "grad_norm": 1.9710349136351437,
+      "learning_rate": 1.3737403882762914e-05,
+      "loss": 0.9161,
+      "step": 3791
+    },
+    {
+      "epoch": 0.3968602825745683,
+      "grad_norm": 2.0619488055452067,
+      "learning_rate": 1.3734259589588165e-05,
+      "loss": 0.8067,
+      "step": 3792
+    },
+    {
+      "epoch": 0.3969649398220827,
+      "grad_norm": 2.1265304547515607,
+      "learning_rate": 1.3731114867339418e-05,
+      "loss": 1.1066,
+      "step": 3793
+    },
+    {
+      "epoch": 0.39706959706959705,
+      "grad_norm": 2.1845388692035135,
+      "learning_rate": 1.3727969716378012e-05,
+      "loss": 1.0085,
+      "step": 3794
+    },
+    {
+      "epoch": 0.39717425431711145,
+      "grad_norm": 2.025263339801749,
+      "learning_rate": 1.3724824137065325e-05,
+      "loss": 0.9922,
+      "step": 3795
+    },
+    {
+      "epoch": 0.39727891156462586,
+      "grad_norm": 1.9103665780889987,
+      "learning_rate": 1.3721678129762792e-05,
+      "loss": 1.0307,
+      "step": 3796
+    },
+    {
+      "epoch": 0.39738356881214026,
+      "grad_norm": 2.420546254970621,
+      "learning_rate": 1.3718531694831903e-05,
+      "loss": 0.9924,
+      "step": 3797
+    },
+    {
+      "epoch": 0.3974882260596546,
+      "grad_norm": 1.9578789743491487,
+      "learning_rate": 1.371538483263418e-05,
+      "loss": 0.9418,
+      "step": 3798
+    },
+    {
+      "epoch": 0.397592883307169,
+      "grad_norm": 2.2999551242197107,
+      "learning_rate": 1.3712237543531208e-05,
+      "loss": 1.0449,
+      "step": 3799
+    },
+    {
+      "epoch": 0.3976975405546834,
+      "grad_norm": 1.9021993595587816,
+      "learning_rate": 1.3709089827884616e-05,
+      "loss": 0.8744,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3978021978021978,
+      "grad_norm": 2.0695970711361444,
+      "learning_rate": 1.3705941686056086e-05,
+      "loss": 1.0607,
+      "step": 3801
+    },
+    {
+      "epoch": 0.3979068550497122,
+      "grad_norm": 1.8591241528827835,
+      "learning_rate": 1.3702793118407345e-05,
+      "loss": 0.9284,
+      "step": 3802
+    },
+    {
+      "epoch": 0.3980115122972266,
+      "grad_norm": 1.87633082654073,
+      "learning_rate": 1.3699644125300165e-05,
+      "loss": 0.9277,
+      "step": 3803
+    },
+    {
+      "epoch": 0.398116169544741,
+      "grad_norm": 2.1413866117330675,
+      "learning_rate": 1.3696494707096374e-05,
+      "loss": 0.9777,
+      "step": 3804
+    },
+    {
+      "epoch": 0.39822082679225534,
+      "grad_norm": 2.029453889878876,
+      "learning_rate": 1.369334486415785e-05,
+      "loss": 0.9102,
+      "step": 3805
+    },
+    {
+      "epoch": 0.39832548403976975,
+      "grad_norm": 1.914840282215036,
+      "learning_rate": 1.3690194596846516e-05,
+      "loss": 0.8982,
+      "step": 3806
+    },
+    {
+      "epoch": 0.39843014128728416,
+      "grad_norm": 2.2156107459198267,
+      "learning_rate": 1.3687043905524337e-05,
+      "loss": 0.9374,
+      "step": 3807
+    },
+    {
+      "epoch": 0.39853479853479856,
+      "grad_norm": 2.353831603694141,
+      "learning_rate": 1.3683892790553344e-05,
+      "loss": 1.0299,
+      "step": 3808
+    },
+    {
+      "epoch": 0.3986394557823129,
+      "grad_norm": 2.2201207812404333,
+      "learning_rate": 1.3680741252295597e-05,
+      "loss": 0.9126,
+      "step": 3809
+    },
+    {
+      "epoch": 0.3987441130298273,
+      "grad_norm": 2.0962471812380006,
+      "learning_rate": 1.367758929111322e-05,
+      "loss": 0.9077,
+      "step": 3810
+    },
+    {
+      "epoch": 0.3988487702773417,
+      "grad_norm": 2.21964726775872,
+      "learning_rate": 1.3674436907368377e-05,
+      "loss": 1.1087,
+      "step": 3811
+    },
+    {
+      "epoch": 0.3989534275248561,
+      "grad_norm": 2.5819871421432286,
+      "learning_rate": 1.3671284101423288e-05,
+      "loss": 0.9415,
+      "step": 3812
+    },
+    {
+      "epoch": 0.3990580847723705,
+      "grad_norm": 1.7198628116479615,
+      "learning_rate": 1.3668130873640211e-05,
+      "loss": 0.8898,
+      "step": 3813
+    },
+    {
+      "epoch": 0.3991627420198849,
+      "grad_norm": 1.8597498193101791,
+      "learning_rate": 1.3664977224381461e-05,
+      "loss": 0.8606,
+      "step": 3814
+    },
+    {
+      "epoch": 0.3992673992673993,
+      "grad_norm": 2.6740292842251234,
+      "learning_rate": 1.3661823154009397e-05,
+      "loss": 1.0387,
+      "step": 3815
+    },
+    {
+      "epoch": 0.39937205651491364,
+      "grad_norm": 2.3003617051512735,
+      "learning_rate": 1.3658668662886432e-05,
+      "loss": 1.1021,
+      "step": 3816
+    },
+    {
+      "epoch": 0.39947671376242805,
+      "grad_norm": 2.3043271725583963,
+      "learning_rate": 1.3655513751375017e-05,
+      "loss": 0.9864,
+      "step": 3817
+    },
+    {
+      "epoch": 0.39958137100994245,
+      "grad_norm": 1.741403290306905,
+      "learning_rate": 1.3652358419837664e-05,
+      "loss": 0.9179,
+      "step": 3818
+    },
+    {
+      "epoch": 0.3996860282574568,
+      "grad_norm": 2.143710219932102,
+      "learning_rate": 1.3649202668636923e-05,
+      "loss": 0.8958,
+      "step": 3819
+    },
+    {
+      "epoch": 0.3997906855049712,
+      "grad_norm": 2.029828508661703,
+      "learning_rate": 1.36460464981354e-05,
+      "loss": 0.9657,
+      "step": 3820
+    },
+    {
+      "epoch": 0.3998953427524856,
+      "grad_norm": 1.9833644803425685,
+      "learning_rate": 1.3642889908695742e-05,
+      "loss": 0.9026,
+      "step": 3821
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.210053513135352,
+      "learning_rate": 1.3639732900680646e-05,
+      "loss": 1.0516,
+      "step": 3822
+    },
+    {
+      "epoch": 0.4001046572475144,
+      "grad_norm": 2.3239837565552985,
+      "learning_rate": 1.3636575474452865e-05,
+      "loss": 0.9058,
+      "step": 3823
+    },
+    {
+      "epoch": 0.4002093144950288,
+      "grad_norm": 2.0848417519232454,
+      "learning_rate": 1.3633417630375188e-05,
+      "loss": 0.9769,
+      "step": 3824
+    },
+    {
+      "epoch": 0.4003139717425432,
+      "grad_norm": 1.953543514114223,
+      "learning_rate": 1.3630259368810461e-05,
+      "loss": 0.9432,
+      "step": 3825
+    },
+    {
+      "epoch": 0.40041862899005753,
+      "grad_norm": 2.1198900948291928,
+      "learning_rate": 1.3627100690121571e-05,
+      "loss": 0.9108,
+      "step": 3826
+    },
+    {
+      "epoch": 0.40052328623757194,
+      "grad_norm": 2.1614969325672377,
+      "learning_rate": 1.362394159467146e-05,
+      "loss": 0.9297,
+      "step": 3827
+    },
+    {
+      "epoch": 0.40062794348508635,
+      "grad_norm": 2.2955139319444418,
+      "learning_rate": 1.3620782082823115e-05,
+      "loss": 0.9899,
+      "step": 3828
+    },
+    {
+      "epoch": 0.40073260073260075,
+      "grad_norm": 2.2734494900620947,
+      "learning_rate": 1.3617622154939565e-05,
+      "loss": 0.8636,
+      "step": 3829
+    },
+    {
+      "epoch": 0.4008372579801151,
+      "grad_norm": 2.3826089612311607,
+      "learning_rate": 1.3614461811383897e-05,
+      "loss": 0.9787,
+      "step": 3830
+    },
+    {
+      "epoch": 0.4009419152276295,
+      "grad_norm": 2.233808465500224,
+      "learning_rate": 1.3611301052519242e-05,
+      "loss": 1.1001,
+      "step": 3831
+    },
+    {
+      "epoch": 0.4010465724751439,
+      "grad_norm": 2.4947957206362976,
+      "learning_rate": 1.360813987870877e-05,
+      "loss": 1.0562,
+      "step": 3832
+    },
+    {
+      "epoch": 0.4011512297226583,
+      "grad_norm": 2.1160921112573807,
+      "learning_rate": 1.3604978290315717e-05,
+      "loss": 0.9986,
+      "step": 3833
+    },
+    {
+      "epoch": 0.40125588697017267,
+      "grad_norm": 2.310532165642057,
+      "learning_rate": 1.360181628770335e-05,
+      "loss": 0.9502,
+      "step": 3834
+    },
+    {
+      "epoch": 0.4013605442176871,
+      "grad_norm": 2.3190655097380106,
+      "learning_rate": 1.3598653871234986e-05,
+      "loss": 1.0915,
+      "step": 3835
+    },
+    {
+      "epoch": 0.4014652014652015,
+      "grad_norm": 2.108855272952508,
+      "learning_rate": 1.3595491041273999e-05,
+      "loss": 0.8137,
+      "step": 3836
+    },
+    {
+      "epoch": 0.40156985871271583,
+      "grad_norm": 2.0659350062129263,
+      "learning_rate": 1.3592327798183802e-05,
+      "loss": 1.0533,
+      "step": 3837
+    },
+    {
+      "epoch": 0.40167451596023024,
+      "grad_norm": 1.9843597273847278,
+      "learning_rate": 1.3589164142327863e-05,
+      "loss": 0.9426,
+      "step": 3838
+    },
+    {
+      "epoch": 0.40177917320774464,
+      "grad_norm": 2.074638091286358,
+      "learning_rate": 1.3586000074069679e-05,
+      "loss": 1.0032,
+      "step": 3839
+    },
+    {
+      "epoch": 0.40188383045525905,
+      "grad_norm": 1.8783659326755497,
+      "learning_rate": 1.3582835593772822e-05,
+      "loss": 0.9629,
+      "step": 3840
+    },
+    {
+      "epoch": 0.4019884877027734,
+      "grad_norm": 2.1806512008747188,
+      "learning_rate": 1.3579670701800893e-05,
+      "loss": 1.0359,
+      "step": 3841
+    },
+    {
+      "epoch": 0.4020931449502878,
+      "grad_norm": 1.8780201218056713,
+      "learning_rate": 1.3576505398517546e-05,
+      "loss": 0.8616,
+      "step": 3842
+    },
+    {
+      "epoch": 0.4021978021978022,
+      "grad_norm": 2.16324024690094,
+      "learning_rate": 1.3573339684286472e-05,
+      "loss": 0.9952,
+      "step": 3843
+    },
+    {
+      "epoch": 0.40230245944531656,
+      "grad_norm": 2.07630913117498,
+      "learning_rate": 1.3570173559471427e-05,
+      "loss": 1.0358,
+      "step": 3844
+    },
+    {
+      "epoch": 0.40240711669283097,
+      "grad_norm": 2.062316520805037,
+      "learning_rate": 1.3567007024436206e-05,
+      "loss": 0.9196,
+      "step": 3845
+    },
+    {
+      "epoch": 0.4025117739403454,
+      "grad_norm": 2.252816836018333,
+      "learning_rate": 1.3563840079544642e-05,
+      "loss": 0.9085,
+      "step": 3846
+    },
+    {
+      "epoch": 0.4026164311878598,
+      "grad_norm": 2.0407067151488087,
+      "learning_rate": 1.3560672725160631e-05,
+      "loss": 1.0633,
+      "step": 3847
+    },
+    {
+      "epoch": 0.40272108843537413,
+      "grad_norm": 2.1656996102091126,
+      "learning_rate": 1.3557504961648102e-05,
+      "loss": 1.046,
+      "step": 3848
+    },
+    {
+      "epoch": 0.40282574568288854,
+      "grad_norm": 1.904183686696131,
+      "learning_rate": 1.3554336789371046e-05,
+      "loss": 0.9816,
+      "step": 3849
+    },
+    {
+      "epoch": 0.40293040293040294,
+      "grad_norm": 2.0311197336923623,
+      "learning_rate": 1.3551168208693486e-05,
+      "loss": 0.9372,
+      "step": 3850
+    },
+    {
+      "epoch": 0.40303506017791735,
+      "grad_norm": 2.1048762647281403,
+      "learning_rate": 1.35479992199795e-05,
+      "loss": 1.0747,
+      "step": 3851
+    },
+    {
+      "epoch": 0.4031397174254317,
+      "grad_norm": 2.219151744165179,
+      "learning_rate": 1.3544829823593208e-05,
+      "loss": 1.0558,
+      "step": 3852
+    },
+    {
+      "epoch": 0.4032443746729461,
+      "grad_norm": 1.8701877669813198,
+      "learning_rate": 1.3541660019898784e-05,
+      "loss": 1.0087,
+      "step": 3853
+    },
+    {
+      "epoch": 0.4033490319204605,
+      "grad_norm": 1.8767505975265026,
+      "learning_rate": 1.3538489809260447e-05,
+      "loss": 0.8751,
+      "step": 3854
+    },
+    {
+      "epoch": 0.40345368916797486,
+      "grad_norm": 1.949016676761772,
+      "learning_rate": 1.3535319192042455e-05,
+      "loss": 1.0158,
+      "step": 3855
+    },
+    {
+      "epoch": 0.40355834641548927,
+      "grad_norm": 1.9762439941084444,
+      "learning_rate": 1.3532148168609123e-05,
+      "loss": 1.0097,
+      "step": 3856
+    },
+    {
+      "epoch": 0.40366300366300367,
+      "grad_norm": 2.490558941314484,
+      "learning_rate": 1.3528976739324807e-05,
+      "loss": 0.9381,
+      "step": 3857
+    },
+    {
+      "epoch": 0.4037676609105181,
+      "grad_norm": 2.3815368861074737,
+      "learning_rate": 1.3525804904553906e-05,
+      "loss": 0.994,
+      "step": 3858
+    },
+    {
+      "epoch": 0.40387231815803243,
+      "grad_norm": 2.224663411985367,
+      "learning_rate": 1.3522632664660878e-05,
+      "loss": 1.0024,
+      "step": 3859
+    },
+    {
+      "epoch": 0.40397697540554683,
+      "grad_norm": 2.070837677342153,
+      "learning_rate": 1.351946002001021e-05,
+      "loss": 0.9638,
+      "step": 3860
+    },
+    {
+      "epoch": 0.40408163265306124,
+      "grad_norm": 2.0394436878689084,
+      "learning_rate": 1.3516286970966454e-05,
+      "loss": 1.0819,
+      "step": 3861
+    },
+    {
+      "epoch": 0.4041862899005756,
+      "grad_norm": 2.104205603022299,
+      "learning_rate": 1.35131135178942e-05,
+      "loss": 0.9877,
+      "step": 3862
+    },
+    {
+      "epoch": 0.40429094714809,
+      "grad_norm": 2.2820283958841325,
+      "learning_rate": 1.3509939661158078e-05,
+      "loss": 1.0813,
+      "step": 3863
+    },
+    {
+      "epoch": 0.4043956043956044,
+      "grad_norm": 1.8165693573504802,
+      "learning_rate": 1.3506765401122774e-05,
+      "loss": 0.9544,
+      "step": 3864
+    },
+    {
+      "epoch": 0.4045002616431188,
+      "grad_norm": 1.8593322923606608,
+      "learning_rate": 1.3503590738153015e-05,
+      "loss": 0.8905,
+      "step": 3865
+    },
+    {
+      "epoch": 0.40460491889063316,
+      "grad_norm": 1.9003454201874181,
+      "learning_rate": 1.3500415672613578e-05,
+      "loss": 0.9831,
+      "step": 3866
+    },
+    {
+      "epoch": 0.40470957613814756,
+      "grad_norm": 2.0723713810408553,
+      "learning_rate": 1.3497240204869287e-05,
+      "loss": 0.9065,
+      "step": 3867
+    },
+    {
+      "epoch": 0.40481423338566197,
+      "grad_norm": 2.0143809176887264,
+      "learning_rate": 1.3494064335285004e-05,
+      "loss": 0.9393,
+      "step": 3868
+    },
+    {
+      "epoch": 0.4049188906331763,
+      "grad_norm": 2.297759622527477,
+      "learning_rate": 1.3490888064225645e-05,
+      "loss": 0.9157,
+      "step": 3869
+    },
+    {
+      "epoch": 0.4050235478806907,
+      "grad_norm": 2.047595647603498,
+      "learning_rate": 1.348771139205617e-05,
+      "loss": 1.0448,
+      "step": 3870
+    },
+    {
+      "epoch": 0.40512820512820513,
+      "grad_norm": 2.304679069365117,
+      "learning_rate": 1.3484534319141592e-05,
+      "loss": 1.0223,
+      "step": 3871
+    },
+    {
+      "epoch": 0.40523286237571954,
+      "grad_norm": 2.2989950063633264,
+      "learning_rate": 1.3481356845846951e-05,
+      "loss": 0.9272,
+      "step": 3872
+    },
+    {
+      "epoch": 0.4053375196232339,
+      "grad_norm": 2.202029868599979,
+      "learning_rate": 1.3478178972537354e-05,
+      "loss": 0.9782,
+      "step": 3873
+    },
+    {
+      "epoch": 0.4054421768707483,
+      "grad_norm": 2.271682115831657,
+      "learning_rate": 1.347500069957794e-05,
+      "loss": 0.9876,
+      "step": 3874
+    },
+    {
+      "epoch": 0.4055468341182627,
+      "grad_norm": 2.1087888782860222,
+      "learning_rate": 1.3471822027333901e-05,
+      "loss": 1.1044,
+      "step": 3875
+    },
+    {
+      "epoch": 0.4056514913657771,
+      "grad_norm": 1.9199653211400192,
+      "learning_rate": 1.3468642956170474e-05,
+      "loss": 0.9796,
+      "step": 3876
+    },
+    {
+      "epoch": 0.40575614861329146,
+      "grad_norm": 2.2590968517959737,
+      "learning_rate": 1.3465463486452942e-05,
+      "loss": 0.8456,
+      "step": 3877
+    },
+    {
+      "epoch": 0.40586080586080586,
+      "grad_norm": 2.4486677027922283,
+      "learning_rate": 1.3462283618546623e-05,
+      "loss": 1.0471,
+      "step": 3878
+    },
+    {
+      "epoch": 0.40596546310832027,
+      "grad_norm": 2.1883091875894745,
+      "learning_rate": 1.3459103352816905e-05,
+      "loss": 1.0535,
+      "step": 3879
+    },
+    {
+      "epoch": 0.4060701203558346,
+      "grad_norm": 1.9403740977737154,
+      "learning_rate": 1.3455922689629195e-05,
+      "loss": 0.9947,
+      "step": 3880
+    },
+    {
+      "epoch": 0.406174777603349,
+      "grad_norm": 2.1633648926775164,
+      "learning_rate": 1.3452741629348966e-05,
+      "loss": 0.9966,
+      "step": 3881
+    },
+    {
+      "epoch": 0.40627943485086343,
+      "grad_norm": 2.1785103292518944,
+      "learning_rate": 1.3449560172341718e-05,
+      "loss": 0.9293,
+      "step": 3882
+    },
+    {
+      "epoch": 0.40638409209837784,
+      "grad_norm": 2.0241202507761953,
+      "learning_rate": 1.344637831897302e-05,
+      "loss": 0.9471,
+      "step": 3883
+    },
+    {
+      "epoch": 0.4064887493458922,
+      "grad_norm": 2.0299069179358016,
+      "learning_rate": 1.3443196069608462e-05,
+      "loss": 0.9633,
+      "step": 3884
+    },
+    {
+      "epoch": 0.4065934065934066,
+      "grad_norm": 1.8230670234656603,
+      "learning_rate": 1.3440013424613699e-05,
+      "loss": 0.8566,
+      "step": 3885
+    },
+    {
+      "epoch": 0.406698063840921,
+      "grad_norm": 2.265586535508352,
+      "learning_rate": 1.3436830384354417e-05,
+      "loss": 1.1007,
+      "step": 3886
+    },
+    {
+      "epoch": 0.40680272108843535,
+      "grad_norm": 2.0941195839844484,
+      "learning_rate": 1.3433646949196354e-05,
+      "loss": 0.9932,
+      "step": 3887
+    },
+    {
+      "epoch": 0.40690737833594975,
+      "grad_norm": 2.0164409933200034,
+      "learning_rate": 1.3430463119505304e-05,
+      "loss": 0.9493,
+      "step": 3888
+    },
+    {
+      "epoch": 0.40701203558346416,
+      "grad_norm": 2.1172725827353522,
+      "learning_rate": 1.3427278895647083e-05,
+      "loss": 1.0876,
+      "step": 3889
+    },
+    {
+      "epoch": 0.40711669283097857,
+      "grad_norm": 2.1818693573193984,
+      "learning_rate": 1.342409427798757e-05,
+      "loss": 0.9947,
+      "step": 3890
+    },
+    {
+      "epoch": 0.4072213500784929,
+      "grad_norm": 1.8963779475096132,
+      "learning_rate": 1.3420909266892679e-05,
+      "loss": 1.092,
+      "step": 3891
+    },
+    {
+      "epoch": 0.4073260073260073,
+      "grad_norm": 2.388464219116994,
+      "learning_rate": 1.341772386272838e-05,
+      "loss": 1.0153,
+      "step": 3892
+    },
+    {
+      "epoch": 0.40743066457352173,
+      "grad_norm": 2.0886208182575445,
+      "learning_rate": 1.3414538065860685e-05,
+      "loss": 1.0086,
+      "step": 3893
+    },
+    {
+      "epoch": 0.40753532182103613,
+      "grad_norm": 1.8683253041052692,
+      "learning_rate": 1.341135187665564e-05,
+      "loss": 1.0031,
+      "step": 3894
+    },
+    {
+      "epoch": 0.4076399790685505,
+      "grad_norm": 1.9302119953502512,
+      "learning_rate": 1.340816529547935e-05,
+      "loss": 0.9738,
+      "step": 3895
+    },
+    {
+      "epoch": 0.4077446363160649,
+      "grad_norm": 2.076097227173372,
+      "learning_rate": 1.340497832269796e-05,
+      "loss": 0.9809,
+      "step": 3896
+    },
+    {
+      "epoch": 0.4078492935635793,
+      "grad_norm": 1.9832695165206116,
+      "learning_rate": 1.3401790958677653e-05,
+      "loss": 0.875,
+      "step": 3897
+    },
+    {
+      "epoch": 0.40795395081109365,
+      "grad_norm": 1.9007778657336216,
+      "learning_rate": 1.3398603203784678e-05,
+      "loss": 0.9117,
+      "step": 3898
+    },
+    {
+      "epoch": 0.40805860805860805,
+      "grad_norm": 2.1000112600332077,
+      "learning_rate": 1.3395415058385297e-05,
+      "loss": 0.9539,
+      "step": 3899
+    },
+    {
+      "epoch": 0.40816326530612246,
+      "grad_norm": 2.128028337883832,
+      "learning_rate": 1.3392226522845843e-05,
+      "loss": 0.9829,
+      "step": 3900
+    },
+    {
+      "epoch": 0.40826792255363686,
+      "grad_norm": 2.3062569400722808,
+      "learning_rate": 1.3389037597532689e-05,
+      "loss": 0.9517,
+      "step": 3901
+    },
+    {
+      "epoch": 0.4083725798011512,
+      "grad_norm": 2.2026909714583254,
+      "learning_rate": 1.3385848282812242e-05,
+      "loss": 0.9477,
+      "step": 3902
+    },
+    {
+      "epoch": 0.4084772370486656,
+      "grad_norm": 1.8784837438489583,
+      "learning_rate": 1.3382658579050964e-05,
+      "loss": 0.8672,
+      "step": 3903
+    },
+    {
+      "epoch": 0.40858189429618,
+      "grad_norm": 1.9667452879169667,
+      "learning_rate": 1.3379468486615357e-05,
+      "loss": 0.9565,
+      "step": 3904
+    },
+    {
+      "epoch": 0.4086865515436944,
+      "grad_norm": 1.9420479807212934,
+      "learning_rate": 1.3376278005871972e-05,
+      "loss": 0.8869,
+      "step": 3905
+    },
+    {
+      "epoch": 0.4087912087912088,
+      "grad_norm": 2.1557813800708057,
+      "learning_rate": 1.33730871371874e-05,
+      "loss": 0.9541,
+      "step": 3906
+    },
+    {
+      "epoch": 0.4088958660387232,
+      "grad_norm": 2.0260654319412574,
+      "learning_rate": 1.3369895880928277e-05,
+      "loss": 1.0403,
+      "step": 3907
+    },
+    {
+      "epoch": 0.4090005232862376,
+      "grad_norm": 2.1520685254880885,
+      "learning_rate": 1.3366704237461283e-05,
+      "loss": 0.9977,
+      "step": 3908
+    },
+    {
+      "epoch": 0.40910518053375194,
+      "grad_norm": 2.0017224417903794,
+      "learning_rate": 1.336351220715315e-05,
+      "loss": 1.0277,
+      "step": 3909
+    },
+    {
+      "epoch": 0.40920983778126635,
+      "grad_norm": 2.085668966136735,
+      "learning_rate": 1.3360319790370646e-05,
+      "loss": 0.9904,
+      "step": 3910
+    },
+    {
+      "epoch": 0.40931449502878076,
+      "grad_norm": 2.233810172732676,
+      "learning_rate": 1.3357126987480587e-05,
+      "loss": 1.0078,
+      "step": 3911
+    },
+    {
+      "epoch": 0.4094191522762951,
+      "grad_norm": 2.456133446509698,
+      "learning_rate": 1.335393379884983e-05,
+      "loss": 1.1043,
+      "step": 3912
+    },
+    {
+      "epoch": 0.4095238095238095,
+      "grad_norm": 1.9446442501126147,
+      "learning_rate": 1.3350740224845277e-05,
+      "loss": 0.9661,
+      "step": 3913
+    },
+    {
+      "epoch": 0.4096284667713239,
+      "grad_norm": 2.139239033329667,
+      "learning_rate": 1.3347546265833885e-05,
+      "loss": 1.0164,
+      "step": 3914
+    },
+    {
+      "epoch": 0.4097331240188383,
+      "grad_norm": 2.414435331126368,
+      "learning_rate": 1.3344351922182642e-05,
+      "loss": 0.9962,
+      "step": 3915
+    },
+    {
+      "epoch": 0.4098377812663527,
+      "grad_norm": 2.2612369566120236,
+      "learning_rate": 1.3341157194258578e-05,
+      "loss": 0.8626,
+      "step": 3916
+    },
+    {
+      "epoch": 0.4099424385138671,
+      "grad_norm": 2.007696031752954,
+      "learning_rate": 1.3337962082428782e-05,
+      "loss": 0.9339,
+      "step": 3917
+    },
+    {
+      "epoch": 0.4100470957613815,
+      "grad_norm": 2.1776335671383453,
+      "learning_rate": 1.3334766587060372e-05,
+      "loss": 0.9919,
+      "step": 3918
+    },
+    {
+      "epoch": 0.4101517530088959,
+      "grad_norm": 2.4159392572041267,
+      "learning_rate": 1.3331570708520527e-05,
+      "loss": 0.8825,
+      "step": 3919
+    },
+    {
+      "epoch": 0.41025641025641024,
+      "grad_norm": 2.2382392402884173,
+      "learning_rate": 1.3328374447176448e-05,
+      "loss": 0.9198,
+      "step": 3920
+    },
+    {
+      "epoch": 0.41036106750392465,
+      "grad_norm": 2.011649128934293,
+      "learning_rate": 1.3325177803395402e-05,
+      "loss": 0.958,
+      "step": 3921
+    },
+    {
+      "epoch": 0.41046572475143905,
+      "grad_norm": 2.1286275554802456,
+      "learning_rate": 1.3321980777544686e-05,
+      "loss": 0.9464,
+      "step": 3922
+    },
+    {
+      "epoch": 0.4105703819989534,
+      "grad_norm": 1.892321406734822,
+      "learning_rate": 1.331878336999164e-05,
+      "loss": 0.9145,
+      "step": 3923
+    },
+    {
+      "epoch": 0.4106750392464678,
+      "grad_norm": 1.8424101188930837,
+      "learning_rate": 1.3315585581103663e-05,
+      "loss": 0.8959,
+      "step": 3924
+    },
+    {
+      "epoch": 0.4107796964939822,
+      "grad_norm": 2.126780664206234,
+      "learning_rate": 1.3312387411248175e-05,
+      "loss": 1.0817,
+      "step": 3925
+    },
+    {
+      "epoch": 0.4108843537414966,
+      "grad_norm": 2.0459834656964384,
+      "learning_rate": 1.330918886079266e-05,
+      "loss": 1.0016,
+      "step": 3926
+    },
+    {
+      "epoch": 0.41098901098901097,
+      "grad_norm": 1.9966709058022425,
+      "learning_rate": 1.3305989930104639e-05,
+      "loss": 0.9498,
+      "step": 3927
+    },
+    {
+      "epoch": 0.4110936682365254,
+      "grad_norm": 1.9943061429132058,
+      "learning_rate": 1.3302790619551673e-05,
+      "loss": 1.0189,
+      "step": 3928
+    },
+    {
+      "epoch": 0.4111983254840398,
+      "grad_norm": 2.2183362311043466,
+      "learning_rate": 1.3299590929501369e-05,
+      "loss": 0.9745,
+      "step": 3929
+    },
+    {
+      "epoch": 0.41130298273155413,
+      "grad_norm": 2.1746772650891826,
+      "learning_rate": 1.3296390860321376e-05,
+      "loss": 0.9183,
+      "step": 3930
+    },
+    {
+      "epoch": 0.41140763997906854,
+      "grad_norm": 1.9839787897777303,
+      "learning_rate": 1.3293190412379392e-05,
+      "loss": 0.9328,
+      "step": 3931
+    },
+    {
+      "epoch": 0.41151229722658295,
+      "grad_norm": 2.106457836929408,
+      "learning_rate": 1.3289989586043153e-05,
+      "loss": 0.9307,
+      "step": 3932
+    },
+    {
+      "epoch": 0.41161695447409735,
+      "grad_norm": 2.00060665012128,
+      "learning_rate": 1.3286788381680444e-05,
+      "loss": 1.0349,
+      "step": 3933
+    },
+    {
+      "epoch": 0.4117216117216117,
+      "grad_norm": 2.1356259208594497,
+      "learning_rate": 1.3283586799659083e-05,
+      "loss": 1.0561,
+      "step": 3934
+    },
+    {
+      "epoch": 0.4118262689691261,
+      "grad_norm": 2.0728612956807138,
+      "learning_rate": 1.3280384840346942e-05,
+      "loss": 0.9573,
+      "step": 3935
+    },
+    {
+      "epoch": 0.4119309262166405,
+      "grad_norm": 2.114394535148644,
+      "learning_rate": 1.3277182504111936e-05,
+      "loss": 0.9407,
+      "step": 3936
+    },
+    {
+      "epoch": 0.4120355834641549,
+      "grad_norm": 2.3060951675811867,
+      "learning_rate": 1.3273979791322016e-05,
+      "loss": 1.0736,
+      "step": 3937
+    },
+    {
+      "epoch": 0.41214024071166927,
+      "grad_norm": 2.2146993223832774,
+      "learning_rate": 1.3270776702345182e-05,
+      "loss": 0.8368,
+      "step": 3938
+    },
+    {
+      "epoch": 0.4122448979591837,
+      "grad_norm": 1.9659217726233473,
+      "learning_rate": 1.3267573237549471e-05,
+      "loss": 0.9114,
+      "step": 3939
+    },
+    {
+      "epoch": 0.4123495552066981,
+      "grad_norm": 2.1795203888166546,
+      "learning_rate": 1.3264369397302974e-05,
+      "loss": 1.0584,
+      "step": 3940
+    },
+    {
+      "epoch": 0.41245421245421243,
+      "grad_norm": 2.2274974384458184,
+      "learning_rate": 1.3261165181973814e-05,
+      "loss": 0.941,
+      "step": 3941
+    },
+    {
+      "epoch": 0.41255886970172684,
+      "grad_norm": 2.0938075682129784,
+      "learning_rate": 1.3257960591930164e-05,
+      "loss": 0.9865,
+      "step": 3942
+    },
+    {
+      "epoch": 0.41266352694924124,
+      "grad_norm": 2.252476991607518,
+      "learning_rate": 1.3254755627540237e-05,
+      "loss": 0.8637,
+      "step": 3943
+    },
+    {
+      "epoch": 0.41276818419675565,
+      "grad_norm": 1.9902821855928159,
+      "learning_rate": 1.3251550289172293e-05,
+      "loss": 0.8776,
+      "step": 3944
+    },
+    {
+      "epoch": 0.41287284144427,
+      "grad_norm": 2.4524909184883628,
+      "learning_rate": 1.3248344577194625e-05,
+      "loss": 0.9399,
+      "step": 3945
+    },
+    {
+      "epoch": 0.4129774986917844,
+      "grad_norm": 2.113793753822339,
+      "learning_rate": 1.3245138491975582e-05,
+      "loss": 0.9082,
+      "step": 3946
+    },
+    {
+      "epoch": 0.4130821559392988,
+      "grad_norm": 2.122883266073302,
+      "learning_rate": 1.3241932033883544e-05,
+      "loss": 0.9503,
+      "step": 3947
+    },
+    {
+      "epoch": 0.41318681318681316,
+      "grad_norm": 1.750988968337933,
+      "learning_rate": 1.3238725203286951e-05,
+      "loss": 0.8322,
+      "step": 3948
+    },
+    {
+      "epoch": 0.41329147043432757,
+      "grad_norm": 2.1864026140069694,
+      "learning_rate": 1.323551800055426e-05,
+      "loss": 1.0516,
+      "step": 3949
+    },
+    {
+      "epoch": 0.413396127681842,
+      "grad_norm": 2.1920452942734316,
+      "learning_rate": 1.3232310426053996e-05,
+      "loss": 0.974,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4135007849293564,
+      "grad_norm": 2.531072565644497,
+      "learning_rate": 1.3229102480154708e-05,
+      "loss": 0.9761,
+      "step": 3951
+    },
+    {
+      "epoch": 0.41360544217687073,
+      "grad_norm": 2.1000618795832526,
+      "learning_rate": 1.3225894163224999e-05,
+      "loss": 1.0461,
+      "step": 3952
+    },
+    {
+      "epoch": 0.41371009942438514,
+      "grad_norm": 2.4031213533226694,
+      "learning_rate": 1.3222685475633515e-05,
+      "loss": 0.945,
+      "step": 3953
+    },
+    {
+      "epoch": 0.41381475667189954,
+      "grad_norm": 2.1941396313682606,
+      "learning_rate": 1.3219476417748934e-05,
+      "loss": 1.0891,
+      "step": 3954
+    },
+    {
+      "epoch": 0.4139194139194139,
+      "grad_norm": 1.8725259557278697,
+      "learning_rate": 1.3216266989939987e-05,
+      "loss": 0.9485,
+      "step": 3955
+    },
+    {
+      "epoch": 0.4140240711669283,
+      "grad_norm": 2.0302624935221814,
+      "learning_rate": 1.321305719257544e-05,
+      "loss": 0.9908,
+      "step": 3956
+    },
+    {
+      "epoch": 0.4141287284144427,
+      "grad_norm": 2.1705816692660274,
+      "learning_rate": 1.3209847026024112e-05,
+      "loss": 0.9744,
+      "step": 3957
+    },
+    {
+      "epoch": 0.4142333856619571,
+      "grad_norm": 1.958836116018982,
+      "learning_rate": 1.3206636490654851e-05,
+      "loss": 0.8859,
+      "step": 3958
+    },
+    {
+      "epoch": 0.41433804290947146,
+      "grad_norm": 1.9407740225892292,
+      "learning_rate": 1.3203425586836558e-05,
+      "loss": 0.9871,
+      "step": 3959
+    },
+    {
+      "epoch": 0.41444270015698587,
+      "grad_norm": 2.356610680228509,
+      "learning_rate": 1.3200214314938169e-05,
+      "loss": 0.8781,
+      "step": 3960
+    },
+    {
+      "epoch": 0.41454735740450027,
+      "grad_norm": 1.9211681674761067,
+      "learning_rate": 1.3197002675328669e-05,
+      "loss": 1.0168,
+      "step": 3961
+    },
+    {
+      "epoch": 0.4146520146520147,
+      "grad_norm": 1.9962879034585508,
+      "learning_rate": 1.3193790668377082e-05,
+      "loss": 0.9542,
+      "step": 3962
+    },
+    {
+      "epoch": 0.41475667189952903,
+      "grad_norm": 2.236000041036604,
+      "learning_rate": 1.3190578294452471e-05,
+      "loss": 1.0742,
+      "step": 3963
+    },
+    {
+      "epoch": 0.41486132914704343,
+      "grad_norm": 1.9426088190833695,
+      "learning_rate": 1.3187365553923948e-05,
+      "loss": 1.0262,
+      "step": 3964
+    },
+    {
+      "epoch": 0.41496598639455784,
+      "grad_norm": 2.35541736370535,
+      "learning_rate": 1.3184152447160658e-05,
+      "loss": 1.0172,
+      "step": 3965
+    },
+    {
+      "epoch": 0.4150706436420722,
+      "grad_norm": 2.088936709244117,
+      "learning_rate": 1.3180938974531799e-05,
+      "loss": 1.0208,
+      "step": 3966
+    },
+    {
+      "epoch": 0.4151753008895866,
+      "grad_norm": 1.8738282005240101,
+      "learning_rate": 1.3177725136406605e-05,
+      "loss": 0.9879,
+      "step": 3967
+    },
+    {
+      "epoch": 0.415279958137101,
+      "grad_norm": 1.9962816803985017,
+      "learning_rate": 1.3174510933154348e-05,
+      "loss": 1.0465,
+      "step": 3968
+    },
+    {
+      "epoch": 0.4153846153846154,
+      "grad_norm": 2.0688274484112705,
+      "learning_rate": 1.3171296365144351e-05,
+      "loss": 0.9194,
+      "step": 3969
+    },
+    {
+      "epoch": 0.41548927263212976,
+      "grad_norm": 1.778517909912603,
+      "learning_rate": 1.3168081432745971e-05,
+      "loss": 0.9127,
+      "step": 3970
+    },
+    {
+      "epoch": 0.41559392987964416,
+      "grad_norm": 2.0828264061542083,
+      "learning_rate": 1.3164866136328613e-05,
+      "loss": 0.8817,
+      "step": 3971
+    },
+    {
+      "epoch": 0.41569858712715857,
+      "grad_norm": 2.229721638477741,
+      "learning_rate": 1.3161650476261723e-05,
+      "loss": 1.0506,
+      "step": 3972
+    },
+    {
+      "epoch": 0.4158032443746729,
+      "grad_norm": 1.9680337726426218,
+      "learning_rate": 1.315843445291478e-05,
+      "loss": 0.9459,
+      "step": 3973
+    },
+    {
+      "epoch": 0.4159079016221873,
+      "grad_norm": 2.14107133654801,
+      "learning_rate": 1.3155218066657315e-05,
+      "loss": 1.0548,
+      "step": 3974
+    },
+    {
+      "epoch": 0.41601255886970173,
+      "grad_norm": 2.013522258681747,
+      "learning_rate": 1.3152001317858902e-05,
+      "loss": 1.0813,
+      "step": 3975
+    },
+    {
+      "epoch": 0.41611721611721614,
+      "grad_norm": 2.1608038286842595,
+      "learning_rate": 1.3148784206889144e-05,
+      "loss": 1.0775,
+      "step": 3976
+    },
+    {
+      "epoch": 0.4162218733647305,
+      "grad_norm": 2.4659678787298094,
+      "learning_rate": 1.3145566734117701e-05,
+      "loss": 1.079,
+      "step": 3977
+    },
+    {
+      "epoch": 0.4163265306122449,
+      "grad_norm": 2.2519431817057,
+      "learning_rate": 1.3142348899914259e-05,
+      "loss": 1.0753,
+      "step": 3978
+    },
+    {
+      "epoch": 0.4164311878597593,
+      "grad_norm": 1.9039732617022176,
+      "learning_rate": 1.3139130704648562e-05,
+      "loss": 0.9651,
+      "step": 3979
+    },
+    {
+      "epoch": 0.4165358451072737,
+      "grad_norm": 2.0825234022954486,
+      "learning_rate": 1.313591214869038e-05,
+      "loss": 0.9306,
+      "step": 3980
+    },
+    {
+      "epoch": 0.41664050235478806,
+      "grad_norm": 1.8388913462765786,
+      "learning_rate": 1.3132693232409538e-05,
+      "loss": 0.7844,
+      "step": 3981
+    },
+    {
+      "epoch": 0.41674515960230246,
+      "grad_norm": 1.9839223196264366,
+      "learning_rate": 1.3129473956175893e-05,
+      "loss": 1.0239,
+      "step": 3982
+    },
+    {
+      "epoch": 0.41684981684981687,
+      "grad_norm": 2.1149890042363855,
+      "learning_rate": 1.3126254320359344e-05,
+      "loss": 1.1038,
+      "step": 3983
+    },
+    {
+      "epoch": 0.4169544740973312,
+      "grad_norm": 2.3388045331623553,
+      "learning_rate": 1.3123034325329838e-05,
+      "loss": 1.0914,
+      "step": 3984
+    },
+    {
+      "epoch": 0.4170591313448456,
+      "grad_norm": 2.3672832146022524,
+      "learning_rate": 1.3119813971457356e-05,
+      "loss": 1.0583,
+      "step": 3985
+    },
+    {
+      "epoch": 0.41716378859236003,
+      "grad_norm": 2.198594788955325,
+      "learning_rate": 1.3116593259111922e-05,
+      "loss": 1.0107,
+      "step": 3986
+    },
+    {
+      "epoch": 0.41726844583987444,
+      "grad_norm": 2.066530857997824,
+      "learning_rate": 1.311337218866361e-05,
+      "loss": 1.0408,
+      "step": 3987
+    },
+    {
+      "epoch": 0.4173731030873888,
+      "grad_norm": 2.1047196181740557,
+      "learning_rate": 1.3110150760482518e-05,
+      "loss": 1.0169,
+      "step": 3988
+    },
+    {
+      "epoch": 0.4174777603349032,
+      "grad_norm": 1.9723714808567003,
+      "learning_rate": 1.3106928974938801e-05,
+      "loss": 0.9556,
+      "step": 3989
+    },
+    {
+      "epoch": 0.4175824175824176,
+      "grad_norm": 2.264669592855649,
+      "learning_rate": 1.3103706832402643e-05,
+      "loss": 0.9994,
+      "step": 3990
+    },
+    {
+      "epoch": 0.41768707482993195,
+      "grad_norm": 1.9073420765404958,
+      "learning_rate": 1.3100484333244282e-05,
+      "loss": 0.8359,
+      "step": 3991
+    },
+    {
+      "epoch": 0.41779173207744635,
+      "grad_norm": 1.8596817557441248,
+      "learning_rate": 1.3097261477833985e-05,
+      "loss": 0.9506,
+      "step": 3992
+    },
+    {
+      "epoch": 0.41789638932496076,
+      "grad_norm": 2.363734651762851,
+      "learning_rate": 1.3094038266542062e-05,
+      "loss": 0.8798,
+      "step": 3993
+    },
+    {
+      "epoch": 0.41800104657247517,
+      "grad_norm": 2.0676118196497733,
+      "learning_rate": 1.3090814699738875e-05,
+      "loss": 0.8891,
+      "step": 3994
+    },
+    {
+      "epoch": 0.4181057038199895,
+      "grad_norm": 2.515973859229479,
+      "learning_rate": 1.3087590777794805e-05,
+      "loss": 0.9715,
+      "step": 3995
+    },
+    {
+      "epoch": 0.4182103610675039,
+      "grad_norm": 2.5102079979788647,
+      "learning_rate": 1.3084366501080304e-05,
+      "loss": 0.9184,
+      "step": 3996
+    },
+    {
+      "epoch": 0.4183150183150183,
+      "grad_norm": 2.0704341062792397,
+      "learning_rate": 1.3081141869965834e-05,
+      "loss": 0.9982,
+      "step": 3997
+    },
+    {
+      "epoch": 0.4184196755625327,
+      "grad_norm": 2.255147279683263,
+      "learning_rate": 1.3077916884821922e-05,
+      "loss": 0.9624,
+      "step": 3998
+    },
+    {
+      "epoch": 0.4185243328100471,
+      "grad_norm": 2.1492963447091644,
+      "learning_rate": 1.3074691546019116e-05,
+      "loss": 0.8961,
+      "step": 3999
+    },
+    {
+      "epoch": 0.4186289900575615,
+      "grad_norm": 2.0388505483770194,
+      "learning_rate": 1.3071465853928018e-05,
+      "loss": 0.9276,
+      "step": 4000
+    },
+    {
+      "epoch": 0.4187336473050759,
+      "grad_norm": 1.866837078619738,
+      "learning_rate": 1.3068239808919271e-05,
+      "loss": 0.8709,
+      "step": 4001
+    },
+    {
+      "epoch": 0.41883830455259025,
+      "grad_norm": 1.986139574512811,
+      "learning_rate": 1.3065013411363548e-05,
+      "loss": 0.9092,
+      "step": 4002
+    },
+    {
+      "epoch": 0.41894296180010465,
+      "grad_norm": 1.8886956223922813,
+      "learning_rate": 1.306178666163157e-05,
+      "loss": 0.9102,
+      "step": 4003
+    },
+    {
+      "epoch": 0.41904761904761906,
+      "grad_norm": 1.9906538968606682,
+      "learning_rate": 1.3058559560094097e-05,
+      "loss": 0.9435,
+      "step": 4004
+    },
+    {
+      "epoch": 0.41915227629513346,
+      "grad_norm": 2.1441939858155425,
+      "learning_rate": 1.3055332107121932e-05,
+      "loss": 1.0081,
+      "step": 4005
+    },
+    {
+      "epoch": 0.4192569335426478,
+      "grad_norm": 2.3353197187128836,
+      "learning_rate": 1.3052104303085915e-05,
+      "loss": 1.064,
+      "step": 4006
+    },
+    {
+      "epoch": 0.4193615907901622,
+      "grad_norm": 1.83565573131337,
+      "learning_rate": 1.3048876148356924e-05,
+      "loss": 0.8715,
+      "step": 4007
+    },
+    {
+      "epoch": 0.4194662480376766,
+      "grad_norm": 2.3218482775446474,
+      "learning_rate": 1.3045647643305883e-05,
+      "loss": 1.0566,
+      "step": 4008
+    },
+    {
+      "epoch": 0.419570905285191,
+      "grad_norm": 1.9284607667631406,
+      "learning_rate": 1.3042418788303752e-05,
+      "loss": 0.9192,
+      "step": 4009
+    },
+    {
+      "epoch": 0.4196755625327054,
+      "grad_norm": 2.061609946734348,
+      "learning_rate": 1.303918958372154e-05,
+      "loss": 0.9721,
+      "step": 4010
+    },
+    {
+      "epoch": 0.4197802197802198,
+      "grad_norm": 1.8955350228878054,
+      "learning_rate": 1.303596002993028e-05,
+      "loss": 0.9905,
+      "step": 4011
+    },
+    {
+      "epoch": 0.4198848770277342,
+      "grad_norm": 2.3133480934371526,
+      "learning_rate": 1.3032730127301056e-05,
+      "loss": 0.8862,
+      "step": 4012
+    },
+    {
+      "epoch": 0.41998953427524854,
+      "grad_norm": 2.0623630507668596,
+      "learning_rate": 1.3029499876204996e-05,
+      "loss": 0.9311,
+      "step": 4013
+    },
+    {
+      "epoch": 0.42009419152276295,
+      "grad_norm": 1.9829665019642981,
+      "learning_rate": 1.3026269277013256e-05,
+      "loss": 1.0163,
+      "step": 4014
+    },
+    {
+      "epoch": 0.42019884877027736,
+      "grad_norm": 2.626803548305666,
+      "learning_rate": 1.3023038330097047e-05,
+      "loss": 1.0789,
+      "step": 4015
+    },
+    {
+      "epoch": 0.4203035060177917,
+      "grad_norm": 2.229456363139348,
+      "learning_rate": 1.3019807035827599e-05,
+      "loss": 1.0455,
+      "step": 4016
+    },
+    {
+      "epoch": 0.4204081632653061,
+      "grad_norm": 1.945363260039406,
+      "learning_rate": 1.3016575394576204e-05,
+      "loss": 1.0051,
+      "step": 4017
+    },
+    {
+      "epoch": 0.4205128205128205,
+      "grad_norm": 1.8657699677861808,
+      "learning_rate": 1.3013343406714181e-05,
+      "loss": 0.9354,
+      "step": 4018
+    },
+    {
+      "epoch": 0.4206174777603349,
+      "grad_norm": 2.13301131218169,
+      "learning_rate": 1.3010111072612888e-05,
+      "loss": 0.9128,
+      "step": 4019
+    },
+    {
+      "epoch": 0.4207221350078493,
+      "grad_norm": 2.0809458736583637,
+      "learning_rate": 1.3006878392643736e-05,
+      "loss": 0.9963,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4208267922553637,
+      "grad_norm": 2.2025198937923784,
+      "learning_rate": 1.3003645367178158e-05,
+      "loss": 1.0341,
+      "step": 4021
+    },
+    {
+      "epoch": 0.4209314495028781,
+      "grad_norm": 1.8303178475154966,
+      "learning_rate": 1.300041199658764e-05,
+      "loss": 0.9209,
+      "step": 4022
+    },
+    {
+      "epoch": 0.4210361067503925,
+      "grad_norm": 2.0995406687477423,
+      "learning_rate": 1.2997178281243698e-05,
+      "loss": 1.0328,
+      "step": 4023
+    },
+    {
+      "epoch": 0.42114076399790684,
+      "grad_norm": 2.133177530342129,
+      "learning_rate": 1.2993944221517898e-05,
+      "loss": 0.918,
+      "step": 4024
+    },
+    {
+      "epoch": 0.42124542124542125,
+      "grad_norm": 1.913400092526822,
+      "learning_rate": 1.2990709817781839e-05,
+      "loss": 0.8177,
+      "step": 4025
+    },
+    {
+      "epoch": 0.42135007849293565,
+      "grad_norm": 1.8746443011882772,
+      "learning_rate": 1.2987475070407154e-05,
+      "loss": 0.9147,
+      "step": 4026
+    },
+    {
+      "epoch": 0.42145473574045,
+      "grad_norm": 2.01816221130607,
+      "learning_rate": 1.2984239979765531e-05,
+      "loss": 1.0686,
+      "step": 4027
+    },
+    {
+      "epoch": 0.4215593929879644,
+      "grad_norm": 2.1127015875668365,
+      "learning_rate": 1.2981004546228685e-05,
+      "loss": 1.0448,
+      "step": 4028
+    },
+    {
+      "epoch": 0.4216640502354788,
+      "grad_norm": 2.1501657966221526,
+      "learning_rate": 1.2977768770168374e-05,
+      "loss": 1.0125,
+      "step": 4029
+    },
+    {
+      "epoch": 0.4217687074829932,
+      "grad_norm": 1.8600363609360118,
+      "learning_rate": 1.2974532651956393e-05,
+      "loss": 1.0479,
+      "step": 4030
+    },
+    {
+      "epoch": 0.42187336473050757,
+      "grad_norm": 2.4569790644754996,
+      "learning_rate": 1.297129619196458e-05,
+      "loss": 0.9219,
+      "step": 4031
+    },
+    {
+      "epoch": 0.421978021978022,
+      "grad_norm": 2.47767730012033,
+      "learning_rate": 1.2968059390564813e-05,
+      "loss": 0.9192,
+      "step": 4032
+    },
+    {
+      "epoch": 0.4220826792255364,
+      "grad_norm": 2.0774703638862415,
+      "learning_rate": 1.2964822248129003e-05,
+      "loss": 0.9324,
+      "step": 4033
+    },
+    {
+      "epoch": 0.42218733647305073,
+      "grad_norm": 2.5616604388013524,
+      "learning_rate": 1.2961584765029107e-05,
+      "loss": 0.9997,
+      "step": 4034
+    },
+    {
+      "epoch": 0.42229199372056514,
+      "grad_norm": 2.3036391018677227,
+      "learning_rate": 1.2958346941637119e-05,
+      "loss": 0.9766,
+      "step": 4035
+    },
+    {
+      "epoch": 0.42239665096807955,
+      "grad_norm": 2.4294354631420885,
+      "learning_rate": 1.295510877832507e-05,
+      "loss": 0.8532,
+      "step": 4036
+    },
+    {
+      "epoch": 0.42250130821559395,
+      "grad_norm": 2.262444310074482,
+      "learning_rate": 1.2951870275465033e-05,
+      "loss": 1.0549,
+      "step": 4037
+    },
+    {
+      "epoch": 0.4226059654631083,
+      "grad_norm": 2.1283035221559654,
+      "learning_rate": 1.2948631433429114e-05,
+      "loss": 1.0502,
+      "step": 4038
+    },
+    {
+      "epoch": 0.4227106227106227,
+      "grad_norm": 2.214881325613008,
+      "learning_rate": 1.2945392252589466e-05,
+      "loss": 0.9077,
+      "step": 4039
+    },
+    {
+      "epoch": 0.4228152799581371,
+      "grad_norm": 2.369040409827918,
+      "learning_rate": 1.2942152733318278e-05,
+      "loss": 0.9257,
+      "step": 4040
+    },
+    {
+      "epoch": 0.42291993720565146,
+      "grad_norm": 1.6747394664320463,
+      "learning_rate": 1.2938912875987776e-05,
+      "loss": 0.9738,
+      "step": 4041
+    },
+    {
+      "epoch": 0.42302459445316587,
+      "grad_norm": 1.8358883748542993,
+      "learning_rate": 1.2935672680970231e-05,
+      "loss": 0.8974,
+      "step": 4042
+    },
+    {
+      "epoch": 0.4231292517006803,
+      "grad_norm": 1.9700721618750736,
+      "learning_rate": 1.2932432148637937e-05,
+      "loss": 0.8539,
+      "step": 4043
+    },
+    {
+      "epoch": 0.4232339089481947,
+      "grad_norm": 1.7081738341481085,
+      "learning_rate": 1.2929191279363248e-05,
+      "loss": 0.9157,
+      "step": 4044
+    },
+    {
+      "epoch": 0.42333856619570903,
+      "grad_norm": 1.9888702561572258,
+      "learning_rate": 1.2925950073518543e-05,
+      "loss": 0.8719,
+      "step": 4045
+    },
+    {
+      "epoch": 0.42344322344322344,
+      "grad_norm": 2.1023936971584494,
+      "learning_rate": 1.2922708531476245e-05,
+      "loss": 0.9698,
+      "step": 4046
+    },
+    {
+      "epoch": 0.42354788069073784,
+      "grad_norm": 2.065987444351679,
+      "learning_rate": 1.2919466653608807e-05,
+      "loss": 1.0014,
+      "step": 4047
+    },
+    {
+      "epoch": 0.42365253793825225,
+      "grad_norm": 2.1081730464926682,
+      "learning_rate": 1.2916224440288736e-05,
+      "loss": 0.8841,
+      "step": 4048
+    },
+    {
+      "epoch": 0.4237571951857666,
+      "grad_norm": 1.9254261511446567,
+      "learning_rate": 1.2912981891888566e-05,
+      "loss": 1.0805,
+      "step": 4049
+    },
+    {
+      "epoch": 0.423861852433281,
+      "grad_norm": 1.8610671376702372,
+      "learning_rate": 1.2909739008780866e-05,
+      "loss": 1.0011,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4239665096807954,
+      "grad_norm": 1.8387402872473773,
+      "learning_rate": 1.290649579133826e-05,
+      "loss": 1.0046,
+      "step": 4051
+    },
+    {
+      "epoch": 0.42407116692830976,
+      "grad_norm": 1.9929812428438674,
+      "learning_rate": 1.2903252239933393e-05,
+      "loss": 0.8668,
+      "step": 4052
+    },
+    {
+      "epoch": 0.42417582417582417,
+      "grad_norm": 2.140516044935987,
+      "learning_rate": 1.290000835493896e-05,
+      "loss": 1.0137,
+      "step": 4053
+    },
+    {
+      "epoch": 0.4242804814233386,
+      "grad_norm": 2.271991022792856,
+      "learning_rate": 1.2896764136727687e-05,
+      "loss": 0.9522,
+      "step": 4054
+    },
+    {
+      "epoch": 0.424385138670853,
+      "grad_norm": 1.8813686606836715,
+      "learning_rate": 1.2893519585672342e-05,
+      "loss": 0.986,
+      "step": 4055
+    },
+    {
+      "epoch": 0.42448979591836733,
+      "grad_norm": 1.974109588184302,
+      "learning_rate": 1.289027470214573e-05,
+      "loss": 1.0201,
+      "step": 4056
+    },
+    {
+      "epoch": 0.42459445316588174,
+      "grad_norm": 2.0120661694359607,
+      "learning_rate": 1.2887029486520697e-05,
+      "loss": 0.9441,
+      "step": 4057
+    },
+    {
+      "epoch": 0.42469911041339614,
+      "grad_norm": 2.141950924459728,
+      "learning_rate": 1.2883783939170125e-05,
+      "loss": 0.9926,
+      "step": 4058
+    },
+    {
+      "epoch": 0.4248037676609105,
+      "grad_norm": 2.5623343563978915,
+      "learning_rate": 1.2880538060466931e-05,
+      "loss": 0.9227,
+      "step": 4059
+    },
+    {
+      "epoch": 0.4249084249084249,
+      "grad_norm": 1.9750731791255807,
+      "learning_rate": 1.287729185078407e-05,
+      "loss": 1.0129,
+      "step": 4060
+    },
+    {
+      "epoch": 0.4250130821559393,
+      "grad_norm": 2.0054191639646692,
+      "learning_rate": 1.2874045310494549e-05,
+      "loss": 0.9068,
+      "step": 4061
+    },
+    {
+      "epoch": 0.4251177394034537,
+      "grad_norm": 1.9724083424778205,
+      "learning_rate": 1.2870798439971393e-05,
+      "loss": 0.862,
+      "step": 4062
+    },
+    {
+      "epoch": 0.42522239665096806,
+      "grad_norm": 2.319494091259867,
+      "learning_rate": 1.2867551239587677e-05,
+      "loss": 1.0978,
+      "step": 4063
+    },
+    {
+      "epoch": 0.42532705389848247,
+      "grad_norm": 2.319487442343795,
+      "learning_rate": 1.286430370971651e-05,
+      "loss": 1.0296,
+      "step": 4064
+    },
+    {
+      "epoch": 0.42543171114599687,
+      "grad_norm": 2.0836979512763336,
+      "learning_rate": 1.2861055850731039e-05,
+      "loss": 0.972,
+      "step": 4065
+    },
+    {
+      "epoch": 0.4255363683935113,
+      "grad_norm": 2.552512993363399,
+      "learning_rate": 1.2857807663004453e-05,
+      "loss": 0.8689,
+      "step": 4066
+    },
+    {
+      "epoch": 0.4256410256410256,
+      "grad_norm": 2.1467202430472763,
+      "learning_rate": 1.285455914690997e-05,
+      "loss": 0.9987,
+      "step": 4067
+    },
+    {
+      "epoch": 0.42574568288854003,
+      "grad_norm": 2.1204522151177474,
+      "learning_rate": 1.285131030282086e-05,
+      "loss": 1.0174,
+      "step": 4068
+    },
+    {
+      "epoch": 0.42585034013605444,
+      "grad_norm": 2.3582725345374027,
+      "learning_rate": 1.284806113111041e-05,
+      "loss": 0.9143,
+      "step": 4069
+    },
+    {
+      "epoch": 0.4259549973835688,
+      "grad_norm": 2.0482186802071287,
+      "learning_rate": 1.2844811632151964e-05,
+      "loss": 0.9063,
+      "step": 4070
+    },
+    {
+      "epoch": 0.4260596546310832,
+      "grad_norm": 2.0895838757594696,
+      "learning_rate": 1.2841561806318897e-05,
+      "loss": 1.052,
+      "step": 4071
+    },
+    {
+      "epoch": 0.4261643118785976,
+      "grad_norm": 2.0282129968162246,
+      "learning_rate": 1.2838311653984615e-05,
+      "loss": 0.9849,
+      "step": 4072
+    },
+    {
+      "epoch": 0.426268969126112,
+      "grad_norm": 2.5256239159226745,
+      "learning_rate": 1.2835061175522569e-05,
+      "loss": 1.102,
+      "step": 4073
+    },
+    {
+      "epoch": 0.42637362637362636,
+      "grad_norm": 2.0879581133791687,
+      "learning_rate": 1.2831810371306247e-05,
+      "loss": 0.9754,
+      "step": 4074
+    },
+    {
+      "epoch": 0.42647828362114076,
+      "grad_norm": 1.8366234669283463,
+      "learning_rate": 1.2828559241709177e-05,
+      "loss": 0.7956,
+      "step": 4075
+    },
+    {
+      "epoch": 0.42658294086865517,
+      "grad_norm": 1.8797736027924894,
+      "learning_rate": 1.2825307787104917e-05,
+      "loss": 0.9943,
+      "step": 4076
+    },
+    {
+      "epoch": 0.4266875981161695,
+      "grad_norm": 2.092067441756589,
+      "learning_rate": 1.282205600786706e-05,
+      "loss": 1.0322,
+      "step": 4077
+    },
+    {
+      "epoch": 0.4267922553636839,
+      "grad_norm": 2.080732968369759,
+      "learning_rate": 1.2818803904369248e-05,
+      "loss": 0.9993,
+      "step": 4078
+    },
+    {
+      "epoch": 0.42689691261119833,
+      "grad_norm": 1.8121577795467074,
+      "learning_rate": 1.2815551476985151e-05,
+      "loss": 0.8637,
+      "step": 4079
+    },
+    {
+      "epoch": 0.42700156985871274,
+      "grad_norm": 1.8717669220769215,
+      "learning_rate": 1.2812298726088486e-05,
+      "loss": 0.8787,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4271062271062271,
+      "grad_norm": 2.179750288411442,
+      "learning_rate": 1.2809045652052992e-05,
+      "loss": 0.8961,
+      "step": 4081
+    },
+    {
+      "epoch": 0.4272108843537415,
+      "grad_norm": 2.0838225686688263,
+      "learning_rate": 1.280579225525246e-05,
+      "loss": 0.9929,
+      "step": 4082
+    },
+    {
+      "epoch": 0.4273155416012559,
+      "grad_norm": 2.275118316941627,
+      "learning_rate": 1.2802538536060712e-05,
+      "loss": 1.0092,
+      "step": 4083
+    },
+    {
+      "epoch": 0.42742019884877025,
+      "grad_norm": 2.130895970762807,
+      "learning_rate": 1.2799284494851602e-05,
+      "loss": 0.8968,
+      "step": 4084
+    },
+    {
+      "epoch": 0.42752485609628466,
+      "grad_norm": 1.8836199307900716,
+      "learning_rate": 1.279603013199903e-05,
+      "loss": 0.9125,
+      "step": 4085
+    },
+    {
+      "epoch": 0.42762951334379906,
+      "grad_norm": 2.458988223380957,
+      "learning_rate": 1.2792775447876928e-05,
+      "loss": 1.0487,
+      "step": 4086
+    },
+    {
+      "epoch": 0.42773417059131347,
+      "grad_norm": 1.9971378709402376,
+      "learning_rate": 1.2789520442859265e-05,
+      "loss": 0.863,
+      "step": 4087
+    },
+    {
+      "epoch": 0.4278388278388278,
+      "grad_norm": 2.007011122353183,
+      "learning_rate": 1.2786265117320047e-05,
+      "loss": 1.0189,
+      "step": 4088
+    },
+    {
+      "epoch": 0.4279434850863422,
+      "grad_norm": 2.1617166345362673,
+      "learning_rate": 1.2783009471633321e-05,
+      "loss": 0.9055,
+      "step": 4089
+    },
+    {
+      "epoch": 0.42804814233385663,
+      "grad_norm": 2.5367597991139115,
+      "learning_rate": 1.2779753506173163e-05,
+      "loss": 1.0275,
+      "step": 4090
+    },
+    {
+      "epoch": 0.42815279958137104,
+      "grad_norm": 1.9826469266228839,
+      "learning_rate": 1.277649722131369e-05,
+      "loss": 0.893,
+      "step": 4091
+    },
+    {
+      "epoch": 0.4282574568288854,
+      "grad_norm": 2.038739638755624,
+      "learning_rate": 1.277324061742906e-05,
+      "loss": 1.0419,
+      "step": 4092
+    },
+    {
+      "epoch": 0.4283621140763998,
+      "grad_norm": 2.5868448257014705,
+      "learning_rate": 1.2769983694893461e-05,
+      "loss": 1.0708,
+      "step": 4093
+    },
+    {
+      "epoch": 0.4284667713239142,
+      "grad_norm": 1.952363024664158,
+      "learning_rate": 1.2766726454081124e-05,
+      "loss": 0.8693,
+      "step": 4094
+    },
+    {
+      "epoch": 0.42857142857142855,
+      "grad_norm": 2.537669019028795,
+      "learning_rate": 1.2763468895366304e-05,
+      "loss": 0.8952,
+      "step": 4095
+    },
+    {
+      "epoch": 0.42867608581894295,
+      "grad_norm": 2.1143917570175166,
+      "learning_rate": 1.2760211019123307e-05,
+      "loss": 0.9822,
+      "step": 4096
+    },
+    {
+      "epoch": 0.42878074306645736,
+      "grad_norm": 2.0948895182122915,
+      "learning_rate": 1.2756952825726469e-05,
+      "loss": 0.9116,
+      "step": 4097
+    },
+    {
+      "epoch": 0.42888540031397177,
+      "grad_norm": 2.222927984303674,
+      "learning_rate": 1.2753694315550163e-05,
+      "loss": 0.9202,
+      "step": 4098
+    },
+    {
+      "epoch": 0.4289900575614861,
+      "grad_norm": 2.127865487750365,
+      "learning_rate": 1.2750435488968793e-05,
+      "loss": 1.0449,
+      "step": 4099
+    },
+    {
+      "epoch": 0.4290947148090005,
+      "grad_norm": 1.9788232612921248,
+      "learning_rate": 1.2747176346356815e-05,
+      "loss": 0.9387,
+      "step": 4100
+    },
+    {
+      "epoch": 0.4291993720565149,
+      "grad_norm": 1.9016287167665056,
+      "learning_rate": 1.2743916888088706e-05,
+      "loss": 0.8767,
+      "step": 4101
+    },
+    {
+      "epoch": 0.4293040293040293,
+      "grad_norm": 2.1605389550086196,
+      "learning_rate": 1.274065711453898e-05,
+      "loss": 1.0258,
+      "step": 4102
+    },
+    {
+      "epoch": 0.4294086865515437,
+      "grad_norm": 2.5323063383357463,
+      "learning_rate": 1.2737397026082199e-05,
+      "loss": 0.9743,
+      "step": 4103
+    },
+    {
+      "epoch": 0.4295133437990581,
+      "grad_norm": 1.9921710514030042,
+      "learning_rate": 1.2734136623092948e-05,
+      "loss": 1.0633,
+      "step": 4104
+    },
+    {
+      "epoch": 0.4296180010465725,
+      "grad_norm": 2.054756726068761,
+      "learning_rate": 1.273087590594586e-05,
+      "loss": 1.0129,
+      "step": 4105
+    },
+    {
+      "epoch": 0.42972265829408685,
+      "grad_norm": 2.1517769857523517,
+      "learning_rate": 1.272761487501559e-05,
+      "loss": 0.9996,
+      "step": 4106
+    },
+    {
+      "epoch": 0.42982731554160125,
+      "grad_norm": 2.0631947885960598,
+      "learning_rate": 1.2724353530676842e-05,
+      "loss": 1.0288,
+      "step": 4107
+    },
+    {
+      "epoch": 0.42993197278911566,
+      "grad_norm": 2.0629095347948625,
+      "learning_rate": 1.2721091873304351e-05,
+      "loss": 0.9055,
+      "step": 4108
+    },
+    {
+      "epoch": 0.43003663003663006,
+      "grad_norm": 2.1035215723837544,
+      "learning_rate": 1.2717829903272889e-05,
+      "loss": 0.9523,
+      "step": 4109
+    },
+    {
+      "epoch": 0.4301412872841444,
+      "grad_norm": 2.1532839005297766,
+      "learning_rate": 1.2714567620957262e-05,
+      "loss": 0.9819,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4302459445316588,
+      "grad_norm": 2.4892633893694778,
+      "learning_rate": 1.271130502673231e-05,
+      "loss": 0.9969,
+      "step": 4111
+    },
+    {
+      "epoch": 0.4303506017791732,
+      "grad_norm": 2.0071464961288243,
+      "learning_rate": 1.2708042120972914e-05,
+      "loss": 1.0227,
+      "step": 4112
+    },
+    {
+      "epoch": 0.4304552590266876,
+      "grad_norm": 2.155107165578016,
+      "learning_rate": 1.270477890405399e-05,
+      "loss": 0.9929,
+      "step": 4113
+    },
+    {
+      "epoch": 0.430559916274202,
+      "grad_norm": 2.169518740126109,
+      "learning_rate": 1.2701515376350489e-05,
+      "loss": 0.9272,
+      "step": 4114
+    },
+    {
+      "epoch": 0.4306645735217164,
+      "grad_norm": 2.739527866572885,
+      "learning_rate": 1.2698251538237389e-05,
+      "loss": 0.9999,
+      "step": 4115
+    },
+    {
+      "epoch": 0.4307692307692308,
+      "grad_norm": 2.513640769034525,
+      "learning_rate": 1.2694987390089723e-05,
+      "loss": 1.0112,
+      "step": 4116
+    },
+    {
+      "epoch": 0.43087388801674514,
+      "grad_norm": 2.0925444858975326,
+      "learning_rate": 1.269172293228254e-05,
+      "loss": 0.9724,
+      "step": 4117
+    },
+    {
+      "epoch": 0.43097854526425955,
+      "grad_norm": 2.1067772462429883,
+      "learning_rate": 1.2688458165190933e-05,
+      "loss": 0.9119,
+      "step": 4118
+    },
+    {
+      "epoch": 0.43108320251177396,
+      "grad_norm": 2.0030783316902165,
+      "learning_rate": 1.2685193089190038e-05,
+      "loss": 1.0959,
+      "step": 4119
+    },
+    {
+      "epoch": 0.4311878597592883,
+      "grad_norm": 2.1073270714006034,
+      "learning_rate": 1.2681927704655008e-05,
+      "loss": 1.0327,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4312925170068027,
+      "grad_norm": 2.137450289870023,
+      "learning_rate": 1.2678662011961051e-05,
+      "loss": 0.9802,
+      "step": 4121
+    },
+    {
+      "epoch": 0.4313971742543171,
+      "grad_norm": 2.0752705351892673,
+      "learning_rate": 1.2675396011483398e-05,
+      "loss": 0.9755,
+      "step": 4122
+    },
+    {
+      "epoch": 0.4315018315018315,
+      "grad_norm": 2.2558252138712582,
+      "learning_rate": 1.2672129703597321e-05,
+      "loss": 1.0148,
+      "step": 4123
+    },
+    {
+      "epoch": 0.4316064887493459,
+      "grad_norm": 1.9748773751441262,
+      "learning_rate": 1.2668863088678124e-05,
+      "loss": 0.9922,
+      "step": 4124
+    },
+    {
+      "epoch": 0.4317111459968603,
+      "grad_norm": 1.8299642601513824,
+      "learning_rate": 1.2665596167101144e-05,
+      "loss": 0.8775,
+      "step": 4125
+    },
+    {
+      "epoch": 0.4318158032443747,
+      "grad_norm": 2.2374692170514527,
+      "learning_rate": 1.2662328939241766e-05,
+      "loss": 1.0831,
+      "step": 4126
+    },
+    {
+      "epoch": 0.4319204604918891,
+      "grad_norm": 2.1794273635322376,
+      "learning_rate": 1.2659061405475392e-05,
+      "loss": 0.9673,
+      "step": 4127
+    },
+    {
+      "epoch": 0.43202511773940344,
+      "grad_norm": 1.939053173523464,
+      "learning_rate": 1.2655793566177475e-05,
+      "loss": 0.8898,
+      "step": 4128
+    },
+    {
+      "epoch": 0.43212977498691785,
+      "grad_norm": 2.0573506268728328,
+      "learning_rate": 1.2652525421723489e-05,
+      "loss": 0.8148,
+      "step": 4129
+    },
+    {
+      "epoch": 0.43223443223443225,
+      "grad_norm": 1.8819328528426567,
+      "learning_rate": 1.2649256972488956e-05,
+      "loss": 1.0517,
+      "step": 4130
+    },
+    {
+      "epoch": 0.4323390894819466,
+      "grad_norm": 1.9984163786543963,
+      "learning_rate": 1.2645988218849432e-05,
+      "loss": 0.908,
+      "step": 4131
+    },
+    {
+      "epoch": 0.432443746729461,
+      "grad_norm": 1.8773422683017276,
+      "learning_rate": 1.264271916118049e-05,
+      "loss": 0.8387,
+      "step": 4132
+    },
+    {
+      "epoch": 0.4325484039769754,
+      "grad_norm": 2.0149662866765863,
+      "learning_rate": 1.2639449799857766e-05,
+      "loss": 0.9982,
+      "step": 4133
+    },
+    {
+      "epoch": 0.4326530612244898,
+      "grad_norm": 2.174292902817768,
+      "learning_rate": 1.2636180135256904e-05,
+      "loss": 1.1082,
+      "step": 4134
+    },
+    {
+      "epoch": 0.43275771847200417,
+      "grad_norm": 2.2208332106910897,
+      "learning_rate": 1.2632910167753601e-05,
+      "loss": 0.8762,
+      "step": 4135
+    },
+    {
+      "epoch": 0.4328623757195186,
+      "grad_norm": 2.181100971590527,
+      "learning_rate": 1.2629639897723585e-05,
+      "loss": 0.975,
+      "step": 4136
+    },
+    {
+      "epoch": 0.432967032967033,
+      "grad_norm": 1.931613675855517,
+      "learning_rate": 1.262636932554261e-05,
+      "loss": 0.8959,
+      "step": 4137
+    },
+    {
+      "epoch": 0.43307169021454733,
+      "grad_norm": 2.2708040082763943,
+      "learning_rate": 1.2623098451586477e-05,
+      "loss": 1.0665,
+      "step": 4138
+    },
+    {
+      "epoch": 0.43317634746206174,
+      "grad_norm": 1.9224472683374367,
+      "learning_rate": 1.2619827276231017e-05,
+      "loss": 0.9472,
+      "step": 4139
+    },
+    {
+      "epoch": 0.43328100470957615,
+      "grad_norm": 2.2865237719592275,
+      "learning_rate": 1.261655579985209e-05,
+      "loss": 0.8564,
+      "step": 4140
+    },
+    {
+      "epoch": 0.43338566195709055,
+      "grad_norm": 1.937006380048429,
+      "learning_rate": 1.2613284022825596e-05,
+      "loss": 1.0623,
+      "step": 4141
+    },
+    {
+      "epoch": 0.4334903192046049,
+      "grad_norm": 2.5686460907237096,
+      "learning_rate": 1.2610011945527471e-05,
+      "loss": 1.0288,
+      "step": 4142
+    },
+    {
+      "epoch": 0.4335949764521193,
+      "grad_norm": 2.4290445999367405,
+      "learning_rate": 1.2606739568333684e-05,
+      "loss": 1.0339,
+      "step": 4143
+    },
+    {
+      "epoch": 0.4336996336996337,
+      "grad_norm": 2.009556122935142,
+      "learning_rate": 1.2603466891620232e-05,
+      "loss": 0.8586,
+      "step": 4144
+    },
+    {
+      "epoch": 0.43380429094714806,
+      "grad_norm": 1.7741581344189656,
+      "learning_rate": 1.2600193915763162e-05,
+      "loss": 0.8876,
+      "step": 4145
+    },
+    {
+      "epoch": 0.43390894819466247,
+      "grad_norm": 2.4372967065545317,
+      "learning_rate": 1.2596920641138538e-05,
+      "loss": 1.0327,
+      "step": 4146
+    },
+    {
+      "epoch": 0.4340136054421769,
+      "grad_norm": 2.0894117465917215,
+      "learning_rate": 1.2593647068122465e-05,
+      "loss": 1.0019,
+      "step": 4147
+    },
+    {
+      "epoch": 0.4341182626896913,
+      "grad_norm": 2.123368845322283,
+      "learning_rate": 1.259037319709109e-05,
+      "loss": 0.9752,
+      "step": 4148
+    },
+    {
+      "epoch": 0.43422291993720563,
+      "grad_norm": 2.041973404473876,
+      "learning_rate": 1.258709902842058e-05,
+      "loss": 1.0645,
+      "step": 4149
+    },
+    {
+      "epoch": 0.43432757718472004,
+      "grad_norm": 2.2360927293499544,
+      "learning_rate": 1.258382456248715e-05,
+      "loss": 0.9802,
+      "step": 4150
+    },
+    {
+      "epoch": 0.43443223443223444,
+      "grad_norm": 1.9616686428786763,
+      "learning_rate": 1.2580549799667034e-05,
+      "loss": 0.8553,
+      "step": 4151
+    },
+    {
+      "epoch": 0.43453689167974885,
+      "grad_norm": 2.2952336749644804,
+      "learning_rate": 1.2577274740336523e-05,
+      "loss": 1.0347,
+      "step": 4152
+    },
+    {
+      "epoch": 0.4346415489272632,
+      "grad_norm": 2.370451934856473,
+      "learning_rate": 1.2573999384871916e-05,
+      "loss": 0.9924,
+      "step": 4153
+    },
+    {
+      "epoch": 0.4347462061747776,
+      "grad_norm": 2.0939601713417577,
+      "learning_rate": 1.2570723733649558e-05,
+      "loss": 0.9574,
+      "step": 4154
+    },
+    {
+      "epoch": 0.434850863422292,
+      "grad_norm": 2.388950829479333,
+      "learning_rate": 1.256744778704584e-05,
+      "loss": 0.9239,
+      "step": 4155
+    },
+    {
+      "epoch": 0.43495552066980636,
+      "grad_norm": 1.978177781520638,
+      "learning_rate": 1.2564171545437156e-05,
+      "loss": 0.9031,
+      "step": 4156
+    },
+    {
+      "epoch": 0.43506017791732077,
+      "grad_norm": 2.112737659944072,
+      "learning_rate": 1.2560895009199972e-05,
+      "loss": 0.9537,
+      "step": 4157
+    },
+    {
+      "epoch": 0.4351648351648352,
+      "grad_norm": 2.033176365951217,
+      "learning_rate": 1.2557618178710756e-05,
+      "loss": 1.0335,
+      "step": 4158
+    },
+    {
+      "epoch": 0.4352694924123496,
+      "grad_norm": 2.1148782403285784,
+      "learning_rate": 1.2554341054346027e-05,
+      "loss": 0.7705,
+      "step": 4159
+    },
+    {
+      "epoch": 0.43537414965986393,
+      "grad_norm": 1.6826596407900571,
+      "learning_rate": 1.2551063636482332e-05,
+      "loss": 0.7461,
+      "step": 4160
+    },
+    {
+      "epoch": 0.43547880690737834,
+      "grad_norm": 2.0821373935182623,
+      "learning_rate": 1.2547785925496255e-05,
+      "loss": 0.8977,
+      "step": 4161
+    },
+    {
+      "epoch": 0.43558346415489274,
+      "grad_norm": 2.7138514710502126,
+      "learning_rate": 1.254450792176441e-05,
+      "loss": 1.0286,
+      "step": 4162
+    },
+    {
+      "epoch": 0.4356881214024071,
+      "grad_norm": 1.928568939003756,
+      "learning_rate": 1.2541229625663445e-05,
+      "loss": 0.8945,
+      "step": 4163
+    },
+    {
+      "epoch": 0.4357927786499215,
+      "grad_norm": 2.5708479006499085,
+      "learning_rate": 1.2537951037570047e-05,
+      "loss": 0.9281,
+      "step": 4164
+    },
+    {
+      "epoch": 0.4358974358974359,
+      "grad_norm": 2.7402121626792773,
+      "learning_rate": 1.2534672157860928e-05,
+      "loss": 0.9111,
+      "step": 4165
+    },
+    {
+      "epoch": 0.4360020931449503,
+      "grad_norm": 2.1020926928873744,
+      "learning_rate": 1.2531392986912839e-05,
+      "loss": 1.0641,
+      "step": 4166
+    },
+    {
+      "epoch": 0.43610675039246466,
+      "grad_norm": 2.0147104900016153,
+      "learning_rate": 1.2528113525102568e-05,
+      "loss": 0.9251,
+      "step": 4167
+    },
+    {
+      "epoch": 0.43621140763997907,
+      "grad_norm": 2.1361681568328463,
+      "learning_rate": 1.2524833772806924e-05,
+      "loss": 1.0695,
+      "step": 4168
+    },
+    {
+      "epoch": 0.43631606488749347,
+      "grad_norm": 1.9844534323381833,
+      "learning_rate": 1.2521553730402762e-05,
+      "loss": 1.022,
+      "step": 4169
+    },
+    {
+      "epoch": 0.4364207221350079,
+      "grad_norm": 2.1536611659943854,
+      "learning_rate": 1.2518273398266963e-05,
+      "loss": 1.0211,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4365253793825222,
+      "grad_norm": 1.9066376292841682,
+      "learning_rate": 1.251499277677645e-05,
+      "loss": 0.9381,
+      "step": 4171
+    },
+    {
+      "epoch": 0.43663003663003663,
+      "grad_norm": 1.979655024358628,
+      "learning_rate": 1.2511711866308167e-05,
+      "loss": 0.9354,
+      "step": 4172
+    },
+    {
+      "epoch": 0.43673469387755104,
+      "grad_norm": 1.930714983152759,
+      "learning_rate": 1.2508430667239094e-05,
+      "loss": 0.8527,
+      "step": 4173
+    },
+    {
+      "epoch": 0.4368393511250654,
+      "grad_norm": 2.1508753326787144,
+      "learning_rate": 1.2505149179946257e-05,
+      "loss": 1.0292,
+      "step": 4174
+    },
+    {
+      "epoch": 0.4369440083725798,
+      "grad_norm": 1.799153917872675,
+      "learning_rate": 1.25018674048067e-05,
+      "loss": 0.9594,
+      "step": 4175
+    },
+    {
+      "epoch": 0.4370486656200942,
+      "grad_norm": 2.1158310072558493,
+      "learning_rate": 1.2498585342197507e-05,
+      "loss": 1.0612,
+      "step": 4176
+    },
+    {
+      "epoch": 0.4371533228676086,
+      "grad_norm": 1.784434212580124,
+      "learning_rate": 1.2495302992495793e-05,
+      "loss": 1.0151,
+      "step": 4177
+    },
+    {
+      "epoch": 0.43725798011512296,
+      "grad_norm": 1.9237924946084939,
+      "learning_rate": 1.2492020356078705e-05,
+      "loss": 0.8651,
+      "step": 4178
+    },
+    {
+      "epoch": 0.43736263736263736,
+      "grad_norm": 2.5836868280522576,
+      "learning_rate": 1.2488737433323429e-05,
+      "loss": 0.9226,
+      "step": 4179
+    },
+    {
+      "epoch": 0.43746729461015177,
+      "grad_norm": 2.019400496982174,
+      "learning_rate": 1.2485454224607171e-05,
+      "loss": 0.8531,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4375719518576661,
+      "grad_norm": 1.8357957654319403,
+      "learning_rate": 1.2482170730307191e-05,
+      "loss": 0.9594,
+      "step": 4181
+    },
+    {
+      "epoch": 0.4376766091051805,
+      "grad_norm": 2.2211596017891084,
+      "learning_rate": 1.2478886950800757e-05,
+      "loss": 0.9483,
+      "step": 4182
+    },
+    {
+      "epoch": 0.43778126635269493,
+      "grad_norm": 2.0328108335435093,
+      "learning_rate": 1.2475602886465187e-05,
+      "loss": 1.0499,
+      "step": 4183
+    },
+    {
+      "epoch": 0.43788592360020934,
+      "grad_norm": 1.9848307097983084,
+      "learning_rate": 1.2472318537677833e-05,
+      "loss": 0.868,
+      "step": 4184
+    },
+    {
+      "epoch": 0.4379905808477237,
+      "grad_norm": 1.9830682161242188,
+      "learning_rate": 1.2469033904816062e-05,
+      "loss": 0.952,
+      "step": 4185
+    },
+    {
+      "epoch": 0.4380952380952381,
+      "grad_norm": 1.8538521533923138,
+      "learning_rate": 1.246574898825729e-05,
+      "loss": 0.8598,
+      "step": 4186
+    },
+    {
+      "epoch": 0.4381998953427525,
+      "grad_norm": 2.3076134727295843,
+      "learning_rate": 1.246246378837896e-05,
+      "loss": 0.8155,
+      "step": 4187
+    },
+    {
+      "epoch": 0.43830455259026685,
+      "grad_norm": 2.3923552507037185,
+      "learning_rate": 1.2459178305558553e-05,
+      "loss": 1.0912,
+      "step": 4188
+    },
+    {
+      "epoch": 0.43840920983778126,
+      "grad_norm": 1.8990741957670583,
+      "learning_rate": 1.2455892540173572e-05,
+      "loss": 0.8969,
+      "step": 4189
+    },
+    {
+      "epoch": 0.43851386708529566,
+      "grad_norm": 1.8395674522822645,
+      "learning_rate": 1.2452606492601556e-05,
+      "loss": 0.905,
+      "step": 4190
+    },
+    {
+      "epoch": 0.43861852433281007,
+      "grad_norm": 1.859136871339698,
+      "learning_rate": 1.2449320163220087e-05,
+      "loss": 0.9013,
+      "step": 4191
+    },
+    {
+      "epoch": 0.4387231815803244,
+      "grad_norm": 2.407371693080437,
+      "learning_rate": 1.2446033552406763e-05,
+      "loss": 0.8304,
+      "step": 4192
+    },
+    {
+      "epoch": 0.4388278388278388,
+      "grad_norm": 2.388853746571948,
+      "learning_rate": 1.2442746660539226e-05,
+      "loss": 1.089,
+      "step": 4193
+    },
+    {
+      "epoch": 0.43893249607535323,
+      "grad_norm": 1.9365985500657477,
+      "learning_rate": 1.2439459487995145e-05,
+      "loss": 0.9722,
+      "step": 4194
+    },
+    {
+      "epoch": 0.43903715332286763,
+      "grad_norm": 2.963598627913234,
+      "learning_rate": 1.2436172035152223e-05,
+      "loss": 0.8233,
+      "step": 4195
+    },
+    {
+      "epoch": 0.439141810570382,
+      "grad_norm": 2.2703125367436363,
+      "learning_rate": 1.2432884302388198e-05,
+      "loss": 1.0116,
+      "step": 4196
+    },
+    {
+      "epoch": 0.4392464678178964,
+      "grad_norm": 2.129118913151718,
+      "learning_rate": 1.2429596290080832e-05,
+      "loss": 1.0584,
+      "step": 4197
+    },
+    {
+      "epoch": 0.4393511250654108,
+      "grad_norm": 2.0256017953605383,
+      "learning_rate": 1.2426307998607929e-05,
+      "loss": 1.008,
+      "step": 4198
+    },
+    {
+      "epoch": 0.43945578231292515,
+      "grad_norm": 2.028462889521559,
+      "learning_rate": 1.2423019428347316e-05,
+      "loss": 0.9788,
+      "step": 4199
+    },
+    {
+      "epoch": 0.43956043956043955,
+      "grad_norm": 2.1206707709023003,
+      "learning_rate": 1.2419730579676861e-05,
+      "loss": 0.8672,
+      "step": 4200
+    },
+    {
+      "epoch": 0.43966509680795396,
+      "grad_norm": 2.0885573335422705,
+      "learning_rate": 1.2416441452974457e-05,
+      "loss": 0.934,
+      "step": 4201
+    },
+    {
+      "epoch": 0.43976975405546836,
+      "grad_norm": 1.9950363704570153,
+      "learning_rate": 1.241315204861803e-05,
+      "loss": 0.9081,
+      "step": 4202
+    },
+    {
+      "epoch": 0.4398744113029827,
+      "grad_norm": 1.8813239970215452,
+      "learning_rate": 1.2409862366985541e-05,
+      "loss": 0.8674,
+      "step": 4203
+    },
+    {
+      "epoch": 0.4399790685504971,
+      "grad_norm": 2.1510971349415757,
+      "learning_rate": 1.240657240845498e-05,
+      "loss": 0.9775,
+      "step": 4204
+    },
+    {
+      "epoch": 0.4400837257980115,
+      "grad_norm": 2.455900196950652,
+      "learning_rate": 1.2403282173404373e-05,
+      "loss": 0.8979,
+      "step": 4205
+    },
+    {
+      "epoch": 0.4401883830455259,
+      "grad_norm": 2.003615035493078,
+      "learning_rate": 1.239999166221177e-05,
+      "loss": 0.949,
+      "step": 4206
+    },
+    {
+      "epoch": 0.4402930402930403,
+      "grad_norm": 2.122829687717139,
+      "learning_rate": 1.2396700875255263e-05,
+      "loss": 1.0648,
+      "step": 4207
+    },
+    {
+      "epoch": 0.4403976975405547,
+      "grad_norm": 1.7612998154135466,
+      "learning_rate": 1.2393409812912963e-05,
+      "loss": 0.8243,
+      "step": 4208
+    },
+    {
+      "epoch": 0.4405023547880691,
+      "grad_norm": 2.151390451235141,
+      "learning_rate": 1.2390118475563024e-05,
+      "loss": 1.1014,
+      "step": 4209
+    },
+    {
+      "epoch": 0.44060701203558345,
+      "grad_norm": 1.769570106535323,
+      "learning_rate": 1.238682686358363e-05,
+      "loss": 0.9075,
+      "step": 4210
+    },
+    {
+      "epoch": 0.44071166928309785,
+      "grad_norm": 1.8741126127698302,
+      "learning_rate": 1.2383534977352989e-05,
+      "loss": 0.9628,
+      "step": 4211
+    },
+    {
+      "epoch": 0.44081632653061226,
+      "grad_norm": 2.11490519415388,
+      "learning_rate": 1.2380242817249343e-05,
+      "loss": 1.0346,
+      "step": 4212
+    },
+    {
+      "epoch": 0.44092098377812666,
+      "grad_norm": 1.9472990388460194,
+      "learning_rate": 1.2376950383650978e-05,
+      "loss": 0.9551,
+      "step": 4213
+    },
+    {
+      "epoch": 0.441025641025641,
+      "grad_norm": 2.1912277529516357,
+      "learning_rate": 1.2373657676936194e-05,
+      "loss": 0.9105,
+      "step": 4214
+    },
+    {
+      "epoch": 0.4411302982731554,
+      "grad_norm": 2.216215119841429,
+      "learning_rate": 1.237036469748333e-05,
+      "loss": 1.0117,
+      "step": 4215
+    },
+    {
+      "epoch": 0.4412349555206698,
+      "grad_norm": 1.9395445115395555,
+      "learning_rate": 1.2367071445670755e-05,
+      "loss": 0.9748,
+      "step": 4216
+    },
+    {
+      "epoch": 0.4413396127681842,
+      "grad_norm": 2.4154728984909717,
+      "learning_rate": 1.2363777921876876e-05,
+      "loss": 1.048,
+      "step": 4217
+    },
+    {
+      "epoch": 0.4414442700156986,
+      "grad_norm": 2.2295606753839006,
+      "learning_rate": 1.2360484126480121e-05,
+      "loss": 1.1085,
+      "step": 4218
+    },
+    {
+      "epoch": 0.441548927263213,
+      "grad_norm": 21.310321961686725,
+      "learning_rate": 1.2357190059858955e-05,
+      "loss": 0.9542,
+      "step": 4219
+    },
+    {
+      "epoch": 0.4416535845107274,
+      "grad_norm": 2.275313654143178,
+      "learning_rate": 1.2353895722391873e-05,
+      "loss": 1.0865,
+      "step": 4220
+    },
+    {
+      "epoch": 0.44175824175824174,
+      "grad_norm": 2.008270022618239,
+      "learning_rate": 1.2350601114457397e-05,
+      "loss": 0.8425,
+      "step": 4221
+    },
+    {
+      "epoch": 0.44186289900575615,
+      "grad_norm": 30.912506956421982,
+      "learning_rate": 1.2347306236434091e-05,
+      "loss": 1.8411,
+      "step": 4222
+    },
+    {
+      "epoch": 0.44196755625327055,
+      "grad_norm": 14.423513861175078,
+      "learning_rate": 1.234401108870054e-05,
+      "loss": 1.2736,
+      "step": 4223
+    },
+    {
+      "epoch": 0.4420722135007849,
+      "grad_norm": 2.5430152347578816,
+      "learning_rate": 1.2340715671635365e-05,
+      "loss": 0.9757,
+      "step": 4224
+    },
+    {
+      "epoch": 0.4421768707482993,
+      "grad_norm": 18.390708578682016,
+      "learning_rate": 1.2337419985617209e-05,
+      "loss": 1.0966,
+      "step": 4225
+    },
+    {
+      "epoch": 0.4422815279958137,
+      "grad_norm": 2.1563225849286964,
+      "learning_rate": 1.2334124031024759e-05,
+      "loss": 0.9753,
+      "step": 4226
+    },
+    {
+      "epoch": 0.4423861852433281,
+      "grad_norm": 2.209454133083187,
+      "learning_rate": 1.2330827808236729e-05,
+      "loss": 1.0031,
+      "step": 4227
+    },
+    {
+      "epoch": 0.4424908424908425,
+      "grad_norm": 2.0533065013534006,
+      "learning_rate": 1.2327531317631858e-05,
+      "loss": 1.0234,
+      "step": 4228
+    },
+    {
+      "epoch": 0.4425954997383569,
+      "grad_norm": 2.267657972321477,
+      "learning_rate": 1.232423455958892e-05,
+      "loss": 0.9795,
+      "step": 4229
+    },
+    {
+      "epoch": 0.4427001569858713,
+      "grad_norm": 2.3724169669212163,
+      "learning_rate": 1.232093753448672e-05,
+      "loss": 1.0609,
+      "step": 4230
+    },
+    {
+      "epoch": 0.44280481423338564,
+      "grad_norm": 2.2647618736763864,
+      "learning_rate": 1.2317640242704089e-05,
+      "loss": 0.9277,
+      "step": 4231
+    },
+    {
+      "epoch": 0.44290947148090004,
+      "grad_norm": 2.426127300716815,
+      "learning_rate": 1.23143426846199e-05,
+      "loss": 0.9978,
+      "step": 4232
+    },
+    {
+      "epoch": 0.44301412872841445,
+      "grad_norm": 12.748064877007662,
+      "learning_rate": 1.2311044860613044e-05,
+      "loss": 1.059,
+      "step": 4233
+    },
+    {
+      "epoch": 0.44311878597592885,
+      "grad_norm": 2.2159616926614953,
+      "learning_rate": 1.2307746771062445e-05,
+      "loss": 0.9901,
+      "step": 4234
+    },
+    {
+      "epoch": 0.4432234432234432,
+      "grad_norm": 1.9980360395518977,
+      "learning_rate": 1.2304448416347066e-05,
+      "loss": 0.8674,
+      "step": 4235
+    },
+    {
+      "epoch": 0.4433281004709576,
+      "grad_norm": 2.2959534557033416,
+      "learning_rate": 1.2301149796845895e-05,
+      "loss": 1.0436,
+      "step": 4236
+    },
+    {
+      "epoch": 0.443432757718472,
+      "grad_norm": 2.449657687751539,
+      "learning_rate": 1.2297850912937946e-05,
+      "loss": 1.1151,
+      "step": 4237
+    },
+    {
+      "epoch": 0.4435374149659864,
+      "grad_norm": 1.852361721600591,
+      "learning_rate": 1.2294551765002265e-05,
+      "loss": 0.8302,
+      "step": 4238
+    },
+    {
+      "epoch": 0.44364207221350077,
+      "grad_norm": 1.7397853007381971,
+      "learning_rate": 1.229125235341794e-05,
+      "loss": 0.839,
+      "step": 4239
+    },
+    {
+      "epoch": 0.4437467294610152,
+      "grad_norm": 2.309355901645421,
+      "learning_rate": 1.228795267856407e-05,
+      "loss": 1.0563,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4438513867085296,
+      "grad_norm": 2.1118423440247627,
+      "learning_rate": 1.2284652740819803e-05,
+      "loss": 1.0022,
+      "step": 4241
+    },
+    {
+      "epoch": 0.44395604395604393,
+      "grad_norm": 2.214872624726237,
+      "learning_rate": 1.2281352540564302e-05,
+      "loss": 0.9913,
+      "step": 4242
+    },
+    {
+      "epoch": 0.44406070120355834,
+      "grad_norm": 2.239625811455294,
+      "learning_rate": 1.227805207817677e-05,
+      "loss": 0.9527,
+      "step": 4243
+    },
+    {
+      "epoch": 0.44416535845107274,
+      "grad_norm": 31.368906412723593,
+      "learning_rate": 1.2274751354036439e-05,
+      "loss": 1.574,
+      "step": 4244
+    },
+    {
+      "epoch": 0.44427001569858715,
+      "grad_norm": 1.9338765858884037,
+      "learning_rate": 1.2271450368522562e-05,
+      "loss": 0.9677,
+      "step": 4245
+    },
+    {
+      "epoch": 0.4443746729461015,
+      "grad_norm": 2.531637449689326,
+      "learning_rate": 1.2268149122014436e-05,
+      "loss": 1.0082,
+      "step": 4246
+    },
+    {
+      "epoch": 0.4444793301936159,
+      "grad_norm": 12.934239661888748,
+      "learning_rate": 1.2264847614891373e-05,
+      "loss": 1.0943,
+      "step": 4247
+    },
+    {
+      "epoch": 0.4445839874411303,
+      "grad_norm": 2.519052355502292,
+      "learning_rate": 1.2261545847532731e-05,
+      "loss": 0.958,
+      "step": 4248
+    },
+    {
+      "epoch": 0.44468864468864466,
+      "grad_norm": 109.81451808318943,
+      "learning_rate": 1.225824382031789e-05,
+      "loss": 2.7126,
+      "step": 4249
+    },
+    {
+      "epoch": 0.44479330193615907,
+      "grad_norm": 2.461447085373303,
+      "learning_rate": 1.2254941533626253e-05,
+      "loss": 0.9693,
+      "step": 4250
+    },
+    {
+      "epoch": 0.4448979591836735,
+      "grad_norm": 2.9594119092355458,
+      "learning_rate": 1.225163898783726e-05,
+      "loss": 0.8497,
+      "step": 4251
+    },
+    {
+      "epoch": 0.4450026164311879,
+      "grad_norm": 5.254203001278294,
+      "learning_rate": 1.2248336183330384e-05,
+      "loss": 1.0249,
+      "step": 4252
+    },
+    {
+      "epoch": 0.44510727367870223,
+      "grad_norm": 4.105904824801587,
+      "learning_rate": 1.2245033120485127e-05,
+      "loss": 0.9186,
+      "step": 4253
+    },
+    {
+      "epoch": 0.44521193092621664,
+      "grad_norm": 2.945698769993366,
+      "learning_rate": 1.2241729799681008e-05,
+      "loss": 0.8103,
+      "step": 4254
+    },
+    {
+      "epoch": 0.44531658817373104,
+      "grad_norm": 5.8775538590842995,
+      "learning_rate": 1.223842622129759e-05,
+      "loss": 1.147,
+      "step": 4255
+    },
+    {
+      "epoch": 0.44542124542124545,
+      "grad_norm": 30.161697236880144,
+      "learning_rate": 1.2235122385714467e-05,
+      "loss": 4.1431,
+      "step": 4256
+    },
+    {
+      "epoch": 0.4455259026687598,
+      "grad_norm": 8.326858087895113,
+      "learning_rate": 1.2231818293311246e-05,
+      "loss": 1.1568,
+      "step": 4257
+    },
+    {
+      "epoch": 0.4456305599162742,
+      "grad_norm": 9.194801521266452,
+      "learning_rate": 1.2228513944467581e-05,
+      "loss": 1.2268,
+      "step": 4258
+    },
+    {
+      "epoch": 0.4457352171637886,
+      "grad_norm": 7.57812928636459,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 1.1154,
+      "step": 4259
+    },
+    {
+      "epoch": 0.44583987441130296,
+      "grad_norm": 6.130800488712524,
+      "learning_rate": 1.2221904478977646e-05,
+      "loss": 0.9692,
+      "step": 4260
+    },
+    {
+      "epoch": 0.44594453165881737,
+      "grad_norm": 4.833189877025352,
+      "learning_rate": 1.2218599363090818e-05,
+      "loss": 0.9329,
+      "step": 4261
+    },
+    {
+      "epoch": 0.4460491889063318,
+      "grad_norm": 3.840930122922004,
+      "learning_rate": 1.2215293992282424e-05,
+      "loss": 1.103,
+      "step": 4262
+    },
+    {
+      "epoch": 0.4461538461538462,
+      "grad_norm": 3.8094608876035267,
+      "learning_rate": 1.2211988366932262e-05,
+      "loss": 1.0614,
+      "step": 4263
+    },
+    {
+      "epoch": 0.44625850340136053,
+      "grad_norm": 4.16696149043726,
+      "learning_rate": 1.2208682487420149e-05,
+      "loss": 0.9907,
+      "step": 4264
+    },
+    {
+      "epoch": 0.44636316064887493,
+      "grad_norm": 4.495255491410338,
+      "learning_rate": 1.2205376354125943e-05,
+      "loss": 1.1127,
+      "step": 4265
+    },
+    {
+      "epoch": 0.44646781789638934,
+      "grad_norm": 3.289846433479056,
+      "learning_rate": 1.2202069967429524e-05,
+      "loss": 1.0025,
+      "step": 4266
+    },
+    {
+      "epoch": 0.4465724751439037,
+      "grad_norm": 2.922002497701106,
+      "learning_rate": 1.2198763327710801e-05,
+      "loss": 1.0079,
+      "step": 4267
+    },
+    {
+      "epoch": 0.4466771323914181,
+      "grad_norm": 2.33163105345098,
+      "learning_rate": 1.2195456435349714e-05,
+      "loss": 1.0025,
+      "step": 4268
+    },
+    {
+      "epoch": 0.4467817896389325,
+      "grad_norm": 2.993882810202546,
+      "learning_rate": 1.219214929072623e-05,
+      "loss": 1.0342,
+      "step": 4269
+    },
+    {
+      "epoch": 0.4468864468864469,
+      "grad_norm": 3.138840354515787,
+      "learning_rate": 1.2188841894220351e-05,
+      "loss": 1.0363,
+      "step": 4270
+    },
+    {
+      "epoch": 0.44699110413396126,
+      "grad_norm": 2.8709505565686144,
+      "learning_rate": 1.21855342462121e-05,
+      "loss": 0.9547,
+      "step": 4271
+    },
+    {
+      "epoch": 0.44709576138147566,
+      "grad_norm": 3.5060252310763858,
+      "learning_rate": 1.2182226347081536e-05,
+      "loss": 1.0352,
+      "step": 4272
+    },
+    {
+      "epoch": 0.44720041862899007,
+      "grad_norm": 3.067631908770296,
+      "learning_rate": 1.2178918197208737e-05,
+      "loss": 0.936,
+      "step": 4273
+    },
+    {
+      "epoch": 0.4473050758765044,
+      "grad_norm": 3.033451592129497,
+      "learning_rate": 1.2175609796973822e-05,
+      "loss": 0.9364,
+      "step": 4274
+    },
+    {
+      "epoch": 0.4474097331240188,
+      "grad_norm": 2.0393393340393087,
+      "learning_rate": 1.2172301146756933e-05,
+      "loss": 0.9712,
+      "step": 4275
+    },
+    {
+      "epoch": 0.44751439037153323,
+      "grad_norm": 38.81812474011344,
+      "learning_rate": 1.2168992246938235e-05,
+      "loss": 2.2501,
+      "step": 4276
+    },
+    {
+      "epoch": 0.44761904761904764,
+      "grad_norm": 2.253226221283191,
+      "learning_rate": 1.2165683097897931e-05,
+      "loss": 1.1345,
+      "step": 4277
+    },
+    {
+      "epoch": 0.447723704866562,
+      "grad_norm": 2.3205323315504813,
+      "learning_rate": 1.2162373700016253e-05,
+      "loss": 1.0651,
+      "step": 4278
+    },
+    {
+      "epoch": 0.4478283621140764,
+      "grad_norm": 4.7092329513884295,
+      "learning_rate": 1.2159064053673449e-05,
+      "loss": 0.9147,
+      "step": 4279
+    },
+    {
+      "epoch": 0.4479330193615908,
+      "grad_norm": 29.970922275864194,
+      "learning_rate": 1.2155754159249811e-05,
+      "loss": 1.2273,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4480376766091052,
+      "grad_norm": 2.187035261760947,
+      "learning_rate": 1.2152444017125647e-05,
+      "loss": 0.9912,
+      "step": 4281
+    },
+    {
+      "epoch": 0.44814233385661956,
+      "grad_norm": 3.2809808282344446,
+      "learning_rate": 1.2149133627681303e-05,
+      "loss": 1.1204,
+      "step": 4282
+    },
+    {
+      "epoch": 0.44824699110413396,
+      "grad_norm": 2.045684946775204,
+      "learning_rate": 1.2145822991297148e-05,
+      "loss": 1.0637,
+      "step": 4283
+    },
+    {
+      "epoch": 0.44835164835164837,
+      "grad_norm": 2.000665560599059,
+      "learning_rate": 1.2142512108353583e-05,
+      "loss": 1.0627,
+      "step": 4284
+    },
+    {
+      "epoch": 0.4484563055991627,
+      "grad_norm": 1.879007116893966,
+      "learning_rate": 1.2139200979231033e-05,
+      "loss": 0.9889,
+      "step": 4285
+    },
+    {
+      "epoch": 0.4485609628466771,
+      "grad_norm": 2.191861241077447,
+      "learning_rate": 1.2135889604309952e-05,
+      "loss": 0.9967,
+      "step": 4286
+    },
+    {
+      "epoch": 0.44866562009419153,
+      "grad_norm": 2.090962524194402,
+      "learning_rate": 1.2132577983970828e-05,
+      "loss": 0.9794,
+      "step": 4287
+    },
+    {
+      "epoch": 0.44877027734170594,
+      "grad_norm": 1.88873946046564,
+      "learning_rate": 1.212926611859417e-05,
+      "loss": 0.9027,
+      "step": 4288
+    },
+    {
+      "epoch": 0.4488749345892203,
+      "grad_norm": 2.2151145974628763,
+      "learning_rate": 1.212595400856052e-05,
+      "loss": 0.9032,
+      "step": 4289
+    },
+    {
+      "epoch": 0.4489795918367347,
+      "grad_norm": 2.130497424667586,
+      "learning_rate": 1.212264165425044e-05,
+      "loss": 0.9292,
+      "step": 4290
+    },
+    {
+      "epoch": 0.4490842490842491,
+      "grad_norm": 2.0469637222247554,
+      "learning_rate": 1.2119329056044533e-05,
+      "loss": 1.0018,
+      "step": 4291
+    },
+    {
+      "epoch": 0.44918890633176345,
+      "grad_norm": 2.5073666390831466,
+      "learning_rate": 1.2116016214323423e-05,
+      "loss": 1.0831,
+      "step": 4292
+    },
+    {
+      "epoch": 0.44929356357927785,
+      "grad_norm": 2.1859064919513522,
+      "learning_rate": 1.2112703129467758e-05,
+      "loss": 1.0436,
+      "step": 4293
+    },
+    {
+      "epoch": 0.44939822082679226,
+      "grad_norm": 2.0934320510193682,
+      "learning_rate": 1.2109389801858225e-05,
+      "loss": 0.9668,
+      "step": 4294
+    },
+    {
+      "epoch": 0.44950287807430667,
+      "grad_norm": 2.1756826044133004,
+      "learning_rate": 1.2106076231875524e-05,
+      "loss": 0.9583,
+      "step": 4295
+    },
+    {
+      "epoch": 0.449607535321821,
+      "grad_norm": 2.353709737315531,
+      "learning_rate": 1.2102762419900391e-05,
+      "loss": 1.0629,
+      "step": 4296
+    },
+    {
+      "epoch": 0.4497121925693354,
+      "grad_norm": 1.952613682916308,
+      "learning_rate": 1.20994483663136e-05,
+      "loss": 0.9243,
+      "step": 4297
+    },
+    {
+      "epoch": 0.44981684981684983,
+      "grad_norm": 2.2467039184701445,
+      "learning_rate": 1.209613407149593e-05,
+      "loss": 1.0368,
+      "step": 4298
+    },
+    {
+      "epoch": 0.44992150706436423,
+      "grad_norm": 2.0181287771348284,
+      "learning_rate": 1.2092819535828208e-05,
+      "loss": 0.8519,
+      "step": 4299
+    },
+    {
+      "epoch": 0.4500261643118786,
+      "grad_norm": 1.9314596098382928,
+      "learning_rate": 1.208950475969128e-05,
+      "loss": 0.8568,
+      "step": 4300
+    },
+    {
+      "epoch": 0.450130821559393,
+      "grad_norm": 1.832701400236754,
+      "learning_rate": 1.208618974346602e-05,
+      "loss": 0.927,
+      "step": 4301
+    },
+    {
+      "epoch": 0.4502354788069074,
+      "grad_norm": 1.9459296613855224,
+      "learning_rate": 1.2082874487533324e-05,
+      "loss": 1.0155,
+      "step": 4302
+    },
+    {
+      "epoch": 0.45034013605442175,
+      "grad_norm": 2.3792299560839667,
+      "learning_rate": 1.2079558992274131e-05,
+      "loss": 1.0069,
+      "step": 4303
+    },
+    {
+      "epoch": 0.45044479330193615,
+      "grad_norm": 2.170868389797402,
+      "learning_rate": 1.2076243258069396e-05,
+      "loss": 1.0151,
+      "step": 4304
+    },
+    {
+      "epoch": 0.45054945054945056,
+      "grad_norm": 2.2013372328892458,
+      "learning_rate": 1.2072927285300099e-05,
+      "loss": 0.9911,
+      "step": 4305
+    },
+    {
+      "epoch": 0.45065410779696496,
+      "grad_norm": 2.2705803608142863,
+      "learning_rate": 1.2069611074347255e-05,
+      "loss": 1.0045,
+      "step": 4306
+    },
+    {
+      "epoch": 0.4507587650444793,
+      "grad_norm": 1.9277985272803437,
+      "learning_rate": 1.2066294625591903e-05,
+      "loss": 0.8673,
+      "step": 4307
+    },
+    {
+      "epoch": 0.4508634222919937,
+      "grad_norm": 2.0440176202060156,
+      "learning_rate": 1.206297793941511e-05,
+      "loss": 1.0108,
+      "step": 4308
+    },
+    {
+      "epoch": 0.4509680795395081,
+      "grad_norm": 2.5462646813427097,
+      "learning_rate": 1.205966101619797e-05,
+      "loss": 0.859,
+      "step": 4309
+    },
+    {
+      "epoch": 0.4510727367870225,
+      "grad_norm": 2.248355643793843,
+      "learning_rate": 1.2056343856321605e-05,
+      "loss": 1.1021,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4511773940345369,
+      "grad_norm": 4.908250414144032,
+      "learning_rate": 1.2053026460167158e-05,
+      "loss": 1.0568,
+      "step": 4311
+    },
+    {
+      "epoch": 0.4512820512820513,
+      "grad_norm": 2.248952995961383,
+      "learning_rate": 1.2049708828115811e-05,
+      "loss": 0.9196,
+      "step": 4312
+    },
+    {
+      "epoch": 0.4513867085295657,
+      "grad_norm": 1.9290283733393867,
+      "learning_rate": 1.2046390960548765e-05,
+      "loss": 1.0563,
+      "step": 4313
+    },
+    {
+      "epoch": 0.45149136577708004,
+      "grad_norm": 2.0660674453272336,
+      "learning_rate": 1.204307285784725e-05,
+      "loss": 0.9519,
+      "step": 4314
+    },
+    {
+      "epoch": 0.45159602302459445,
+      "grad_norm": 1.996879072685833,
+      "learning_rate": 1.2039754520392522e-05,
+      "loss": 1.0722,
+      "step": 4315
+    },
+    {
+      "epoch": 0.45170068027210886,
+      "grad_norm": 2.5022938436613624,
+      "learning_rate": 1.2036435948565863e-05,
+      "loss": 1.0778,
+      "step": 4316
+    },
+    {
+      "epoch": 0.4518053375196232,
+      "grad_norm": 2.2109024944276885,
+      "learning_rate": 1.2033117142748585e-05,
+      "loss": 1.0058,
+      "step": 4317
+    },
+    {
+      "epoch": 0.4519099947671376,
+      "grad_norm": 2.2693762543652674,
+      "learning_rate": 1.2029798103322028e-05,
+      "loss": 0.9833,
+      "step": 4318
+    },
+    {
+      "epoch": 0.452014652014652,
+      "grad_norm": 2.116453756308576,
+      "learning_rate": 1.2026478830667551e-05,
+      "loss": 0.9363,
+      "step": 4319
+    },
+    {
+      "epoch": 0.4521193092621664,
+      "grad_norm": 3.342959773212483,
+      "learning_rate": 1.2023159325166551e-05,
+      "loss": 0.8688,
+      "step": 4320
+    },
+    {
+      "epoch": 0.4522239665096808,
+      "grad_norm": 2.3163640304872812,
+      "learning_rate": 1.2019839587200442e-05,
+      "loss": 0.8287,
+      "step": 4321
+    },
+    {
+      "epoch": 0.4523286237571952,
+      "grad_norm": 1.9681851990856516,
+      "learning_rate": 1.2016519617150668e-05,
+      "loss": 0.9896,
+      "step": 4322
+    },
+    {
+      "epoch": 0.4524332810047096,
+      "grad_norm": 2.143371304104883,
+      "learning_rate": 1.2013199415398706e-05,
+      "loss": 0.8437,
+      "step": 4323
+    },
+    {
+      "epoch": 0.452537938252224,
+      "grad_norm": 2.4121771940664587,
+      "learning_rate": 1.2009878982326047e-05,
+      "loss": 0.898,
+      "step": 4324
+    },
+    {
+      "epoch": 0.45264259549973834,
+      "grad_norm": 2.2077705572979847,
+      "learning_rate": 1.2006558318314219e-05,
+      "loss": 0.981,
+      "step": 4325
+    },
+    {
+      "epoch": 0.45274725274725275,
+      "grad_norm": 1.8137959654643108,
+      "learning_rate": 1.2003237423744773e-05,
+      "loss": 0.9494,
+      "step": 4326
+    },
+    {
+      "epoch": 0.45285190999476715,
+      "grad_norm": 2.043816728834952,
+      "learning_rate": 1.1999916298999285e-05,
+      "loss": 1.0507,
+      "step": 4327
+    },
+    {
+      "epoch": 0.4529565672422815,
+      "grad_norm": 1.8399500676215097,
+      "learning_rate": 1.1996594944459363e-05,
+      "loss": 0.9921,
+      "step": 4328
+    },
+    {
+      "epoch": 0.4530612244897959,
+      "grad_norm": 1.8888260838202213,
+      "learning_rate": 1.199327336050663e-05,
+      "loss": 1.0088,
+      "step": 4329
+    },
+    {
+      "epoch": 0.4531658817373103,
+      "grad_norm": 2.2320849845692474,
+      "learning_rate": 1.198995154752275e-05,
+      "loss": 1.0454,
+      "step": 4330
+    },
+    {
+      "epoch": 0.4532705389848247,
+      "grad_norm": 2.315466026238162,
+      "learning_rate": 1.1986629505889402e-05,
+      "loss": 0.8162,
+      "step": 4331
+    },
+    {
+      "epoch": 0.4533751962323391,
+      "grad_norm": 2.0448363790200745,
+      "learning_rate": 1.1983307235988297e-05,
+      "loss": 0.92,
+      "step": 4332
+    },
+    {
+      "epoch": 0.4534798534798535,
+      "grad_norm": 1.9783944834680685,
+      "learning_rate": 1.1979984738201172e-05,
+      "loss": 0.9508,
+      "step": 4333
+    },
+    {
+      "epoch": 0.4535845107273679,
+      "grad_norm": 2.1566268663323167,
+      "learning_rate": 1.1976662012909781e-05,
+      "loss": 0.9213,
+      "step": 4334
+    },
+    {
+      "epoch": 0.45368916797488223,
+      "grad_norm": 2.088854843791031,
+      "learning_rate": 1.1973339060495923e-05,
+      "loss": 0.9617,
+      "step": 4335
+    },
+    {
+      "epoch": 0.45379382522239664,
+      "grad_norm": 2.0416167816146977,
+      "learning_rate": 1.1970015881341405e-05,
+      "loss": 0.9292,
+      "step": 4336
+    },
+    {
+      "epoch": 0.45389848246991105,
+      "grad_norm": 2.0230522167032206,
+      "learning_rate": 1.196669247582807e-05,
+      "loss": 0.8476,
+      "step": 4337
+    },
+    {
+      "epoch": 0.45400313971742545,
+      "grad_norm": 2.1730001331296704,
+      "learning_rate": 1.1963368844337779e-05,
+      "loss": 0.9025,
+      "step": 4338
+    },
+    {
+      "epoch": 0.4541077969649398,
+      "grad_norm": 1.8508787301092124,
+      "learning_rate": 1.196004498725243e-05,
+      "loss": 0.8897,
+      "step": 4339
+    },
+    {
+      "epoch": 0.4542124542124542,
+      "grad_norm": 2.2654361716293336,
+      "learning_rate": 1.195672090495394e-05,
+      "loss": 1.0252,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4543171114599686,
+      "grad_norm": 1.8928691469715988,
+      "learning_rate": 1.1953396597824249e-05,
+      "loss": 0.9031,
+      "step": 4341
+    },
+    {
+      "epoch": 0.454421768707483,
+      "grad_norm": 2.3645611004316702,
+      "learning_rate": 1.1950072066245328e-05,
+      "loss": 1.0823,
+      "step": 4342
+    },
+    {
+      "epoch": 0.45452642595499737,
+      "grad_norm": 1.7075263528138125,
+      "learning_rate": 1.1946747310599176e-05,
+      "loss": 0.9554,
+      "step": 4343
+    },
+    {
+      "epoch": 0.4546310832025118,
+      "grad_norm": 2.3091455726092884,
+      "learning_rate": 1.1943422331267811e-05,
+      "loss": 1.0067,
+      "step": 4344
+    },
+    {
+      "epoch": 0.4547357404500262,
+      "grad_norm": 2.098765765313036,
+      "learning_rate": 1.194009712863328e-05,
+      "loss": 0.9867,
+      "step": 4345
+    },
+    {
+      "epoch": 0.45484039769754053,
+      "grad_norm": 1.7727533702277667,
+      "learning_rate": 1.1936771703077656e-05,
+      "loss": 0.9671,
+      "step": 4346
+    },
+    {
+      "epoch": 0.45494505494505494,
+      "grad_norm": 2.074696485047073,
+      "learning_rate": 1.1933446054983035e-05,
+      "loss": 0.9257,
+      "step": 4347
+    },
+    {
+      "epoch": 0.45504971219256934,
+      "grad_norm": 2.039328732869746,
+      "learning_rate": 1.1930120184731544e-05,
+      "loss": 0.9301,
+      "step": 4348
+    },
+    {
+      "epoch": 0.45515436944008375,
+      "grad_norm": 2.1160277945116586,
+      "learning_rate": 1.1926794092705335e-05,
+      "loss": 0.9874,
+      "step": 4349
+    },
+    {
+      "epoch": 0.4552590266875981,
+      "grad_norm": 1.8772607321330006,
+      "learning_rate": 1.1923467779286573e-05,
+      "loss": 0.941,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4553636839351125,
+      "grad_norm": 1.9022941147099999,
+      "learning_rate": 1.1920141244857466e-05,
+      "loss": 0.908,
+      "step": 4351
+    },
+    {
+      "epoch": 0.4554683411826269,
+      "grad_norm": 2.3523711222096404,
+      "learning_rate": 1.191681448980024e-05,
+      "loss": 0.8785,
+      "step": 4352
+    },
+    {
+      "epoch": 0.45557299843014126,
+      "grad_norm": 2.002200825932997,
+      "learning_rate": 1.191348751449714e-05,
+      "loss": 1.0048,
+      "step": 4353
+    },
+    {
+      "epoch": 0.45567765567765567,
+      "grad_norm": 2.0270694451502282,
+      "learning_rate": 1.1910160319330446e-05,
+      "loss": 1.0233,
+      "step": 4354
+    },
+    {
+      "epoch": 0.4557823129251701,
+      "grad_norm": 2.0541524762488814,
+      "learning_rate": 1.1906832904682459e-05,
+      "loss": 0.7993,
+      "step": 4355
+    },
+    {
+      "epoch": 0.4558869701726845,
+      "grad_norm": 1.8638621351295752,
+      "learning_rate": 1.1903505270935507e-05,
+      "loss": 0.8905,
+      "step": 4356
+    },
+    {
+      "epoch": 0.45599162742019883,
+      "grad_norm": 1.9736021287043854,
+      "learning_rate": 1.190017741847194e-05,
+      "loss": 1.0373,
+      "step": 4357
+    },
+    {
+      "epoch": 0.45609628466771324,
+      "grad_norm": 1.8797795224789213,
+      "learning_rate": 1.1896849347674135e-05,
+      "loss": 0.9821,
+      "step": 4358
+    },
+    {
+      "epoch": 0.45620094191522764,
+      "grad_norm": 1.9368205053092564,
+      "learning_rate": 1.1893521058924497e-05,
+      "loss": 1.0668,
+      "step": 4359
+    },
+    {
+      "epoch": 0.456305599162742,
+      "grad_norm": 2.132381992165538,
+      "learning_rate": 1.1890192552605447e-05,
+      "loss": 0.9218,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4564102564102564,
+      "grad_norm": 2.0456266510770056,
+      "learning_rate": 1.1886863829099442e-05,
+      "loss": 0.8965,
+      "step": 4361
+    },
+    {
+      "epoch": 0.4565149136577708,
+      "grad_norm": 2.2461716024926823,
+      "learning_rate": 1.1883534888788961e-05,
+      "loss": 1.072,
+      "step": 4362
+    },
+    {
+      "epoch": 0.4566195709052852,
+      "grad_norm": 1.8114815853829138,
+      "learning_rate": 1.18802057320565e-05,
+      "loss": 0.9784,
+      "step": 4363
+    },
+    {
+      "epoch": 0.45672422815279956,
+      "grad_norm": 2.074839338100702,
+      "learning_rate": 1.1876876359284589e-05,
+      "loss": 0.9458,
+      "step": 4364
+    },
+    {
+      "epoch": 0.45682888540031397,
+      "grad_norm": 3.272621021673472,
+      "learning_rate": 1.187354677085578e-05,
+      "loss": 1.0344,
+      "step": 4365
+    },
+    {
+      "epoch": 0.4569335426478284,
+      "grad_norm": 1.8423824960811983,
+      "learning_rate": 1.1870216967152649e-05,
+      "loss": 0.9991,
+      "step": 4366
+    },
+    {
+      "epoch": 0.4570381998953428,
+      "grad_norm": 2.0436626042627735,
+      "learning_rate": 1.1866886948557796e-05,
+      "loss": 0.9367,
+      "step": 4367
+    },
+    {
+      "epoch": 0.45714285714285713,
+      "grad_norm": 2.243427552622139,
+      "learning_rate": 1.1863556715453847e-05,
+      "loss": 1.081,
+      "step": 4368
+    },
+    {
+      "epoch": 0.45724751439037153,
+      "grad_norm": 2.1396337485328965,
+      "learning_rate": 1.1860226268223459e-05,
+      "loss": 1.0269,
+      "step": 4369
+    },
+    {
+      "epoch": 0.45735217163788594,
+      "grad_norm": 2.0912618560301715,
+      "learning_rate": 1.1856895607249297e-05,
+      "loss": 0.9635,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4574568288854003,
+      "grad_norm": 1.8134320104293313,
+      "learning_rate": 1.1853564732914066e-05,
+      "loss": 0.8776,
+      "step": 4371
+    },
+    {
+      "epoch": 0.4575614861329147,
+      "grad_norm": 1.873748145270781,
+      "learning_rate": 1.185023364560049e-05,
+      "loss": 0.8853,
+      "step": 4372
+    },
+    {
+      "epoch": 0.4576661433804291,
+      "grad_norm": 1.8136741058042303,
+      "learning_rate": 1.1846902345691317e-05,
+      "loss": 0.98,
+      "step": 4373
+    },
+    {
+      "epoch": 0.4577708006279435,
+      "grad_norm": 1.7292941149458805,
+      "learning_rate": 1.1843570833569324e-05,
+      "loss": 0.9699,
+      "step": 4374
+    },
+    {
+      "epoch": 0.45787545787545786,
+      "grad_norm": 2.6314289006744587,
+      "learning_rate": 1.1840239109617302e-05,
+      "loss": 1.1551,
+      "step": 4375
+    },
+    {
+      "epoch": 0.45798011512297226,
+      "grad_norm": 22.007184596120965,
+      "learning_rate": 1.1836907174218078e-05,
+      "loss": 1.8465,
+      "step": 4376
+    },
+    {
+      "epoch": 0.45808477237048667,
+      "grad_norm": 2.295041122834174,
+      "learning_rate": 1.183357502775449e-05,
+      "loss": 1.0055,
+      "step": 4377
+    },
+    {
+      "epoch": 0.458189429618001,
+      "grad_norm": 31.559871912501077,
+      "learning_rate": 1.1830242670609417e-05,
+      "loss": 1.7263,
+      "step": 4378
+    },
+    {
+      "epoch": 0.4582940868655154,
+      "grad_norm": 2.0871333437628055,
+      "learning_rate": 1.1826910103165753e-05,
+      "loss": 0.8551,
+      "step": 4379
+    },
+    {
+      "epoch": 0.45839874411302983,
+      "grad_norm": 2.726282817146809,
+      "learning_rate": 1.1823577325806414e-05,
+      "loss": 1.0138,
+      "step": 4380
+    },
+    {
+      "epoch": 0.45850340136054424,
+      "grad_norm": 2.1086177426453374,
+      "learning_rate": 1.1820244338914341e-05,
+      "loss": 0.9771,
+      "step": 4381
+    },
+    {
+      "epoch": 0.4586080586080586,
+      "grad_norm": 2.3764884198932,
+      "learning_rate": 1.1816911142872507e-05,
+      "loss": 1.0197,
+      "step": 4382
+    },
+    {
+      "epoch": 0.458712715855573,
+      "grad_norm": 2.045556899175085,
+      "learning_rate": 1.1813577738063897e-05,
+      "loss": 0.9103,
+      "step": 4383
+    },
+    {
+      "epoch": 0.4588173731030874,
+      "grad_norm": 2.5972265367678564,
+      "learning_rate": 1.1810244124871528e-05,
+      "loss": 1.0018,
+      "step": 4384
+    },
+    {
+      "epoch": 0.4589220303506018,
+      "grad_norm": 2.551709456154248,
+      "learning_rate": 1.1806910303678443e-05,
+      "loss": 1.0322,
+      "step": 4385
+    },
+    {
+      "epoch": 0.45902668759811616,
+      "grad_norm": 2.1748517522412407,
+      "learning_rate": 1.1803576274867698e-05,
+      "loss": 0.9927,
+      "step": 4386
+    },
+    {
+      "epoch": 0.45913134484563056,
+      "grad_norm": 3.812812110949521,
+      "learning_rate": 1.1800242038822382e-05,
+      "loss": 0.9246,
+      "step": 4387
+    },
+    {
+      "epoch": 0.45923600209314497,
+      "grad_norm": 2.376394222425811,
+      "learning_rate": 1.1796907595925608e-05,
+      "loss": 0.9445,
+      "step": 4388
+    },
+    {
+      "epoch": 0.4593406593406593,
+      "grad_norm": 2.5161863830002735,
+      "learning_rate": 1.179357294656051e-05,
+      "loss": 1.0452,
+      "step": 4389
+    },
+    {
+      "epoch": 0.4594453165881737,
+      "grad_norm": 2.6069106789560488,
+      "learning_rate": 1.179023809111024e-05,
+      "loss": 1.1245,
+      "step": 4390
+    },
+    {
+      "epoch": 0.45954997383568813,
+      "grad_norm": 2.4849062318250397,
+      "learning_rate": 1.1786903029957987e-05,
+      "loss": 1.0237,
+      "step": 4391
+    },
+    {
+      "epoch": 0.45965463108320254,
+      "grad_norm": 2.1071131463822357,
+      "learning_rate": 1.1783567763486954e-05,
+      "loss": 0.9831,
+      "step": 4392
+    },
+    {
+      "epoch": 0.4597592883307169,
+      "grad_norm": 2.7912058793141115,
+      "learning_rate": 1.1780232292080371e-05,
+      "loss": 0.9761,
+      "step": 4393
+    },
+    {
+      "epoch": 0.4598639455782313,
+      "grad_norm": 2.033218484100733,
+      "learning_rate": 1.1776896616121483e-05,
+      "loss": 0.8881,
+      "step": 4394
+    },
+    {
+      "epoch": 0.4599686028257457,
+      "grad_norm": 2.044503825084946,
+      "learning_rate": 1.1773560735993576e-05,
+      "loss": 1.0179,
+      "step": 4395
+    },
+    {
+      "epoch": 0.46007326007326005,
+      "grad_norm": 2.2543377501620308,
+      "learning_rate": 1.1770224652079943e-05,
+      "loss": 0.823,
+      "step": 4396
+    },
+    {
+      "epoch": 0.46017791732077445,
+      "grad_norm": 2.3807439687952217,
+      "learning_rate": 1.1766888364763912e-05,
+      "loss": 0.8314,
+      "step": 4397
+    },
+    {
+      "epoch": 0.46028257456828886,
+      "grad_norm": 1.970046805954839,
+      "learning_rate": 1.1763551874428823e-05,
+      "loss": 0.879,
+      "step": 4398
+    },
+    {
+      "epoch": 0.46038723181580327,
+      "grad_norm": 2.1330010409516116,
+      "learning_rate": 1.1760215181458048e-05,
+      "loss": 1.1079,
+      "step": 4399
+    },
+    {
+      "epoch": 0.4604918890633176,
+      "grad_norm": 1.9239002886217489,
+      "learning_rate": 1.1756878286234983e-05,
+      "loss": 0.9412,
+      "step": 4400
+    },
+    {
+      "epoch": 0.460596546310832,
+      "grad_norm": 2.1793334815489693,
+      "learning_rate": 1.1753541189143039e-05,
+      "loss": 0.8828,
+      "step": 4401
+    },
+    {
+      "epoch": 0.46070120355834643,
+      "grad_norm": 2.254420506227355,
+      "learning_rate": 1.1750203890565661e-05,
+      "loss": 0.8499,
+      "step": 4402
+    },
+    {
+      "epoch": 0.4608058608058608,
+      "grad_norm": 2.1754811115427843,
+      "learning_rate": 1.1746866390886304e-05,
+      "loss": 0.924,
+      "step": 4403
+    },
+    {
+      "epoch": 0.4609105180533752,
+      "grad_norm": 2.396703022879703,
+      "learning_rate": 1.1743528690488457e-05,
+      "loss": 0.936,
+      "step": 4404
+    },
+    {
+      "epoch": 0.4610151753008896,
+      "grad_norm": 1.8475247870093567,
+      "learning_rate": 1.1740190789755635e-05,
+      "loss": 0.9359,
+      "step": 4405
+    },
+    {
+      "epoch": 0.461119832548404,
+      "grad_norm": 1.9948668165368353,
+      "learning_rate": 1.1736852689071357e-05,
+      "loss": 0.9159,
+      "step": 4406
+    },
+    {
+      "epoch": 0.46122448979591835,
+      "grad_norm": 2.3311014731095474,
+      "learning_rate": 1.1733514388819186e-05,
+      "loss": 1.0174,
+      "step": 4407
+    },
+    {
+      "epoch": 0.46132914704343275,
+      "grad_norm": 2.657571922556327,
+      "learning_rate": 1.1730175889382697e-05,
+      "loss": 0.9448,
+      "step": 4408
+    },
+    {
+      "epoch": 0.46143380429094716,
+      "grad_norm": 1.9668833389759839,
+      "learning_rate": 1.1726837191145492e-05,
+      "loss": 0.9506,
+      "step": 4409
+    },
+    {
+      "epoch": 0.46153846153846156,
+      "grad_norm": 161.64348480668815,
+      "learning_rate": 1.1723498294491196e-05,
+      "loss": 3.7708,
+      "step": 4410
+    },
+    {
+      "epoch": 0.4616431187859759,
+      "grad_norm": 2.1254582340869836,
+      "learning_rate": 1.1720159199803447e-05,
+      "loss": 1.0292,
+      "step": 4411
+    },
+    {
+      "epoch": 0.4617477760334903,
+      "grad_norm": 3.149115475190535,
+      "learning_rate": 1.1716819907465922e-05,
+      "loss": 1.0485,
+      "step": 4412
+    },
+    {
+      "epoch": 0.4618524332810047,
+      "grad_norm": 2.7285678675780627,
+      "learning_rate": 1.1713480417862306e-05,
+      "loss": 1.0172,
+      "step": 4413
+    },
+    {
+      "epoch": 0.4619570905285191,
+      "grad_norm": 2.2641817986994957,
+      "learning_rate": 1.171014073137632e-05,
+      "loss": 0.829,
+      "step": 4414
+    },
+    {
+      "epoch": 0.4620617477760335,
+      "grad_norm": 21.38524678867409,
+      "learning_rate": 1.1706800848391694e-05,
+      "loss": 2.726,
+      "step": 4415
+    },
+    {
+      "epoch": 0.4621664050235479,
+      "grad_norm": 2.442948735347796,
+      "learning_rate": 1.1703460769292192e-05,
+      "loss": 0.9794,
+      "step": 4416
+    },
+    {
+      "epoch": 0.4622710622710623,
+      "grad_norm": 2.762838250862807,
+      "learning_rate": 1.1700120494461595e-05,
+      "loss": 1.0578,
+      "step": 4417
+    },
+    {
+      "epoch": 0.46237571951857664,
+      "grad_norm": 3.0564755310218885,
+      "learning_rate": 1.1696780024283708e-05,
+      "loss": 0.912,
+      "step": 4418
+    },
+    {
+      "epoch": 0.46248037676609105,
+      "grad_norm": 4.2059114992513935,
+      "learning_rate": 1.1693439359142356e-05,
+      "loss": 1.211,
+      "step": 4419
+    },
+    {
+      "epoch": 0.46258503401360546,
+      "grad_norm": 3.764016376227401,
+      "learning_rate": 1.1690098499421384e-05,
+      "loss": 1.0091,
+      "step": 4420
+    },
+    {
+      "epoch": 0.4626896912611198,
+      "grad_norm": 3.1796219459492705,
+      "learning_rate": 1.1686757445504673e-05,
+      "loss": 1.0657,
+      "step": 4421
+    },
+    {
+      "epoch": 0.4627943485086342,
+      "grad_norm": 2.579766465159248,
+      "learning_rate": 1.1683416197776113e-05,
+      "loss": 0.9986,
+      "step": 4422
+    },
+    {
+      "epoch": 0.4628990057561486,
+      "grad_norm": 2.122008428264213,
+      "learning_rate": 1.1680074756619614e-05,
+      "loss": 1.0272,
+      "step": 4423
+    },
+    {
+      "epoch": 0.463003663003663,
+      "grad_norm": 2.26997985317698,
+      "learning_rate": 1.1676733122419124e-05,
+      "loss": 0.9752,
+      "step": 4424
+    },
+    {
+      "epoch": 0.4631083202511774,
+      "grad_norm": 2.2969054913130313,
+      "learning_rate": 1.1673391295558596e-05,
+      "loss": 0.9876,
+      "step": 4425
+    },
+    {
+      "epoch": 0.4632129774986918,
+      "grad_norm": 2.7300799446270467,
+      "learning_rate": 1.1670049276422017e-05,
+      "loss": 1.006,
+      "step": 4426
+    },
+    {
+      "epoch": 0.4633176347462062,
+      "grad_norm": 2.843913255345786,
+      "learning_rate": 1.166670706539339e-05,
+      "loss": 1.0773,
+      "step": 4427
+    },
+    {
+      "epoch": 0.4634222919937206,
+      "grad_norm": 2.661110599574868,
+      "learning_rate": 1.1663364662856747e-05,
+      "loss": 0.9482,
+      "step": 4428
+    },
+    {
+      "epoch": 0.46352694924123494,
+      "grad_norm": 2.2380660123868115,
+      "learning_rate": 1.1660022069196126e-05,
+      "loss": 0.8998,
+      "step": 4429
+    },
+    {
+      "epoch": 0.46363160648874935,
+      "grad_norm": 2.178013787617105,
+      "learning_rate": 1.1656679284795607e-05,
+      "loss": 0.9949,
+      "step": 4430
+    },
+    {
+      "epoch": 0.46373626373626375,
+      "grad_norm": 2.1088259317784295,
+      "learning_rate": 1.165333631003928e-05,
+      "loss": 0.9476,
+      "step": 4431
+    },
+    {
+      "epoch": 0.4638409209837781,
+      "grad_norm": 2.185800456163915,
+      "learning_rate": 1.1649993145311258e-05,
+      "loss": 1.0237,
+      "step": 4432
+    },
+    {
+      "epoch": 0.4639455782312925,
+      "grad_norm": 2.148724627286947,
+      "learning_rate": 1.1646649790995679e-05,
+      "loss": 0.9447,
+      "step": 4433
+    },
+    {
+      "epoch": 0.4640502354788069,
+      "grad_norm": 2.4213799049050566,
+      "learning_rate": 1.1643306247476702e-05,
+      "loss": 0.9744,
+      "step": 4434
+    },
+    {
+      "epoch": 0.4641548927263213,
+      "grad_norm": 2.088289187776693,
+      "learning_rate": 1.1639962515138506e-05,
+      "loss": 0.9488,
+      "step": 4435
+    },
+    {
+      "epoch": 0.4642595499738357,
+      "grad_norm": 2.1310001246837653,
+      "learning_rate": 1.1636618594365291e-05,
+      "loss": 0.9596,
+      "step": 4436
+    },
+    {
+      "epoch": 0.4643642072213501,
+      "grad_norm": 2.4081476306831053,
+      "learning_rate": 1.1633274485541284e-05,
+      "loss": 0.9266,
+      "step": 4437
+    },
+    {
+      "epoch": 0.4644688644688645,
+      "grad_norm": 2.247732950837275,
+      "learning_rate": 1.1629930189050724e-05,
+      "loss": 1.141,
+      "step": 4438
+    },
+    {
+      "epoch": 0.46457352171637883,
+      "grad_norm": 2.7553884410693534,
+      "learning_rate": 1.1626585705277886e-05,
+      "loss": 0.977,
+      "step": 4439
+    },
+    {
+      "epoch": 0.46467817896389324,
+      "grad_norm": 2.093553580818768,
+      "learning_rate": 1.1623241034607049e-05,
+      "loss": 0.92,
+      "step": 4440
+    },
+    {
+      "epoch": 0.46478283621140765,
+      "grad_norm": 2.102413811719979,
+      "learning_rate": 1.161989617742253e-05,
+      "loss": 0.9748,
+      "step": 4441
+    },
+    {
+      "epoch": 0.46488749345892205,
+      "grad_norm": 2.1474470371663523,
+      "learning_rate": 1.1616551134108649e-05,
+      "loss": 0.8978,
+      "step": 4442
+    },
+    {
+      "epoch": 0.4649921507064364,
+      "grad_norm": 1.8755006879257996,
+      "learning_rate": 1.1613205905049774e-05,
+      "loss": 1.0037,
+      "step": 4443
+    },
+    {
+      "epoch": 0.4650968079539508,
+      "grad_norm": 12.367373504928828,
+      "learning_rate": 1.1609860490630265e-05,
+      "loss": 1.7202,
+      "step": 4444
+    },
+    {
+      "epoch": 0.4652014652014652,
+      "grad_norm": 2.434919186952808,
+      "learning_rate": 1.1606514891234526e-05,
+      "loss": 1.0571,
+      "step": 4445
+    },
+    {
+      "epoch": 0.46530612244897956,
+      "grad_norm": 8.610048201919636,
+      "learning_rate": 1.1603169107246966e-05,
+      "loss": 1.6151,
+      "step": 4446
+    },
+    {
+      "epoch": 0.46541077969649397,
+      "grad_norm": 1.9525960112801914,
+      "learning_rate": 1.1599823139052027e-05,
+      "loss": 0.875,
+      "step": 4447
+    },
+    {
+      "epoch": 0.4655154369440084,
+      "grad_norm": 2.2632342098228775,
+      "learning_rate": 1.1596476987034167e-05,
+      "loss": 1.0215,
+      "step": 4448
+    },
+    {
+      "epoch": 0.4656200941915228,
+      "grad_norm": 2.2366838939003335,
+      "learning_rate": 1.1593130651577861e-05,
+      "loss": 1.0237,
+      "step": 4449
+    },
+    {
+      "epoch": 0.46572475143903713,
+      "grad_norm": 38.16329198450561,
+      "learning_rate": 1.1589784133067619e-05,
+      "loss": 1.9354,
+      "step": 4450
+    },
+    {
+      "epoch": 0.46582940868655154,
+      "grad_norm": 2.3716405493770485,
+      "learning_rate": 1.1586437431887954e-05,
+      "loss": 0.8736,
+      "step": 4451
+    },
+    {
+      "epoch": 0.46593406593406594,
+      "grad_norm": 2.125681534199429,
+      "learning_rate": 1.158309054842341e-05,
+      "loss": 1.0289,
+      "step": 4452
+    },
+    {
+      "epoch": 0.46603872318158035,
+      "grad_norm": 2.3919411314017536,
+      "learning_rate": 1.1579743483058556e-05,
+      "loss": 0.9878,
+      "step": 4453
+    },
+    {
+      "epoch": 0.4661433804290947,
+      "grad_norm": 2.39638097963942,
+      "learning_rate": 1.1576396236177974e-05,
+      "loss": 1.0409,
+      "step": 4454
+    },
+    {
+      "epoch": 0.4662480376766091,
+      "grad_norm": 2.5103806573334215,
+      "learning_rate": 1.1573048808166265e-05,
+      "loss": 0.9669,
+      "step": 4455
+    },
+    {
+      "epoch": 0.4663526949241235,
+      "grad_norm": 19.54312815228122,
+      "learning_rate": 1.1569701199408063e-05,
+      "loss": 1.6532,
+      "step": 4456
+    },
+    {
+      "epoch": 0.46645735217163786,
+      "grad_norm": 2.118501244763855,
+      "learning_rate": 1.1566353410288009e-05,
+      "loss": 1.0849,
+      "step": 4457
+    },
+    {
+      "epoch": 0.46656200941915227,
+      "grad_norm": 1.7858040697407882,
+      "learning_rate": 1.1563005441190776e-05,
+      "loss": 0.8408,
+      "step": 4458
+    },
+    {
+      "epoch": 0.4666666666666667,
+      "grad_norm": 2.2969793541608166,
+      "learning_rate": 1.1559657292501044e-05,
+      "loss": 0.9529,
+      "step": 4459
+    },
+    {
+      "epoch": 0.4667713239141811,
+      "grad_norm": 2.5581364794755355,
+      "learning_rate": 1.1556308964603531e-05,
+      "loss": 0.8612,
+      "step": 4460
+    },
+    {
+      "epoch": 0.46687598116169543,
+      "grad_norm": 2.300009884961941,
+      "learning_rate": 1.155296045788296e-05,
+      "loss": 1.1076,
+      "step": 4461
+    },
+    {
+      "epoch": 0.46698063840920984,
+      "grad_norm": 1.9569146697798587,
+      "learning_rate": 1.1549611772724087e-05,
+      "loss": 0.9219,
+      "step": 4462
+    },
+    {
+      "epoch": 0.46708529565672424,
+      "grad_norm": 2.0103709930682876,
+      "learning_rate": 1.1546262909511679e-05,
+      "loss": 1.0447,
+      "step": 4463
+    },
+    {
+      "epoch": 0.4671899529042386,
+      "grad_norm": 2.2197999718443238,
+      "learning_rate": 1.1542913868630527e-05,
+      "loss": 1.0264,
+      "step": 4464
+    },
+    {
+      "epoch": 0.467294610151753,
+      "grad_norm": 1.8177368735546657,
+      "learning_rate": 1.1539564650465447e-05,
+      "loss": 0.9716,
+      "step": 4465
+    },
+    {
+      "epoch": 0.4673992673992674,
+      "grad_norm": 2.1059310144937617,
+      "learning_rate": 1.1536215255401265e-05,
+      "loss": 1.0372,
+      "step": 4466
+    },
+    {
+      "epoch": 0.4675039246467818,
+      "grad_norm": 2.197191188748699,
+      "learning_rate": 1.1532865683822836e-05,
+      "loss": 1.0703,
+      "step": 4467
+    },
+    {
+      "epoch": 0.46760858189429616,
+      "grad_norm": 2.6122799139573423,
+      "learning_rate": 1.1529515936115033e-05,
+      "loss": 1.0208,
+      "step": 4468
+    },
+    {
+      "epoch": 0.46771323914181057,
+      "grad_norm": 2.093417244729284,
+      "learning_rate": 1.1526166012662748e-05,
+      "loss": 0.8367,
+      "step": 4469
+    },
+    {
+      "epoch": 0.46781789638932497,
+      "grad_norm": 2.135644048890782,
+      "learning_rate": 1.1522815913850896e-05,
+      "loss": 0.905,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4679225536368394,
+      "grad_norm": 2.045207253531856,
+      "learning_rate": 1.1519465640064405e-05,
+      "loss": 0.8331,
+      "step": 4471
+    },
+    {
+      "epoch": 0.46802721088435373,
+      "grad_norm": 7.896758500820517,
+      "learning_rate": 1.1516115191688232e-05,
+      "loss": 1.3637,
+      "step": 4472
+    },
+    {
+      "epoch": 0.46813186813186813,
+      "grad_norm": 2.143393184203902,
+      "learning_rate": 1.151276456910735e-05,
+      "loss": 1.0432,
+      "step": 4473
+    },
+    {
+      "epoch": 0.46823652537938254,
+      "grad_norm": 2.075941729013746,
+      "learning_rate": 1.1509413772706757e-05,
+      "loss": 1.1271,
+      "step": 4474
+    },
+    {
+      "epoch": 0.4683411826268969,
+      "grad_norm": 1.863301564352949,
+      "learning_rate": 1.150606280287146e-05,
+      "loss": 0.9209,
+      "step": 4475
+    },
+    {
+      "epoch": 0.4684458398744113,
+      "grad_norm": 2.298175682660464,
+      "learning_rate": 1.1502711659986491e-05,
+      "loss": 0.9713,
+      "step": 4476
+    },
+    {
+      "epoch": 0.4685504971219257,
+      "grad_norm": 2.4343036503123994,
+      "learning_rate": 1.1499360344436908e-05,
+      "loss": 0.9463,
+      "step": 4477
+    },
+    {
+      "epoch": 0.4686551543694401,
+      "grad_norm": 2.1765146376199946,
+      "learning_rate": 1.1496008856607779e-05,
+      "loss": 0.9651,
+      "step": 4478
+    },
+    {
+      "epoch": 0.46875981161695446,
+      "grad_norm": 1.8298464336339146,
+      "learning_rate": 1.1492657196884203e-05,
+      "loss": 0.9422,
+      "step": 4479
+    },
+    {
+      "epoch": 0.46886446886446886,
+      "grad_norm": 1.945013493191885,
+      "learning_rate": 1.1489305365651286e-05,
+      "loss": 1.0195,
+      "step": 4480
+    },
+    {
+      "epoch": 0.46896912611198327,
+      "grad_norm": 1.9336766437798703,
+      "learning_rate": 1.1485953363294167e-05,
+      "loss": 1.0002,
+      "step": 4481
+    },
+    {
+      "epoch": 0.4690737833594976,
+      "grad_norm": 1.9030733453182003,
+      "learning_rate": 1.1482601190197993e-05,
+      "loss": 0.8946,
+      "step": 4482
+    },
+    {
+      "epoch": 0.469178440607012,
+      "grad_norm": 4.129348859610016,
+      "learning_rate": 1.1479248846747934e-05,
+      "loss": 1.1738,
+      "step": 4483
+    },
+    {
+      "epoch": 0.46928309785452643,
+      "grad_norm": 2.041585816161266,
+      "learning_rate": 1.1475896333329188e-05,
+      "loss": 0.9353,
+      "step": 4484
+    },
+    {
+      "epoch": 0.46938775510204084,
+      "grad_norm": 1.8593945186994896,
+      "learning_rate": 1.1472543650326959e-05,
+      "loss": 0.9923,
+      "step": 4485
+    },
+    {
+      "epoch": 0.4694924123495552,
+      "grad_norm": 5.031841167564904,
+      "learning_rate": 1.1469190798126479e-05,
+      "loss": 1.179,
+      "step": 4486
+    },
+    {
+      "epoch": 0.4695970695970696,
+      "grad_norm": 2.1889564063975135,
+      "learning_rate": 1.1465837777113e-05,
+      "loss": 0.9345,
+      "step": 4487
+    },
+    {
+      "epoch": 0.469701726844584,
+      "grad_norm": 2.0960505576447868,
+      "learning_rate": 1.1462484587671788e-05,
+      "loss": 1.057,
+      "step": 4488
+    },
+    {
+      "epoch": 0.46980638409209835,
+      "grad_norm": 2.122413193941446,
+      "learning_rate": 1.1459131230188132e-05,
+      "loss": 0.9431,
+      "step": 4489
+    },
+    {
+      "epoch": 0.46991104133961276,
+      "grad_norm": 1.8665692849116569,
+      "learning_rate": 1.1455777705047337e-05,
+      "loss": 0.8859,
+      "step": 4490
+    },
+    {
+      "epoch": 0.47001569858712716,
+      "grad_norm": 2.2256281211232967,
+      "learning_rate": 1.1452424012634734e-05,
+      "loss": 0.9491,
+      "step": 4491
+    },
+    {
+      "epoch": 0.47012035583464157,
+      "grad_norm": 2.3097076298000068,
+      "learning_rate": 1.1449070153335669e-05,
+      "loss": 1.0756,
+      "step": 4492
+    },
+    {
+      "epoch": 0.4702250130821559,
+      "grad_norm": 5.195580495117896,
+      "learning_rate": 1.1445716127535506e-05,
+      "loss": 1.1469,
+      "step": 4493
+    },
+    {
+      "epoch": 0.4703296703296703,
+      "grad_norm": 1.8105322884131707,
+      "learning_rate": 1.1442361935619628e-05,
+      "loss": 0.9045,
+      "step": 4494
+    },
+    {
+      "epoch": 0.47043432757718473,
+      "grad_norm": 2.01942155404332,
+      "learning_rate": 1.143900757797344e-05,
+      "loss": 0.9083,
+      "step": 4495
+    },
+    {
+      "epoch": 0.47053898482469914,
+      "grad_norm": 1.932387279006279,
+      "learning_rate": 1.1435653054982365e-05,
+      "loss": 0.9502,
+      "step": 4496
+    },
+    {
+      "epoch": 0.4706436420722135,
+      "grad_norm": 4.1256458431646355,
+      "learning_rate": 1.1432298367031843e-05,
+      "loss": 1.0059,
+      "step": 4497
+    },
+    {
+      "epoch": 0.4707482993197279,
+      "grad_norm": 1.942103396468479,
+      "learning_rate": 1.1428943514507334e-05,
+      "loss": 1.0407,
+      "step": 4498
+    },
+    {
+      "epoch": 0.4708529565672423,
+      "grad_norm": 2.273840379567588,
+      "learning_rate": 1.1425588497794319e-05,
+      "loss": 1.0614,
+      "step": 4499
+    },
+    {
+      "epoch": 0.47095761381475665,
+      "grad_norm": 2.1200667200778907,
+      "learning_rate": 1.1422233317278295e-05,
+      "loss": 0.9771,
+      "step": 4500
+    },
+    {
+      "epoch": 0.47106227106227105,
+      "grad_norm": 2.083148734600215,
+      "learning_rate": 1.1418877973344781e-05,
+      "loss": 0.8671,
+      "step": 4501
+    },
+    {
+      "epoch": 0.47116692830978546,
+      "grad_norm": 2.2953006328435674,
+      "learning_rate": 1.1415522466379312e-05,
+      "loss": 0.9418,
+      "step": 4502
+    },
+    {
+      "epoch": 0.47127158555729987,
+      "grad_norm": 2.189812095653051,
+      "learning_rate": 1.1412166796767442e-05,
+      "loss": 0.8781,
+      "step": 4503
+    },
+    {
+      "epoch": 0.4713762428048142,
+      "grad_norm": 2.2314434200482407,
+      "learning_rate": 1.1408810964894743e-05,
+      "loss": 0.9772,
+      "step": 4504
+    },
+    {
+      "epoch": 0.4714809000523286,
+      "grad_norm": 2.0090730558505205,
+      "learning_rate": 1.1405454971146808e-05,
+      "loss": 0.9608,
+      "step": 4505
+    },
+    {
+      "epoch": 0.47158555729984303,
+      "grad_norm": 2.2175393347464314,
+      "learning_rate": 1.140209881590925e-05,
+      "loss": 1.0648,
+      "step": 4506
+    },
+    {
+      "epoch": 0.4716902145473574,
+      "grad_norm": 1.8968894676133286,
+      "learning_rate": 1.1398742499567689e-05,
+      "loss": 0.9704,
+      "step": 4507
+    },
+    {
+      "epoch": 0.4717948717948718,
+      "grad_norm": 2.2753935373797916,
+      "learning_rate": 1.1395386022507786e-05,
+      "loss": 0.9457,
+      "step": 4508
+    },
+    {
+      "epoch": 0.4718995290423862,
+      "grad_norm": 2.2158138128200533,
+      "learning_rate": 1.13920293851152e-05,
+      "loss": 0.939,
+      "step": 4509
+    },
+    {
+      "epoch": 0.4720041862899006,
+      "grad_norm": 2.149361055179521,
+      "learning_rate": 1.1388672587775615e-05,
+      "loss": 0.8684,
+      "step": 4510
+    },
+    {
+      "epoch": 0.47210884353741495,
+      "grad_norm": 2.242585827443349,
+      "learning_rate": 1.1385315630874732e-05,
+      "loss": 0.9222,
+      "step": 4511
+    },
+    {
+      "epoch": 0.47221350078492935,
+      "grad_norm": 2.0858170359584878,
+      "learning_rate": 1.1381958514798277e-05,
+      "loss": 1.0521,
+      "step": 4512
+    },
+    {
+      "epoch": 0.47231815803244376,
+      "grad_norm": 2.3578219833534635,
+      "learning_rate": 1.1378601239931988e-05,
+      "loss": 0.9778,
+      "step": 4513
+    },
+    {
+      "epoch": 0.47242281527995816,
+      "grad_norm": 2.318116192208368,
+      "learning_rate": 1.1375243806661617e-05,
+      "loss": 0.9584,
+      "step": 4514
+    },
+    {
+      "epoch": 0.4725274725274725,
+      "grad_norm": 1.8752032576547455,
+      "learning_rate": 1.1371886215372952e-05,
+      "loss": 1.0631,
+      "step": 4515
+    },
+    {
+      "epoch": 0.4726321297749869,
+      "grad_norm": 2.4461026659114067,
+      "learning_rate": 1.1368528466451774e-05,
+      "loss": 1.0267,
+      "step": 4516
+    },
+    {
+      "epoch": 0.4727367870225013,
+      "grad_norm": 2.0213212076244638,
+      "learning_rate": 1.1365170560283903e-05,
+      "loss": 0.9101,
+      "step": 4517
+    },
+    {
+      "epoch": 0.4728414442700157,
+      "grad_norm": 1.8825133489006614,
+      "learning_rate": 1.1361812497255167e-05,
+      "loss": 1.0026,
+      "step": 4518
+    },
+    {
+      "epoch": 0.4729461015175301,
+      "grad_norm": 2.2245743255363286,
+      "learning_rate": 1.1358454277751415e-05,
+      "loss": 0.9904,
+      "step": 4519
+    },
+    {
+      "epoch": 0.4730507587650445,
+      "grad_norm": 1.9671246859720815,
+      "learning_rate": 1.1355095902158507e-05,
+      "loss": 0.8788,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4731554160125589,
+      "grad_norm": 2.045528524105347,
+      "learning_rate": 1.1351737370862339e-05,
+      "loss": 1.0552,
+      "step": 4521
+    },
+    {
+      "epoch": 0.47326007326007324,
+      "grad_norm": 2.076578499573767,
+      "learning_rate": 1.1348378684248806e-05,
+      "loss": 0.9716,
+      "step": 4522
+    },
+    {
+      "epoch": 0.47336473050758765,
+      "grad_norm": 2.1579134355425764,
+      "learning_rate": 1.134501984270383e-05,
+      "loss": 1.1023,
+      "step": 4523
+    },
+    {
+      "epoch": 0.47346938775510206,
+      "grad_norm": 2.340911759330156,
+      "learning_rate": 1.1341660846613343e-05,
+      "loss": 0.9794,
+      "step": 4524
+    },
+    {
+      "epoch": 0.4735740450026164,
+      "grad_norm": 1.9924334227820981,
+      "learning_rate": 1.133830169636331e-05,
+      "loss": 1.0848,
+      "step": 4525
+    },
+    {
+      "epoch": 0.4736787022501308,
+      "grad_norm": 1.962966847625367,
+      "learning_rate": 1.1334942392339693e-05,
+      "loss": 1.0938,
+      "step": 4526
+    },
+    {
+      "epoch": 0.4737833594976452,
+      "grad_norm": 2.178454110615139,
+      "learning_rate": 1.1331582934928495e-05,
+      "loss": 1.0561,
+      "step": 4527
+    },
+    {
+      "epoch": 0.4738880167451596,
+      "grad_norm": 2.368913775696146,
+      "learning_rate": 1.1328223324515714e-05,
+      "loss": 0.8537,
+      "step": 4528
+    },
+    {
+      "epoch": 0.473992673992674,
+      "grad_norm": 1.9280380818904281,
+      "learning_rate": 1.1324863561487383e-05,
+      "loss": 0.8553,
+      "step": 4529
+    },
+    {
+      "epoch": 0.4740973312401884,
+      "grad_norm": 1.962816395052731,
+      "learning_rate": 1.1321503646229543e-05,
+      "loss": 0.8091,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4742019884877028,
+      "grad_norm": 2.068969926585598,
+      "learning_rate": 1.1318143579128254e-05,
+      "loss": 0.9076,
+      "step": 4531
+    },
+    {
+      "epoch": 0.47430664573521714,
+      "grad_norm": 1.9731106713105366,
+      "learning_rate": 1.1314783360569597e-05,
+      "loss": 0.9354,
+      "step": 4532
+    },
+    {
+      "epoch": 0.47441130298273154,
+      "grad_norm": 1.9249243484641931,
+      "learning_rate": 1.1311422990939664e-05,
+      "loss": 0.8459,
+      "step": 4533
+    },
+    {
+      "epoch": 0.47451596023024595,
+      "grad_norm": 2.014143494128217,
+      "learning_rate": 1.1308062470624572e-05,
+      "loss": 1.0709,
+      "step": 4534
+    },
+    {
+      "epoch": 0.47462061747776035,
+      "grad_norm": 1.790897621437224,
+      "learning_rate": 1.1304701800010453e-05,
+      "loss": 0.9237,
+      "step": 4535
+    },
+    {
+      "epoch": 0.4747252747252747,
+      "grad_norm": 1.8946726238557325,
+      "learning_rate": 1.1301340979483448e-05,
+      "loss": 0.9411,
+      "step": 4536
+    },
+    {
+      "epoch": 0.4748299319727891,
+      "grad_norm": 1.8356376485135328,
+      "learning_rate": 1.129798000942973e-05,
+      "loss": 0.958,
+      "step": 4537
+    },
+    {
+      "epoch": 0.4749345892203035,
+      "grad_norm": 2.0838881862402765,
+      "learning_rate": 1.1294618890235474e-05,
+      "loss": 1.0522,
+      "step": 4538
+    },
+    {
+      "epoch": 0.4750392464678179,
+      "grad_norm": 2.1995593212183264,
+      "learning_rate": 1.1291257622286887e-05,
+      "loss": 0.9551,
+      "step": 4539
+    },
+    {
+      "epoch": 0.47514390371533227,
+      "grad_norm": 1.8441762381994096,
+      "learning_rate": 1.1287896205970178e-05,
+      "loss": 0.8266,
+      "step": 4540
+    },
+    {
+      "epoch": 0.4752485609628467,
+      "grad_norm": 4.422227979112271,
+      "learning_rate": 1.1284534641671588e-05,
+      "loss": 1.1611,
+      "step": 4541
+    },
+    {
+      "epoch": 0.4753532182103611,
+      "grad_norm": 2.3144636336989945,
+      "learning_rate": 1.128117292977736e-05,
+      "loss": 0.9875,
+      "step": 4542
+    },
+    {
+      "epoch": 0.47545787545787543,
+      "grad_norm": 2.10412297009563,
+      "learning_rate": 1.1277811070673765e-05,
+      "loss": 0.9113,
+      "step": 4543
+    },
+    {
+      "epoch": 0.47556253270538984,
+      "grad_norm": 2.215518349186847,
+      "learning_rate": 1.127444906474709e-05,
+      "loss": 0.8954,
+      "step": 4544
+    },
+    {
+      "epoch": 0.47566718995290425,
+      "grad_norm": 2.30718848055899,
+      "learning_rate": 1.1271086912383633e-05,
+      "loss": 0.9245,
+      "step": 4545
+    },
+    {
+      "epoch": 0.47577184720041865,
+      "grad_norm": 2.028770055513572,
+      "learning_rate": 1.1267724613969712e-05,
+      "loss": 0.9973,
+      "step": 4546
+    },
+    {
+      "epoch": 0.475876504447933,
+      "grad_norm": 2.2846525131478437,
+      "learning_rate": 1.1264362169891664e-05,
+      "loss": 1.0271,
+      "step": 4547
+    },
+    {
+      "epoch": 0.4759811616954474,
+      "grad_norm": 2.0978658851996257,
+      "learning_rate": 1.1260999580535842e-05,
+      "loss": 0.9312,
+      "step": 4548
+    },
+    {
+      "epoch": 0.4760858189429618,
+      "grad_norm": 3.541394648972959,
+      "learning_rate": 1.125763684628861e-05,
+      "loss": 1.0477,
+      "step": 4549
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 2.317340528280266,
+      "learning_rate": 1.1254273967536352e-05,
+      "loss": 0.9305,
+      "step": 4550
+    },
+    {
+      "epoch": 0.47629513343799057,
+      "grad_norm": 2.5592216831346093,
+      "learning_rate": 1.1250910944665475e-05,
+      "loss": 0.9933,
+      "step": 4551
+    },
+    {
+      "epoch": 0.476399790685505,
+      "grad_norm": 2.3033433240600942,
+      "learning_rate": 1.1247547778062394e-05,
+      "loss": 1.0646,
+      "step": 4552
+    },
+    {
+      "epoch": 0.4765044479330194,
+      "grad_norm": 1.782189325078589,
+      "learning_rate": 1.1244184468113544e-05,
+      "loss": 0.8961,
+      "step": 4553
+    },
+    {
+      "epoch": 0.47660910518053373,
+      "grad_norm": 2.3437218275063914,
+      "learning_rate": 1.1240821015205378e-05,
+      "loss": 0.9471,
+      "step": 4554
+    },
+    {
+      "epoch": 0.47671376242804814,
+      "grad_norm": 2.0231024397759563,
+      "learning_rate": 1.1237457419724357e-05,
+      "loss": 0.9574,
+      "step": 4555
+    },
+    {
+      "epoch": 0.47681841967556254,
+      "grad_norm": 1.9992919395657165,
+      "learning_rate": 1.1234093682056976e-05,
+      "loss": 0.8381,
+      "step": 4556
+    },
+    {
+      "epoch": 0.47692307692307695,
+      "grad_norm": 2.2448074144884433,
+      "learning_rate": 1.1230729802589727e-05,
+      "loss": 1.005,
+      "step": 4557
+    },
+    {
+      "epoch": 0.4770277341705913,
+      "grad_norm": 2.094998090831663,
+      "learning_rate": 1.1227365781709129e-05,
+      "loss": 0.8991,
+      "step": 4558
+    },
+    {
+      "epoch": 0.4771323914181057,
+      "grad_norm": 2.1848516156132227,
+      "learning_rate": 1.1224001619801711e-05,
+      "loss": 0.9532,
+      "step": 4559
+    },
+    {
+      "epoch": 0.4772370486656201,
+      "grad_norm": 1.9856619963316688,
+      "learning_rate": 1.122063731725403e-05,
+      "loss": 0.9081,
+      "step": 4560
+    },
+    {
+      "epoch": 0.47734170591313446,
+      "grad_norm": 2.5402648997557367,
+      "learning_rate": 1.1217272874452647e-05,
+      "loss": 1.0494,
+      "step": 4561
+    },
+    {
+      "epoch": 0.47744636316064887,
+      "grad_norm": 2.129703098208944,
+      "learning_rate": 1.121390829178414e-05,
+      "loss": 0.8839,
+      "step": 4562
+    },
+    {
+      "epoch": 0.4775510204081633,
+      "grad_norm": 2.207630890065239,
+      "learning_rate": 1.1210543569635113e-05,
+      "loss": 1.0222,
+      "step": 4563
+    },
+    {
+      "epoch": 0.4776556776556777,
+      "grad_norm": 2.3323434569261425,
+      "learning_rate": 1.1207178708392173e-05,
+      "loss": 0.8753,
+      "step": 4564
+    },
+    {
+      "epoch": 0.47776033490319203,
+      "grad_norm": 2.0722087950689616,
+      "learning_rate": 1.1203813708441953e-05,
+      "loss": 1.0534,
+      "step": 4565
+    },
+    {
+      "epoch": 0.47786499215070644,
+      "grad_norm": 1.9543831145256765,
+      "learning_rate": 1.1200448570171099e-05,
+      "loss": 1.0708,
+      "step": 4566
+    },
+    {
+      "epoch": 0.47796964939822084,
+      "grad_norm": 2.2669487242998487,
+      "learning_rate": 1.1197083293966267e-05,
+      "loss": 0.901,
+      "step": 4567
+    },
+    {
+      "epoch": 0.4780743066457352,
+      "grad_norm": 1.950939820948088,
+      "learning_rate": 1.1193717880214139e-05,
+      "loss": 0.8952,
+      "step": 4568
+    },
+    {
+      "epoch": 0.4781789638932496,
+      "grad_norm": 2.1867834960215737,
+      "learning_rate": 1.1190352329301408e-05,
+      "loss": 0.9256,
+      "step": 4569
+    },
+    {
+      "epoch": 0.478283621140764,
+      "grad_norm": 1.9100913001648718,
+      "learning_rate": 1.118698664161478e-05,
+      "loss": 0.9184,
+      "step": 4570
+    },
+    {
+      "epoch": 0.4783882783882784,
+      "grad_norm": 2.1641906169210765,
+      "learning_rate": 1.1183620817540985e-05,
+      "loss": 0.7917,
+      "step": 4571
+    },
+    {
+      "epoch": 0.47849293563579276,
+      "grad_norm": 1.964578596649523,
+      "learning_rate": 1.1180254857466753e-05,
+      "loss": 0.9407,
+      "step": 4572
+    },
+    {
+      "epoch": 0.47859759288330717,
+      "grad_norm": 1.9010529569383956,
+      "learning_rate": 1.1176888761778848e-05,
+      "loss": 0.9953,
+      "step": 4573
+    },
+    {
+      "epoch": 0.47870225013082157,
+      "grad_norm": 1.933991820551362,
+      "learning_rate": 1.1173522530864036e-05,
+      "loss": 1.007,
+      "step": 4574
+    },
+    {
+      "epoch": 0.478806907378336,
+      "grad_norm": 1.972812990270169,
+      "learning_rate": 1.117015616510911e-05,
+      "loss": 0.9306,
+      "step": 4575
+    },
+    {
+      "epoch": 0.47891156462585033,
+      "grad_norm": 2.0473114243949815,
+      "learning_rate": 1.1166789664900866e-05,
+      "loss": 1.005,
+      "step": 4576
+    },
+    {
+      "epoch": 0.47901622187336473,
+      "grad_norm": 2.1372399776159168,
+      "learning_rate": 1.1163423030626123e-05,
+      "loss": 0.9524,
+      "step": 4577
+    },
+    {
+      "epoch": 0.47912087912087914,
+      "grad_norm": 2.1890057903200826,
+      "learning_rate": 1.1160056262671718e-05,
+      "loss": 0.8805,
+      "step": 4578
+    },
+    {
+      "epoch": 0.4792255363683935,
+      "grad_norm": 2.2821866808196196,
+      "learning_rate": 1.1156689361424496e-05,
+      "loss": 0.931,
+      "step": 4579
+    },
+    {
+      "epoch": 0.4793301936159079,
+      "grad_norm": 1.92030262663086,
+      "learning_rate": 1.1153322327271324e-05,
+      "loss": 1.0229,
+      "step": 4580
+    },
+    {
+      "epoch": 0.4794348508634223,
+      "grad_norm": 2.264619342723187,
+      "learning_rate": 1.1149955160599073e-05,
+      "loss": 1.0943,
+      "step": 4581
+    },
+    {
+      "epoch": 0.4795395081109367,
+      "grad_norm": 2.4491546035308986,
+      "learning_rate": 1.1146587861794645e-05,
+      "loss": 0.8622,
+      "step": 4582
+    },
+    {
+      "epoch": 0.47964416535845106,
+      "grad_norm": 2.700636952916195,
+      "learning_rate": 1.114322043124495e-05,
+      "loss": 0.9035,
+      "step": 4583
+    },
+    {
+      "epoch": 0.47974882260596546,
+      "grad_norm": 5.883040265797574,
+      "learning_rate": 1.1139852869336907e-05,
+      "loss": 1.0753,
+      "step": 4584
+    },
+    {
+      "epoch": 0.47985347985347987,
+      "grad_norm": 2.2890986655042234,
+      "learning_rate": 1.113648517645746e-05,
+      "loss": 1.0844,
+      "step": 4585
+    },
+    {
+      "epoch": 0.4799581371009942,
+      "grad_norm": 1.8524051771038759,
+      "learning_rate": 1.1133117352993559e-05,
+      "loss": 0.9628,
+      "step": 4586
+    },
+    {
+      "epoch": 0.4800627943485086,
+      "grad_norm": 2.2350258071349263,
+      "learning_rate": 1.1129749399332182e-05,
+      "loss": 0.9315,
+      "step": 4587
+    },
+    {
+      "epoch": 0.48016745159602303,
+      "grad_norm": 3.5795609315488726,
+      "learning_rate": 1.1126381315860306e-05,
+      "loss": 1.1486,
+      "step": 4588
+    },
+    {
+      "epoch": 0.48027210884353744,
+      "grad_norm": 2.0026623409331696,
+      "learning_rate": 1.1123013102964934e-05,
+      "loss": 1.0134,
+      "step": 4589
+    },
+    {
+      "epoch": 0.4803767660910518,
+      "grad_norm": 2.44461585559246,
+      "learning_rate": 1.1119644761033079e-05,
+      "loss": 1.0314,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4804814233385662,
+      "grad_norm": 1.9386712926730827,
+      "learning_rate": 1.1116276290451773e-05,
+      "loss": 1.0203,
+      "step": 4591
+    },
+    {
+      "epoch": 0.4805860805860806,
+      "grad_norm": 1.9273580062176585,
+      "learning_rate": 1.1112907691608059e-05,
+      "loss": 0.9667,
+      "step": 4592
+    },
+    {
+      "epoch": 0.48069073783359495,
+      "grad_norm": 2.2393335832841483,
+      "learning_rate": 1.1109538964888993e-05,
+      "loss": 0.9332,
+      "step": 4593
+    },
+    {
+      "epoch": 0.48079539508110936,
+      "grad_norm": 2.2101404361932397,
+      "learning_rate": 1.110617011068165e-05,
+      "loss": 0.8757,
+      "step": 4594
+    },
+    {
+      "epoch": 0.48090005232862376,
+      "grad_norm": 2.07916151536985,
+      "learning_rate": 1.110280112937312e-05,
+      "loss": 0.8394,
+      "step": 4595
+    },
+    {
+      "epoch": 0.48100470957613817,
+      "grad_norm": 2.4008188696017077,
+      "learning_rate": 1.1099432021350504e-05,
+      "loss": 0.9937,
+      "step": 4596
+    },
+    {
+      "epoch": 0.4811093668236525,
+      "grad_norm": 1.98281407430736,
+      "learning_rate": 1.1096062787000923e-05,
+      "loss": 0.9293,
+      "step": 4597
+    },
+    {
+      "epoch": 0.4812140240711669,
+      "grad_norm": 1.6750650744945574,
+      "learning_rate": 1.1092693426711501e-05,
+      "loss": 0.8145,
+      "step": 4598
+    },
+    {
+      "epoch": 0.48131868131868133,
+      "grad_norm": 2.1258343895313137,
+      "learning_rate": 1.1089323940869392e-05,
+      "loss": 1.0777,
+      "step": 4599
+    },
+    {
+      "epoch": 0.48142333856619574,
+      "grad_norm": 2.3850630786161227,
+      "learning_rate": 1.1085954329861754e-05,
+      "loss": 1.1615,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4815279958137101,
+      "grad_norm": 2.041493035918398,
+      "learning_rate": 1.1082584594075762e-05,
+      "loss": 0.9125,
+      "step": 4601
+    },
+    {
+      "epoch": 0.4816326530612245,
+      "grad_norm": 2.102783748957608,
+      "learning_rate": 1.1079214733898606e-05,
+      "loss": 0.9909,
+      "step": 4602
+    },
+    {
+      "epoch": 0.4817373103087389,
+      "grad_norm": 1.6810209540981638,
+      "learning_rate": 1.1075844749717486e-05,
+      "loss": 0.9937,
+      "step": 4603
+    },
+    {
+      "epoch": 0.48184196755625325,
+      "grad_norm": 1.9499292480253156,
+      "learning_rate": 1.1072474641919626e-05,
+      "loss": 1.0339,
+      "step": 4604
+    },
+    {
+      "epoch": 0.48194662480376765,
+      "grad_norm": 1.9692957698439062,
+      "learning_rate": 1.1069104410892255e-05,
+      "loss": 0.9911,
+      "step": 4605
+    },
+    {
+      "epoch": 0.48205128205128206,
+      "grad_norm": 1.9832587584087509,
+      "learning_rate": 1.1065734057022622e-05,
+      "loss": 0.9872,
+      "step": 4606
+    },
+    {
+      "epoch": 0.48215593929879647,
+      "grad_norm": 1.9753953053703226,
+      "learning_rate": 1.1062363580697982e-05,
+      "loss": 0.8596,
+      "step": 4607
+    },
+    {
+      "epoch": 0.4822605965463108,
+      "grad_norm": 2.2053778552706387,
+      "learning_rate": 1.1058992982305613e-05,
+      "loss": 0.9707,
+      "step": 4608
+    },
+    {
+      "epoch": 0.4823652537938252,
+      "grad_norm": 4.451868601106817,
+      "learning_rate": 1.1055622262232808e-05,
+      "loss": 1.0707,
+      "step": 4609
+    },
+    {
+      "epoch": 0.4824699110413396,
+      "grad_norm": 2.427513648919258,
+      "learning_rate": 1.105225142086686e-05,
+      "loss": 1.093,
+      "step": 4610
+    },
+    {
+      "epoch": 0.482574568288854,
+      "grad_norm": 1.8975018847651661,
+      "learning_rate": 1.1048880458595093e-05,
+      "loss": 0.9038,
+      "step": 4611
+    },
+    {
+      "epoch": 0.4826792255363684,
+      "grad_norm": 2.101137234743737,
+      "learning_rate": 1.1045509375804835e-05,
+      "loss": 0.9936,
+      "step": 4612
+    },
+    {
+      "epoch": 0.4827838827838828,
+      "grad_norm": 2.116146425294967,
+      "learning_rate": 1.104213817288343e-05,
+      "loss": 0.9991,
+      "step": 4613
+    },
+    {
+      "epoch": 0.4828885400313972,
+      "grad_norm": 7.379023528902324,
+      "learning_rate": 1.1038766850218237e-05,
+      "loss": 1.0964,
+      "step": 4614
+    },
+    {
+      "epoch": 0.48299319727891155,
+      "grad_norm": 2.303483496597832,
+      "learning_rate": 1.1035395408196625e-05,
+      "loss": 1.0693,
+      "step": 4615
+    },
+    {
+      "epoch": 0.48309785452642595,
+      "grad_norm": 1.8368414922882212,
+      "learning_rate": 1.103202384720598e-05,
+      "loss": 0.9938,
+      "step": 4616
+    },
+    {
+      "epoch": 0.48320251177394036,
+      "grad_norm": 1.939124879567201,
+      "learning_rate": 1.1028652167633706e-05,
+      "loss": 0.9361,
+      "step": 4617
+    },
+    {
+      "epoch": 0.48330716902145476,
+      "grad_norm": 1.931462825025141,
+      "learning_rate": 1.1025280369867216e-05,
+      "loss": 0.996,
+      "step": 4618
+    },
+    {
+      "epoch": 0.4834118262689691,
+      "grad_norm": 1.8660312624366957,
+      "learning_rate": 1.102190845429393e-05,
+      "loss": 0.8925,
+      "step": 4619
+    },
+    {
+      "epoch": 0.4835164835164835,
+      "grad_norm": 1.9759725551334741,
+      "learning_rate": 1.1018536421301287e-05,
+      "loss": 0.8974,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4836211407639979,
+      "grad_norm": 2.042722969414731,
+      "learning_rate": 1.1015164271276747e-05,
+      "loss": 1.0188,
+      "step": 4621
+    },
+    {
+      "epoch": 0.4837257980115123,
+      "grad_norm": 2.038746548550091,
+      "learning_rate": 1.1011792004607778e-05,
+      "loss": 0.875,
+      "step": 4622
+    },
+    {
+      "epoch": 0.4838304552590267,
+      "grad_norm": 2.104636214257139,
+      "learning_rate": 1.1008419621681854e-05,
+      "loss": 0.9584,
+      "step": 4623
+    },
+    {
+      "epoch": 0.4839351125065411,
+      "grad_norm": 1.8747597757848282,
+      "learning_rate": 1.100504712288647e-05,
+      "loss": 0.9251,
+      "step": 4624
+    },
+    {
+      "epoch": 0.4840397697540555,
+      "grad_norm": 2.2211808317732418,
+      "learning_rate": 1.1001674508609135e-05,
+      "loss": 1.0446,
+      "step": 4625
+    },
+    {
+      "epoch": 0.48414442700156984,
+      "grad_norm": 2.3946698315504005,
+      "learning_rate": 1.0998301779237367e-05,
+      "loss": 1.0903,
+      "step": 4626
+    },
+    {
+      "epoch": 0.48424908424908425,
+      "grad_norm": 2.2432670399188472,
+      "learning_rate": 1.0994928935158703e-05,
+      "loss": 0.9233,
+      "step": 4627
+    },
+    {
+      "epoch": 0.48435374149659866,
+      "grad_norm": 1.8197431587694193,
+      "learning_rate": 1.0991555976760688e-05,
+      "loss": 0.9245,
+      "step": 4628
+    },
+    {
+      "epoch": 0.484458398744113,
+      "grad_norm": 2.025709348707062,
+      "learning_rate": 1.098818290443088e-05,
+      "loss": 0.9994,
+      "step": 4629
+    },
+    {
+      "epoch": 0.4845630559916274,
+      "grad_norm": 2.065981035535976,
+      "learning_rate": 1.098480971855685e-05,
+      "loss": 1.0224,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4846677132391418,
+      "grad_norm": 5.77757048663166,
+      "learning_rate": 1.0981436419526193e-05,
+      "loss": 1.1061,
+      "step": 4631
+    },
+    {
+      "epoch": 0.4847723704866562,
+      "grad_norm": 1.870408334687293,
+      "learning_rate": 1.0978063007726495e-05,
+      "loss": 0.7735,
+      "step": 4632
+    },
+    {
+      "epoch": 0.4848770277341706,
+      "grad_norm": 3.3267436752556065,
+      "learning_rate": 1.0974689483545378e-05,
+      "loss": 1.0417,
+      "step": 4633
+    },
+    {
+      "epoch": 0.484981684981685,
+      "grad_norm": 1.9245609462036217,
+      "learning_rate": 1.0971315847370463e-05,
+      "loss": 0.8995,
+      "step": 4634
+    },
+    {
+      "epoch": 0.4850863422291994,
+      "grad_norm": 1.9011101446154672,
+      "learning_rate": 1.0967942099589391e-05,
+      "loss": 0.8339,
+      "step": 4635
+    },
+    {
+      "epoch": 0.48519099947671374,
+      "grad_norm": 1.9115078692573162,
+      "learning_rate": 1.0964568240589808e-05,
+      "loss": 0.7742,
+      "step": 4636
+    },
+    {
+      "epoch": 0.48529565672422814,
+      "grad_norm": 1.9022548969400057,
+      "learning_rate": 1.0961194270759375e-05,
+      "loss": 0.9253,
+      "step": 4637
+    },
+    {
+      "epoch": 0.48540031397174255,
+      "grad_norm": 2.0318224541882657,
+      "learning_rate": 1.0957820190485775e-05,
+      "loss": 0.9972,
+      "step": 4638
+    },
+    {
+      "epoch": 0.48550497121925695,
+      "grad_norm": 2.0727823257764544,
+      "learning_rate": 1.0954446000156693e-05,
+      "loss": 1.058,
+      "step": 4639
+    },
+    {
+      "epoch": 0.4856096284667713,
+      "grad_norm": 1.9956018186674882,
+      "learning_rate": 1.0951071700159833e-05,
+      "loss": 0.8057,
+      "step": 4640
+    },
+    {
+      "epoch": 0.4857142857142857,
+      "grad_norm": 2.20621162649101,
+      "learning_rate": 1.0947697290882903e-05,
+      "loss": 0.834,
+      "step": 4641
+    },
+    {
+      "epoch": 0.4858189429618001,
+      "grad_norm": 2.051794455421428,
+      "learning_rate": 1.0944322772713634e-05,
+      "loss": 0.9757,
+      "step": 4642
+    },
+    {
+      "epoch": 0.4859236002093145,
+      "grad_norm": 2.1409550137945974,
+      "learning_rate": 1.0940948146039768e-05,
+      "loss": 0.974,
+      "step": 4643
+    },
+    {
+      "epoch": 0.48602825745682887,
+      "grad_norm": 1.948251250014736,
+      "learning_rate": 1.0937573411249047e-05,
+      "loss": 0.9849,
+      "step": 4644
+    },
+    {
+      "epoch": 0.4861329147043433,
+      "grad_norm": 1.9818999450949544,
+      "learning_rate": 1.0934198568729245e-05,
+      "loss": 0.9915,
+      "step": 4645
+    },
+    {
+      "epoch": 0.4862375719518577,
+      "grad_norm": 2.056814682650398,
+      "learning_rate": 1.0930823618868128e-05,
+      "loss": 0.9572,
+      "step": 4646
+    },
+    {
+      "epoch": 0.48634222919937203,
+      "grad_norm": 1.93186679850152,
+      "learning_rate": 1.0927448562053494e-05,
+      "loss": 0.9447,
+      "step": 4647
+    },
+    {
+      "epoch": 0.48644688644688644,
+      "grad_norm": 1.9807004018502983,
+      "learning_rate": 1.092407339867314e-05,
+      "loss": 0.9509,
+      "step": 4648
+    },
+    {
+      "epoch": 0.48655154369440085,
+      "grad_norm": 2.2561421939508324,
+      "learning_rate": 1.0920698129114877e-05,
+      "loss": 0.9115,
+      "step": 4649
+    },
+    {
+      "epoch": 0.48665620094191525,
+      "grad_norm": 1.7925812958550895,
+      "learning_rate": 1.0917322753766536e-05,
+      "loss": 0.8921,
+      "step": 4650
+    },
+    {
+      "epoch": 0.4867608581894296,
+      "grad_norm": 1.980186761047103,
+      "learning_rate": 1.0913947273015948e-05,
+      "loss": 0.9832,
+      "step": 4651
+    },
+    {
+      "epoch": 0.486865515436944,
+      "grad_norm": 2.3562237026686343,
+      "learning_rate": 1.0910571687250965e-05,
+      "loss": 0.9792,
+      "step": 4652
+    },
+    {
+      "epoch": 0.4869701726844584,
+      "grad_norm": 5.0353868336406915,
+      "learning_rate": 1.0907195996859453e-05,
+      "loss": 0.9708,
+      "step": 4653
+    },
+    {
+      "epoch": 0.48707482993197276,
+      "grad_norm": 1.9131908425142234,
+      "learning_rate": 1.090382020222928e-05,
+      "loss": 0.9326,
+      "step": 4654
+    },
+    {
+      "epoch": 0.48717948717948717,
+      "grad_norm": 1.790537959582867,
+      "learning_rate": 1.0900444303748333e-05,
+      "loss": 0.9168,
+      "step": 4655
+    },
+    {
+      "epoch": 0.4872841444270016,
+      "grad_norm": 2.140607384118724,
+      "learning_rate": 1.089706830180451e-05,
+      "loss": 1.1137,
+      "step": 4656
+    },
+    {
+      "epoch": 0.487388801674516,
+      "grad_norm": 1.891634591818433,
+      "learning_rate": 1.0893692196785722e-05,
+      "loss": 0.8324,
+      "step": 4657
+    },
+    {
+      "epoch": 0.48749345892203033,
+      "grad_norm": 2.09720854094086,
+      "learning_rate": 1.0890315989079887e-05,
+      "loss": 0.9571,
+      "step": 4658
+    },
+    {
+      "epoch": 0.48759811616954474,
+      "grad_norm": 2.332233200389157,
+      "learning_rate": 1.0886939679074939e-05,
+      "loss": 1.0647,
+      "step": 4659
+    },
+    {
+      "epoch": 0.48770277341705914,
+      "grad_norm": 2.3675130191516014,
+      "learning_rate": 1.0883563267158827e-05,
+      "loss": 0.8783,
+      "step": 4660
+    },
+    {
+      "epoch": 0.48780743066457355,
+      "grad_norm": 2.067847253311974,
+      "learning_rate": 1.0880186753719505e-05,
+      "loss": 0.9496,
+      "step": 4661
+    },
+    {
+      "epoch": 0.4879120879120879,
+      "grad_norm": 2.059957357047224,
+      "learning_rate": 1.087681013914494e-05,
+      "loss": 1.0097,
+      "step": 4662
+    },
+    {
+      "epoch": 0.4880167451596023,
+      "grad_norm": 1.7780085625194908,
+      "learning_rate": 1.0873433423823108e-05,
+      "loss": 0.9731,
+      "step": 4663
+    },
+    {
+      "epoch": 0.4881214024071167,
+      "grad_norm": 2.2693994526383183,
+      "learning_rate": 1.087005660814201e-05,
+      "loss": 0.9633,
+      "step": 4664
+    },
+    {
+      "epoch": 0.48822605965463106,
+      "grad_norm": 2.2058172273341343,
+      "learning_rate": 1.0866679692489643e-05,
+      "loss": 0.8353,
+      "step": 4665
+    },
+    {
+      "epoch": 0.48833071690214547,
+      "grad_norm": 2.1061269956596846,
+      "learning_rate": 1.0863302677254021e-05,
+      "loss": 0.9219,
+      "step": 4666
+    },
+    {
+      "epoch": 0.4884353741496599,
+      "grad_norm": 2.000729147710908,
+      "learning_rate": 1.0859925562823172e-05,
+      "loss": 0.9055,
+      "step": 4667
+    },
+    {
+      "epoch": 0.4885400313971743,
+      "grad_norm": 1.9868769740264263,
+      "learning_rate": 1.0856548349585132e-05,
+      "loss": 0.9625,
+      "step": 4668
+    },
+    {
+      "epoch": 0.48864468864468863,
+      "grad_norm": 1.9715388535082388,
+      "learning_rate": 1.0853171037927952e-05,
+      "loss": 0.9459,
+      "step": 4669
+    },
+    {
+      "epoch": 0.48874934589220304,
+      "grad_norm": 2.1247424417156147,
+      "learning_rate": 1.0849793628239687e-05,
+      "loss": 1.0457,
+      "step": 4670
+    },
+    {
+      "epoch": 0.48885400313971744,
+      "grad_norm": 2.0181718772211426,
+      "learning_rate": 1.0846416120908416e-05,
+      "loss": 0.8496,
+      "step": 4671
+    },
+    {
+      "epoch": 0.4889586603872318,
+      "grad_norm": 1.858935680154186,
+      "learning_rate": 1.0843038516322213e-05,
+      "loss": 1.0369,
+      "step": 4672
+    },
+    {
+      "epoch": 0.4890633176347462,
+      "grad_norm": 1.9353831132266497,
+      "learning_rate": 1.0839660814869174e-05,
+      "loss": 1.0288,
+      "step": 4673
+    },
+    {
+      "epoch": 0.4891679748822606,
+      "grad_norm": 1.92877995038633,
+      "learning_rate": 1.083628301693741e-05,
+      "loss": 0.9495,
+      "step": 4674
+    },
+    {
+      "epoch": 0.489272632129775,
+      "grad_norm": 1.8743301066850448,
+      "learning_rate": 1.083290512291503e-05,
+      "loss": 0.9853,
+      "step": 4675
+    },
+    {
+      "epoch": 0.48937728937728936,
+      "grad_norm": 2.166368633461282,
+      "learning_rate": 1.0829527133190163e-05,
+      "loss": 0.9273,
+      "step": 4676
+    },
+    {
+      "epoch": 0.48948194662480377,
+      "grad_norm": 2.626516377134673,
+      "learning_rate": 1.0826149048150947e-05,
+      "loss": 0.9658,
+      "step": 4677
+    },
+    {
+      "epoch": 0.48958660387231817,
+      "grad_norm": 2.007119331685366,
+      "learning_rate": 1.082277086818553e-05,
+      "loss": 0.9525,
+      "step": 4678
+    },
+    {
+      "epoch": 0.4896912611198325,
+      "grad_norm": 2.3214333736204043,
+      "learning_rate": 1.0819392593682074e-05,
+      "loss": 0.9228,
+      "step": 4679
+    },
+    {
+      "epoch": 0.4897959183673469,
+      "grad_norm": 2.1792774915549464,
+      "learning_rate": 1.0816014225028745e-05,
+      "loss": 1.001,
+      "step": 4680
+    },
+    {
+      "epoch": 0.48990057561486133,
+      "grad_norm": 1.871180713970224,
+      "learning_rate": 1.0812635762613728e-05,
+      "loss": 0.9649,
+      "step": 4681
+    },
+    {
+      "epoch": 0.49000523286237574,
+      "grad_norm": 2.415124070494655,
+      "learning_rate": 1.0809257206825214e-05,
+      "loss": 0.8755,
+      "step": 4682
+    },
+    {
+      "epoch": 0.4901098901098901,
+      "grad_norm": 1.972768747408386,
+      "learning_rate": 1.080587855805141e-05,
+      "loss": 0.9227,
+      "step": 4683
+    },
+    {
+      "epoch": 0.4902145473574045,
+      "grad_norm": 2.0814935675576196,
+      "learning_rate": 1.0802499816680525e-05,
+      "loss": 0.9524,
+      "step": 4684
+    },
+    {
+      "epoch": 0.4903192046049189,
+      "grad_norm": 2.1273011146323424,
+      "learning_rate": 1.079912098310078e-05,
+      "loss": 0.9118,
+      "step": 4685
+    },
+    {
+      "epoch": 0.4904238618524333,
+      "grad_norm": 2.3374604970516932,
+      "learning_rate": 1.0795742057700419e-05,
+      "loss": 1.0986,
+      "step": 4686
+    },
+    {
+      "epoch": 0.49052851909994766,
+      "grad_norm": 2.362107494802518,
+      "learning_rate": 1.0792363040867677e-05,
+      "loss": 0.941,
+      "step": 4687
+    },
+    {
+      "epoch": 0.49063317634746206,
+      "grad_norm": 1.7863431015815339,
+      "learning_rate": 1.0788983932990821e-05,
+      "loss": 0.8811,
+      "step": 4688
+    },
+    {
+      "epoch": 0.49073783359497647,
+      "grad_norm": 1.9476171669722804,
+      "learning_rate": 1.0785604734458108e-05,
+      "loss": 0.9976,
+      "step": 4689
+    },
+    {
+      "epoch": 0.4908424908424908,
+      "grad_norm": 1.9535812880238408,
+      "learning_rate": 1.0782225445657818e-05,
+      "loss": 0.9657,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4909471480900052,
+      "grad_norm": 2.09502949679378,
+      "learning_rate": 1.0778846066978244e-05,
+      "loss": 1.0246,
+      "step": 4691
+    },
+    {
+      "epoch": 0.49105180533751963,
+      "grad_norm": 1.9911290871631744,
+      "learning_rate": 1.0775466598807673e-05,
+      "loss": 0.9342,
+      "step": 4692
+    },
+    {
+      "epoch": 0.49115646258503404,
+      "grad_norm": 2.0405251799273074,
+      "learning_rate": 1.077208704153442e-05,
+      "loss": 0.8705,
+      "step": 4693
+    },
+    {
+      "epoch": 0.4912611198325484,
+      "grad_norm": 2.230283723602765,
+      "learning_rate": 1.07687073955468e-05,
+      "loss": 0.9413,
+      "step": 4694
+    },
+    {
+      "epoch": 0.4913657770800628,
+      "grad_norm": 2.0963860345437997,
+      "learning_rate": 1.076532766123314e-05,
+      "loss": 1.0367,
+      "step": 4695
+    },
+    {
+      "epoch": 0.4914704343275772,
+      "grad_norm": 2.0670944200322845,
+      "learning_rate": 1.0761947838981784e-05,
+      "loss": 0.9836,
+      "step": 4696
+    },
+    {
+      "epoch": 0.49157509157509155,
+      "grad_norm": 1.9165456806981649,
+      "learning_rate": 1.0758567929181074e-05,
+      "loss": 0.959,
+      "step": 4697
+    },
+    {
+      "epoch": 0.49167974882260596,
+      "grad_norm": 2.228963424636731,
+      "learning_rate": 1.075518793221937e-05,
+      "loss": 0.9847,
+      "step": 4698
+    },
+    {
+      "epoch": 0.49178440607012036,
+      "grad_norm": 2.3859878281796747,
+      "learning_rate": 1.0751807848485043e-05,
+      "loss": 0.9669,
+      "step": 4699
+    },
+    {
+      "epoch": 0.49188906331763477,
+      "grad_norm": 1.8576336024035527,
+      "learning_rate": 1.0748427678366474e-05,
+      "loss": 0.9819,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4919937205651491,
+      "grad_norm": 1.9665979871967703,
+      "learning_rate": 1.0745047422252041e-05,
+      "loss": 0.9886,
+      "step": 4701
+    },
+    {
+      "epoch": 0.4920983778126635,
+      "grad_norm": 1.9561562986500205,
+      "learning_rate": 1.0741667080530152e-05,
+      "loss": 0.9116,
+      "step": 4702
+    },
+    {
+      "epoch": 0.49220303506017793,
+      "grad_norm": 1.84642427868714,
+      "learning_rate": 1.0738286653589215e-05,
+      "loss": 0.9101,
+      "step": 4703
+    },
+    {
+      "epoch": 0.49230769230769234,
+      "grad_norm": 2.2932509082000823,
+      "learning_rate": 1.0734906141817638e-05,
+      "loss": 1.0601,
+      "step": 4704
+    },
+    {
+      "epoch": 0.4924123495552067,
+      "grad_norm": 2.1010715133795403,
+      "learning_rate": 1.0731525545603863e-05,
+      "loss": 0.9556,
+      "step": 4705
+    },
+    {
+      "epoch": 0.4925170068027211,
+      "grad_norm": 2.0849794034429117,
+      "learning_rate": 1.0728144865336315e-05,
+      "loss": 0.8108,
+      "step": 4706
+    },
+    {
+      "epoch": 0.4926216640502355,
+      "grad_norm": 6.181013708885361,
+      "learning_rate": 1.0724764101403445e-05,
+      "loss": 0.9935,
+      "step": 4707
+    },
+    {
+      "epoch": 0.49272632129774985,
+      "grad_norm": 2.045937670470183,
+      "learning_rate": 1.0721383254193714e-05,
+      "loss": 0.9471,
+      "step": 4708
+    },
+    {
+      "epoch": 0.49283097854526425,
+      "grad_norm": 2.7772247895749613,
+      "learning_rate": 1.0718002324095582e-05,
+      "loss": 0.9565,
+      "step": 4709
+    },
+    {
+      "epoch": 0.49293563579277866,
+      "grad_norm": 1.890604125310268,
+      "learning_rate": 1.071462131149753e-05,
+      "loss": 0.9257,
+      "step": 4710
+    },
+    {
+      "epoch": 0.49304029304029307,
+      "grad_norm": 2.489027702426938,
+      "learning_rate": 1.0711240216788036e-05,
+      "loss": 0.9638,
+      "step": 4711
+    },
+    {
+      "epoch": 0.4931449502878074,
+      "grad_norm": 2.455683040276122,
+      "learning_rate": 1.0707859040355603e-05,
+      "loss": 1.0213,
+      "step": 4712
+    },
+    {
+      "epoch": 0.4932496075353218,
+      "grad_norm": 2.143398110677024,
+      "learning_rate": 1.0704477782588733e-05,
+      "loss": 0.9781,
+      "step": 4713
+    },
+    {
+      "epoch": 0.4933542647828362,
+      "grad_norm": 1.9663410678077706,
+      "learning_rate": 1.0701096443875936e-05,
+      "loss": 0.9224,
+      "step": 4714
+    },
+    {
+      "epoch": 0.4934589220303506,
+      "grad_norm": 2.1480345691790705,
+      "learning_rate": 1.0697715024605735e-05,
+      "loss": 0.9175,
+      "step": 4715
+    },
+    {
+      "epoch": 0.493563579277865,
+      "grad_norm": 9.477655375017251,
+      "learning_rate": 1.0694333525166662e-05,
+      "loss": 0.978,
+      "step": 4716
+    },
+    {
+      "epoch": 0.4936682365253794,
+      "grad_norm": 1.80863597665584,
+      "learning_rate": 1.0690951945947263e-05,
+      "loss": 0.8808,
+      "step": 4717
+    },
+    {
+      "epoch": 0.4937728937728938,
+      "grad_norm": 1.886985809641338,
+      "learning_rate": 1.0687570287336082e-05,
+      "loss": 0.8756,
+      "step": 4718
+    },
+    {
+      "epoch": 0.49387755102040815,
+      "grad_norm": 2.0556937103470143,
+      "learning_rate": 1.0684188549721683e-05,
+      "loss": 0.9244,
+      "step": 4719
+    },
+    {
+      "epoch": 0.49398220826792255,
+      "grad_norm": 2.0110374519014154,
+      "learning_rate": 1.0680806733492632e-05,
+      "loss": 1.0597,
+      "step": 4720
+    },
+    {
+      "epoch": 0.49408686551543696,
+      "grad_norm": 1.7973197164685586,
+      "learning_rate": 1.0677424839037506e-05,
+      "loss": 0.9803,
+      "step": 4721
+    },
+    {
+      "epoch": 0.4941915227629513,
+      "grad_norm": 2.186835840861067,
+      "learning_rate": 1.0674042866744896e-05,
+      "loss": 0.9843,
+      "step": 4722
+    },
+    {
+      "epoch": 0.4942961800104657,
+      "grad_norm": 1.9713736175942005,
+      "learning_rate": 1.0670660817003391e-05,
+      "loss": 1.0191,
+      "step": 4723
+    },
+    {
+      "epoch": 0.4944008372579801,
+      "grad_norm": 2.1417694138266845,
+      "learning_rate": 1.0667278690201597e-05,
+      "loss": 0.9995,
+      "step": 4724
+    },
+    {
+      "epoch": 0.4945054945054945,
+      "grad_norm": 2.8859513823150724,
+      "learning_rate": 1.0663896486728134e-05,
+      "loss": 1.1489,
+      "step": 4725
+    },
+    {
+      "epoch": 0.4946101517530089,
+      "grad_norm": 1.9628740590998055,
+      "learning_rate": 1.0660514206971615e-05,
+      "loss": 0.9062,
+      "step": 4726
+    },
+    {
+      "epoch": 0.4947148090005233,
+      "grad_norm": 1.8512096693230278,
+      "learning_rate": 1.0657131851320677e-05,
+      "loss": 0.8803,
+      "step": 4727
+    },
+    {
+      "epoch": 0.4948194662480377,
+      "grad_norm": 2.433646998663187,
+      "learning_rate": 1.0653749420163953e-05,
+      "loss": 0.9568,
+      "step": 4728
+    },
+    {
+      "epoch": 0.4949241234955521,
+      "grad_norm": 2.0345077036147656,
+      "learning_rate": 1.06503669138901e-05,
+      "loss": 0.7353,
+      "step": 4729
+    },
+    {
+      "epoch": 0.49502878074306644,
+      "grad_norm": 1.9424603161161924,
+      "learning_rate": 1.0646984332887766e-05,
+      "loss": 0.968,
+      "step": 4730
+    },
+    {
+      "epoch": 0.49513343799058085,
+      "grad_norm": 2.3399922101360553,
+      "learning_rate": 1.0643601677545626e-05,
+      "loss": 0.9581,
+      "step": 4731
+    },
+    {
+      "epoch": 0.49523809523809526,
+      "grad_norm": 1.8649126698331577,
+      "learning_rate": 1.0640218948252345e-05,
+      "loss": 0.7876,
+      "step": 4732
+    },
+    {
+      "epoch": 0.4953427524856096,
+      "grad_norm": 2.331216885557431,
+      "learning_rate": 1.0636836145396607e-05,
+      "loss": 0.9769,
+      "step": 4733
+    },
+    {
+      "epoch": 0.495447409733124,
+      "grad_norm": 2.163582657761894,
+      "learning_rate": 1.063345326936711e-05,
+      "loss": 0.9926,
+      "step": 4734
+    },
+    {
+      "epoch": 0.4955520669806384,
+      "grad_norm": 2.081349677166352,
+      "learning_rate": 1.0630070320552544e-05,
+      "loss": 1.0198,
+      "step": 4735
+    },
+    {
+      "epoch": 0.4956567242281528,
+      "grad_norm": 1.7921126006767771,
+      "learning_rate": 1.0626687299341622e-05,
+      "loss": 0.8941,
+      "step": 4736
+    },
+    {
+      "epoch": 0.4957613814756672,
+      "grad_norm": 1.7566094714296194,
+      "learning_rate": 1.062330420612306e-05,
+      "loss": 0.9175,
+      "step": 4737
+    },
+    {
+      "epoch": 0.4958660387231816,
+      "grad_norm": 2.3106445373652496,
+      "learning_rate": 1.0619921041285579e-05,
+      "loss": 0.9188,
+      "step": 4738
+    },
+    {
+      "epoch": 0.495970695970696,
+      "grad_norm": 2.8264044010883413,
+      "learning_rate": 1.0616537805217918e-05,
+      "loss": 0.8804,
+      "step": 4739
+    },
+    {
+      "epoch": 0.49607535321821034,
+      "grad_norm": 1.9520944530357538,
+      "learning_rate": 1.0613154498308808e-05,
+      "loss": 0.957,
+      "step": 4740
+    },
+    {
+      "epoch": 0.49618001046572474,
+      "grad_norm": 2.3861280899553905,
+      "learning_rate": 1.0609771120947005e-05,
+      "loss": 0.9467,
+      "step": 4741
+    },
+    {
+      "epoch": 0.49628466771323915,
+      "grad_norm": 2.0740843031163045,
+      "learning_rate": 1.0606387673521265e-05,
+      "loss": 0.9223,
+      "step": 4742
+    },
+    {
+      "epoch": 0.49638932496075355,
+      "grad_norm": 9.399003361754902,
+      "learning_rate": 1.0603004156420354e-05,
+      "loss": 1.2298,
+      "step": 4743
+    },
+    {
+      "epoch": 0.4964939822082679,
+      "grad_norm": 2.0945292190249747,
+      "learning_rate": 1.059962057003304e-05,
+      "loss": 1.0156,
+      "step": 4744
+    },
+    {
+      "epoch": 0.4965986394557823,
+      "grad_norm": 2.1545476746682453,
+      "learning_rate": 1.0596236914748107e-05,
+      "loss": 1.0291,
+      "step": 4745
+    },
+    {
+      "epoch": 0.4967032967032967,
+      "grad_norm": 1.8857323776138402,
+      "learning_rate": 1.0592853190954345e-05,
+      "loss": 0.9664,
+      "step": 4746
+    },
+    {
+      "epoch": 0.4968079539508111,
+      "grad_norm": 2.095800214096105,
+      "learning_rate": 1.0589469399040546e-05,
+      "loss": 0.9522,
+      "step": 4747
+    },
+    {
+      "epoch": 0.49691261119832547,
+      "grad_norm": 2.0247693141512055,
+      "learning_rate": 1.0586085539395523e-05,
+      "loss": 1.0192,
+      "step": 4748
+    },
+    {
+      "epoch": 0.4970172684458399,
+      "grad_norm": 1.8584866253076624,
+      "learning_rate": 1.0582701612408083e-05,
+      "loss": 0.9195,
+      "step": 4749
+    },
+    {
+      "epoch": 0.4971219256933543,
+      "grad_norm": 1.931335358563096,
+      "learning_rate": 1.0579317618467043e-05,
+      "loss": 0.9267,
+      "step": 4750
+    },
+    {
+      "epoch": 0.49722658294086863,
+      "grad_norm": 2.097579996549894,
+      "learning_rate": 1.0575933557961238e-05,
+      "loss": 0.9581,
+      "step": 4751
+    },
+    {
+      "epoch": 0.49733124018838304,
+      "grad_norm": 2.033270771923552,
+      "learning_rate": 1.0572549431279495e-05,
+      "loss": 0.9175,
+      "step": 4752
+    },
+    {
+      "epoch": 0.49743589743589745,
+      "grad_norm": 2.2607341333747772,
+      "learning_rate": 1.0569165238810666e-05,
+      "loss": 1.0193,
+      "step": 4753
+    },
+    {
+      "epoch": 0.49754055468341185,
+      "grad_norm": 1.8647489455827506,
+      "learning_rate": 1.0565780980943596e-05,
+      "loss": 0.9772,
+      "step": 4754
+    },
+    {
+      "epoch": 0.4976452119309262,
+      "grad_norm": 1.9057129930055388,
+      "learning_rate": 1.0562396658067142e-05,
+      "loss": 0.9925,
+      "step": 4755
+    },
+    {
+      "epoch": 0.4977498691784406,
+      "grad_norm": 2.8885566484907823,
+      "learning_rate": 1.0559012270570177e-05,
+      "loss": 0.9656,
+      "step": 4756
+    },
+    {
+      "epoch": 0.497854526425955,
+      "grad_norm": 1.895968677495429,
+      "learning_rate": 1.0555627818841563e-05,
+      "loss": 0.8475,
+      "step": 4757
+    },
+    {
+      "epoch": 0.49795918367346936,
+      "grad_norm": 2.019680946799338,
+      "learning_rate": 1.055224330327019e-05,
+      "loss": 0.983,
+      "step": 4758
+    },
+    {
+      "epoch": 0.49806384092098377,
+      "grad_norm": 1.9749773215931818,
+      "learning_rate": 1.0548858724244939e-05,
+      "loss": 0.948,
+      "step": 4759
+    },
+    {
+      "epoch": 0.4981684981684982,
+      "grad_norm": 1.9535266279189702,
+      "learning_rate": 1.0545474082154711e-05,
+      "loss": 0.9621,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4982731554160126,
+      "grad_norm": 3.9985738676646103,
+      "learning_rate": 1.0542089377388406e-05,
+      "loss": 0.9936,
+      "step": 4761
+    },
+    {
+      "epoch": 0.49837781266352693,
+      "grad_norm": 2.2584714009921356,
+      "learning_rate": 1.0538704610334929e-05,
+      "loss": 0.871,
+      "step": 4762
+    },
+    {
+      "epoch": 0.49848246991104134,
+      "grad_norm": 2.2312381629656404,
+      "learning_rate": 1.0535319781383202e-05,
+      "loss": 1.001,
+      "step": 4763
+    },
+    {
+      "epoch": 0.49858712715855574,
+      "grad_norm": 2.0424072639034816,
+      "learning_rate": 1.0531934890922147e-05,
+      "loss": 0.9762,
+      "step": 4764
+    },
+    {
+      "epoch": 0.4986917844060701,
+      "grad_norm": 2.1580470907610887,
+      "learning_rate": 1.0528549939340695e-05,
+      "loss": 0.9045,
+      "step": 4765
+    },
+    {
+      "epoch": 0.4987964416535845,
+      "grad_norm": 2.2467255664294528,
+      "learning_rate": 1.0525164927027783e-05,
+      "loss": 1.0969,
+      "step": 4766
+    },
+    {
+      "epoch": 0.4989010989010989,
+      "grad_norm": 1.982846113442221,
+      "learning_rate": 1.0521779854372353e-05,
+      "loss": 1.038,
+      "step": 4767
+    },
+    {
+      "epoch": 0.4990057561486133,
+      "grad_norm": 2.3854040119508233,
+      "learning_rate": 1.0518394721763365e-05,
+      "loss": 1.0746,
+      "step": 4768
+    },
+    {
+      "epoch": 0.49911041339612766,
+      "grad_norm": 1.9707658343258043,
+      "learning_rate": 1.051500952958977e-05,
+      "loss": 0.8012,
+      "step": 4769
+    },
+    {
+      "epoch": 0.49921507064364207,
+      "grad_norm": 2.1046625201814746,
+      "learning_rate": 1.051162427824054e-05,
+      "loss": 0.9312,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4993197278911565,
+      "grad_norm": 1.6661194083882027,
+      "learning_rate": 1.0508238968104638e-05,
+      "loss": 0.7945,
+      "step": 4771
+    },
+    {
+      "epoch": 0.4994243851386709,
+      "grad_norm": 2.083849894430902,
+      "learning_rate": 1.050485359957105e-05,
+      "loss": 0.8743,
+      "step": 4772
+    },
+    {
+      "epoch": 0.49952904238618523,
+      "grad_norm": 2.2924292467485974,
+      "learning_rate": 1.0501468173028762e-05,
+      "loss": 1.0575,
+      "step": 4773
+    },
+    {
+      "epoch": 0.49963369963369964,
+      "grad_norm": 1.9733099739758075,
+      "learning_rate": 1.0498082688866762e-05,
+      "loss": 0.9393,
+      "step": 4774
+    },
+    {
+      "epoch": 0.49973835688121404,
+      "grad_norm": 1.9074190492709282,
+      "learning_rate": 1.0494697147474053e-05,
+      "loss": 0.9925,
+      "step": 4775
+    },
+    {
+      "epoch": 0.4998430141287284,
+      "grad_norm": 1.859529806411289,
+      "learning_rate": 1.0491311549239635e-05,
+      "loss": 1.0015,
+      "step": 4776
+    },
+    {
+      "epoch": 0.4999476713762428,
+      "grad_norm": 1.8791033216965147,
+      "learning_rate": 1.0487925894552529e-05,
+      "loss": 1.0242,
+      "step": 4777
+    },
+    {
+      "epoch": 0.5000523286237571,
+      "grad_norm": 1.9342110638713974,
+      "learning_rate": 1.0484540183801744e-05,
+      "loss": 0.9966,
+      "step": 4778
+    },
+    {
+      "epoch": 0.5001569858712716,
+      "grad_norm": 1.7724120947349102,
+      "learning_rate": 1.0481154417376313e-05,
+      "loss": 0.9939,
+      "step": 4779
+    },
+    {
+      "epoch": 0.500261643118786,
+      "grad_norm": 2.154486064088719,
+      "learning_rate": 1.0477768595665263e-05,
+      "loss": 1.0931,
+      "step": 4780
+    },
+    {
+      "epoch": 0.5003663003663004,
+      "grad_norm": 2.1518890500826506,
+      "learning_rate": 1.0474382719057633e-05,
+      "loss": 0.9757,
+      "step": 4781
+    },
+    {
+      "epoch": 0.5004709576138148,
+      "grad_norm": 1.8763970416206928,
+      "learning_rate": 1.0470996787942466e-05,
+      "loss": 0.9491,
+      "step": 4782
+    },
+    {
+      "epoch": 0.5005756148613292,
+      "grad_norm": 2.3823417687483732,
+      "learning_rate": 1.0467610802708814e-05,
+      "loss": 1.045,
+      "step": 4783
+    },
+    {
+      "epoch": 0.5006802721088436,
+      "grad_norm": 2.0877142215744087,
+      "learning_rate": 1.0464224763745737e-05,
+      "loss": 0.8905,
+      "step": 4784
+    },
+    {
+      "epoch": 0.5007849293563579,
+      "grad_norm": 1.8525479505640599,
+      "learning_rate": 1.0460838671442286e-05,
+      "loss": 0.9826,
+      "step": 4785
+    },
+    {
+      "epoch": 0.5008895866038723,
+      "grad_norm": 2.029476451987434,
+      "learning_rate": 1.0457452526187541e-05,
+      "loss": 1.035,
+      "step": 4786
+    },
+    {
+      "epoch": 0.5009942438513867,
+      "grad_norm": 2.0510409778147842,
+      "learning_rate": 1.0454066328370573e-05,
+      "loss": 1.0014,
+      "step": 4787
+    },
+    {
+      "epoch": 0.5010989010989011,
+      "grad_norm": 2.050804457353104,
+      "learning_rate": 1.0450680078380465e-05,
+      "loss": 1.0106,
+      "step": 4788
+    },
+    {
+      "epoch": 0.5012035583464155,
+      "grad_norm": 1.9228482758753471,
+      "learning_rate": 1.04472937766063e-05,
+      "loss": 1.0036,
+      "step": 4789
+    },
+    {
+      "epoch": 0.5013082155939299,
+      "grad_norm": 1.8304036460035518,
+      "learning_rate": 1.0443907423437177e-05,
+      "loss": 0.9524,
+      "step": 4790
+    },
+    {
+      "epoch": 0.5014128728414443,
+      "grad_norm": 2.2850976774419918,
+      "learning_rate": 1.0440521019262187e-05,
+      "loss": 0.8331,
+      "step": 4791
+    },
+    {
+      "epoch": 0.5015175300889586,
+      "grad_norm": 2.038030446976533,
+      "learning_rate": 1.0437134564470442e-05,
+      "loss": 0.9628,
+      "step": 4792
+    },
+    {
+      "epoch": 0.501622187336473,
+      "grad_norm": 2.113415421458547,
+      "learning_rate": 1.0433748059451045e-05,
+      "loss": 1.0095,
+      "step": 4793
+    },
+    {
+      "epoch": 0.5017268445839874,
+      "grad_norm": 2.1782622269854675,
+      "learning_rate": 1.0430361504593118e-05,
+      "loss": 0.9503,
+      "step": 4794
+    },
+    {
+      "epoch": 0.5018315018315018,
+      "grad_norm": 2.238496790294893,
+      "learning_rate": 1.0426974900285784e-05,
+      "loss": 1.0506,
+      "step": 4795
+    },
+    {
+      "epoch": 0.5019361590790162,
+      "grad_norm": 2.1891901037629746,
+      "learning_rate": 1.0423588246918168e-05,
+      "loss": 1.0063,
+      "step": 4796
+    },
+    {
+      "epoch": 0.5020408163265306,
+      "grad_norm": 2.092133045742852,
+      "learning_rate": 1.0420201544879405e-05,
+      "loss": 1.0078,
+      "step": 4797
+    },
+    {
+      "epoch": 0.502145473574045,
+      "grad_norm": 1.9013198519529835,
+      "learning_rate": 1.0416814794558628e-05,
+      "loss": 0.9786,
+      "step": 4798
+    },
+    {
+      "epoch": 0.5022501308215594,
+      "grad_norm": 2.1099077143715914,
+      "learning_rate": 1.0413427996344991e-05,
+      "loss": 0.9702,
+      "step": 4799
+    },
+    {
+      "epoch": 0.5023547880690737,
+      "grad_norm": 1.7239541182847857,
+      "learning_rate": 1.0410041150627636e-05,
+      "loss": 0.9089,
+      "step": 4800
+    },
+    {
+      "epoch": 0.5024594453165881,
+      "grad_norm": 2.3381648619636137,
+      "learning_rate": 1.0406654257795723e-05,
+      "loss": 0.7951,
+      "step": 4801
+    },
+    {
+      "epoch": 0.5025641025641026,
+      "grad_norm": 2.282729233569848,
+      "learning_rate": 1.040326731823841e-05,
+      "loss": 1.0216,
+      "step": 4802
+    },
+    {
+      "epoch": 0.502668759811617,
+      "grad_norm": 2.2977040163903264,
+      "learning_rate": 1.0399880332344865e-05,
+      "loss": 1.0084,
+      "step": 4803
+    },
+    {
+      "epoch": 0.5027734170591314,
+      "grad_norm": 2.006735467260644,
+      "learning_rate": 1.0396493300504263e-05,
+      "loss": 0.9048,
+      "step": 4804
+    },
+    {
+      "epoch": 0.5028780743066458,
+      "grad_norm": 1.8040980766553913,
+      "learning_rate": 1.0393106223105771e-05,
+      "loss": 0.9794,
+      "step": 4805
+    },
+    {
+      "epoch": 0.5029827315541602,
+      "grad_norm": 2.131327716426053,
+      "learning_rate": 1.0389719100538583e-05,
+      "loss": 0.9558,
+      "step": 4806
+    },
+    {
+      "epoch": 0.5030873888016745,
+      "grad_norm": 2.0367496240204246,
+      "learning_rate": 1.0386331933191874e-05,
+      "loss": 0.8524,
+      "step": 4807
+    },
+    {
+      "epoch": 0.5031920460491889,
+      "grad_norm": 1.927303561475201,
+      "learning_rate": 1.0382944721454847e-05,
+      "loss": 0.8965,
+      "step": 4808
+    },
+    {
+      "epoch": 0.5032967032967033,
+      "grad_norm": 2.4684209425092485,
+      "learning_rate": 1.0379557465716696e-05,
+      "loss": 0.8173,
+      "step": 4809
+    },
+    {
+      "epoch": 0.5034013605442177,
+      "grad_norm": 1.9985491017896084,
+      "learning_rate": 1.0376170166366619e-05,
+      "loss": 1.0035,
+      "step": 4810
+    },
+    {
+      "epoch": 0.5035060177917321,
+      "grad_norm": 1.8490510380262115,
+      "learning_rate": 1.037278282379383e-05,
+      "loss": 0.889,
+      "step": 4811
+    },
+    {
+      "epoch": 0.5036106750392465,
+      "grad_norm": 1.9348394119418788,
+      "learning_rate": 1.0369395438387535e-05,
+      "loss": 0.8902,
+      "step": 4812
+    },
+    {
+      "epoch": 0.5037153322867609,
+      "grad_norm": 2.08472258783819,
+      "learning_rate": 1.0366008010536962e-05,
+      "loss": 0.8893,
+      "step": 4813
+    },
+    {
+      "epoch": 0.5038199895342752,
+      "grad_norm": 1.8325956711014155,
+      "learning_rate": 1.036262054063132e-05,
+      "loss": 0.9288,
+      "step": 4814
+    },
+    {
+      "epoch": 0.5039246467817896,
+      "grad_norm": 1.9933367795637793,
+      "learning_rate": 1.0359233029059845e-05,
+      "loss": 0.886,
+      "step": 4815
+    },
+    {
+      "epoch": 0.504029304029304,
+      "grad_norm": 2.1844459941163157,
+      "learning_rate": 1.035584547621177e-05,
+      "loss": 0.8664,
+      "step": 4816
+    },
+    {
+      "epoch": 0.5041339612768184,
+      "grad_norm": 2.124796059309372,
+      "learning_rate": 1.0352457882476326e-05,
+      "loss": 0.901,
+      "step": 4817
+    },
+    {
+      "epoch": 0.5042386185243328,
+      "grad_norm": 2.0620543862987395,
+      "learning_rate": 1.0349070248242756e-05,
+      "loss": 1.0371,
+      "step": 4818
+    },
+    {
+      "epoch": 0.5043432757718472,
+      "grad_norm": 1.9219187469619123,
+      "learning_rate": 1.0345682573900306e-05,
+      "loss": 0.8801,
+      "step": 4819
+    },
+    {
+      "epoch": 0.5044479330193616,
+      "grad_norm": 1.9754947289145173,
+      "learning_rate": 1.0342294859838228e-05,
+      "loss": 0.9137,
+      "step": 4820
+    },
+    {
+      "epoch": 0.5045525902668759,
+      "grad_norm": 1.9746909929500338,
+      "learning_rate": 1.0338907106445781e-05,
+      "loss": 1.008,
+      "step": 4821
+    },
+    {
+      "epoch": 0.5046572475143903,
+      "grad_norm": 1.827473906585448,
+      "learning_rate": 1.0335519314112217e-05,
+      "loss": 0.8131,
+      "step": 4822
+    },
+    {
+      "epoch": 0.5047619047619047,
+      "grad_norm": 2.196935612027101,
+      "learning_rate": 1.0332131483226805e-05,
+      "loss": 1.0269,
+      "step": 4823
+    },
+    {
+      "epoch": 0.5048665620094192,
+      "grad_norm": 2.0147704818493,
+      "learning_rate": 1.0328743614178807e-05,
+      "loss": 0.8966,
+      "step": 4824
+    },
+    {
+      "epoch": 0.5049712192569336,
+      "grad_norm": 1.7839156040063775,
+      "learning_rate": 1.0325355707357507e-05,
+      "loss": 0.9519,
+      "step": 4825
+    },
+    {
+      "epoch": 0.505075876504448,
+      "grad_norm": 2.299507673691946,
+      "learning_rate": 1.0321967763152176e-05,
+      "loss": 0.959,
+      "step": 4826
+    },
+    {
+      "epoch": 0.5051805337519624,
+      "grad_norm": 2.2267555529088825,
+      "learning_rate": 1.0318579781952095e-05,
+      "loss": 0.8851,
+      "step": 4827
+    },
+    {
+      "epoch": 0.5052851909994767,
+      "grad_norm": 2.102441977176161,
+      "learning_rate": 1.0315191764146551e-05,
+      "loss": 0.9512,
+      "step": 4828
+    },
+    {
+      "epoch": 0.5053898482469911,
+      "grad_norm": 1.9711747773484447,
+      "learning_rate": 1.0311803710124832e-05,
+      "loss": 1.0464,
+      "step": 4829
+    },
+    {
+      "epoch": 0.5054945054945055,
+      "grad_norm": 2.1197570532175285,
+      "learning_rate": 1.030841562027624e-05,
+      "loss": 0.914,
+      "step": 4830
+    },
+    {
+      "epoch": 0.5055991627420199,
+      "grad_norm": 2.2152984722921327,
+      "learning_rate": 1.0305027494990065e-05,
+      "loss": 0.9595,
+      "step": 4831
+    },
+    {
+      "epoch": 0.5057038199895343,
+      "grad_norm": 2.006104338040643,
+      "learning_rate": 1.0301639334655612e-05,
+      "loss": 0.9543,
+      "step": 4832
+    },
+    {
+      "epoch": 0.5058084772370487,
+      "grad_norm": 2.233956630960079,
+      "learning_rate": 1.029825113966219e-05,
+      "loss": 1.0661,
+      "step": 4833
+    },
+    {
+      "epoch": 0.5059131344845631,
+      "grad_norm": 1.8056443704209098,
+      "learning_rate": 1.0294862910399106e-05,
+      "loss": 0.778,
+      "step": 4834
+    },
+    {
+      "epoch": 0.5060177917320774,
+      "grad_norm": 1.8210897127610195,
+      "learning_rate": 1.0291474647255677e-05,
+      "loss": 1.0009,
+      "step": 4835
+    },
+    {
+      "epoch": 0.5061224489795918,
+      "grad_norm": 2.265489705370913,
+      "learning_rate": 1.0288086350621219e-05,
+      "loss": 0.8936,
+      "step": 4836
+    },
+    {
+      "epoch": 0.5062271062271062,
+      "grad_norm": 2.2329819145492724,
+      "learning_rate": 1.0284698020885054e-05,
+      "loss": 0.8475,
+      "step": 4837
+    },
+    {
+      "epoch": 0.5063317634746206,
+      "grad_norm": 1.9140187907534734,
+      "learning_rate": 1.0281309658436514e-05,
+      "loss": 0.9804,
+      "step": 4838
+    },
+    {
+      "epoch": 0.506436420722135,
+      "grad_norm": 2.1495837609117077,
+      "learning_rate": 1.0277921263664918e-05,
+      "loss": 0.9649,
+      "step": 4839
+    },
+    {
+      "epoch": 0.5065410779696494,
+      "grad_norm": 2.147434202179756,
+      "learning_rate": 1.027453283695961e-05,
+      "loss": 0.9257,
+      "step": 4840
+    },
+    {
+      "epoch": 0.5066457352171638,
+      "grad_norm": 2.1508850173428504,
+      "learning_rate": 1.0271144378709919e-05,
+      "loss": 1.0815,
+      "step": 4841
+    },
+    {
+      "epoch": 0.5067503924646782,
+      "grad_norm": 1.890843092683489,
+      "learning_rate": 1.026775588930519e-05,
+      "loss": 0.9078,
+      "step": 4842
+    },
+    {
+      "epoch": 0.5068550497121925,
+      "grad_norm": 1.718743135555168,
+      "learning_rate": 1.0264367369134767e-05,
+      "loss": 0.8615,
+      "step": 4843
+    },
+    {
+      "epoch": 0.5069597069597069,
+      "grad_norm": 2.1338027250972718,
+      "learning_rate": 1.0260978818588e-05,
+      "loss": 0.9485,
+      "step": 4844
+    },
+    {
+      "epoch": 0.5070643642072213,
+      "grad_norm": 2.3030713079734197,
+      "learning_rate": 1.0257590238054238e-05,
+      "loss": 1.0152,
+      "step": 4845
+    },
+    {
+      "epoch": 0.5071690214547357,
+      "grad_norm": 2.108066963927156,
+      "learning_rate": 1.0254201627922829e-05,
+      "loss": 0.8987,
+      "step": 4846
+    },
+    {
+      "epoch": 0.5072736787022502,
+      "grad_norm": 1.9348248604669798,
+      "learning_rate": 1.0250812988583144e-05,
+      "loss": 0.8945,
+      "step": 4847
+    },
+    {
+      "epoch": 0.5073783359497646,
+      "grad_norm": 2.0214829385791426,
+      "learning_rate": 1.0247424320424534e-05,
+      "loss": 0.9324,
+      "step": 4848
+    },
+    {
+      "epoch": 0.507482993197279,
+      "grad_norm": 2.103062514985118,
+      "learning_rate": 1.024403562383637e-05,
+      "loss": 0.9869,
+      "step": 4849
+    },
+    {
+      "epoch": 0.5075876504447933,
+      "grad_norm": 2.0388689399057633,
+      "learning_rate": 1.0240646899208018e-05,
+      "loss": 1.0085,
+      "step": 4850
+    },
+    {
+      "epoch": 0.5076923076923077,
+      "grad_norm": 2.2019460294281243,
+      "learning_rate": 1.0237258146928849e-05,
+      "loss": 0.9457,
+      "step": 4851
+    },
+    {
+      "epoch": 0.5077969649398221,
+      "grad_norm": 1.8404221416654807,
+      "learning_rate": 1.023386936738824e-05,
+      "loss": 0.9133,
+      "step": 4852
+    },
+    {
+      "epoch": 0.5079016221873365,
+      "grad_norm": 1.9361603963782874,
+      "learning_rate": 1.0230480560975564e-05,
+      "loss": 0.9412,
+      "step": 4853
+    },
+    {
+      "epoch": 0.5080062794348509,
+      "grad_norm": 2.0201875001702807,
+      "learning_rate": 1.0227091728080203e-05,
+      "loss": 0.9903,
+      "step": 4854
+    },
+    {
+      "epoch": 0.5081109366823653,
+      "grad_norm": 2.174119911436257,
+      "learning_rate": 1.0223702869091548e-05,
+      "loss": 0.9529,
+      "step": 4855
+    },
+    {
+      "epoch": 0.5082155939298797,
+      "grad_norm": 2.1378859332077047,
+      "learning_rate": 1.0220313984398972e-05,
+      "loss": 1.0162,
+      "step": 4856
+    },
+    {
+      "epoch": 0.508320251177394,
+      "grad_norm": 1.9197826619859129,
+      "learning_rate": 1.0216925074391879e-05,
+      "loss": 0.9307,
+      "step": 4857
+    },
+    {
+      "epoch": 0.5084249084249084,
+      "grad_norm": 2.2932973290164096,
+      "learning_rate": 1.0213536139459651e-05,
+      "loss": 1.0056,
+      "step": 4858
+    },
+    {
+      "epoch": 0.5085295656724228,
+      "grad_norm": 2.2317819242913695,
+      "learning_rate": 1.021014717999169e-05,
+      "loss": 0.9703,
+      "step": 4859
+    },
+    {
+      "epoch": 0.5086342229199372,
+      "grad_norm": 2.1281106553255675,
+      "learning_rate": 1.0206758196377391e-05,
+      "loss": 1.0403,
+      "step": 4860
+    },
+    {
+      "epoch": 0.5087388801674516,
+      "grad_norm": 2.171215516994493,
+      "learning_rate": 1.0203369189006157e-05,
+      "loss": 0.8991,
+      "step": 4861
+    },
+    {
+      "epoch": 0.508843537414966,
+      "grad_norm": 2.255784354708711,
+      "learning_rate": 1.019998015826739e-05,
+      "loss": 0.8639,
+      "step": 4862
+    },
+    {
+      "epoch": 0.5089481946624804,
+      "grad_norm": 2.2311385876757166,
+      "learning_rate": 1.0196591104550497e-05,
+      "loss": 0.9527,
+      "step": 4863
+    },
+    {
+      "epoch": 0.5090528519099947,
+      "grad_norm": 2.1216367946605614,
+      "learning_rate": 1.0193202028244891e-05,
+      "loss": 0.7992,
+      "step": 4864
+    },
+    {
+      "epoch": 0.5091575091575091,
+      "grad_norm": 2.2266924800513928,
+      "learning_rate": 1.0189812929739976e-05,
+      "loss": 0.9312,
+      "step": 4865
+    },
+    {
+      "epoch": 0.5092621664050235,
+      "grad_norm": 2.184861742273941,
+      "learning_rate": 1.0186423809425175e-05,
+      "loss": 0.9421,
+      "step": 4866
+    },
+    {
+      "epoch": 0.5093668236525379,
+      "grad_norm": 2.2359774420575045,
+      "learning_rate": 1.0183034667689898e-05,
+      "loss": 0.8726,
+      "step": 4867
+    },
+    {
+      "epoch": 0.5094714809000523,
+      "grad_norm": 2.251077097785011,
+      "learning_rate": 1.0179645504923565e-05,
+      "loss": 0.9215,
+      "step": 4868
+    },
+    {
+      "epoch": 0.5095761381475667,
+      "grad_norm": 1.962671210761371,
+      "learning_rate": 1.0176256321515601e-05,
+      "loss": 0.9715,
+      "step": 4869
+    },
+    {
+      "epoch": 0.5096807953950812,
+      "grad_norm": 2.1786034572062687,
+      "learning_rate": 1.0172867117855428e-05,
+      "loss": 0.9004,
+      "step": 4870
+    },
+    {
+      "epoch": 0.5097854526425954,
+      "grad_norm": 1.984923185389256,
+      "learning_rate": 1.0169477894332473e-05,
+      "loss": 1.0999,
+      "step": 4871
+    },
+    {
+      "epoch": 0.5098901098901099,
+      "grad_norm": 2.3249280492593107,
+      "learning_rate": 1.016608865133616e-05,
+      "loss": 1.0306,
+      "step": 4872
+    },
+    {
+      "epoch": 0.5099947671376243,
+      "grad_norm": 2.631735537137025,
+      "learning_rate": 1.0162699389255933e-05,
+      "loss": 0.7941,
+      "step": 4873
+    },
+    {
+      "epoch": 0.5100994243851387,
+      "grad_norm": 2.074101642604738,
+      "learning_rate": 1.0159310108481212e-05,
+      "loss": 0.8317,
+      "step": 4874
+    },
+    {
+      "epoch": 0.5102040816326531,
+      "grad_norm": 2.0707425216929094,
+      "learning_rate": 1.0155920809401437e-05,
+      "loss": 0.9869,
+      "step": 4875
+    },
+    {
+      "epoch": 0.5103087388801675,
+      "grad_norm": 1.9855437335816901,
+      "learning_rate": 1.015253149240604e-05,
+      "loss": 0.8376,
+      "step": 4876
+    },
+    {
+      "epoch": 0.5104133961276819,
+      "grad_norm": 2.125627681296742,
+      "learning_rate": 1.014914215788447e-05,
+      "loss": 0.8265,
+      "step": 4877
+    },
+    {
+      "epoch": 0.5105180533751962,
+      "grad_norm": 2.0097025281038183,
+      "learning_rate": 1.0145752806226165e-05,
+      "loss": 1.0651,
+      "step": 4878
+    },
+    {
+      "epoch": 0.5106227106227106,
+      "grad_norm": 1.9381824884213676,
+      "learning_rate": 1.0142363437820566e-05,
+      "loss": 0.9377,
+      "step": 4879
+    },
+    {
+      "epoch": 0.510727367870225,
+      "grad_norm": 1.8057048082927534,
+      "learning_rate": 1.0138974053057118e-05,
+      "loss": 1.0258,
+      "step": 4880
+    },
+    {
+      "epoch": 0.5108320251177394,
+      "grad_norm": 1.9284372353034598,
+      "learning_rate": 1.013558465232527e-05,
+      "loss": 0.9767,
+      "step": 4881
+    },
+    {
+      "epoch": 0.5109366823652538,
+      "grad_norm": 2.0912145996565807,
+      "learning_rate": 1.0132195236014471e-05,
+      "loss": 0.9621,
+      "step": 4882
+    },
+    {
+      "epoch": 0.5110413396127682,
+      "grad_norm": 2.4074410237986203,
+      "learning_rate": 1.0128805804514173e-05,
+      "loss": 0.9953,
+      "step": 4883
+    },
+    {
+      "epoch": 0.5111459968602826,
+      "grad_norm": 2.1350122261578948,
+      "learning_rate": 1.0125416358213825e-05,
+      "loss": 0.9114,
+      "step": 4884
+    },
+    {
+      "epoch": 0.511250654107797,
+      "grad_norm": 2.1766627738568824,
+      "learning_rate": 1.0122026897502882e-05,
+      "loss": 0.9303,
+      "step": 4885
+    },
+    {
+      "epoch": 0.5113553113553113,
+      "grad_norm": 1.659780196796726,
+      "learning_rate": 1.0118637422770803e-05,
+      "loss": 0.8643,
+      "step": 4886
+    },
+    {
+      "epoch": 0.5114599686028257,
+      "grad_norm": 1.8759283670973985,
+      "learning_rate": 1.0115247934407045e-05,
+      "loss": 0.9288,
+      "step": 4887
+    },
+    {
+      "epoch": 0.5115646258503401,
+      "grad_norm": 1.981277049724218,
+      "learning_rate": 1.0111858432801063e-05,
+      "loss": 0.807,
+      "step": 4888
+    },
+    {
+      "epoch": 0.5116692830978545,
+      "grad_norm": 2.104064781470603,
+      "learning_rate": 1.0108468918342317e-05,
+      "loss": 0.9398,
+      "step": 4889
+    },
+    {
+      "epoch": 0.5117739403453689,
+      "grad_norm": 1.862863320953064,
+      "learning_rate": 1.0105079391420281e-05,
+      "loss": 0.8971,
+      "step": 4890
+    },
+    {
+      "epoch": 0.5118785975928833,
+      "grad_norm": 2.0632811238069872,
+      "learning_rate": 1.0101689852424404e-05,
+      "loss": 0.9229,
+      "step": 4891
+    },
+    {
+      "epoch": 0.5119832548403978,
+      "grad_norm": 2.0328293416753787,
+      "learning_rate": 1.0098300301744159e-05,
+      "loss": 0.8541,
+      "step": 4892
+    },
+    {
+      "epoch": 0.512087912087912,
+      "grad_norm": 2.2183959790280094,
+      "learning_rate": 1.0094910739769007e-05,
+      "loss": 0.9588,
+      "step": 4893
+    },
+    {
+      "epoch": 0.5121925693354265,
+      "grad_norm": 2.0086969358639597,
+      "learning_rate": 1.0091521166888422e-05,
+      "loss": 0.9431,
+      "step": 4894
+    },
+    {
+      "epoch": 0.5122972265829409,
+      "grad_norm": 1.8566480087720016,
+      "learning_rate": 1.0088131583491869e-05,
+      "loss": 0.9237,
+      "step": 4895
+    },
+    {
+      "epoch": 0.5124018838304553,
+      "grad_norm": 1.978008384805466,
+      "learning_rate": 1.0084741989968818e-05,
+      "loss": 0.9383,
+      "step": 4896
+    },
+    {
+      "epoch": 0.5125065410779697,
+      "grad_norm": 2.2899059061417146,
+      "learning_rate": 1.008135238670874e-05,
+      "loss": 1.0155,
+      "step": 4897
+    },
+    {
+      "epoch": 0.5126111983254841,
+      "grad_norm": 2.1435767412931215,
+      "learning_rate": 1.0077962774101108e-05,
+      "loss": 0.7721,
+      "step": 4898
+    },
+    {
+      "epoch": 0.5127158555729985,
+      "grad_norm": 1.9651414340794402,
+      "learning_rate": 1.0074573152535397e-05,
+      "loss": 0.8895,
+      "step": 4899
+    },
+    {
+      "epoch": 0.5128205128205128,
+      "grad_norm": 1.9700996694269328,
+      "learning_rate": 1.0071183522401078e-05,
+      "loss": 0.8386,
+      "step": 4900
+    },
+    {
+      "epoch": 0.5129251700680272,
+      "grad_norm": 2.043250711393673,
+      "learning_rate": 1.0067793884087626e-05,
+      "loss": 1.0259,
+      "step": 4901
+    },
+    {
+      "epoch": 0.5130298273155416,
+      "grad_norm": 2.2215963106479064,
+      "learning_rate": 1.0064404237984522e-05,
+      "loss": 1.0239,
+      "step": 4902
+    },
+    {
+      "epoch": 0.513134484563056,
+      "grad_norm": 2.043312538764727,
+      "learning_rate": 1.0061014584481242e-05,
+      "loss": 0.9675,
+      "step": 4903
+    },
+    {
+      "epoch": 0.5132391418105704,
+      "grad_norm": 2.017720307338605,
+      "learning_rate": 1.0057624923967258e-05,
+      "loss": 1.0402,
+      "step": 4904
+    },
+    {
+      "epoch": 0.5133437990580848,
+      "grad_norm": 1.9522219810870778,
+      "learning_rate": 1.0054235256832057e-05,
+      "loss": 0.9143,
+      "step": 4905
+    },
+    {
+      "epoch": 0.5134484563055992,
+      "grad_norm": 2.3060614281471423,
+      "learning_rate": 1.0050845583465112e-05,
+      "loss": 0.8615,
+      "step": 4906
+    },
+    {
+      "epoch": 0.5135531135531135,
+      "grad_norm": 2.070929318747068,
+      "learning_rate": 1.004745590425591e-05,
+      "loss": 0.8553,
+      "step": 4907
+    },
+    {
+      "epoch": 0.5136577708006279,
+      "grad_norm": 2.198912283882256,
+      "learning_rate": 1.0044066219593925e-05,
+      "loss": 0.9677,
+      "step": 4908
+    },
+    {
+      "epoch": 0.5137624280481423,
+      "grad_norm": 2.2590338340709204,
+      "learning_rate": 1.0040676529868644e-05,
+      "loss": 1.0453,
+      "step": 4909
+    },
+    {
+      "epoch": 0.5138670852956567,
+      "grad_norm": 2.3124159355511766,
+      "learning_rate": 1.0037286835469546e-05,
+      "loss": 1.1255,
+      "step": 4910
+    },
+    {
+      "epoch": 0.5139717425431711,
+      "grad_norm": 1.8769919169356388,
+      "learning_rate": 1.0033897136786115e-05,
+      "loss": 0.8565,
+      "step": 4911
+    },
+    {
+      "epoch": 0.5140763997906855,
+      "grad_norm": 2.032937691495361,
+      "learning_rate": 1.0030507434207836e-05,
+      "loss": 0.9021,
+      "step": 4912
+    },
+    {
+      "epoch": 0.5141810570381999,
+      "grad_norm": 1.948805286567476,
+      "learning_rate": 1.002711772812419e-05,
+      "loss": 0.9255,
+      "step": 4913
+    },
+    {
+      "epoch": 0.5142857142857142,
+      "grad_norm": 1.8811051091573516,
+      "learning_rate": 1.0023728018924663e-05,
+      "loss": 0.9634,
+      "step": 4914
+    },
+    {
+      "epoch": 0.5143903715332286,
+      "grad_norm": 1.895037622840537,
+      "learning_rate": 1.0020338306998739e-05,
+      "loss": 0.7764,
+      "step": 4915
+    },
+    {
+      "epoch": 0.514495028780743,
+      "grad_norm": 2.3622734576508417,
+      "learning_rate": 1.0016948592735898e-05,
+      "loss": 0.9616,
+      "step": 4916
+    },
+    {
+      "epoch": 0.5145996860282575,
+      "grad_norm": 2.1068542919039004,
+      "learning_rate": 1.0013558876525635e-05,
+      "loss": 0.9764,
+      "step": 4917
+    },
+    {
+      "epoch": 0.5147043432757719,
+      "grad_norm": 2.0518452033894428,
+      "learning_rate": 1.0010169158757425e-05,
+      "loss": 0.9017,
+      "step": 4918
+    },
+    {
+      "epoch": 0.5148090005232863,
+      "grad_norm": 2.268883115799465,
+      "learning_rate": 1.000677943982076e-05,
+      "loss": 0.8253,
+      "step": 4919
+    },
+    {
+      "epoch": 0.5149136577708007,
+      "grad_norm": 1.9101165871841854,
+      "learning_rate": 1.000338972010512e-05,
+      "loss": 1.0396,
+      "step": 4920
+    },
+    {
+      "epoch": 0.515018315018315,
+      "grad_norm": 1.9018696065166185,
+      "learning_rate": 1e-05,
+      "loss": 0.8675,
+      "step": 4921
+    },
+    {
+      "epoch": 0.5151229722658294,
+      "grad_norm": 1.9135895846787352,
+      "learning_rate": 9.99661027989488e-06,
+      "loss": 0.953,
+      "step": 4922
+    },
+    {
+      "epoch": 0.5152276295133438,
+      "grad_norm": 2.2944292672819744,
+      "learning_rate": 9.993220560179244e-06,
+      "loss": 0.9654,
+      "step": 4923
+    },
+    {
+      "epoch": 0.5153322867608582,
+      "grad_norm": 2.3373997913546187,
+      "learning_rate": 9.98983084124258e-06,
+      "loss": 1.0137,
+      "step": 4924
+    },
+    {
+      "epoch": 0.5154369440083726,
+      "grad_norm": 2.1760314698321697,
+      "learning_rate": 9.986441123474371e-06,
+      "loss": 1.0149,
+      "step": 4925
+    },
+    {
+      "epoch": 0.515541601255887,
+      "grad_norm": 2.0995522000742635,
+      "learning_rate": 9.983051407264102e-06,
+      "loss": 0.994,
+      "step": 4926
+    },
+    {
+      "epoch": 0.5156462585034014,
+      "grad_norm": 2.135301385827847,
+      "learning_rate": 9.979661693001267e-06,
+      "loss": 0.9958,
+      "step": 4927
+    },
+    {
+      "epoch": 0.5157509157509158,
+      "grad_norm": 2.016177869195349,
+      "learning_rate": 9.976271981075339e-06,
+      "loss": 1.0032,
+      "step": 4928
+    },
+    {
+      "epoch": 0.5158555729984301,
+      "grad_norm": 2.198242514789201,
+      "learning_rate": 9.972882271875814e-06,
+      "loss": 0.9751,
+      "step": 4929
+    },
+    {
+      "epoch": 0.5159602302459445,
+      "grad_norm": 2.568304071771772,
+      "learning_rate": 9.969492565792166e-06,
+      "loss": 0.8212,
+      "step": 4930
+    },
+    {
+      "epoch": 0.5160648874934589,
+      "grad_norm": 2.150285720935691,
+      "learning_rate": 9.966102863213889e-06,
+      "loss": 0.8911,
+      "step": 4931
+    },
+    {
+      "epoch": 0.5161695447409733,
+      "grad_norm": 1.8427410847871126,
+      "learning_rate": 9.962713164530457e-06,
+      "loss": 0.9333,
+      "step": 4932
+    },
+    {
+      "epoch": 0.5162742019884877,
+      "grad_norm": 1.9777153731928134,
+      "learning_rate": 9.95932347013136e-06,
+      "loss": 0.9572,
+      "step": 4933
+    },
+    {
+      "epoch": 0.5163788592360021,
+      "grad_norm": 2.023147402384463,
+      "learning_rate": 9.955933780406079e-06,
+      "loss": 0.9403,
+      "step": 4934
+    },
+    {
+      "epoch": 0.5164835164835165,
+      "grad_norm": 2.339289650649997,
+      "learning_rate": 9.952544095744092e-06,
+      "loss": 1.012,
+      "step": 4935
+    },
+    {
+      "epoch": 0.5165881737310308,
+      "grad_norm": 2.0703824315481447,
+      "learning_rate": 9.94915441653489e-06,
+      "loss": 0.8705,
+      "step": 4936
+    },
+    {
+      "epoch": 0.5166928309785452,
+      "grad_norm": 2.1054511171333252,
+      "learning_rate": 9.945764743167947e-06,
+      "loss": 0.8875,
+      "step": 4937
+    },
+    {
+      "epoch": 0.5167974882260596,
+      "grad_norm": 2.0257304303475108,
+      "learning_rate": 9.942375076032745e-06,
+      "loss": 0.792,
+      "step": 4938
+    },
+    {
+      "epoch": 0.516902145473574,
+      "grad_norm": 2.1389835239632298,
+      "learning_rate": 9.938985415518763e-06,
+      "loss": 0.8896,
+      "step": 4939
+    },
+    {
+      "epoch": 0.5170068027210885,
+      "grad_norm": 2.2239270796318156,
+      "learning_rate": 9.935595762015481e-06,
+      "loss": 0.9462,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5171114599686029,
+      "grad_norm": 1.922334466980784,
+      "learning_rate": 9.932206115912379e-06,
+      "loss": 0.9672,
+      "step": 4941
+    },
+    {
+      "epoch": 0.5172161172161173,
+      "grad_norm": 2.0497231083957583,
+      "learning_rate": 9.928816477598927e-06,
+      "loss": 0.9285,
+      "step": 4942
+    },
+    {
+      "epoch": 0.5173207744636316,
+      "grad_norm": 2.036361821419443,
+      "learning_rate": 9.925426847464605e-06,
+      "loss": 0.9388,
+      "step": 4943
+    },
+    {
+      "epoch": 0.517425431711146,
+      "grad_norm": 2.0485533284742767,
+      "learning_rate": 9.922037225898893e-06,
+      "loss": 1.0012,
+      "step": 4944
+    },
+    {
+      "epoch": 0.5175300889586604,
+      "grad_norm": 1.781792397254675,
+      "learning_rate": 9.918647613291262e-06,
+      "loss": 0.9396,
+      "step": 4945
+    },
+    {
+      "epoch": 0.5176347462061748,
+      "grad_norm": 2.6325182895039045,
+      "learning_rate": 9.915258010031184e-06,
+      "loss": 1.0054,
+      "step": 4946
+    },
+    {
+      "epoch": 0.5177394034536892,
+      "grad_norm": 2.24395320097852,
+      "learning_rate": 9.911868416508133e-06,
+      "loss": 0.8993,
+      "step": 4947
+    },
+    {
+      "epoch": 0.5178440607012036,
+      "grad_norm": 1.950860927917329,
+      "learning_rate": 9.908478833111581e-06,
+      "loss": 0.7653,
+      "step": 4948
+    },
+    {
+      "epoch": 0.517948717948718,
+      "grad_norm": 2.3503369439679136,
+      "learning_rate": 9.905089260230994e-06,
+      "loss": 0.8748,
+      "step": 4949
+    },
+    {
+      "epoch": 0.5180533751962323,
+      "grad_norm": 1.8774277642314408,
+      "learning_rate": 9.901699698255846e-06,
+      "loss": 0.989,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5181580324437467,
+      "grad_norm": 1.8303975022931225,
+      "learning_rate": 9.8983101475756e-06,
+      "loss": 1.0733,
+      "step": 4951
+    },
+    {
+      "epoch": 0.5182626896912611,
+      "grad_norm": 2.317212810335493,
+      "learning_rate": 9.89492060857972e-06,
+      "loss": 1.1076,
+      "step": 4952
+    },
+    {
+      "epoch": 0.5183673469387755,
+      "grad_norm": 2.1808943237436993,
+      "learning_rate": 9.891531081657681e-06,
+      "loss": 0.9556,
+      "step": 4953
+    },
+    {
+      "epoch": 0.5184720041862899,
+      "grad_norm": 2.029241896808696,
+      "learning_rate": 9.888141567198938e-06,
+      "loss": 1.0964,
+      "step": 4954
+    },
+    {
+      "epoch": 0.5185766614338043,
+      "grad_norm": 1.9949608216514931,
+      "learning_rate": 9.88475206559296e-06,
+      "loss": 0.9498,
+      "step": 4955
+    },
+    {
+      "epoch": 0.5186813186813187,
+      "grad_norm": 2.2830478335421174,
+      "learning_rate": 9.881362577229199e-06,
+      "loss": 1.0157,
+      "step": 4956
+    },
+    {
+      "epoch": 0.518785975928833,
+      "grad_norm": 1.9690660007947534,
+      "learning_rate": 9.87797310249712e-06,
+      "loss": 1.0411,
+      "step": 4957
+    },
+    {
+      "epoch": 0.5188906331763474,
+      "grad_norm": 1.977273523937895,
+      "learning_rate": 9.874583641786178e-06,
+      "loss": 0.8597,
+      "step": 4958
+    },
+    {
+      "epoch": 0.5189952904238618,
+      "grad_norm": 1.9243362489372648,
+      "learning_rate": 9.871194195485833e-06,
+      "loss": 0.9745,
+      "step": 4959
+    },
+    {
+      "epoch": 0.5190999476713762,
+      "grad_norm": 2.1432993449566737,
+      "learning_rate": 9.867804763985534e-06,
+      "loss": 1.0604,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5192046049188906,
+      "grad_norm": 2.0876636574915595,
+      "learning_rate": 9.864415347674732e-06,
+      "loss": 0.9654,
+      "step": 4961
+    },
+    {
+      "epoch": 0.519309262166405,
+      "grad_norm": 1.9248359573202058,
+      "learning_rate": 9.861025946942883e-06,
+      "loss": 0.8849,
+      "step": 4962
+    },
+    {
+      "epoch": 0.5194139194139195,
+      "grad_norm": 2.151905236783193,
+      "learning_rate": 9.857636562179439e-06,
+      "loss": 0.917,
+      "step": 4963
+    },
+    {
+      "epoch": 0.5195185766614338,
+      "grad_norm": 2.40363372620985,
+      "learning_rate": 9.854247193773837e-06,
+      "loss": 0.8764,
+      "step": 4964
+    },
+    {
+      "epoch": 0.5196232339089482,
+      "grad_norm": 2.2427852684647425,
+      "learning_rate": 9.850857842115533e-06,
+      "loss": 1.0175,
+      "step": 4965
+    },
+    {
+      "epoch": 0.5197278911564626,
+      "grad_norm": 1.9138156790964573,
+      "learning_rate": 9.847468507593961e-06,
+      "loss": 0.8898,
+      "step": 4966
+    },
+    {
+      "epoch": 0.519832548403977,
+      "grad_norm": 2.0160695177069603,
+      "learning_rate": 9.844079190598568e-06,
+      "loss": 0.9839,
+      "step": 4967
+    },
+    {
+      "epoch": 0.5199372056514914,
+      "grad_norm": 1.8516868195463738,
+      "learning_rate": 9.840689891518793e-06,
+      "loss": 1.0036,
+      "step": 4968
+    },
+    {
+      "epoch": 0.5200418628990058,
+      "grad_norm": 2.1447596627697254,
+      "learning_rate": 9.837300610744069e-06,
+      "loss": 0.8792,
+      "step": 4969
+    },
+    {
+      "epoch": 0.5201465201465202,
+      "grad_norm": 2.4054927998015447,
+      "learning_rate": 9.833911348663838e-06,
+      "loss": 1.0158,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5202511773940346,
+      "grad_norm": 2.228363478525115,
+      "learning_rate": 9.83052210566753e-06,
+      "loss": 0.9684,
+      "step": 4971
+    },
+    {
+      "epoch": 0.5203558346415489,
+      "grad_norm": 1.8149258250247282,
+      "learning_rate": 9.827132882144576e-06,
+      "loss": 0.9464,
+      "step": 4972
+    },
+    {
+      "epoch": 0.5204604918890633,
+      "grad_norm": 2.380722994225285,
+      "learning_rate": 9.8237436784844e-06,
+      "loss": 1.027,
+      "step": 4973
+    },
+    {
+      "epoch": 0.5205651491365777,
+      "grad_norm": 1.7873907854560183,
+      "learning_rate": 9.820354495076439e-06,
+      "loss": 0.769,
+      "step": 4974
+    },
+    {
+      "epoch": 0.5206698063840921,
+      "grad_norm": 2.2802115646024133,
+      "learning_rate": 9.816965332310106e-06,
+      "loss": 0.9555,
+      "step": 4975
+    },
+    {
+      "epoch": 0.5207744636316065,
+      "grad_norm": 2.038426429660569,
+      "learning_rate": 9.81357619057483e-06,
+      "loss": 0.9761,
+      "step": 4976
+    },
+    {
+      "epoch": 0.5208791208791209,
+      "grad_norm": 2.264959189290348,
+      "learning_rate": 9.810187070260029e-06,
+      "loss": 0.8613,
+      "step": 4977
+    },
+    {
+      "epoch": 0.5209837781266353,
+      "grad_norm": 2.3352359731890537,
+      "learning_rate": 9.80679797175511e-06,
+      "loss": 0.9207,
+      "step": 4978
+    },
+    {
+      "epoch": 0.5210884353741496,
+      "grad_norm": 1.813637189730288,
+      "learning_rate": 9.803408895449502e-06,
+      "loss": 0.8949,
+      "step": 4979
+    },
+    {
+      "epoch": 0.521193092621664,
+      "grad_norm": 2.0367522181120985,
+      "learning_rate": 9.800019841732613e-06,
+      "loss": 0.9757,
+      "step": 4980
+    },
+    {
+      "epoch": 0.5212977498691784,
+      "grad_norm": 2.098582096085878,
+      "learning_rate": 9.796630810993844e-06,
+      "loss": 1.0026,
+      "step": 4981
+    },
+    {
+      "epoch": 0.5214024071166928,
+      "grad_norm": 1.9944200217385741,
+      "learning_rate": 9.793241803622612e-06,
+      "loss": 0.8153,
+      "step": 4982
+    },
+    {
+      "epoch": 0.5215070643642072,
+      "grad_norm": 2.0471451034556147,
+      "learning_rate": 9.789852820008312e-06,
+      "loss": 0.782,
+      "step": 4983
+    },
+    {
+      "epoch": 0.5216117216117216,
+      "grad_norm": 2.199390774507097,
+      "learning_rate": 9.786463860540352e-06,
+      "loss": 0.8614,
+      "step": 4984
+    },
+    {
+      "epoch": 0.521716378859236,
+      "grad_norm": 2.071453495859607,
+      "learning_rate": 9.783074925608126e-06,
+      "loss": 0.8793,
+      "step": 4985
+    },
+    {
+      "epoch": 0.5218210361067503,
+      "grad_norm": 2.3490583249057893,
+      "learning_rate": 9.779686015601031e-06,
+      "loss": 1.0274,
+      "step": 4986
+    },
+    {
+      "epoch": 0.5219256933542648,
+      "grad_norm": 1.959370734056799,
+      "learning_rate": 9.776297130908456e-06,
+      "loss": 0.9316,
+      "step": 4987
+    },
+    {
+      "epoch": 0.5220303506017792,
+      "grad_norm": 2.2160442511485825,
+      "learning_rate": 9.772908271919797e-06,
+      "loss": 1.0251,
+      "step": 4988
+    },
+    {
+      "epoch": 0.5221350078492936,
+      "grad_norm": 1.968606658582411,
+      "learning_rate": 9.76951943902444e-06,
+      "loss": 0.944,
+      "step": 4989
+    },
+    {
+      "epoch": 0.522239665096808,
+      "grad_norm": 2.4107803674809736,
+      "learning_rate": 9.766130632611762e-06,
+      "loss": 0.9429,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5223443223443224,
+      "grad_norm": 2.0652044268509693,
+      "learning_rate": 9.762741853071153e-06,
+      "loss": 0.8697,
+      "step": 4991
+    },
+    {
+      "epoch": 0.5224489795918368,
+      "grad_norm": 2.062990165066534,
+      "learning_rate": 9.759353100791986e-06,
+      "loss": 1.0389,
+      "step": 4992
+    },
+    {
+      "epoch": 0.5225536368393511,
+      "grad_norm": 1.9364323150880431,
+      "learning_rate": 9.755964376163633e-06,
+      "loss": 0.9711,
+      "step": 4993
+    },
+    {
+      "epoch": 0.5226582940868655,
+      "grad_norm": 2.4876581630290273,
+      "learning_rate": 9.752575679575469e-06,
+      "loss": 0.8624,
+      "step": 4994
+    },
+    {
+      "epoch": 0.5227629513343799,
+      "grad_norm": 1.8708610631837064,
+      "learning_rate": 9.749187011416858e-06,
+      "loss": 0.8952,
+      "step": 4995
+    },
+    {
+      "epoch": 0.5228676085818943,
+      "grad_norm": 2.0610166226921467,
+      "learning_rate": 9.745798372077173e-06,
+      "loss": 1.0246,
+      "step": 4996
+    },
+    {
+      "epoch": 0.5229722658294087,
+      "grad_norm": 2.091437740906605,
+      "learning_rate": 9.742409761945765e-06,
+      "loss": 0.9257,
+      "step": 4997
+    },
+    {
+      "epoch": 0.5230769230769231,
+      "grad_norm": 3.1322596751647236,
+      "learning_rate": 9.739021181412003e-06,
+      "loss": 0.9185,
+      "step": 4998
+    },
+    {
+      "epoch": 0.5231815803244375,
+      "grad_norm": 2.983387965543901,
+      "learning_rate": 9.735632630865234e-06,
+      "loss": 0.8107,
+      "step": 4999
+    },
+    {
+      "epoch": 0.5232862375719518,
+      "grad_norm": 2.3964467020952425,
+      "learning_rate": 9.732244110694813e-06,
+      "loss": 0.9936,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5233908948194662,
+      "grad_norm": 2.1655991778042893,
+      "learning_rate": 9.728855621290086e-06,
+      "loss": 0.881,
+      "step": 5001
+    },
+    {
+      "epoch": 0.5234955520669806,
+      "grad_norm": 2.323301195527726,
+      "learning_rate": 9.725467163040395e-06,
+      "loss": 1.0034,
+      "step": 5002
+    },
+    {
+      "epoch": 0.523600209314495,
+      "grad_norm": 2.1683566774088194,
+      "learning_rate": 9.722078736335087e-06,
+      "loss": 0.9435,
+      "step": 5003
+    },
+    {
+      "epoch": 0.5237048665620094,
+      "grad_norm": 1.9021798201348261,
+      "learning_rate": 9.71869034156349e-06,
+      "loss": 0.9163,
+      "step": 5004
+    },
+    {
+      "epoch": 0.5238095238095238,
+      "grad_norm": 2.1575573537521278,
+      "learning_rate": 9.715301979114946e-06,
+      "loss": 0.9994,
+      "step": 5005
+    },
+    {
+      "epoch": 0.5239141810570382,
+      "grad_norm": 2.1460735218265174,
+      "learning_rate": 9.711913649378785e-06,
+      "loss": 0.9674,
+      "step": 5006
+    },
+    {
+      "epoch": 0.5240188383045525,
+      "grad_norm": 2.0470323099465486,
+      "learning_rate": 9.708525352744325e-06,
+      "loss": 0.9025,
+      "step": 5007
+    },
+    {
+      "epoch": 0.5241234955520669,
+      "grad_norm": 2.0043888800821907,
+      "learning_rate": 9.705137089600898e-06,
+      "loss": 0.943,
+      "step": 5008
+    },
+    {
+      "epoch": 0.5242281527995813,
+      "grad_norm": 2.011463626160743,
+      "learning_rate": 9.701748860337812e-06,
+      "loss": 0.935,
+      "step": 5009
+    },
+    {
+      "epoch": 0.5243328100470958,
+      "grad_norm": 2.050505275508048,
+      "learning_rate": 9.698360665344391e-06,
+      "loss": 0.9326,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5244374672946102,
+      "grad_norm": 2.144129516792977,
+      "learning_rate": 9.694972505009938e-06,
+      "loss": 1.0178,
+      "step": 5011
+    },
+    {
+      "epoch": 0.5245421245421246,
+      "grad_norm": 1.9003656125321036,
+      "learning_rate": 9.691584379723766e-06,
+      "loss": 0.9608,
+      "step": 5012
+    },
+    {
+      "epoch": 0.524646781789639,
+      "grad_norm": 2.2612768774796654,
+      "learning_rate": 9.688196289875168e-06,
+      "loss": 0.9817,
+      "step": 5013
+    },
+    {
+      "epoch": 0.5247514390371534,
+      "grad_norm": 2.290640926354991,
+      "learning_rate": 9.68480823585345e-06,
+      "loss": 1.1034,
+      "step": 5014
+    },
+    {
+      "epoch": 0.5248560962846677,
+      "grad_norm": 2.4607191178201475,
+      "learning_rate": 9.681420218047909e-06,
+      "loss": 0.9906,
+      "step": 5015
+    },
+    {
+      "epoch": 0.5249607535321821,
+      "grad_norm": 1.9026816954278603,
+      "learning_rate": 9.678032236847827e-06,
+      "loss": 0.8915,
+      "step": 5016
+    },
+    {
+      "epoch": 0.5250654107796965,
+      "grad_norm": 1.9421672245762756,
+      "learning_rate": 9.674644292642496e-06,
+      "loss": 0.9105,
+      "step": 5017
+    },
+    {
+      "epoch": 0.5251700680272109,
+      "grad_norm": 2.4702863347349084,
+      "learning_rate": 9.671256385821196e-06,
+      "loss": 1.0937,
+      "step": 5018
+    },
+    {
+      "epoch": 0.5252747252747253,
+      "grad_norm": 1.876747432470777,
+      "learning_rate": 9.6678685167732e-06,
+      "loss": 0.8908,
+      "step": 5019
+    },
+    {
+      "epoch": 0.5253793825222397,
+      "grad_norm": 2.9876022744753934,
+      "learning_rate": 9.66448068588779e-06,
+      "loss": 0.9807,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5254840397697541,
+      "grad_norm": 1.9265704361668035,
+      "learning_rate": 9.661092893554222e-06,
+      "loss": 0.8965,
+      "step": 5021
+    },
+    {
+      "epoch": 0.5255886970172684,
+      "grad_norm": 2.092597658236264,
+      "learning_rate": 9.657705140161772e-06,
+      "loss": 0.9318,
+      "step": 5022
+    },
+    {
+      "epoch": 0.5256933542647828,
+      "grad_norm": 1.9546071290696667,
+      "learning_rate": 9.654317426099695e-06,
+      "loss": 0.9737,
+      "step": 5023
+    },
+    {
+      "epoch": 0.5257980115122972,
+      "grad_norm": 1.9365552977725047,
+      "learning_rate": 9.650929751757247e-06,
+      "loss": 1.0049,
+      "step": 5024
+    },
+    {
+      "epoch": 0.5259026687598116,
+      "grad_norm": 1.967006406063824,
+      "learning_rate": 9.647542117523679e-06,
+      "loss": 0.9506,
+      "step": 5025
+    },
+    {
+      "epoch": 0.526007326007326,
+      "grad_norm": 2.3155282840857847,
+      "learning_rate": 9.644154523788233e-06,
+      "loss": 0.7686,
+      "step": 5026
+    },
+    {
+      "epoch": 0.5261119832548404,
+      "grad_norm": 2.157265802659998,
+      "learning_rate": 9.640766970940157e-06,
+      "loss": 0.9982,
+      "step": 5027
+    },
+    {
+      "epoch": 0.5262166405023548,
+      "grad_norm": 2.1389647147739823,
+      "learning_rate": 9.637379459368682e-06,
+      "loss": 0.8971,
+      "step": 5028
+    },
+    {
+      "epoch": 0.5263212977498691,
+      "grad_norm": 2.0363015164707607,
+      "learning_rate": 9.633991989463043e-06,
+      "loss": 1.0196,
+      "step": 5029
+    },
+    {
+      "epoch": 0.5264259549973835,
+      "grad_norm": 1.9294652609584093,
+      "learning_rate": 9.630604561612465e-06,
+      "loss": 0.7722,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5265306122448979,
+      "grad_norm": 2.342838953335117,
+      "learning_rate": 9.627217176206172e-06,
+      "loss": 0.9184,
+      "step": 5031
+    },
+    {
+      "epoch": 0.5266352694924124,
+      "grad_norm": 1.9756381699879217,
+      "learning_rate": 9.623829833633384e-06,
+      "loss": 1.0176,
+      "step": 5032
+    },
+    {
+      "epoch": 0.5267399267399268,
+      "grad_norm": 2.03546759419274,
+      "learning_rate": 9.620442534283308e-06,
+      "loss": 0.9261,
+      "step": 5033
+    },
+    {
+      "epoch": 0.5268445839874412,
+      "grad_norm": 2.288452566068461,
+      "learning_rate": 9.617055278545155e-06,
+      "loss": 0.9991,
+      "step": 5034
+    },
+    {
+      "epoch": 0.5269492412349556,
+      "grad_norm": 2.057195359204059,
+      "learning_rate": 9.61366806680813e-06,
+      "loss": 0.9043,
+      "step": 5035
+    },
+    {
+      "epoch": 0.5270538984824699,
+      "grad_norm": 2.0034791907235134,
+      "learning_rate": 9.610280899461422e-06,
+      "loss": 0.8265,
+      "step": 5036
+    },
+    {
+      "epoch": 0.5271585557299843,
+      "grad_norm": 2.300980115407969,
+      "learning_rate": 9.606893776894232e-06,
+      "loss": 0.972,
+      "step": 5037
+    },
+    {
+      "epoch": 0.5272632129774987,
+      "grad_norm": 2.1374768924728715,
+      "learning_rate": 9.603506699495742e-06,
+      "loss": 0.9861,
+      "step": 5038
+    },
+    {
+      "epoch": 0.5273678702250131,
+      "grad_norm": 1.901539200611645,
+      "learning_rate": 9.600119667655135e-06,
+      "loss": 0.9153,
+      "step": 5039
+    },
+    {
+      "epoch": 0.5274725274725275,
+      "grad_norm": 1.898313837762518,
+      "learning_rate": 9.596732681761591e-06,
+      "loss": 0.9927,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5275771847200419,
+      "grad_norm": 2.203116494419673,
+      "learning_rate": 9.593345742204278e-06,
+      "loss": 0.9173,
+      "step": 5041
+    },
+    {
+      "epoch": 0.5276818419675563,
+      "grad_norm": 1.6960490365340934,
+      "learning_rate": 9.589958849372367e-06,
+      "loss": 0.9086,
+      "step": 5042
+    },
+    {
+      "epoch": 0.5277864992150706,
+      "grad_norm": 2.2070200124057133,
+      "learning_rate": 9.586572003655012e-06,
+      "loss": 0.8878,
+      "step": 5043
+    },
+    {
+      "epoch": 0.527891156462585,
+      "grad_norm": 1.954022666257893,
+      "learning_rate": 9.583185205441376e-06,
+      "loss": 0.9223,
+      "step": 5044
+    },
+    {
+      "epoch": 0.5279958137100994,
+      "grad_norm": 2.0745859713137866,
+      "learning_rate": 9.5797984551206e-06,
+      "loss": 1.0438,
+      "step": 5045
+    },
+    {
+      "epoch": 0.5281004709576138,
+      "grad_norm": 2.265741127424256,
+      "learning_rate": 9.576411753081837e-06,
+      "loss": 0.9013,
+      "step": 5046
+    },
+    {
+      "epoch": 0.5282051282051282,
+      "grad_norm": 1.8371450812640804,
+      "learning_rate": 9.573025099714218e-06,
+      "loss": 1.0232,
+      "step": 5047
+    },
+    {
+      "epoch": 0.5283097854526426,
+      "grad_norm": 2.033374031917156,
+      "learning_rate": 9.569638495406882e-06,
+      "loss": 0.8909,
+      "step": 5048
+    },
+    {
+      "epoch": 0.528414442700157,
+      "grad_norm": 2.1076096190375697,
+      "learning_rate": 9.566251940548957e-06,
+      "loss": 0.9968,
+      "step": 5049
+    },
+    {
+      "epoch": 0.5285190999476713,
+      "grad_norm": 1.941537663106642,
+      "learning_rate": 9.562865435529563e-06,
+      "loss": 0.884,
+      "step": 5050
+    },
+    {
+      "epoch": 0.5286237571951857,
+      "grad_norm": 1.993306824385622,
+      "learning_rate": 9.559478980737817e-06,
+      "loss": 0.9804,
+      "step": 5051
+    },
+    {
+      "epoch": 0.5287284144427001,
+      "grad_norm": 1.8310381477052629,
+      "learning_rate": 9.556092576562828e-06,
+      "loss": 0.8713,
+      "step": 5052
+    },
+    {
+      "epoch": 0.5288330716902145,
+      "grad_norm": 2.1706814859478643,
+      "learning_rate": 9.552706223393704e-06,
+      "loss": 0.953,
+      "step": 5053
+    },
+    {
+      "epoch": 0.528937728937729,
+      "grad_norm": 2.0181734603318917,
+      "learning_rate": 9.54931992161954e-06,
+      "loss": 1.0406,
+      "step": 5054
+    },
+    {
+      "epoch": 0.5290423861852434,
+      "grad_norm": 2.1729682113864537,
+      "learning_rate": 9.54593367162943e-06,
+      "loss": 0.9522,
+      "step": 5055
+    },
+    {
+      "epoch": 0.5291470434327578,
+      "grad_norm": 1.9793023318843381,
+      "learning_rate": 9.54254747381246e-06,
+      "loss": 0.8517,
+      "step": 5056
+    },
+    {
+      "epoch": 0.5292517006802722,
+      "grad_norm": 1.956291432464463,
+      "learning_rate": 9.539161328557716e-06,
+      "loss": 0.9517,
+      "step": 5057
+    },
+    {
+      "epoch": 0.5293563579277865,
+      "grad_norm": 2.033082190907283,
+      "learning_rate": 9.535775236254268e-06,
+      "loss": 0.8648,
+      "step": 5058
+    },
+    {
+      "epoch": 0.5294610151753009,
+      "grad_norm": 2.13270173634752,
+      "learning_rate": 9.532389197291187e-06,
+      "loss": 0.9357,
+      "step": 5059
+    },
+    {
+      "epoch": 0.5295656724228153,
+      "grad_norm": 1.91157758973134,
+      "learning_rate": 9.529003212057536e-06,
+      "loss": 0.9013,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5296703296703297,
+      "grad_norm": 1.832312512024985,
+      "learning_rate": 9.52561728094237e-06,
+      "loss": 0.9018,
+      "step": 5061
+    },
+    {
+      "epoch": 0.5297749869178441,
+      "grad_norm": 2.316257080957267,
+      "learning_rate": 9.52223140433474e-06,
+      "loss": 0.7679,
+      "step": 5062
+    },
+    {
+      "epoch": 0.5298796441653585,
+      "grad_norm": 1.8496246062520718,
+      "learning_rate": 9.518845582623692e-06,
+      "loss": 0.9617,
+      "step": 5063
+    },
+    {
+      "epoch": 0.5299843014128729,
+      "grad_norm": 2.0807591771920197,
+      "learning_rate": 9.515459816198258e-06,
+      "loss": 0.9035,
+      "step": 5064
+    },
+    {
+      "epoch": 0.5300889586603872,
+      "grad_norm": 1.8998070623413483,
+      "learning_rate": 9.512074105447473e-06,
+      "loss": 0.9626,
+      "step": 5065
+    },
+    {
+      "epoch": 0.5301936159079016,
+      "grad_norm": 1.9812096820937861,
+      "learning_rate": 9.508688450760367e-06,
+      "loss": 0.9601,
+      "step": 5066
+    },
+    {
+      "epoch": 0.530298273155416,
+      "grad_norm": 2.147757889449252,
+      "learning_rate": 9.50530285252595e-06,
+      "loss": 1.0095,
+      "step": 5067
+    },
+    {
+      "epoch": 0.5304029304029304,
+      "grad_norm": 1.8908877406835036,
+      "learning_rate": 9.501917311133241e-06,
+      "loss": 0.9554,
+      "step": 5068
+    },
+    {
+      "epoch": 0.5305075876504448,
+      "grad_norm": 2.006086304429676,
+      "learning_rate": 9.498531826971242e-06,
+      "loss": 0.9882,
+      "step": 5069
+    },
+    {
+      "epoch": 0.5306122448979592,
+      "grad_norm": 1.7842507152981133,
+      "learning_rate": 9.495146400428953e-06,
+      "loss": 0.9366,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5307169021454736,
+      "grad_norm": 2.046078498853944,
+      "learning_rate": 9.491761031895364e-06,
+      "loss": 0.876,
+      "step": 5071
+    },
+    {
+      "epoch": 0.5308215593929879,
+      "grad_norm": 1.9166303496266188,
+      "learning_rate": 9.488375721759467e-06,
+      "loss": 0.8727,
+      "step": 5072
+    },
+    {
+      "epoch": 0.5309262166405023,
+      "grad_norm": 1.8738691308012003,
+      "learning_rate": 9.48499047041023e-06,
+      "loss": 0.9591,
+      "step": 5073
+    },
+    {
+      "epoch": 0.5310308738880167,
+      "grad_norm": 2.107065364309017,
+      "learning_rate": 9.481605278236636e-06,
+      "loss": 0.8822,
+      "step": 5074
+    },
+    {
+      "epoch": 0.5311355311355311,
+      "grad_norm": 2.327347288791375,
+      "learning_rate": 9.478220145627645e-06,
+      "loss": 1.0899,
+      "step": 5075
+    },
+    {
+      "epoch": 0.5312401883830455,
+      "grad_norm": 2.2223497049637713,
+      "learning_rate": 9.47483507297222e-06,
+      "loss": 0.8937,
+      "step": 5076
+    },
+    {
+      "epoch": 0.53134484563056,
+      "grad_norm": 2.256580775897136,
+      "learning_rate": 9.471450060659308e-06,
+      "loss": 1.0627,
+      "step": 5077
+    },
+    {
+      "epoch": 0.5314495028780744,
+      "grad_norm": 2.1603161263939628,
+      "learning_rate": 9.468065109077858e-06,
+      "loss": 0.8642,
+      "step": 5078
+    },
+    {
+      "epoch": 0.5315541601255886,
+      "grad_norm": 2.021675968414649,
+      "learning_rate": 9.464680218616801e-06,
+      "loss": 0.9218,
+      "step": 5079
+    },
+    {
+      "epoch": 0.531658817373103,
+      "grad_norm": 2.027162622909609,
+      "learning_rate": 9.461295389665075e-06,
+      "loss": 0.8784,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5317634746206175,
+      "grad_norm": 2.1737803034324896,
+      "learning_rate": 9.457910622611599e-06,
+      "loss": 0.9499,
+      "step": 5081
+    },
+    {
+      "epoch": 0.5318681318681319,
+      "grad_norm": 2.1361694404225573,
+      "learning_rate": 9.454525917845289e-06,
+      "loss": 0.9557,
+      "step": 5082
+    },
+    {
+      "epoch": 0.5319727891156463,
+      "grad_norm": 1.9555190602414432,
+      "learning_rate": 9.451141275755061e-06,
+      "loss": 1.0161,
+      "step": 5083
+    },
+    {
+      "epoch": 0.5320774463631607,
+      "grad_norm": 2.0343065372865823,
+      "learning_rate": 9.447756696729811e-06,
+      "loss": 0.9702,
+      "step": 5084
+    },
+    {
+      "epoch": 0.5321821036106751,
+      "grad_norm": 1.990208162802974,
+      "learning_rate": 9.444372181158439e-06,
+      "loss": 0.881,
+      "step": 5085
+    },
+    {
+      "epoch": 0.5322867608581894,
+      "grad_norm": 1.9953926441133305,
+      "learning_rate": 9.440987729429827e-06,
+      "loss": 0.8863,
+      "step": 5086
+    },
+    {
+      "epoch": 0.5323914181057038,
+      "grad_norm": 2.3741453667707257,
+      "learning_rate": 9.437603341932861e-06,
+      "loss": 0.8405,
+      "step": 5087
+    },
+    {
+      "epoch": 0.5324960753532182,
+      "grad_norm": 2.0080306933110252,
+      "learning_rate": 9.434219019056407e-06,
+      "loss": 0.8811,
+      "step": 5088
+    },
+    {
+      "epoch": 0.5326007326007326,
+      "grad_norm": 1.9052829376729115,
+      "learning_rate": 9.430834761189339e-06,
+      "loss": 0.9062,
+      "step": 5089
+    },
+    {
+      "epoch": 0.532705389848247,
+      "grad_norm": 2.232762167686221,
+      "learning_rate": 9.427450568720507e-06,
+      "loss": 1.066,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5328100470957614,
+      "grad_norm": 1.7077102794519774,
+      "learning_rate": 9.424066442038764e-06,
+      "loss": 0.998,
+      "step": 5091
+    },
+    {
+      "epoch": 0.5329147043432758,
+      "grad_norm": 2.0986156305726316,
+      "learning_rate": 9.420682381532959e-06,
+      "loss": 1.0063,
+      "step": 5092
+    },
+    {
+      "epoch": 0.5330193615907901,
+      "grad_norm": 1.985756966883505,
+      "learning_rate": 9.417298387591923e-06,
+      "loss": 0.9973,
+      "step": 5093
+    },
+    {
+      "epoch": 0.5331240188383045,
+      "grad_norm": 2.3041795560347555,
+      "learning_rate": 9.41391446060448e-06,
+      "loss": 1.0417,
+      "step": 5094
+    },
+    {
+      "epoch": 0.5332286760858189,
+      "grad_norm": 2.3998949708050135,
+      "learning_rate": 9.410530600959455e-06,
+      "loss": 1.0162,
+      "step": 5095
+    },
+    {
+      "epoch": 0.5333333333333333,
+      "grad_norm": 1.9102226487477791,
+      "learning_rate": 9.407146809045658e-06,
+      "loss": 0.8784,
+      "step": 5096
+    },
+    {
+      "epoch": 0.5334379905808477,
+      "grad_norm": 1.9875471868758454,
+      "learning_rate": 9.403763085251898e-06,
+      "loss": 0.8156,
+      "step": 5097
+    },
+    {
+      "epoch": 0.5335426478283621,
+      "grad_norm": 1.953172303607464,
+      "learning_rate": 9.400379429966965e-06,
+      "loss": 0.8733,
+      "step": 5098
+    },
+    {
+      "epoch": 0.5336473050758765,
+      "grad_norm": 2.2056345622053537,
+      "learning_rate": 9.396995843579648e-06,
+      "loss": 0.9868,
+      "step": 5099
+    },
+    {
+      "epoch": 0.533751962323391,
+      "grad_norm": 2.053690818273544,
+      "learning_rate": 9.393612326478736e-06,
+      "loss": 0.8402,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5338566195709052,
+      "grad_norm": 2.3399486682168775,
+      "learning_rate": 9.390228879052995e-06,
+      "loss": 1.043,
+      "step": 5101
+    },
+    {
+      "epoch": 0.5339612768184197,
+      "grad_norm": 2.075427595574749,
+      "learning_rate": 9.386845501691194e-06,
+      "loss": 0.9583,
+      "step": 5102
+    },
+    {
+      "epoch": 0.5340659340659341,
+      "grad_norm": 2.21323312612293,
+      "learning_rate": 9.383462194782086e-06,
+      "loss": 0.8203,
+      "step": 5103
+    },
+    {
+      "epoch": 0.5341705913134485,
+      "grad_norm": 1.7638697451900374,
+      "learning_rate": 9.380078958714424e-06,
+      "loss": 0.8476,
+      "step": 5104
+    },
+    {
+      "epoch": 0.5342752485609629,
+      "grad_norm": 2.1757947162837636,
+      "learning_rate": 9.376695793876943e-06,
+      "loss": 1.0289,
+      "step": 5105
+    },
+    {
+      "epoch": 0.5343799058084773,
+      "grad_norm": 2.0498873657140675,
+      "learning_rate": 9.373312700658381e-06,
+      "loss": 0.9722,
+      "step": 5106
+    },
+    {
+      "epoch": 0.5344845630559917,
+      "grad_norm": 2.146781521567094,
+      "learning_rate": 9.36992967944746e-06,
+      "loss": 0.9346,
+      "step": 5107
+    },
+    {
+      "epoch": 0.534589220303506,
+      "grad_norm": 2.4924758465124643,
+      "learning_rate": 9.366546730632892e-06,
+      "loss": 0.9294,
+      "step": 5108
+    },
+    {
+      "epoch": 0.5346938775510204,
+      "grad_norm": 2.1871982606239224,
+      "learning_rate": 9.363163854603393e-06,
+      "loss": 0.9268,
+      "step": 5109
+    },
+    {
+      "epoch": 0.5347985347985348,
+      "grad_norm": 2.6130844536579145,
+      "learning_rate": 9.359781051747658e-06,
+      "loss": 0.8488,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5349031920460492,
+      "grad_norm": 2.3176010643788856,
+      "learning_rate": 9.356398322454378e-06,
+      "loss": 1.0246,
+      "step": 5111
+    },
+    {
+      "epoch": 0.5350078492935636,
+      "grad_norm": 1.9976965271715834,
+      "learning_rate": 9.353015667112236e-06,
+      "loss": 0.9324,
+      "step": 5112
+    },
+    {
+      "epoch": 0.535112506541078,
+      "grad_norm": 2.1264457121845073,
+      "learning_rate": 9.349633086109904e-06,
+      "loss": 0.957,
+      "step": 5113
+    },
+    {
+      "epoch": 0.5352171637885924,
+      "grad_norm": 2.1082662476439897,
+      "learning_rate": 9.34625057983605e-06,
+      "loss": 0.8603,
+      "step": 5114
+    },
+    {
+      "epoch": 0.5353218210361067,
+      "grad_norm": 2.081793729672057,
+      "learning_rate": 9.342868148679328e-06,
+      "loss": 0.9384,
+      "step": 5115
+    },
+    {
+      "epoch": 0.5354264782836211,
+      "grad_norm": 1.8602117077983622,
+      "learning_rate": 9.33948579302839e-06,
+      "loss": 0.9818,
+      "step": 5116
+    },
+    {
+      "epoch": 0.5355311355311355,
+      "grad_norm": 2.068748019729851,
+      "learning_rate": 9.33610351327187e-06,
+      "loss": 0.8313,
+      "step": 5117
+    },
+    {
+      "epoch": 0.5356357927786499,
+      "grad_norm": 2.250399416328852,
+      "learning_rate": 9.332721309798403e-06,
+      "loss": 0.9082,
+      "step": 5118
+    },
+    {
+      "epoch": 0.5357404500261643,
+      "grad_norm": 2.0089516460747068,
+      "learning_rate": 9.329339182996614e-06,
+      "loss": 1.0031,
+      "step": 5119
+    },
+    {
+      "epoch": 0.5358451072736787,
+      "grad_norm": 2.1921142529491062,
+      "learning_rate": 9.325957133255107e-06,
+      "loss": 0.9773,
+      "step": 5120
+    },
+    {
+      "epoch": 0.5359497645211931,
+      "grad_norm": 1.7765848948209197,
+      "learning_rate": 9.322575160962497e-06,
+      "loss": 0.8713,
+      "step": 5121
+    },
+    {
+      "epoch": 0.5360544217687074,
+      "grad_norm": 2.2822852338608133,
+      "learning_rate": 9.319193266507371e-06,
+      "loss": 1.0108,
+      "step": 5122
+    },
+    {
+      "epoch": 0.5361590790162218,
+      "grad_norm": 1.9080359520771442,
+      "learning_rate": 9.315811450278322e-06,
+      "loss": 1.0417,
+      "step": 5123
+    },
+    {
+      "epoch": 0.5362637362637362,
+      "grad_norm": 2.126848470721643,
+      "learning_rate": 9.312429712663922e-06,
+      "loss": 1.0819,
+      "step": 5124
+    },
+    {
+      "epoch": 0.5363683935112507,
+      "grad_norm": 1.9448417891971947,
+      "learning_rate": 9.309048054052738e-06,
+      "loss": 1.0022,
+      "step": 5125
+    },
+    {
+      "epoch": 0.5364730507587651,
+      "grad_norm": 2.0110601683746716,
+      "learning_rate": 9.30566647483334e-06,
+      "loss": 0.9406,
+      "step": 5126
+    },
+    {
+      "epoch": 0.5365777080062795,
+      "grad_norm": 2.059272200316169,
+      "learning_rate": 9.302284975394267e-06,
+      "loss": 0.9979,
+      "step": 5127
+    },
+    {
+      "epoch": 0.5366823652537939,
+      "grad_norm": 1.9516944950557442,
+      "learning_rate": 9.298903556124069e-06,
+      "loss": 0.964,
+      "step": 5128
+    },
+    {
+      "epoch": 0.5367870225013082,
+      "grad_norm": 1.9463999538178405,
+      "learning_rate": 9.29552221741127e-06,
+      "loss": 0.9758,
+      "step": 5129
+    },
+    {
+      "epoch": 0.5368916797488226,
+      "grad_norm": 1.98796486302683,
+      "learning_rate": 9.292140959644398e-06,
+      "loss": 0.9545,
+      "step": 5130
+    },
+    {
+      "epoch": 0.536996336996337,
+      "grad_norm": 1.9929189015148039,
+      "learning_rate": 9.288759783211967e-06,
+      "loss": 0.9298,
+      "step": 5131
+    },
+    {
+      "epoch": 0.5371009942438514,
+      "grad_norm": 2.1502952424043196,
+      "learning_rate": 9.285378688502475e-06,
+      "loss": 0.9023,
+      "step": 5132
+    },
+    {
+      "epoch": 0.5372056514913658,
+      "grad_norm": 1.8849112778207193,
+      "learning_rate": 9.281997675904423e-06,
+      "loss": 0.9625,
+      "step": 5133
+    },
+    {
+      "epoch": 0.5373103087388802,
+      "grad_norm": 2.2034978552750926,
+      "learning_rate": 9.27861674580629e-06,
+      "loss": 0.8996,
+      "step": 5134
+    },
+    {
+      "epoch": 0.5374149659863946,
+      "grad_norm": 2.1228025728157514,
+      "learning_rate": 9.275235898596555e-06,
+      "loss": 0.8977,
+      "step": 5135
+    },
+    {
+      "epoch": 0.5375196232339089,
+      "grad_norm": 2.155478741058007,
+      "learning_rate": 9.271855134663689e-06,
+      "loss": 1.0623,
+      "step": 5136
+    },
+    {
+      "epoch": 0.5376242804814233,
+      "grad_norm": 2.017078129664075,
+      "learning_rate": 9.26847445439614e-06,
+      "loss": 0.9586,
+      "step": 5137
+    },
+    {
+      "epoch": 0.5377289377289377,
+      "grad_norm": 2.2424961468183295,
+      "learning_rate": 9.265093858182364e-06,
+      "loss": 1.0009,
+      "step": 5138
+    },
+    {
+      "epoch": 0.5378335949764521,
+      "grad_norm": 1.9391397613594792,
+      "learning_rate": 9.26171334641079e-06,
+      "loss": 0.831,
+      "step": 5139
+    },
+    {
+      "epoch": 0.5379382522239665,
+      "grad_norm": 2.041390251270642,
+      "learning_rate": 9.258332919469851e-06,
+      "loss": 0.9798,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5380429094714809,
+      "grad_norm": 2.086322474319101,
+      "learning_rate": 9.25495257774796e-06,
+      "loss": 1.0148,
+      "step": 5141
+    },
+    {
+      "epoch": 0.5381475667189953,
+      "grad_norm": 1.9006136542065337,
+      "learning_rate": 9.251572321633533e-06,
+      "loss": 0.9227,
+      "step": 5142
+    },
+    {
+      "epoch": 0.5382522239665097,
+      "grad_norm": 1.9414787885775822,
+      "learning_rate": 9.248192151514957e-06,
+      "loss": 0.8972,
+      "step": 5143
+    },
+    {
+      "epoch": 0.538356881214024,
+      "grad_norm": 1.8618340965139868,
+      "learning_rate": 9.24481206778063e-06,
+      "loss": 1.0364,
+      "step": 5144
+    },
+    {
+      "epoch": 0.5384615384615384,
+      "grad_norm": 1.7750231543939474,
+      "learning_rate": 9.241432070818929e-06,
+      "loss": 0.8416,
+      "step": 5145
+    },
+    {
+      "epoch": 0.5385661957090528,
+      "grad_norm": 1.7192970430261034,
+      "learning_rate": 9.23805216101822e-06,
+      "loss": 0.8291,
+      "step": 5146
+    },
+    {
+      "epoch": 0.5386708529565672,
+      "grad_norm": 2.267675160306062,
+      "learning_rate": 9.234672338766863e-06,
+      "loss": 0.8009,
+      "step": 5147
+    },
+    {
+      "epoch": 0.5387755102040817,
+      "grad_norm": 1.9151967169817872,
+      "learning_rate": 9.231292604453203e-06,
+      "loss": 0.983,
+      "step": 5148
+    },
+    {
+      "epoch": 0.5388801674515961,
+      "grad_norm": 1.8860340166769067,
+      "learning_rate": 9.227912958465584e-06,
+      "loss": 0.9827,
+      "step": 5149
+    },
+    {
+      "epoch": 0.5389848246991105,
+      "grad_norm": 2.377402452195343,
+      "learning_rate": 9.224533401192332e-06,
+      "loss": 0.7757,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5390894819466248,
+      "grad_norm": 2.1940029519811444,
+      "learning_rate": 9.221153933021759e-06,
+      "loss": 0.9636,
+      "step": 5151
+    },
+    {
+      "epoch": 0.5391941391941392,
+      "grad_norm": 2.0338541104057266,
+      "learning_rate": 9.21777455434218e-06,
+      "loss": 1.0547,
+      "step": 5152
+    },
+    {
+      "epoch": 0.5392987964416536,
+      "grad_norm": 2.9722612954759255,
+      "learning_rate": 9.214395265541894e-06,
+      "loss": 0.9978,
+      "step": 5153
+    },
+    {
+      "epoch": 0.539403453689168,
+      "grad_norm": 2.1825836270355654,
+      "learning_rate": 9.211016067009182e-06,
+      "loss": 0.8928,
+      "step": 5154
+    },
+    {
+      "epoch": 0.5395081109366824,
+      "grad_norm": 2.0138240313959543,
+      "learning_rate": 9.207636959132325e-06,
+      "loss": 0.8572,
+      "step": 5155
+    },
+    {
+      "epoch": 0.5396127681841968,
+      "grad_norm": 2.3175892149524966,
+      "learning_rate": 9.204257942299585e-06,
+      "loss": 1.0819,
+      "step": 5156
+    },
+    {
+      "epoch": 0.5397174254317112,
+      "grad_norm": 2.235711126597796,
+      "learning_rate": 9.200879016899223e-06,
+      "loss": 0.8762,
+      "step": 5157
+    },
+    {
+      "epoch": 0.5398220826792255,
+      "grad_norm": 2.450747499531445,
+      "learning_rate": 9.197500183319481e-06,
+      "loss": 1.0596,
+      "step": 5158
+    },
+    {
+      "epoch": 0.5399267399267399,
+      "grad_norm": 2.3432608161807513,
+      "learning_rate": 9.194121441948596e-06,
+      "loss": 0.9195,
+      "step": 5159
+    },
+    {
+      "epoch": 0.5400313971742543,
+      "grad_norm": 2.2491282476173775,
+      "learning_rate": 9.190742793174786e-06,
+      "loss": 0.8444,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5401360544217687,
+      "grad_norm": 2.3639647455866895,
+      "learning_rate": 9.187364237386273e-06,
+      "loss": 0.9589,
+      "step": 5161
+    },
+    {
+      "epoch": 0.5402407116692831,
+      "grad_norm": 2.359812971612293,
+      "learning_rate": 9.183985774971258e-06,
+      "loss": 0.8739,
+      "step": 5162
+    },
+    {
+      "epoch": 0.5403453689167975,
+      "grad_norm": 1.7777656507828674,
+      "learning_rate": 9.18060740631793e-06,
+      "loss": 0.8449,
+      "step": 5163
+    },
+    {
+      "epoch": 0.5404500261643119,
+      "grad_norm": 2.173125028700542,
+      "learning_rate": 9.177229131814473e-06,
+      "loss": 0.9589,
+      "step": 5164
+    },
+    {
+      "epoch": 0.5405546834118262,
+      "grad_norm": 2.1583802338493427,
+      "learning_rate": 9.173850951849057e-06,
+      "loss": 1.0773,
+      "step": 5165
+    },
+    {
+      "epoch": 0.5406593406593406,
+      "grad_norm": 1.897622605941461,
+      "learning_rate": 9.170472866809842e-06,
+      "loss": 0.9341,
+      "step": 5166
+    },
+    {
+      "epoch": 0.540763997906855,
+      "grad_norm": 1.884134801109637,
+      "learning_rate": 9.167094877084974e-06,
+      "loss": 0.9404,
+      "step": 5167
+    },
+    {
+      "epoch": 0.5408686551543694,
+      "grad_norm": 2.1870995357335525,
+      "learning_rate": 9.163716983062594e-06,
+      "loss": 0.8965,
+      "step": 5168
+    },
+    {
+      "epoch": 0.5409733124018838,
+      "grad_norm": 2.141040011803557,
+      "learning_rate": 9.160339185130825e-06,
+      "loss": 0.9386,
+      "step": 5169
+    },
+    {
+      "epoch": 0.5410779696493982,
+      "grad_norm": 2.042779608504371,
+      "learning_rate": 9.15696148367779e-06,
+      "loss": 0.8375,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5411826268969127,
+      "grad_norm": 1.8593064018228953,
+      "learning_rate": 9.153583879091588e-06,
+      "loss": 0.8555,
+      "step": 5171
+    },
+    {
+      "epoch": 0.541287284144427,
+      "grad_norm": 2.3293817595274056,
+      "learning_rate": 9.150206371760315e-06,
+      "loss": 0.9834,
+      "step": 5172
+    },
+    {
+      "epoch": 0.5413919413919414,
+      "grad_norm": 2.0079430212166627,
+      "learning_rate": 9.146828962072051e-06,
+      "loss": 0.9392,
+      "step": 5173
+    },
+    {
+      "epoch": 0.5414965986394558,
+      "grad_norm": 2.1119483487510355,
+      "learning_rate": 9.143451650414871e-06,
+      "loss": 1.0077,
+      "step": 5174
+    },
+    {
+      "epoch": 0.5416012558869702,
+      "grad_norm": 1.9355398541105262,
+      "learning_rate": 9.140074437176831e-06,
+      "loss": 0.9627,
+      "step": 5175
+    },
+    {
+      "epoch": 0.5417059131344846,
+      "grad_norm": 2.066119618087384,
+      "learning_rate": 9.136697322745984e-06,
+      "loss": 0.9772,
+      "step": 5176
+    },
+    {
+      "epoch": 0.541810570381999,
+      "grad_norm": 1.8564381011561766,
+      "learning_rate": 9.133320307510359e-06,
+      "loss": 0.8735,
+      "step": 5177
+    },
+    {
+      "epoch": 0.5419152276295134,
+      "grad_norm": 1.9999578854370876,
+      "learning_rate": 9.129943391857992e-06,
+      "loss": 1.0295,
+      "step": 5178
+    },
+    {
+      "epoch": 0.5420198848770277,
+      "grad_norm": 2.264031570128424,
+      "learning_rate": 9.126566576176893e-06,
+      "loss": 0.7778,
+      "step": 5179
+    },
+    {
+      "epoch": 0.5421245421245421,
+      "grad_norm": 1.8699876284645591,
+      "learning_rate": 9.123189860855064e-06,
+      "loss": 0.9129,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5422291993720565,
+      "grad_norm": 1.9231443138469622,
+      "learning_rate": 9.119813246280499e-06,
+      "loss": 0.9216,
+      "step": 5181
+    },
+    {
+      "epoch": 0.5423338566195709,
+      "grad_norm": 1.7836314115264373,
+      "learning_rate": 9.116436732841174e-06,
+      "loss": 0.9005,
+      "step": 5182
+    },
+    {
+      "epoch": 0.5424385138670853,
+      "grad_norm": 2.2350357880443448,
+      "learning_rate": 9.113060320925063e-06,
+      "loss": 0.9321,
+      "step": 5183
+    },
+    {
+      "epoch": 0.5425431711145997,
+      "grad_norm": 2.079863133130194,
+      "learning_rate": 9.109684010920116e-06,
+      "loss": 0.8967,
+      "step": 5184
+    },
+    {
+      "epoch": 0.5426478283621141,
+      "grad_norm": 2.2139801096197553,
+      "learning_rate": 9.106307803214283e-06,
+      "loss": 1.0516,
+      "step": 5185
+    },
+    {
+      "epoch": 0.5427524856096285,
+      "grad_norm": 2.179734467120083,
+      "learning_rate": 9.102931698195491e-06,
+      "loss": 0.7468,
+      "step": 5186
+    },
+    {
+      "epoch": 0.5428571428571428,
+      "grad_norm": 2.157564716376065,
+      "learning_rate": 9.099555696251667e-06,
+      "loss": 0.9357,
+      "step": 5187
+    },
+    {
+      "epoch": 0.5429618001046572,
+      "grad_norm": 2.041100938324909,
+      "learning_rate": 9.096179797770723e-06,
+      "loss": 0.9291,
+      "step": 5188
+    },
+    {
+      "epoch": 0.5430664573521716,
+      "grad_norm": 1.8303406459070106,
+      "learning_rate": 9.09280400314055e-06,
+      "loss": 0.8139,
+      "step": 5189
+    },
+    {
+      "epoch": 0.543171114599686,
+      "grad_norm": 1.8598137235962091,
+      "learning_rate": 9.089428312749036e-06,
+      "loss": 0.8703,
+      "step": 5190
+    },
+    {
+      "epoch": 0.5432757718472004,
+      "grad_norm": 2.0757977617675945,
+      "learning_rate": 9.086052726984057e-06,
+      "loss": 0.9464,
+      "step": 5191
+    },
+    {
+      "epoch": 0.5433804290947148,
+      "grad_norm": 1.891684900539831,
+      "learning_rate": 9.08267724623347e-06,
+      "loss": 0.9067,
+      "step": 5192
+    },
+    {
+      "epoch": 0.5434850863422293,
+      "grad_norm": 1.9766032826136062,
+      "learning_rate": 9.079301870885126e-06,
+      "loss": 0.9417,
+      "step": 5193
+    },
+    {
+      "epoch": 0.5435897435897435,
+      "grad_norm": 2.1848428050243824,
+      "learning_rate": 9.075926601326865e-06,
+      "loss": 0.8109,
+      "step": 5194
+    },
+    {
+      "epoch": 0.543694400837258,
+      "grad_norm": 1.928000464883851,
+      "learning_rate": 9.072551437946508e-06,
+      "loss": 0.9048,
+      "step": 5195
+    },
+    {
+      "epoch": 0.5437990580847724,
+      "grad_norm": 2.2851706994938477,
+      "learning_rate": 9.069176381131874e-06,
+      "loss": 1.0057,
+      "step": 5196
+    },
+    {
+      "epoch": 0.5439037153322868,
+      "grad_norm": 2.2316974873884634,
+      "learning_rate": 9.06580143127076e-06,
+      "loss": 0.9539,
+      "step": 5197
+    },
+    {
+      "epoch": 0.5440083725798012,
+      "grad_norm": 2.431415213869758,
+      "learning_rate": 9.062426588750956e-06,
+      "loss": 0.9679,
+      "step": 5198
+    },
+    {
+      "epoch": 0.5441130298273156,
+      "grad_norm": 1.8303873361505676,
+      "learning_rate": 9.059051853960237e-06,
+      "loss": 1.0003,
+      "step": 5199
+    },
+    {
+      "epoch": 0.54421768707483,
+      "grad_norm": 1.9180411674794864,
+      "learning_rate": 9.05567722728637e-06,
+      "loss": 0.941,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5443223443223443,
+      "grad_norm": 2.1309566319139055,
+      "learning_rate": 9.0523027091171e-06,
+      "loss": 0.9292,
+      "step": 5201
+    },
+    {
+      "epoch": 0.5444270015698587,
+      "grad_norm": 1.9920223205280956,
+      "learning_rate": 9.048928299840172e-06,
+      "loss": 0.9156,
+      "step": 5202
+    },
+    {
+      "epoch": 0.5445316588173731,
+      "grad_norm": 1.9755399701908867,
+      "learning_rate": 9.04555399984331e-06,
+      "loss": 1.0254,
+      "step": 5203
+    },
+    {
+      "epoch": 0.5446363160648875,
+      "grad_norm": 1.867791911625654,
+      "learning_rate": 9.042179809514225e-06,
+      "loss": 0.9065,
+      "step": 5204
+    },
+    {
+      "epoch": 0.5447409733124019,
+      "grad_norm": 2.142420704324338,
+      "learning_rate": 9.038805729240626e-06,
+      "loss": 0.9879,
+      "step": 5205
+    },
+    {
+      "epoch": 0.5448456305599163,
+      "grad_norm": 1.8804509672023069,
+      "learning_rate": 9.035431759410194e-06,
+      "loss": 0.8701,
+      "step": 5206
+    },
+    {
+      "epoch": 0.5449502878074307,
+      "grad_norm": 1.8907052172885805,
+      "learning_rate": 9.032057900410612e-06,
+      "loss": 0.9644,
+      "step": 5207
+    },
+    {
+      "epoch": 0.545054945054945,
+      "grad_norm": 2.3910996543132965,
+      "learning_rate": 9.02868415262954e-06,
+      "loss": 0.9196,
+      "step": 5208
+    },
+    {
+      "epoch": 0.5451596023024594,
+      "grad_norm": 3.681066797684883,
+      "learning_rate": 9.025310516454623e-06,
+      "loss": 1.0363,
+      "step": 5209
+    },
+    {
+      "epoch": 0.5452642595499738,
+      "grad_norm": 1.6994987522917748,
+      "learning_rate": 9.021936992273508e-06,
+      "loss": 0.8546,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5453689167974882,
+      "grad_norm": 2.5337992006092405,
+      "learning_rate": 9.018563580473814e-06,
+      "loss": 0.9382,
+      "step": 5211
+    },
+    {
+      "epoch": 0.5454735740450026,
+      "grad_norm": 1.8493509787452933,
+      "learning_rate": 9.01519028144315e-06,
+      "loss": 0.8746,
+      "step": 5212
+    },
+    {
+      "epoch": 0.545578231292517,
+      "grad_norm": 1.9579976590736436,
+      "learning_rate": 9.011817095569124e-06,
+      "loss": 0.9685,
+      "step": 5213
+    },
+    {
+      "epoch": 0.5456828885400314,
+      "grad_norm": 2.1333444289770087,
+      "learning_rate": 9.008444023239315e-06,
+      "loss": 0.7952,
+      "step": 5214
+    },
+    {
+      "epoch": 0.5457875457875457,
+      "grad_norm": 2.3503433202907478,
+      "learning_rate": 9.005071064841299e-06,
+      "loss": 0.88,
+      "step": 5215
+    },
+    {
+      "epoch": 0.5458922030350601,
+      "grad_norm": 2.2546724085226844,
+      "learning_rate": 9.001698220762634e-06,
+      "loss": 0.9255,
+      "step": 5216
+    },
+    {
+      "epoch": 0.5459968602825745,
+      "grad_norm": 2.356440368535242,
+      "learning_rate": 8.998325491390869e-06,
+      "loss": 1.0099,
+      "step": 5217
+    },
+    {
+      "epoch": 0.546101517530089,
+      "grad_norm": 2.2034666717622033,
+      "learning_rate": 8.994952877113533e-06,
+      "loss": 1.0675,
+      "step": 5218
+    },
+    {
+      "epoch": 0.5462061747776034,
+      "grad_norm": 1.9688211472663686,
+      "learning_rate": 8.991580378318151e-06,
+      "loss": 0.845,
+      "step": 5219
+    },
+    {
+      "epoch": 0.5463108320251178,
+      "grad_norm": 2.1233709050192213,
+      "learning_rate": 8.988207995392227e-06,
+      "loss": 1.1076,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5464154892726322,
+      "grad_norm": 2.2100117848067886,
+      "learning_rate": 8.984835728723251e-06,
+      "loss": 0.9292,
+      "step": 5221
+    },
+    {
+      "epoch": 0.5465201465201465,
+      "grad_norm": 2.00762864700889,
+      "learning_rate": 8.981463578698714e-06,
+      "loss": 0.9889,
+      "step": 5222
+    },
+    {
+      "epoch": 0.5466248037676609,
+      "grad_norm": 2.000506555020569,
+      "learning_rate": 8.978091545706074e-06,
+      "loss": 0.9671,
+      "step": 5223
+    },
+    {
+      "epoch": 0.5467294610151753,
+      "grad_norm": 1.9318729631393265,
+      "learning_rate": 8.974719630132788e-06,
+      "loss": 0.9534,
+      "step": 5224
+    },
+    {
+      "epoch": 0.5468341182626897,
+      "grad_norm": 2.704458307866667,
+      "learning_rate": 8.971347832366296e-06,
+      "loss": 0.9215,
+      "step": 5225
+    },
+    {
+      "epoch": 0.5469387755102041,
+      "grad_norm": 1.8785357601729604,
+      "learning_rate": 8.96797615279402e-06,
+      "loss": 0.8026,
+      "step": 5226
+    },
+    {
+      "epoch": 0.5470434327577185,
+      "grad_norm": 1.982632154167765,
+      "learning_rate": 8.96460459180338e-06,
+      "loss": 0.9695,
+      "step": 5227
+    },
+    {
+      "epoch": 0.5471480900052329,
+      "grad_norm": 2.143564500630684,
+      "learning_rate": 8.961233149781768e-06,
+      "loss": 0.939,
+      "step": 5228
+    },
+    {
+      "epoch": 0.5472527472527473,
+      "grad_norm": 2.203497996122728,
+      "learning_rate": 8.957861827116576e-06,
+      "loss": 0.9991,
+      "step": 5229
+    },
+    {
+      "epoch": 0.5473574045002616,
+      "grad_norm": 1.9379036312231492,
+      "learning_rate": 8.954490624195169e-06,
+      "loss": 0.8894,
+      "step": 5230
+    },
+    {
+      "epoch": 0.547462061747776,
+      "grad_norm": 2.0508974667099245,
+      "learning_rate": 8.951119541404908e-06,
+      "loss": 1.0449,
+      "step": 5231
+    },
+    {
+      "epoch": 0.5475667189952904,
+      "grad_norm": 2.588002315713727,
+      "learning_rate": 8.947748579133143e-06,
+      "loss": 0.9373,
+      "step": 5232
+    },
+    {
+      "epoch": 0.5476713762428048,
+      "grad_norm": 2.38449359167579,
+      "learning_rate": 8.944377737767195e-06,
+      "loss": 1.041,
+      "step": 5233
+    },
+    {
+      "epoch": 0.5477760334903192,
+      "grad_norm": 1.9288662968847525,
+      "learning_rate": 8.94100701769439e-06,
+      "loss": 0.9473,
+      "step": 5234
+    },
+    {
+      "epoch": 0.5478806907378336,
+      "grad_norm": 2.1753526717070155,
+      "learning_rate": 8.93763641930202e-06,
+      "loss": 0.8533,
+      "step": 5235
+    },
+    {
+      "epoch": 0.547985347985348,
+      "grad_norm": 2.214682121752073,
+      "learning_rate": 8.934265942977383e-06,
+      "loss": 1.0522,
+      "step": 5236
+    },
+    {
+      "epoch": 0.5480900052328623,
+      "grad_norm": 1.9884467197972144,
+      "learning_rate": 8.930895589107748e-06,
+      "loss": 1.0617,
+      "step": 5237
+    },
+    {
+      "epoch": 0.5481946624803767,
+      "grad_norm": 1.8273085911630584,
+      "learning_rate": 8.927525358080374e-06,
+      "loss": 0.8846,
+      "step": 5238
+    },
+    {
+      "epoch": 0.5482993197278911,
+      "grad_norm": 2.132364155223627,
+      "learning_rate": 8.924155250282515e-06,
+      "loss": 0.8767,
+      "step": 5239
+    },
+    {
+      "epoch": 0.5484039769754055,
+      "grad_norm": 2.1653790608930557,
+      "learning_rate": 8.920785266101396e-06,
+      "loss": 1.0162,
+      "step": 5240
+    },
+    {
+      "epoch": 0.54850863422292,
+      "grad_norm": 1.9836556420176448,
+      "learning_rate": 8.917415405924241e-06,
+      "loss": 0.9864,
+      "step": 5241
+    },
+    {
+      "epoch": 0.5486132914704344,
+      "grad_norm": 2.497343830248175,
+      "learning_rate": 8.914045670138247e-06,
+      "loss": 1.0194,
+      "step": 5242
+    },
+    {
+      "epoch": 0.5487179487179488,
+      "grad_norm": 2.2357270888989262,
+      "learning_rate": 8.910676059130611e-06,
+      "loss": 1.0598,
+      "step": 5243
+    },
+    {
+      "epoch": 0.5488226059654631,
+      "grad_norm": 2.076847097610531,
+      "learning_rate": 8.907306573288502e-06,
+      "loss": 0.916,
+      "step": 5244
+    },
+    {
+      "epoch": 0.5489272632129775,
+      "grad_norm": 1.9802029893372697,
+      "learning_rate": 8.903937212999082e-06,
+      "loss": 0.843,
+      "step": 5245
+    },
+    {
+      "epoch": 0.5490319204604919,
+      "grad_norm": 2.0192766915384155,
+      "learning_rate": 8.900567978649501e-06,
+      "loss": 0.9069,
+      "step": 5246
+    },
+    {
+      "epoch": 0.5491365777080063,
+      "grad_norm": 2.432583695671836,
+      "learning_rate": 8.897198870626882e-06,
+      "loss": 0.8992,
+      "step": 5247
+    },
+    {
+      "epoch": 0.5492412349555207,
+      "grad_norm": 2.213741486304002,
+      "learning_rate": 8.893829889318352e-06,
+      "loss": 0.8917,
+      "step": 5248
+    },
+    {
+      "epoch": 0.5493458922030351,
+      "grad_norm": 2.135434999613721,
+      "learning_rate": 8.89046103511101e-06,
+      "loss": 0.9377,
+      "step": 5249
+    },
+    {
+      "epoch": 0.5494505494505495,
+      "grad_norm": 2.122303160131647,
+      "learning_rate": 8.887092308391945e-06,
+      "loss": 0.9298,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5495552066980638,
+      "grad_norm": 2.2880405163464532,
+      "learning_rate": 8.883723709548232e-06,
+      "loss": 1.0074,
+      "step": 5251
+    },
+    {
+      "epoch": 0.5496598639455782,
+      "grad_norm": 1.9274854009826472,
+      "learning_rate": 8.880355238966923e-06,
+      "loss": 0.8281,
+      "step": 5252
+    },
+    {
+      "epoch": 0.5497645211930926,
+      "grad_norm": 1.9114757164675305,
+      "learning_rate": 8.87698689703507e-06,
+      "loss": 0.9118,
+      "step": 5253
+    },
+    {
+      "epoch": 0.549869178440607,
+      "grad_norm": 2.1358804324224088,
+      "learning_rate": 8.873618684139697e-06,
+      "loss": 0.9843,
+      "step": 5254
+    },
+    {
+      "epoch": 0.5499738356881214,
+      "grad_norm": 2.1558357260108205,
+      "learning_rate": 8.870250600667823e-06,
+      "loss": 1.0874,
+      "step": 5255
+    },
+    {
+      "epoch": 0.5500784929356358,
+      "grad_norm": 2.4247493566253926,
+      "learning_rate": 8.866882647006441e-06,
+      "loss": 0.9887,
+      "step": 5256
+    },
+    {
+      "epoch": 0.5501831501831502,
+      "grad_norm": 2.028235978688342,
+      "learning_rate": 8.863514823542542e-06,
+      "loss": 0.9167,
+      "step": 5257
+    },
+    {
+      "epoch": 0.5502878074306645,
+      "grad_norm": 1.9043186883483203,
+      "learning_rate": 8.860147130663096e-06,
+      "loss": 0.8844,
+      "step": 5258
+    },
+    {
+      "epoch": 0.5503924646781789,
+      "grad_norm": 2.8270360697499566,
+      "learning_rate": 8.856779568755052e-06,
+      "loss": 0.9991,
+      "step": 5259
+    },
+    {
+      "epoch": 0.5504971219256933,
+      "grad_norm": 1.942578281067632,
+      "learning_rate": 8.853412138205357e-06,
+      "loss": 0.9976,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5506017791732077,
+      "grad_norm": 2.2072497818987467,
+      "learning_rate": 8.85004483940093e-06,
+      "loss": 0.9079,
+      "step": 5261
+    },
+    {
+      "epoch": 0.5507064364207221,
+      "grad_norm": 1.958197617874853,
+      "learning_rate": 8.846677672728683e-06,
+      "loss": 1.0171,
+      "step": 5262
+    },
+    {
+      "epoch": 0.5508110936682366,
+      "grad_norm": 1.946933390805341,
+      "learning_rate": 8.84331063857551e-06,
+      "loss": 0.9166,
+      "step": 5263
+    },
+    {
+      "epoch": 0.550915750915751,
+      "grad_norm": 2.099808823892071,
+      "learning_rate": 8.839943737328282e-06,
+      "loss": 0.9827,
+      "step": 5264
+    },
+    {
+      "epoch": 0.5510204081632653,
+      "grad_norm": 2.018139469461617,
+      "learning_rate": 8.836576969373877e-06,
+      "loss": 0.9721,
+      "step": 5265
+    },
+    {
+      "epoch": 0.5511250654107797,
+      "grad_norm": 2.297998216881771,
+      "learning_rate": 8.833210335099137e-06,
+      "loss": 0.9523,
+      "step": 5266
+    },
+    {
+      "epoch": 0.5512297226582941,
+      "grad_norm": 2.0134991695196005,
+      "learning_rate": 8.829843834890893e-06,
+      "loss": 0.975,
+      "step": 5267
+    },
+    {
+      "epoch": 0.5513343799058085,
+      "grad_norm": 2.150846758924609,
+      "learning_rate": 8.826477469135967e-06,
+      "loss": 0.8611,
+      "step": 5268
+    },
+    {
+      "epoch": 0.5514390371533229,
+      "grad_norm": 1.82333106005884,
+      "learning_rate": 8.823111238221156e-06,
+      "loss": 0.9238,
+      "step": 5269
+    },
+    {
+      "epoch": 0.5515436944008373,
+      "grad_norm": 2.4050851629812655,
+      "learning_rate": 8.819745142533252e-06,
+      "loss": 1.0834,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5516483516483517,
+      "grad_norm": 2.28131876935219,
+      "learning_rate": 8.81637918245902e-06,
+      "loss": 0.9526,
+      "step": 5271
+    },
+    {
+      "epoch": 0.5517530088958661,
+      "grad_norm": 2.1174671106347236,
+      "learning_rate": 8.813013358385223e-06,
+      "loss": 0.9436,
+      "step": 5272
+    },
+    {
+      "epoch": 0.5518576661433804,
+      "grad_norm": 1.9671009087310183,
+      "learning_rate": 8.809647670698594e-06,
+      "loss": 1.0025,
+      "step": 5273
+    },
+    {
+      "epoch": 0.5519623233908948,
+      "grad_norm": 2.052327149623945,
+      "learning_rate": 8.806282119785861e-06,
+      "loss": 0.8248,
+      "step": 5274
+    },
+    {
+      "epoch": 0.5520669806384092,
+      "grad_norm": 2.384339181879582,
+      "learning_rate": 8.802916706033734e-06,
+      "loss": 1.008,
+      "step": 5275
+    },
+    {
+      "epoch": 0.5521716378859236,
+      "grad_norm": 1.8901455647678653,
+      "learning_rate": 8.799551429828906e-06,
+      "loss": 0.8009,
+      "step": 5276
+    },
+    {
+      "epoch": 0.552276295133438,
+      "grad_norm": 1.9504635654515794,
+      "learning_rate": 8.796186291558052e-06,
+      "loss": 0.9597,
+      "step": 5277
+    },
+    {
+      "epoch": 0.5523809523809524,
+      "grad_norm": 2.3275789921544017,
+      "learning_rate": 8.79282129160783e-06,
+      "loss": 0.8494,
+      "step": 5278
+    },
+    {
+      "epoch": 0.5524856096284668,
+      "grad_norm": 1.835713336563086,
+      "learning_rate": 8.789456430364892e-06,
+      "loss": 0.9284,
+      "step": 5279
+    },
+    {
+      "epoch": 0.5525902668759811,
+      "grad_norm": 2.017436152195592,
+      "learning_rate": 8.786091708215863e-06,
+      "loss": 0.8889,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5526949241234955,
+      "grad_norm": 1.959115114298543,
+      "learning_rate": 8.782727125547358e-06,
+      "loss": 0.8949,
+      "step": 5281
+    },
+    {
+      "epoch": 0.5527995813710099,
+      "grad_norm": 2.0528665958582093,
+      "learning_rate": 8.779362682745971e-06,
+      "loss": 0.9803,
+      "step": 5282
+    },
+    {
+      "epoch": 0.5529042386185243,
+      "grad_norm": 1.8604560176356941,
+      "learning_rate": 8.77599838019829e-06,
+      "loss": 0.923,
+      "step": 5283
+    },
+    {
+      "epoch": 0.5530088958660387,
+      "grad_norm": 1.800262625922265,
+      "learning_rate": 8.772634218290874e-06,
+      "loss": 0.9148,
+      "step": 5284
+    },
+    {
+      "epoch": 0.5531135531135531,
+      "grad_norm": 2.4993048236879525,
+      "learning_rate": 8.769270197410277e-06,
+      "loss": 0.9607,
+      "step": 5285
+    },
+    {
+      "epoch": 0.5532182103610676,
+      "grad_norm": 1.8252037891658632,
+      "learning_rate": 8.765906317943027e-06,
+      "loss": 0.8508,
+      "step": 5286
+    },
+    {
+      "epoch": 0.5533228676085818,
+      "grad_norm": 2.238304823724985,
+      "learning_rate": 8.762542580275644e-06,
+      "loss": 0.9228,
+      "step": 5287
+    },
+    {
+      "epoch": 0.5534275248560963,
+      "grad_norm": 2.1072228350806896,
+      "learning_rate": 8.759178984794627e-06,
+      "loss": 0.8588,
+      "step": 5288
+    },
+    {
+      "epoch": 0.5535321821036107,
+      "grad_norm": 1.9622546540770844,
+      "learning_rate": 8.755815531886461e-06,
+      "loss": 0.9111,
+      "step": 5289
+    },
+    {
+      "epoch": 0.5536368393511251,
+      "grad_norm": 2.2818234572001144,
+      "learning_rate": 8.752452221937608e-06,
+      "loss": 1.0615,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5537414965986395,
+      "grad_norm": 2.1750822391853344,
+      "learning_rate": 8.749089055334527e-06,
+      "loss": 0.8601,
+      "step": 5291
+    },
+    {
+      "epoch": 0.5538461538461539,
+      "grad_norm": 2.0616636889051896,
+      "learning_rate": 8.74572603246365e-06,
+      "loss": 1.0359,
+      "step": 5292
+    },
+    {
+      "epoch": 0.5539508110936683,
+      "grad_norm": 2.710571314021102,
+      "learning_rate": 8.742363153711396e-06,
+      "loss": 1.0601,
+      "step": 5293
+    },
+    {
+      "epoch": 0.5540554683411826,
+      "grad_norm": 2.6145632006844886,
+      "learning_rate": 8.739000419464164e-06,
+      "loss": 1.0233,
+      "step": 5294
+    },
+    {
+      "epoch": 0.554160125588697,
+      "grad_norm": 1.9109618531413386,
+      "learning_rate": 8.735637830108337e-06,
+      "loss": 0.9872,
+      "step": 5295
+    },
+    {
+      "epoch": 0.5542647828362114,
+      "grad_norm": 2.1467379854230995,
+      "learning_rate": 8.732275386030291e-06,
+      "loss": 0.9363,
+      "step": 5296
+    },
+    {
+      "epoch": 0.5543694400837258,
+      "grad_norm": 2.0056169978726786,
+      "learning_rate": 8.72891308761637e-06,
+      "loss": 0.989,
+      "step": 5297
+    },
+    {
+      "epoch": 0.5544740973312402,
+      "grad_norm": 2.1780493741262608,
+      "learning_rate": 8.725550935252914e-06,
+      "loss": 1.0403,
+      "step": 5298
+    },
+    {
+      "epoch": 0.5545787545787546,
+      "grad_norm": 2.21960346530747,
+      "learning_rate": 8.722188929326237e-06,
+      "loss": 0.9504,
+      "step": 5299
+    },
+    {
+      "epoch": 0.554683411826269,
+      "grad_norm": 2.1450710730738116,
+      "learning_rate": 8.718827070222642e-06,
+      "loss": 1.1257,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5547880690737833,
+      "grad_norm": 2.228582965255394,
+      "learning_rate": 8.715465358328417e-06,
+      "loss": 0.9697,
+      "step": 5301
+    },
+    {
+      "epoch": 0.5548927263212977,
+      "grad_norm": 2.4830808941857554,
+      "learning_rate": 8.712103794029824e-06,
+      "loss": 0.8945,
+      "step": 5302
+    },
+    {
+      "epoch": 0.5549973835688121,
+      "grad_norm": 2.2847911242856656,
+      "learning_rate": 8.708742377713118e-06,
+      "loss": 0.9817,
+      "step": 5303
+    },
+    {
+      "epoch": 0.5551020408163265,
+      "grad_norm": 2.1604390174895416,
+      "learning_rate": 8.705381109764529e-06,
+      "loss": 0.9853,
+      "step": 5304
+    },
+    {
+      "epoch": 0.5552066980638409,
+      "grad_norm": 2.2396160300426082,
+      "learning_rate": 8.702019990570274e-06,
+      "loss": 0.8813,
+      "step": 5305
+    },
+    {
+      "epoch": 0.5553113553113553,
+      "grad_norm": 2.130623108079099,
+      "learning_rate": 8.698659020516556e-06,
+      "loss": 0.8336,
+      "step": 5306
+    },
+    {
+      "epoch": 0.5554160125588697,
+      "grad_norm": 1.9908826133921063,
+      "learning_rate": 8.695298199989552e-06,
+      "loss": 0.9344,
+      "step": 5307
+    },
+    {
+      "epoch": 0.555520669806384,
+      "grad_norm": 1.9711870901565465,
+      "learning_rate": 8.691937529375428e-06,
+      "loss": 0.9166,
+      "step": 5308
+    },
+    {
+      "epoch": 0.5556253270538984,
+      "grad_norm": 1.9835386623262095,
+      "learning_rate": 8.688577009060338e-06,
+      "loss": 0.9319,
+      "step": 5309
+    },
+    {
+      "epoch": 0.5557299843014128,
+      "grad_norm": 2.3085235533894126,
+      "learning_rate": 8.685216639430405e-06,
+      "loss": 0.8664,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5558346415489273,
+      "grad_norm": 3.3332655720477895,
+      "learning_rate": 8.681856420871747e-06,
+      "loss": 0.8843,
+      "step": 5311
+    },
+    {
+      "epoch": 0.5559392987964417,
+      "grad_norm": 1.8582951572007596,
+      "learning_rate": 8.678496353770458e-06,
+      "loss": 0.8867,
+      "step": 5312
+    },
+    {
+      "epoch": 0.5560439560439561,
+      "grad_norm": 2.173807723457474,
+      "learning_rate": 8.67513643851262e-06,
+      "loss": 0.9826,
+      "step": 5313
+    },
+    {
+      "epoch": 0.5561486132914705,
+      "grad_norm": 1.7581129151706683,
+      "learning_rate": 8.671776675484288e-06,
+      "loss": 0.9056,
+      "step": 5314
+    },
+    {
+      "epoch": 0.5562532705389849,
+      "grad_norm": 2.1180334877984017,
+      "learning_rate": 8.66841706507151e-06,
+      "loss": 0.9616,
+      "step": 5315
+    },
+    {
+      "epoch": 0.5563579277864992,
+      "grad_norm": 2.508246391771904,
+      "learning_rate": 8.665057607660305e-06,
+      "loss": 0.9179,
+      "step": 5316
+    },
+    {
+      "epoch": 0.5564625850340136,
+      "grad_norm": 1.9132610573755788,
+      "learning_rate": 8.661698303636692e-06,
+      "loss": 0.8745,
+      "step": 5317
+    },
+    {
+      "epoch": 0.556567242281528,
+      "grad_norm": 2.4958112187692767,
+      "learning_rate": 8.658339153386658e-06,
+      "loss": 0.9097,
+      "step": 5318
+    },
+    {
+      "epoch": 0.5566718995290424,
+      "grad_norm": 2.249652224152348,
+      "learning_rate": 8.654980157296173e-06,
+      "loss": 1.0163,
+      "step": 5319
+    },
+    {
+      "epoch": 0.5567765567765568,
+      "grad_norm": 2.1636612203919787,
+      "learning_rate": 8.651621315751197e-06,
+      "loss": 1.0146,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5568812140240712,
+      "grad_norm": 2.4873635948113484,
+      "learning_rate": 8.648262629137663e-06,
+      "loss": 1.0301,
+      "step": 5321
+    },
+    {
+      "epoch": 0.5569858712715856,
+      "grad_norm": 2.1298989438781044,
+      "learning_rate": 8.644904097841494e-06,
+      "loss": 0.979,
+      "step": 5322
+    },
+    {
+      "epoch": 0.5570905285190999,
+      "grad_norm": 2.296389158373194,
+      "learning_rate": 8.641545722248592e-06,
+      "loss": 0.9278,
+      "step": 5323
+    },
+    {
+      "epoch": 0.5571951857666143,
+      "grad_norm": 2.2433733862004055,
+      "learning_rate": 8.638187502744838e-06,
+      "loss": 0.8756,
+      "step": 5324
+    },
+    {
+      "epoch": 0.5572998430141287,
+      "grad_norm": 1.9895145461612076,
+      "learning_rate": 8.634829439716099e-06,
+      "loss": 0.9287,
+      "step": 5325
+    },
+    {
+      "epoch": 0.5574045002616431,
+      "grad_norm": 1.8763204041932915,
+      "learning_rate": 8.63147153354823e-06,
+      "loss": 0.875,
+      "step": 5326
+    },
+    {
+      "epoch": 0.5575091575091575,
+      "grad_norm": 2.2272253472444103,
+      "learning_rate": 8.628113784627053e-06,
+      "loss": 0.9695,
+      "step": 5327
+    },
+    {
+      "epoch": 0.5576138147566719,
+      "grad_norm": 1.972305846861178,
+      "learning_rate": 8.624756193338384e-06,
+      "loss": 0.9767,
+      "step": 5328
+    },
+    {
+      "epoch": 0.5577184720041863,
+      "grad_norm": 2.3386342244281124,
+      "learning_rate": 8.621398760068017e-06,
+      "loss": 0.9315,
+      "step": 5329
+    },
+    {
+      "epoch": 0.5578231292517006,
+      "grad_norm": 2.3263287856510773,
+      "learning_rate": 8.618041485201728e-06,
+      "loss": 0.8489,
+      "step": 5330
+    },
+    {
+      "epoch": 0.557927786499215,
+      "grad_norm": 2.1728955209423315,
+      "learning_rate": 8.61468436912527e-06,
+      "loss": 0.8711,
+      "step": 5331
+    },
+    {
+      "epoch": 0.5580324437467294,
+      "grad_norm": 1.9615782696853798,
+      "learning_rate": 8.61132741222439e-06,
+      "loss": 0.9714,
+      "step": 5332
+    },
+    {
+      "epoch": 0.5581371009942439,
+      "grad_norm": 1.997852261589968,
+      "learning_rate": 8.607970614884804e-06,
+      "loss": 0.7903,
+      "step": 5333
+    },
+    {
+      "epoch": 0.5582417582417583,
+      "grad_norm": 2.0318840672120375,
+      "learning_rate": 8.604613977492212e-06,
+      "loss": 0.9649,
+      "step": 5334
+    },
+    {
+      "epoch": 0.5583464154892727,
+      "grad_norm": 2.0231096418522756,
+      "learning_rate": 8.60125750043231e-06,
+      "loss": 0.894,
+      "step": 5335
+    },
+    {
+      "epoch": 0.5584510727367871,
+      "grad_norm": 2.0941739900946663,
+      "learning_rate": 8.597901184090753e-06,
+      "loss": 1.0217,
+      "step": 5336
+    },
+    {
+      "epoch": 0.5585557299843014,
+      "grad_norm": 2.0371217772719383,
+      "learning_rate": 8.594545028853195e-06,
+      "loss": 0.9738,
+      "step": 5337
+    },
+    {
+      "epoch": 0.5586603872318158,
+      "grad_norm": 2.4539134994923724,
+      "learning_rate": 8.59118903510526e-06,
+      "loss": 0.8423,
+      "step": 5338
+    },
+    {
+      "epoch": 0.5587650444793302,
+      "grad_norm": 2.0675918762574557,
+      "learning_rate": 8.587833203232564e-06,
+      "loss": 0.9802,
+      "step": 5339
+    },
+    {
+      "epoch": 0.5588697017268446,
+      "grad_norm": 2.419698934398359,
+      "learning_rate": 8.584477533620693e-06,
+      "loss": 1.0828,
+      "step": 5340
+    },
+    {
+      "epoch": 0.558974358974359,
+      "grad_norm": 1.7962022864475375,
+      "learning_rate": 8.581122026655222e-06,
+      "loss": 0.8116,
+      "step": 5341
+    },
+    {
+      "epoch": 0.5590790162218734,
+      "grad_norm": 2.2433085787237044,
+      "learning_rate": 8.577766682721705e-06,
+      "loss": 0.913,
+      "step": 5342
+    },
+    {
+      "epoch": 0.5591836734693878,
+      "grad_norm": 2.1428509222947656,
+      "learning_rate": 8.574411502205682e-06,
+      "loss": 1.0034,
+      "step": 5343
+    },
+    {
+      "epoch": 0.5592883307169021,
+      "grad_norm": 2.0066030593881035,
+      "learning_rate": 8.571056485492668e-06,
+      "loss": 0.9946,
+      "step": 5344
+    },
+    {
+      "epoch": 0.5593929879644165,
+      "grad_norm": 2.1542448139874164,
+      "learning_rate": 8.56770163296816e-06,
+      "loss": 0.9432,
+      "step": 5345
+    },
+    {
+      "epoch": 0.5594976452119309,
+      "grad_norm": 2.133404649889683,
+      "learning_rate": 8.564346945017637e-06,
+      "loss": 0.8289,
+      "step": 5346
+    },
+    {
+      "epoch": 0.5596023024594453,
+      "grad_norm": 1.9683719436497848,
+      "learning_rate": 8.560992422026563e-06,
+      "loss": 0.9729,
+      "step": 5347
+    },
+    {
+      "epoch": 0.5597069597069597,
+      "grad_norm": 2.0108641044421374,
+      "learning_rate": 8.557638064380375e-06,
+      "loss": 0.9478,
+      "step": 5348
+    },
+    {
+      "epoch": 0.5598116169544741,
+      "grad_norm": 1.9417606569777943,
+      "learning_rate": 8.554283872464499e-06,
+      "loss": 1.1138,
+      "step": 5349
+    },
+    {
+      "epoch": 0.5599162742019885,
+      "grad_norm": 2.0004963619150646,
+      "learning_rate": 8.550929846664333e-06,
+      "loss": 0.832,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5600209314495028,
+      "grad_norm": 1.9750719959886254,
+      "learning_rate": 8.547575987365266e-06,
+      "loss": 0.8858,
+      "step": 5351
+    },
+    {
+      "epoch": 0.5601255886970172,
+      "grad_norm": 1.962371605202089,
+      "learning_rate": 8.544222294952665e-06,
+      "loss": 0.8864,
+      "step": 5352
+    },
+    {
+      "epoch": 0.5602302459445316,
+      "grad_norm": 2.2675771268273506,
+      "learning_rate": 8.540868769811871e-06,
+      "loss": 1.0038,
+      "step": 5353
+    },
+    {
+      "epoch": 0.560334903192046,
+      "grad_norm": 2.0704526254917055,
+      "learning_rate": 8.537515412328216e-06,
+      "loss": 0.9498,
+      "step": 5354
+    },
+    {
+      "epoch": 0.5604395604395604,
+      "grad_norm": 2.073132608363666,
+      "learning_rate": 8.534162222887003e-06,
+      "loss": 0.9982,
+      "step": 5355
+    },
+    {
+      "epoch": 0.5605442176870749,
+      "grad_norm": 2.171878885625719,
+      "learning_rate": 8.530809201873523e-06,
+      "loss": 0.8136,
+      "step": 5356
+    },
+    {
+      "epoch": 0.5606488749345893,
+      "grad_norm": 2.1045724839104367,
+      "learning_rate": 8.527456349673043e-06,
+      "loss": 1.0294,
+      "step": 5357
+    },
+    {
+      "epoch": 0.5607535321821037,
+      "grad_norm": 2.1863543307087987,
+      "learning_rate": 8.524103666670817e-06,
+      "loss": 0.9556,
+      "step": 5358
+    },
+    {
+      "epoch": 0.560858189429618,
+      "grad_norm": 2.0478052085618557,
+      "learning_rate": 8.52075115325207e-06,
+      "loss": 0.9486,
+      "step": 5359
+    },
+    {
+      "epoch": 0.5609628466771324,
+      "grad_norm": 2.186343411269352,
+      "learning_rate": 8.517398809802009e-06,
+      "loss": 1.1094,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5610675039246468,
+      "grad_norm": 2.1314430854613007,
+      "learning_rate": 8.514046636705835e-06,
+      "loss": 0.9564,
+      "step": 5361
+    },
+    {
+      "epoch": 0.5611721611721612,
+      "grad_norm": 2.118979548912484,
+      "learning_rate": 8.510694634348715e-06,
+      "loss": 0.9487,
+      "step": 5362
+    },
+    {
+      "epoch": 0.5612768184196756,
+      "grad_norm": 1.9296804650577581,
+      "learning_rate": 8.507342803115799e-06,
+      "loss": 1.041,
+      "step": 5363
+    },
+    {
+      "epoch": 0.56138147566719,
+      "grad_norm": 1.7942107435770736,
+      "learning_rate": 8.503991143392225e-06,
+      "loss": 0.9425,
+      "step": 5364
+    },
+    {
+      "epoch": 0.5614861329147044,
+      "grad_norm": 2.465347277668139,
+      "learning_rate": 8.500639655563097e-06,
+      "loss": 0.9761,
+      "step": 5365
+    },
+    {
+      "epoch": 0.5615907901622187,
+      "grad_norm": 2.1891772863766774,
+      "learning_rate": 8.497288340013514e-06,
+      "loss": 0.9151,
+      "step": 5366
+    },
+    {
+      "epoch": 0.5616954474097331,
+      "grad_norm": 1.726643594528601,
+      "learning_rate": 8.493937197128546e-06,
+      "loss": 0.959,
+      "step": 5367
+    },
+    {
+      "epoch": 0.5618001046572475,
+      "grad_norm": 1.7289012170427154,
+      "learning_rate": 8.490586227293244e-06,
+      "loss": 0.8114,
+      "step": 5368
+    },
+    {
+      "epoch": 0.5619047619047619,
+      "grad_norm": 1.8444026585255053,
+      "learning_rate": 8.487235430892648e-06,
+      "loss": 0.8111,
+      "step": 5369
+    },
+    {
+      "epoch": 0.5620094191522763,
+      "grad_norm": 2.1272015623519778,
+      "learning_rate": 8.483884808311768e-06,
+      "loss": 0.8135,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5621140763997907,
+      "grad_norm": 2.064621150380741,
+      "learning_rate": 8.480534359935596e-06,
+      "loss": 0.7135,
+      "step": 5371
+    },
+    {
+      "epoch": 0.5622187336473051,
+      "grad_norm": 2.201762241819266,
+      "learning_rate": 8.477184086149109e-06,
+      "loss": 1.0045,
+      "step": 5372
+    },
+    {
+      "epoch": 0.5623233908948194,
+      "grad_norm": 2.331143016583063,
+      "learning_rate": 8.473833987337255e-06,
+      "loss": 0.8975,
+      "step": 5373
+    },
+    {
+      "epoch": 0.5624280481423338,
+      "grad_norm": 2.956408422523764,
+      "learning_rate": 8.470484063884969e-06,
+      "loss": 0.9348,
+      "step": 5374
+    },
+    {
+      "epoch": 0.5625327053898482,
+      "grad_norm": 2.929476780401184,
+      "learning_rate": 8.467134316177167e-06,
+      "loss": 0.9468,
+      "step": 5375
+    },
+    {
+      "epoch": 0.5626373626373626,
+      "grad_norm": 2.022910118977885,
+      "learning_rate": 8.46378474459874e-06,
+      "loss": 0.8546,
+      "step": 5376
+    },
+    {
+      "epoch": 0.562742019884877,
+      "grad_norm": 2.2903787311759123,
+      "learning_rate": 8.460435349534555e-06,
+      "loss": 0.9358,
+      "step": 5377
+    },
+    {
+      "epoch": 0.5628466771323914,
+      "grad_norm": 2.899288834804206,
+      "learning_rate": 8.457086131369475e-06,
+      "loss": 0.938,
+      "step": 5378
+    },
+    {
+      "epoch": 0.5629513343799059,
+      "grad_norm": 1.8808885484596338,
+      "learning_rate": 8.453737090488324e-06,
+      "loss": 0.8926,
+      "step": 5379
+    },
+    {
+      "epoch": 0.5630559916274201,
+      "grad_norm": 2.015376498610437,
+      "learning_rate": 8.450388227275914e-06,
+      "loss": 0.9181,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5631606488749346,
+      "grad_norm": 2.4062132896620576,
+      "learning_rate": 8.447039542117043e-06,
+      "loss": 0.9622,
+      "step": 5381
+    },
+    {
+      "epoch": 0.563265306122449,
+      "grad_norm": 2.163527877551462,
+      "learning_rate": 8.443691035396472e-06,
+      "loss": 0.9314,
+      "step": 5382
+    },
+    {
+      "epoch": 0.5633699633699634,
+      "grad_norm": 1.7474043964097528,
+      "learning_rate": 8.440342707498961e-06,
+      "loss": 0.8849,
+      "step": 5383
+    },
+    {
+      "epoch": 0.5634746206174778,
+      "grad_norm": 2.0844952386004256,
+      "learning_rate": 8.43699455880923e-06,
+      "loss": 0.9463,
+      "step": 5384
+    },
+    {
+      "epoch": 0.5635792778649922,
+      "grad_norm": 2.1603071390316018,
+      "learning_rate": 8.433646589711996e-06,
+      "loss": 0.9383,
+      "step": 5385
+    },
+    {
+      "epoch": 0.5636839351125066,
+      "grad_norm": 2.0181491721198626,
+      "learning_rate": 8.430298800591939e-06,
+      "loss": 0.9197,
+      "step": 5386
+    },
+    {
+      "epoch": 0.5637885923600209,
+      "grad_norm": 1.827058275655275,
+      "learning_rate": 8.426951191833735e-06,
+      "loss": 0.7931,
+      "step": 5387
+    },
+    {
+      "epoch": 0.5638932496075353,
+      "grad_norm": 1.901008131381671,
+      "learning_rate": 8.42360376382203e-06,
+      "loss": 0.9066,
+      "step": 5388
+    },
+    {
+      "epoch": 0.5639979068550497,
+      "grad_norm": 2.061629426104861,
+      "learning_rate": 8.420256516941446e-06,
+      "loss": 0.9644,
+      "step": 5389
+    },
+    {
+      "epoch": 0.5641025641025641,
+      "grad_norm": 2.0842792186189225,
+      "learning_rate": 8.416909451576592e-06,
+      "loss": 0.8135,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5642072213500785,
+      "grad_norm": 2.3449029492925906,
+      "learning_rate": 8.41356256811205e-06,
+      "loss": 0.9087,
+      "step": 5391
+    },
+    {
+      "epoch": 0.5643118785975929,
+      "grad_norm": 2.382812471439128,
+      "learning_rate": 8.410215866932386e-06,
+      "loss": 0.9713,
+      "step": 5392
+    },
+    {
+      "epoch": 0.5644165358451073,
+      "grad_norm": 2.1076018814211466,
+      "learning_rate": 8.40686934842214e-06,
+      "loss": 1.0884,
+      "step": 5393
+    },
+    {
+      "epoch": 0.5645211930926216,
+      "grad_norm": 2.220481614569876,
+      "learning_rate": 8.403523012965835e-06,
+      "loss": 0.9772,
+      "step": 5394
+    },
+    {
+      "epoch": 0.564625850340136,
+      "grad_norm": 2.0659611056947687,
+      "learning_rate": 8.400176860947975e-06,
+      "loss": 1.0665,
+      "step": 5395
+    },
+    {
+      "epoch": 0.5647305075876504,
+      "grad_norm": 1.8672547133425201,
+      "learning_rate": 8.396830892753036e-06,
+      "loss": 0.9571,
+      "step": 5396
+    },
+    {
+      "epoch": 0.5648351648351648,
+      "grad_norm": 2.1394363159336955,
+      "learning_rate": 8.393485108765477e-06,
+      "loss": 0.7481,
+      "step": 5397
+    },
+    {
+      "epoch": 0.5649398220826792,
+      "grad_norm": 2.007437446941716,
+      "learning_rate": 8.390139509369739e-06,
+      "loss": 0.8622,
+      "step": 5398
+    },
+    {
+      "epoch": 0.5650444793301936,
+      "grad_norm": 2.1993539972983154,
+      "learning_rate": 8.386794094950231e-06,
+      "loss": 1.032,
+      "step": 5399
+    },
+    {
+      "epoch": 0.565149136577708,
+      "grad_norm": 1.9010216623009064,
+      "learning_rate": 8.383448865891353e-06,
+      "loss": 0.8278,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5652537938252225,
+      "grad_norm": 1.8728602015988864,
+      "learning_rate": 8.380103822577476e-06,
+      "loss": 0.7388,
+      "step": 5401
+    },
+    {
+      "epoch": 0.5653584510727367,
+      "grad_norm": 1.9149296409387317,
+      "learning_rate": 8.376758965392956e-06,
+      "loss": 0.9607,
+      "step": 5402
+    },
+    {
+      "epoch": 0.5654631083202512,
+      "grad_norm": 1.8429299678951467,
+      "learning_rate": 8.373414294722117e-06,
+      "loss": 0.9577,
+      "step": 5403
+    },
+    {
+      "epoch": 0.5655677655677656,
+      "grad_norm": 2.0351716866666587,
+      "learning_rate": 8.370069810949276e-06,
+      "loss": 0.9344,
+      "step": 5404
+    },
+    {
+      "epoch": 0.56567242281528,
+      "grad_norm": 2.029063031151813,
+      "learning_rate": 8.366725514458719e-06,
+      "loss": 0.8565,
+      "step": 5405
+    },
+    {
+      "epoch": 0.5657770800627944,
+      "grad_norm": 2.25021340401501,
+      "learning_rate": 8.36338140563471e-06,
+      "loss": 0.9405,
+      "step": 5406
+    },
+    {
+      "epoch": 0.5658817373103088,
+      "grad_norm": 1.9493532133820115,
+      "learning_rate": 8.360037484861497e-06,
+      "loss": 1.0342,
+      "step": 5407
+    },
+    {
+      "epoch": 0.5659863945578232,
+      "grad_norm": 1.9189066069765968,
+      "learning_rate": 8.3566937525233e-06,
+      "loss": 0.9019,
+      "step": 5408
+    },
+    {
+      "epoch": 0.5660910518053375,
+      "grad_norm": 2.2068316926644633,
+      "learning_rate": 8.353350209004323e-06,
+      "loss": 1.089,
+      "step": 5409
+    },
+    {
+      "epoch": 0.5661957090528519,
+      "grad_norm": 2.0845230459581163,
+      "learning_rate": 8.350006854688744e-06,
+      "loss": 0.8706,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5663003663003663,
+      "grad_norm": 1.9447916121641364,
+      "learning_rate": 8.346663689960724e-06,
+      "loss": 0.8611,
+      "step": 5411
+    },
+    {
+      "epoch": 0.5664050235478807,
+      "grad_norm": 2.1579606427265885,
+      "learning_rate": 8.343320715204393e-06,
+      "loss": 0.894,
+      "step": 5412
+    },
+    {
+      "epoch": 0.5665096807953951,
+      "grad_norm": 2.0662389913637087,
+      "learning_rate": 8.339977930803872e-06,
+      "loss": 1.0039,
+      "step": 5413
+    },
+    {
+      "epoch": 0.5666143380429095,
+      "grad_norm": 1.9074458194803643,
+      "learning_rate": 8.336635337143257e-06,
+      "loss": 0.9126,
+      "step": 5414
+    },
+    {
+      "epoch": 0.5667189952904239,
+      "grad_norm": 2.260555516177037,
+      "learning_rate": 8.333292934606611e-06,
+      "loss": 0.9502,
+      "step": 5415
+    },
+    {
+      "epoch": 0.5668236525379382,
+      "grad_norm": 2.5104319898551886,
+      "learning_rate": 8.329950723577985e-06,
+      "loss": 0.8688,
+      "step": 5416
+    },
+    {
+      "epoch": 0.5669283097854526,
+      "grad_norm": 1.938544875865869,
+      "learning_rate": 8.326608704441408e-06,
+      "loss": 0.8114,
+      "step": 5417
+    },
+    {
+      "epoch": 0.567032967032967,
+      "grad_norm": 1.873974736150882,
+      "learning_rate": 8.323266877580881e-06,
+      "loss": 0.9325,
+      "step": 5418
+    },
+    {
+      "epoch": 0.5671376242804814,
+      "grad_norm": 2.316623632030778,
+      "learning_rate": 8.31992524338039e-06,
+      "loss": 0.9622,
+      "step": 5419
+    },
+    {
+      "epoch": 0.5672422815279958,
+      "grad_norm": 1.762820617049393,
+      "learning_rate": 8.316583802223892e-06,
+      "loss": 0.9175,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5673469387755102,
+      "grad_norm": 2.0087856448492736,
+      "learning_rate": 8.313242554495328e-06,
+      "loss": 1.0089,
+      "step": 5421
+    },
+    {
+      "epoch": 0.5674515960230246,
+      "grad_norm": 2.0981791703798978,
+      "learning_rate": 8.309901500578617e-06,
+      "loss": 0.9378,
+      "step": 5422
+    },
+    {
+      "epoch": 0.5675562532705389,
+      "grad_norm": 2.0273142602067216,
+      "learning_rate": 8.306560640857649e-06,
+      "loss": 0.896,
+      "step": 5423
+    },
+    {
+      "epoch": 0.5676609105180533,
+      "grad_norm": 2.2220399006402647,
+      "learning_rate": 8.303219975716296e-06,
+      "loss": 0.8499,
+      "step": 5424
+    },
+    {
+      "epoch": 0.5677655677655677,
+      "grad_norm": 1.9703508493150457,
+      "learning_rate": 8.299879505538407e-06,
+      "loss": 0.9592,
+      "step": 5425
+    },
+    {
+      "epoch": 0.5678702250130822,
+      "grad_norm": 2.105250742608055,
+      "learning_rate": 8.29653923070781e-06,
+      "loss": 0.8972,
+      "step": 5426
+    },
+    {
+      "epoch": 0.5679748822605966,
+      "grad_norm": 2.030286562075985,
+      "learning_rate": 8.293199151608307e-06,
+      "loss": 0.9613,
+      "step": 5427
+    },
+    {
+      "epoch": 0.568079539508111,
+      "grad_norm": 2.3379239596475685,
+      "learning_rate": 8.289859268623685e-06,
+      "loss": 0.9236,
+      "step": 5428
+    },
+    {
+      "epoch": 0.5681841967556254,
+      "grad_norm": 1.9841324284776523,
+      "learning_rate": 8.286519582137694e-06,
+      "loss": 1.055,
+      "step": 5429
+    },
+    {
+      "epoch": 0.5682888540031397,
+      "grad_norm": 2.4795485146638647,
+      "learning_rate": 8.28318009253408e-06,
+      "loss": 0.9924,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5683935112506541,
+      "grad_norm": 2.320539277472763,
+      "learning_rate": 8.279840800196555e-06,
+      "loss": 1.0494,
+      "step": 5431
+    },
+    {
+      "epoch": 0.5684981684981685,
+      "grad_norm": 1.8100197495806924,
+      "learning_rate": 8.276501705508808e-06,
+      "loss": 0.8236,
+      "step": 5432
+    },
+    {
+      "epoch": 0.5686028257456829,
+      "grad_norm": 1.9211836188740459,
+      "learning_rate": 8.27316280885451e-06,
+      "loss": 0.9775,
+      "step": 5433
+    },
+    {
+      "epoch": 0.5687074829931973,
+      "grad_norm": 1.7835418034139316,
+      "learning_rate": 8.269824110617306e-06,
+      "loss": 0.9137,
+      "step": 5434
+    },
+    {
+      "epoch": 0.5688121402407117,
+      "grad_norm": 2.4840168814546533,
+      "learning_rate": 8.266485611180816e-06,
+      "loss": 1.0129,
+      "step": 5435
+    },
+    {
+      "epoch": 0.5689167974882261,
+      "grad_norm": 2.124908482971047,
+      "learning_rate": 8.263147310928647e-06,
+      "loss": 1.0432,
+      "step": 5436
+    },
+    {
+      "epoch": 0.5690214547357404,
+      "grad_norm": 1.8840247947452105,
+      "learning_rate": 8.259809210244372e-06,
+      "loss": 0.9159,
+      "step": 5437
+    },
+    {
+      "epoch": 0.5691261119832548,
+      "grad_norm": 2.0456780135127426,
+      "learning_rate": 8.256471309511541e-06,
+      "loss": 0.9619,
+      "step": 5438
+    },
+    {
+      "epoch": 0.5692307692307692,
+      "grad_norm": 1.9569879555280116,
+      "learning_rate": 8.253133609113699e-06,
+      "loss": 0.8922,
+      "step": 5439
+    },
+    {
+      "epoch": 0.5693354264782836,
+      "grad_norm": 2.2202855654973868,
+      "learning_rate": 8.249796109434342e-06,
+      "loss": 0.8476,
+      "step": 5440
+    },
+    {
+      "epoch": 0.569440083725798,
+      "grad_norm": 1.8957910452182647,
+      "learning_rate": 8.246458810856963e-06,
+      "loss": 0.974,
+      "step": 5441
+    },
+    {
+      "epoch": 0.5695447409733124,
+      "grad_norm": 2.0379277438037575,
+      "learning_rate": 8.24312171376502e-06,
+      "loss": 0.9315,
+      "step": 5442
+    },
+    {
+      "epoch": 0.5696493982208268,
+      "grad_norm": 2.3174337089710435,
+      "learning_rate": 8.239784818541955e-06,
+      "loss": 0.9172,
+      "step": 5443
+    },
+    {
+      "epoch": 0.5697540554683412,
+      "grad_norm": 1.8964131960152406,
+      "learning_rate": 8.23644812557118e-06,
+      "loss": 0.9638,
+      "step": 5444
+    },
+    {
+      "epoch": 0.5698587127158555,
+      "grad_norm": 1.8962610994511044,
+      "learning_rate": 8.233111635236093e-06,
+      "loss": 0.8465,
+      "step": 5445
+    },
+    {
+      "epoch": 0.5699633699633699,
+      "grad_norm": 2.0617322216840632,
+      "learning_rate": 8.22977534792006e-06,
+      "loss": 0.9347,
+      "step": 5446
+    },
+    {
+      "epoch": 0.5700680272108843,
+      "grad_norm": 2.010417513155231,
+      "learning_rate": 8.226439264006425e-06,
+      "loss": 0.9013,
+      "step": 5447
+    },
+    {
+      "epoch": 0.5701726844583987,
+      "grad_norm": 1.905616574561522,
+      "learning_rate": 8.223103383878517e-06,
+      "loss": 0.9417,
+      "step": 5448
+    },
+    {
+      "epoch": 0.5702773417059132,
+      "grad_norm": 2.0075024867237548,
+      "learning_rate": 8.219767707919632e-06,
+      "loss": 1.0392,
+      "step": 5449
+    },
+    {
+      "epoch": 0.5703819989534276,
+      "grad_norm": 1.8721709785411973,
+      "learning_rate": 8.216432236513048e-06,
+      "loss": 0.9228,
+      "step": 5450
+    },
+    {
+      "epoch": 0.570486656200942,
+      "grad_norm": 1.9218512333873883,
+      "learning_rate": 8.213096970042015e-06,
+      "loss": 1.0,
+      "step": 5451
+    },
+    {
+      "epoch": 0.5705913134484563,
+      "grad_norm": 2.329819893077006,
+      "learning_rate": 8.209761908889762e-06,
+      "loss": 0.9031,
+      "step": 5452
+    },
+    {
+      "epoch": 0.5706959706959707,
+      "grad_norm": 1.8170918140020158,
+      "learning_rate": 8.206427053439497e-06,
+      "loss": 0.8399,
+      "step": 5453
+    },
+    {
+      "epoch": 0.5708006279434851,
+      "grad_norm": 2.195702490756807,
+      "learning_rate": 8.203092404074395e-06,
+      "loss": 0.902,
+      "step": 5454
+    },
+    {
+      "epoch": 0.5709052851909995,
+      "grad_norm": 1.8273594046031365,
+      "learning_rate": 8.199757961177618e-06,
+      "loss": 0.8918,
+      "step": 5455
+    },
+    {
+      "epoch": 0.5710099424385139,
+      "grad_norm": 2.146875048935083,
+      "learning_rate": 8.196423725132305e-06,
+      "loss": 0.9573,
+      "step": 5456
+    },
+    {
+      "epoch": 0.5711145996860283,
+      "grad_norm": 2.1126148763204022,
+      "learning_rate": 8.19308969632156e-06,
+      "loss": 0.8839,
+      "step": 5457
+    },
+    {
+      "epoch": 0.5712192569335427,
+      "grad_norm": 2.390207657529769,
+      "learning_rate": 8.189755875128474e-06,
+      "loss": 1.0406,
+      "step": 5458
+    },
+    {
+      "epoch": 0.571323914181057,
+      "grad_norm": 2.1517068860045128,
+      "learning_rate": 8.186422261936105e-06,
+      "loss": 0.8052,
+      "step": 5459
+    },
+    {
+      "epoch": 0.5714285714285714,
+      "grad_norm": 2.119466613691594,
+      "learning_rate": 8.183088857127496e-06,
+      "loss": 0.9079,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5715332286760858,
+      "grad_norm": 2.231374940946366,
+      "learning_rate": 8.17975566108566e-06,
+      "loss": 0.9486,
+      "step": 5461
+    },
+    {
+      "epoch": 0.5716378859236002,
+      "grad_norm": 2.073842713633685,
+      "learning_rate": 8.176422674193591e-06,
+      "loss": 0.9796,
+      "step": 5462
+    },
+    {
+      "epoch": 0.5717425431711146,
+      "grad_norm": 1.8850162422585321,
+      "learning_rate": 8.173089896834249e-06,
+      "loss": 0.9233,
+      "step": 5463
+    },
+    {
+      "epoch": 0.571847200418629,
+      "grad_norm": 2.2918049824032605,
+      "learning_rate": 8.169757329390581e-06,
+      "loss": 0.9653,
+      "step": 5464
+    },
+    {
+      "epoch": 0.5719518576661434,
+      "grad_norm": 1.987836099412543,
+      "learning_rate": 8.16642497224551e-06,
+      "loss": 0.8883,
+      "step": 5465
+    },
+    {
+      "epoch": 0.5720565149136577,
+      "grad_norm": 1.9580693885487788,
+      "learning_rate": 8.163092825781927e-06,
+      "loss": 1.0531,
+      "step": 5466
+    },
+    {
+      "epoch": 0.5721611721611721,
+      "grad_norm": 2.5063692651334195,
+      "learning_rate": 8.159760890382703e-06,
+      "loss": 0.9574,
+      "step": 5467
+    },
+    {
+      "epoch": 0.5722658294086865,
+      "grad_norm": 2.2807498841062275,
+      "learning_rate": 8.15642916643068e-06,
+      "loss": 1.09,
+      "step": 5468
+    },
+    {
+      "epoch": 0.5723704866562009,
+      "grad_norm": 1.9133632160896512,
+      "learning_rate": 8.153097654308684e-06,
+      "loss": 0.8269,
+      "step": 5469
+    },
+    {
+      "epoch": 0.5724751439037153,
+      "grad_norm": 1.8332089750958485,
+      "learning_rate": 8.149766354399512e-06,
+      "loss": 0.8831,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5725798011512298,
+      "grad_norm": 1.9413274135809737,
+      "learning_rate": 8.146435267085937e-06,
+      "loss": 0.9679,
+      "step": 5471
+    },
+    {
+      "epoch": 0.5726844583987442,
+      "grad_norm": 1.8297455313311937,
+      "learning_rate": 8.143104392750708e-06,
+      "loss": 0.949,
+      "step": 5472
+    },
+    {
+      "epoch": 0.5727891156462585,
+      "grad_norm": 2.0994208439284052,
+      "learning_rate": 8.139773731776545e-06,
+      "loss": 0.9182,
+      "step": 5473
+    },
+    {
+      "epoch": 0.5728937728937729,
+      "grad_norm": 2.2224164168293887,
+      "learning_rate": 8.136443284546153e-06,
+      "loss": 0.8501,
+      "step": 5474
+    },
+    {
+      "epoch": 0.5729984301412873,
+      "grad_norm": 1.9784406564547241,
+      "learning_rate": 8.133113051442207e-06,
+      "loss": 0.9075,
+      "step": 5475
+    },
+    {
+      "epoch": 0.5731030873888017,
+      "grad_norm": 2.1561299853644162,
+      "learning_rate": 8.129783032847353e-06,
+      "loss": 0.8528,
+      "step": 5476
+    },
+    {
+      "epoch": 0.5732077446363161,
+      "grad_norm": 1.8007982019446203,
+      "learning_rate": 8.126453229144224e-06,
+      "loss": 0.9403,
+      "step": 5477
+    },
+    {
+      "epoch": 0.5733124018838305,
+      "grad_norm": 1.944118541502138,
+      "learning_rate": 8.123123640715415e-06,
+      "loss": 0.9122,
+      "step": 5478
+    },
+    {
+      "epoch": 0.5734170591313449,
+      "grad_norm": 2.0045254601484443,
+      "learning_rate": 8.119794267943506e-06,
+      "loss": 0.8893,
+      "step": 5479
+    },
+    {
+      "epoch": 0.5735217163788592,
+      "grad_norm": 1.908937792505844,
+      "learning_rate": 8.116465111211045e-06,
+      "loss": 0.8759,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5736263736263736,
+      "grad_norm": 1.9982276022368928,
+      "learning_rate": 8.113136170900558e-06,
+      "loss": 0.9882,
+      "step": 5481
+    },
+    {
+      "epoch": 0.573731030873888,
+      "grad_norm": 2.0425269282462803,
+      "learning_rate": 8.109807447394555e-06,
+      "loss": 0.9599,
+      "step": 5482
+    },
+    {
+      "epoch": 0.5738356881214024,
+      "grad_norm": 1.8900946053549617,
+      "learning_rate": 8.106478941075506e-06,
+      "loss": 0.8938,
+      "step": 5483
+    },
+    {
+      "epoch": 0.5739403453689168,
+      "grad_norm": 2.0201948859873022,
+      "learning_rate": 8.103150652325867e-06,
+      "loss": 0.9496,
+      "step": 5484
+    },
+    {
+      "epoch": 0.5740450026164312,
+      "grad_norm": 1.914651039584181,
+      "learning_rate": 8.099822581528061e-06,
+      "loss": 0.8073,
+      "step": 5485
+    },
+    {
+      "epoch": 0.5741496598639456,
+      "grad_norm": 1.9592779998675791,
+      "learning_rate": 8.096494729064497e-06,
+      "loss": 0.9203,
+      "step": 5486
+    },
+    {
+      "epoch": 0.57425431711146,
+      "grad_norm": 1.9404520688455333,
+      "learning_rate": 8.093167095317543e-06,
+      "loss": 0.9277,
+      "step": 5487
+    },
+    {
+      "epoch": 0.5743589743589743,
+      "grad_norm": 2.102772910368145,
+      "learning_rate": 8.089839680669557e-06,
+      "loss": 1.0397,
+      "step": 5488
+    },
+    {
+      "epoch": 0.5744636316064887,
+      "grad_norm": 2.0883574055987197,
+      "learning_rate": 8.086512485502864e-06,
+      "loss": 0.8756,
+      "step": 5489
+    },
+    {
+      "epoch": 0.5745682888540031,
+      "grad_norm": 1.885771295192384,
+      "learning_rate": 8.083185510199762e-06,
+      "loss": 0.9681,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5746729461015175,
+      "grad_norm": 2.025131247967418,
+      "learning_rate": 8.079858755142536e-06,
+      "loss": 0.9212,
+      "step": 5491
+    },
+    {
+      "epoch": 0.5747776033490319,
+      "grad_norm": 2.055060910903538,
+      "learning_rate": 8.076532220713429e-06,
+      "loss": 0.9848,
+      "step": 5492
+    },
+    {
+      "epoch": 0.5748822605965463,
+      "grad_norm": 1.6221987425177355,
+      "learning_rate": 8.073205907294669e-06,
+      "loss": 0.8392,
+      "step": 5493
+    },
+    {
+      "epoch": 0.5749869178440608,
+      "grad_norm": 2.112224534296191,
+      "learning_rate": 8.069879815268458e-06,
+      "loss": 0.9113,
+      "step": 5494
+    },
+    {
+      "epoch": 0.575091575091575,
+      "grad_norm": 1.9240692006326605,
+      "learning_rate": 8.066553945016968e-06,
+      "loss": 0.9778,
+      "step": 5495
+    },
+    {
+      "epoch": 0.5751962323390895,
+      "grad_norm": 1.9511742285030678,
+      "learning_rate": 8.063228296922349e-06,
+      "loss": 0.9533,
+      "step": 5496
+    },
+    {
+      "epoch": 0.5753008895866039,
+      "grad_norm": 2.2922538518406,
+      "learning_rate": 8.059902871366724e-06,
+      "loss": 0.8646,
+      "step": 5497
+    },
+    {
+      "epoch": 0.5754055468341183,
+      "grad_norm": 2.126390921205575,
+      "learning_rate": 8.056577668732196e-06,
+      "loss": 0.7988,
+      "step": 5498
+    },
+    {
+      "epoch": 0.5755102040816327,
+      "grad_norm": 2.0074966511095758,
+      "learning_rate": 8.053252689400826e-06,
+      "loss": 0.9166,
+      "step": 5499
+    },
+    {
+      "epoch": 0.5756148613291471,
+      "grad_norm": 2.0738861300544134,
+      "learning_rate": 8.049927933754672e-06,
+      "loss": 1.006,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5757195185766615,
+      "grad_norm": 2.415212004526124,
+      "learning_rate": 8.046603402175755e-06,
+      "loss": 0.9133,
+      "step": 5501
+    },
+    {
+      "epoch": 0.5758241758241758,
+      "grad_norm": 1.9234109153078869,
+      "learning_rate": 8.043279095046064e-06,
+      "loss": 0.8681,
+      "step": 5502
+    },
+    {
+      "epoch": 0.5759288330716902,
+      "grad_norm": 1.889529105841766,
+      "learning_rate": 8.039955012747573e-06,
+      "loss": 0.9011,
+      "step": 5503
+    },
+    {
+      "epoch": 0.5760334903192046,
+      "grad_norm": 2.034277892033179,
+      "learning_rate": 8.036631155662223e-06,
+      "loss": 1.0261,
+      "step": 5504
+    },
+    {
+      "epoch": 0.576138147566719,
+      "grad_norm": 1.8787654537409386,
+      "learning_rate": 8.033307524171935e-06,
+      "loss": 0.9117,
+      "step": 5505
+    },
+    {
+      "epoch": 0.5762428048142334,
+      "grad_norm": 1.9010114330459553,
+      "learning_rate": 8.029984118658598e-06,
+      "loss": 0.9688,
+      "step": 5506
+    },
+    {
+      "epoch": 0.5763474620617478,
+      "grad_norm": 1.8350448320881683,
+      "learning_rate": 8.026660939504076e-06,
+      "loss": 0.9382,
+      "step": 5507
+    },
+    {
+      "epoch": 0.5764521193092622,
+      "grad_norm": 2.3648628400669183,
+      "learning_rate": 8.023337987090219e-06,
+      "loss": 0.7732,
+      "step": 5508
+    },
+    {
+      "epoch": 0.5765567765567765,
+      "grad_norm": 1.9441530508631961,
+      "learning_rate": 8.02001526179883e-06,
+      "loss": 0.9569,
+      "step": 5509
+    },
+    {
+      "epoch": 0.5766614338042909,
+      "grad_norm": 2.148620913574854,
+      "learning_rate": 8.016692764011705e-06,
+      "loss": 1.017,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5767660910518053,
+      "grad_norm": 2.054332121041,
+      "learning_rate": 8.013370494110601e-06,
+      "loss": 0.9866,
+      "step": 5511
+    },
+    {
+      "epoch": 0.5768707482993197,
+      "grad_norm": 1.7926126111942025,
+      "learning_rate": 8.010048452477253e-06,
+      "loss": 0.9684,
+      "step": 5512
+    },
+    {
+      "epoch": 0.5769754055468341,
+      "grad_norm": 1.9449655810781312,
+      "learning_rate": 8.006726639493374e-06,
+      "loss": 0.8362,
+      "step": 5513
+    },
+    {
+      "epoch": 0.5770800627943485,
+      "grad_norm": 2.1002111681115965,
+      "learning_rate": 8.003405055540642e-06,
+      "loss": 0.8563,
+      "step": 5514
+    },
+    {
+      "epoch": 0.5771847200418629,
+      "grad_norm": 1.8369318308457716,
+      "learning_rate": 8.00008370100072e-06,
+      "loss": 0.9018,
+      "step": 5515
+    },
+    {
+      "epoch": 0.5772893772893772,
+      "grad_norm": 2.2436953873311256,
+      "learning_rate": 7.996762576255229e-06,
+      "loss": 0.8905,
+      "step": 5516
+    },
+    {
+      "epoch": 0.5773940345368916,
+      "grad_norm": 2.1273954997385434,
+      "learning_rate": 7.993441681685783e-06,
+      "loss": 0.9944,
+      "step": 5517
+    },
+    {
+      "epoch": 0.577498691784406,
+      "grad_norm": 2.0031275428527837,
+      "learning_rate": 7.990121017673956e-06,
+      "loss": 0.8795,
+      "step": 5518
+    },
+    {
+      "epoch": 0.5776033490319205,
+      "grad_norm": 2.2855770714536683,
+      "learning_rate": 7.986800584601298e-06,
+      "loss": 1.0042,
+      "step": 5519
+    },
+    {
+      "epoch": 0.5777080062794349,
+      "grad_norm": 1.7868487916899245,
+      "learning_rate": 7.983480382849335e-06,
+      "loss": 0.8817,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5778126635269493,
+      "grad_norm": 2.0217427991401986,
+      "learning_rate": 7.980160412799561e-06,
+      "loss": 0.8515,
+      "step": 5521
+    },
+    {
+      "epoch": 0.5779173207744637,
+      "grad_norm": 2.0627313405323697,
+      "learning_rate": 7.976840674833454e-06,
+      "loss": 1.0603,
+      "step": 5522
+    },
+    {
+      "epoch": 0.578021978021978,
+      "grad_norm": 1.8428954962859756,
+      "learning_rate": 7.973521169332452e-06,
+      "loss": 0.857,
+      "step": 5523
+    },
+    {
+      "epoch": 0.5781266352694924,
+      "grad_norm": 2.075289447886697,
+      "learning_rate": 7.970201896677978e-06,
+      "loss": 0.9109,
+      "step": 5524
+    },
+    {
+      "epoch": 0.5782312925170068,
+      "grad_norm": 1.885372734744151,
+      "learning_rate": 7.966882857251416e-06,
+      "loss": 0.9159,
+      "step": 5525
+    },
+    {
+      "epoch": 0.5783359497645212,
+      "grad_norm": 1.6845318805206528,
+      "learning_rate": 7.963564051434139e-06,
+      "loss": 0.7578,
+      "step": 5526
+    },
+    {
+      "epoch": 0.5784406070120356,
+      "grad_norm": 2.0450170748383014,
+      "learning_rate": 7.960245479607482e-06,
+      "loss": 0.9455,
+      "step": 5527
+    },
+    {
+      "epoch": 0.57854526425955,
+      "grad_norm": 2.2160296759535885,
+      "learning_rate": 7.956927142152751e-06,
+      "loss": 1.004,
+      "step": 5528
+    },
+    {
+      "epoch": 0.5786499215070644,
+      "grad_norm": 1.71231827285406,
+      "learning_rate": 7.953609039451238e-06,
+      "loss": 0.7896,
+      "step": 5529
+    },
+    {
+      "epoch": 0.5787545787545788,
+      "grad_norm": 2.0196918037710567,
+      "learning_rate": 7.950291171884192e-06,
+      "loss": 0.8954,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5788592360020931,
+      "grad_norm": 1.9380102025151027,
+      "learning_rate": 7.946973539832845e-06,
+      "loss": 0.8539,
+      "step": 5531
+    },
+    {
+      "epoch": 0.5789638932496075,
+      "grad_norm": 2.079187841186523,
+      "learning_rate": 7.943656143678402e-06,
+      "loss": 0.9011,
+      "step": 5532
+    },
+    {
+      "epoch": 0.5790685504971219,
+      "grad_norm": 2.0873564237094318,
+      "learning_rate": 7.940338983802032e-06,
+      "loss": 0.859,
+      "step": 5533
+    },
+    {
+      "epoch": 0.5791732077446363,
+      "grad_norm": 3.551683446968433,
+      "learning_rate": 7.937022060584891e-06,
+      "loss": 0.8587,
+      "step": 5534
+    },
+    {
+      "epoch": 0.5792778649921507,
+      "grad_norm": 1.786949141630154,
+      "learning_rate": 7.933705374408099e-06,
+      "loss": 0.8939,
+      "step": 5535
+    },
+    {
+      "epoch": 0.5793825222396651,
+      "grad_norm": 2.307443887128846,
+      "learning_rate": 7.930388925652746e-06,
+      "loss": 0.9096,
+      "step": 5536
+    },
+    {
+      "epoch": 0.5794871794871795,
+      "grad_norm": 2.429912604129563,
+      "learning_rate": 7.927072714699903e-06,
+      "loss": 0.9904,
+      "step": 5537
+    },
+    {
+      "epoch": 0.5795918367346938,
+      "grad_norm": 1.929972910007676,
+      "learning_rate": 7.923756741930607e-06,
+      "loss": 0.9509,
+      "step": 5538
+    },
+    {
+      "epoch": 0.5796964939822082,
+      "grad_norm": 2.0723179929400652,
+      "learning_rate": 7.92044100772587e-06,
+      "loss": 0.9514,
+      "step": 5539
+    },
+    {
+      "epoch": 0.5798011512297226,
+      "grad_norm": 1.9377894076263404,
+      "learning_rate": 7.917125512466677e-06,
+      "loss": 0.9522,
+      "step": 5540
+    },
+    {
+      "epoch": 0.579905808477237,
+      "grad_norm": 1.9466085430494366,
+      "learning_rate": 7.913810256533987e-06,
+      "loss": 1.0248,
+      "step": 5541
+    },
+    {
+      "epoch": 0.5800104657247515,
+      "grad_norm": 1.9946195154612312,
+      "learning_rate": 7.910495240308722e-06,
+      "loss": 0.9148,
+      "step": 5542
+    },
+    {
+      "epoch": 0.5801151229722659,
+      "grad_norm": 2.066175895381359,
+      "learning_rate": 7.907180464171792e-06,
+      "loss": 0.992,
+      "step": 5543
+    },
+    {
+      "epoch": 0.5802197802197803,
+      "grad_norm": 2.091752799122247,
+      "learning_rate": 7.903865928504073e-06,
+      "loss": 0.9818,
+      "step": 5544
+    },
+    {
+      "epoch": 0.5803244374672946,
+      "grad_norm": 2.0815270541413438,
+      "learning_rate": 7.900551633686405e-06,
+      "loss": 1.053,
+      "step": 5545
+    },
+    {
+      "epoch": 0.580429094714809,
+      "grad_norm": 1.8552562530126415,
+      "learning_rate": 7.89723758009961e-06,
+      "loss": 0.8185,
+      "step": 5546
+    },
+    {
+      "epoch": 0.5805337519623234,
+      "grad_norm": 2.3294466334249555,
+      "learning_rate": 7.893923768124481e-06,
+      "loss": 0.9063,
+      "step": 5547
+    },
+    {
+      "epoch": 0.5806384092098378,
+      "grad_norm": 2.883079539310358,
+      "learning_rate": 7.890610198141782e-06,
+      "loss": 0.9315,
+      "step": 5548
+    },
+    {
+      "epoch": 0.5807430664573522,
+      "grad_norm": 1.7119553078829457,
+      "learning_rate": 7.887296870532246e-06,
+      "loss": 0.9367,
+      "step": 5549
+    },
+    {
+      "epoch": 0.5808477237048666,
+      "grad_norm": 2.2200861287766114,
+      "learning_rate": 7.88398378567658e-06,
+      "loss": 0.7905,
+      "step": 5550
+    },
+    {
+      "epoch": 0.580952380952381,
+      "grad_norm": 2.039585468725626,
+      "learning_rate": 7.880670943955467e-06,
+      "loss": 1.0008,
+      "step": 5551
+    },
+    {
+      "epoch": 0.5810570381998953,
+      "grad_norm": 1.9366320130825088,
+      "learning_rate": 7.877358345749562e-06,
+      "loss": 0.8768,
+      "step": 5552
+    },
+    {
+      "epoch": 0.5811616954474097,
+      "grad_norm": 2.056342630692964,
+      "learning_rate": 7.874045991439484e-06,
+      "loss": 0.9288,
+      "step": 5553
+    },
+    {
+      "epoch": 0.5812663526949241,
+      "grad_norm": 2.113800312709259,
+      "learning_rate": 7.870733881405835e-06,
+      "loss": 0.9684,
+      "step": 5554
+    },
+    {
+      "epoch": 0.5813710099424385,
+      "grad_norm": 2.0422803075735403,
+      "learning_rate": 7.867422016029174e-06,
+      "loss": 0.8649,
+      "step": 5555
+    },
+    {
+      "epoch": 0.5814756671899529,
+      "grad_norm": 2.208609946344471,
+      "learning_rate": 7.864110395690051e-06,
+      "loss": 1.0076,
+      "step": 5556
+    },
+    {
+      "epoch": 0.5815803244374673,
+      "grad_norm": 2.2154587482971935,
+      "learning_rate": 7.86079902076897e-06,
+      "loss": 0.7496,
+      "step": 5557
+    },
+    {
+      "epoch": 0.5816849816849817,
+      "grad_norm": 2.0643605224522616,
+      "learning_rate": 7.857487891646422e-06,
+      "loss": 0.8465,
+      "step": 5558
+    },
+    {
+      "epoch": 0.581789638932496,
+      "grad_norm": 1.900572692373446,
+      "learning_rate": 7.854177008702852e-06,
+      "loss": 0.9254,
+      "step": 5559
+    },
+    {
+      "epoch": 0.5818942961800104,
+      "grad_norm": 2.070474926104496,
+      "learning_rate": 7.850866372318697e-06,
+      "loss": 1.0204,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5819989534275248,
+      "grad_norm": 1.9724412431279685,
+      "learning_rate": 7.847555982874355e-06,
+      "loss": 1.0067,
+      "step": 5561
+    },
+    {
+      "epoch": 0.5821036106750392,
+      "grad_norm": 2.0205088862472285,
+      "learning_rate": 7.844245840750194e-06,
+      "loss": 0.9708,
+      "step": 5562
+    },
+    {
+      "epoch": 0.5822082679225536,
+      "grad_norm": 2.1780476787506076,
+      "learning_rate": 7.840935946326554e-06,
+      "loss": 0.8583,
+      "step": 5563
+    },
+    {
+      "epoch": 0.582312925170068,
+      "grad_norm": 2.113023798763882,
+      "learning_rate": 7.83762629998375e-06,
+      "loss": 0.9245,
+      "step": 5564
+    },
+    {
+      "epoch": 0.5824175824175825,
+      "grad_norm": 2.197691507927687,
+      "learning_rate": 7.834316902102072e-06,
+      "loss": 0.8333,
+      "step": 5565
+    },
+    {
+      "epoch": 0.5825222396650968,
+      "grad_norm": 2.1409592615846664,
+      "learning_rate": 7.83100775306177e-06,
+      "loss": 0.9258,
+      "step": 5566
+    },
+    {
+      "epoch": 0.5826268969126112,
+      "grad_norm": 2.045119732160066,
+      "learning_rate": 7.827698853243073e-06,
+      "loss": 0.957,
+      "step": 5567
+    },
+    {
+      "epoch": 0.5827315541601256,
+      "grad_norm": 2.256212640620829,
+      "learning_rate": 7.82439020302618e-06,
+      "loss": 0.9719,
+      "step": 5568
+    },
+    {
+      "epoch": 0.58283621140764,
+      "grad_norm": 1.9793551575101223,
+      "learning_rate": 7.821081802791265e-06,
+      "loss": 1.0033,
+      "step": 5569
+    },
+    {
+      "epoch": 0.5829408686551544,
+      "grad_norm": 1.9336354212815183,
+      "learning_rate": 7.817773652918467e-06,
+      "loss": 0.9265,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5830455259026688,
+      "grad_norm": 2.132258291924315,
+      "learning_rate": 7.814465753787902e-06,
+      "loss": 0.9252,
+      "step": 5571
+    },
+    {
+      "epoch": 0.5831501831501832,
+      "grad_norm": 2.2295413408637854,
+      "learning_rate": 7.81115810577965e-06,
+      "loss": 0.9697,
+      "step": 5572
+    },
+    {
+      "epoch": 0.5832548403976976,
+      "grad_norm": 1.906463312716307,
+      "learning_rate": 7.807850709273772e-06,
+      "loss": 0.8958,
+      "step": 5573
+    },
+    {
+      "epoch": 0.5833594976452119,
+      "grad_norm": 2.1146161586182495,
+      "learning_rate": 7.80454356465029e-06,
+      "loss": 0.9138,
+      "step": 5574
+    },
+    {
+      "epoch": 0.5834641548927263,
+      "grad_norm": 2.3811043954927906,
+      "learning_rate": 7.801236672289204e-06,
+      "loss": 1.0349,
+      "step": 5575
+    },
+    {
+      "epoch": 0.5835688121402407,
+      "grad_norm": 2.108644748299407,
+      "learning_rate": 7.79793003257048e-06,
+      "loss": 0.839,
+      "step": 5576
+    },
+    {
+      "epoch": 0.5836734693877551,
+      "grad_norm": 2.1813751914717234,
+      "learning_rate": 7.794623645874057e-06,
+      "loss": 0.99,
+      "step": 5577
+    },
+    {
+      "epoch": 0.5837781266352695,
+      "grad_norm": 1.8978749173293086,
+      "learning_rate": 7.791317512579853e-06,
+      "loss": 0.9008,
+      "step": 5578
+    },
+    {
+      "epoch": 0.5838827838827839,
+      "grad_norm": 2.0693475512491664,
+      "learning_rate": 7.78801163306774e-06,
+      "loss": 0.9453,
+      "step": 5579
+    },
+    {
+      "epoch": 0.5839874411302983,
+      "grad_norm": 2.3044788296228194,
+      "learning_rate": 7.784706007717578e-06,
+      "loss": 0.948,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5840920983778126,
+      "grad_norm": 1.8045170295820876,
+      "learning_rate": 7.781400636909184e-06,
+      "loss": 0.9382,
+      "step": 5581
+    },
+    {
+      "epoch": 0.584196755625327,
+      "grad_norm": 2.483078766173217,
+      "learning_rate": 7.778095521022357e-06,
+      "loss": 0.8534,
+      "step": 5582
+    },
+    {
+      "epoch": 0.5843014128728414,
+      "grad_norm": 2.1631414208383806,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.8908,
+      "step": 5583
+    },
+    {
+      "epoch": 0.5844060701203558,
+      "grad_norm": 2.168390803098506,
+      "learning_rate": 7.771486055532424e-06,
+      "loss": 0.9576,
+      "step": 5584
+    },
+    {
+      "epoch": 0.5845107273678702,
+      "grad_norm": 2.123503361696419,
+      "learning_rate": 7.768181706688754e-06,
+      "loss": 0.9218,
+      "step": 5585
+    },
+    {
+      "epoch": 0.5846153846153846,
+      "grad_norm": 2.1013306321139487,
+      "learning_rate": 7.764877614285533e-06,
+      "loss": 0.8682,
+      "step": 5586
+    },
+    {
+      "epoch": 0.584720041862899,
+      "grad_norm": 2.255480731355252,
+      "learning_rate": 7.76157377870241e-06,
+      "loss": 1.0501,
+      "step": 5587
+    },
+    {
+      "epoch": 0.5848246991104133,
+      "grad_norm": 1.945562700084701,
+      "learning_rate": 7.758270200318995e-06,
+      "loss": 0.8875,
+      "step": 5588
+    },
+    {
+      "epoch": 0.5849293563579278,
+      "grad_norm": 2.1195133177923235,
+      "learning_rate": 7.754966879514878e-06,
+      "loss": 0.8932,
+      "step": 5589
+    },
+    {
+      "epoch": 0.5850340136054422,
+      "grad_norm": 2.1209423994658114,
+      "learning_rate": 7.75166381666962e-06,
+      "loss": 1.0249,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5851386708529566,
+      "grad_norm": 2.009199929745769,
+      "learning_rate": 7.748361012162743e-06,
+      "loss": 0.9536,
+      "step": 5591
+    },
+    {
+      "epoch": 0.585243328100471,
+      "grad_norm": 1.8131883322530284,
+      "learning_rate": 7.745058466373754e-06,
+      "loss": 0.7916,
+      "step": 5592
+    },
+    {
+      "epoch": 0.5853479853479854,
+      "grad_norm": 2.318244040850203,
+      "learning_rate": 7.741756179682116e-06,
+      "loss": 0.8724,
+      "step": 5593
+    },
+    {
+      "epoch": 0.5854526425954998,
+      "grad_norm": 2.1235713244627523,
+      "learning_rate": 7.738454152467267e-06,
+      "loss": 0.9192,
+      "step": 5594
+    },
+    {
+      "epoch": 0.5855572998430141,
+      "grad_norm": 2.2040042407951987,
+      "learning_rate": 7.735152385108627e-06,
+      "loss": 1.007,
+      "step": 5595
+    },
+    {
+      "epoch": 0.5856619570905285,
+      "grad_norm": 2.292748328386475,
+      "learning_rate": 7.731850877985567e-06,
+      "loss": 0.9257,
+      "step": 5596
+    },
+    {
+      "epoch": 0.5857666143380429,
+      "grad_norm": 2.077810847731391,
+      "learning_rate": 7.728549631477441e-06,
+      "loss": 0.9772,
+      "step": 5597
+    },
+    {
+      "epoch": 0.5858712715855573,
+      "grad_norm": 2.0349358401554727,
+      "learning_rate": 7.725248645963565e-06,
+      "loss": 0.9287,
+      "step": 5598
+    },
+    {
+      "epoch": 0.5859759288330717,
+      "grad_norm": 2.4781346688637935,
+      "learning_rate": 7.721947921823232e-06,
+      "loss": 0.9193,
+      "step": 5599
+    },
+    {
+      "epoch": 0.5860805860805861,
+      "grad_norm": 1.9931375215068494,
+      "learning_rate": 7.7186474594357e-06,
+      "loss": 0.9548,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5861852433281005,
+      "grad_norm": 1.7997207751821085,
+      "learning_rate": 7.715347259180202e-06,
+      "loss": 0.9453,
+      "step": 5601
+    },
+    {
+      "epoch": 0.5862899005756148,
+      "grad_norm": 2.010052931739926,
+      "learning_rate": 7.712047321435932e-06,
+      "loss": 1.0162,
+      "step": 5602
+    },
+    {
+      "epoch": 0.5863945578231292,
+      "grad_norm": 2.333239467555819,
+      "learning_rate": 7.708747646582061e-06,
+      "loss": 0.9328,
+      "step": 5603
+    },
+    {
+      "epoch": 0.5864992150706436,
+      "grad_norm": 2.10594950324619,
+      "learning_rate": 7.705448234997735e-06,
+      "loss": 0.9705,
+      "step": 5604
+    },
+    {
+      "epoch": 0.586603872318158,
+      "grad_norm": 2.166799778137154,
+      "learning_rate": 7.702149087062056e-06,
+      "loss": 0.9954,
+      "step": 5605
+    },
+    {
+      "epoch": 0.5867085295656724,
+      "grad_norm": 2.0918172155366177,
+      "learning_rate": 7.698850203154108e-06,
+      "loss": 0.839,
+      "step": 5606
+    },
+    {
+      "epoch": 0.5868131868131868,
+      "grad_norm": 1.9671646846087965,
+      "learning_rate": 7.695551583652935e-06,
+      "loss": 0.8696,
+      "step": 5607
+    },
+    {
+      "epoch": 0.5869178440607012,
+      "grad_norm": 2.2227576188186267,
+      "learning_rate": 7.692253228937558e-06,
+      "loss": 1.0203,
+      "step": 5608
+    },
+    {
+      "epoch": 0.5870225013082155,
+      "grad_norm": 2.2068097944504164,
+      "learning_rate": 7.688955139386961e-06,
+      "loss": 0.9753,
+      "step": 5609
+    },
+    {
+      "epoch": 0.5871271585557299,
+      "grad_norm": 2.439406039090037,
+      "learning_rate": 7.685657315380105e-06,
+      "loss": 0.9711,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5872318158032444,
+      "grad_norm": 2.1525531876859767,
+      "learning_rate": 7.682359757295911e-06,
+      "loss": 1.005,
+      "step": 5611
+    },
+    {
+      "epoch": 0.5873364730507588,
+      "grad_norm": 2.036134319104039,
+      "learning_rate": 7.679062465513282e-06,
+      "loss": 0.9106,
+      "step": 5612
+    },
+    {
+      "epoch": 0.5874411302982732,
+      "grad_norm": 1.980663244945154,
+      "learning_rate": 7.675765440411082e-06,
+      "loss": 1.0143,
+      "step": 5613
+    },
+    {
+      "epoch": 0.5875457875457876,
+      "grad_norm": 1.9155213798006931,
+      "learning_rate": 7.672468682368145e-06,
+      "loss": 0.8642,
+      "step": 5614
+    },
+    {
+      "epoch": 0.587650444793302,
+      "grad_norm": 2.0336061775870693,
+      "learning_rate": 7.669172191763273e-06,
+      "loss": 0.9126,
+      "step": 5615
+    },
+    {
+      "epoch": 0.5877551020408164,
+      "grad_norm": 1.8849398925253606,
+      "learning_rate": 7.665875968975243e-06,
+      "loss": 0.7622,
+      "step": 5616
+    },
+    {
+      "epoch": 0.5878597592883307,
+      "grad_norm": 2.325465135874042,
+      "learning_rate": 7.662580014382794e-06,
+      "loss": 0.9221,
+      "step": 5617
+    },
+    {
+      "epoch": 0.5879644165358451,
+      "grad_norm": 2.315387773578512,
+      "learning_rate": 7.659284328364642e-06,
+      "loss": 0.854,
+      "step": 5618
+    },
+    {
+      "epoch": 0.5880690737833595,
+      "grad_norm": 2.520796475508092,
+      "learning_rate": 7.655988911299463e-06,
+      "loss": 0.9864,
+      "step": 5619
+    },
+    {
+      "epoch": 0.5881737310308739,
+      "grad_norm": 1.9719906781148255,
+      "learning_rate": 7.652693763565909e-06,
+      "loss": 1.0244,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5882783882783883,
+      "grad_norm": 2.1962848838041085,
+      "learning_rate": 7.649398885542603e-06,
+      "loss": 0.9411,
+      "step": 5621
+    },
+    {
+      "epoch": 0.5883830455259027,
+      "grad_norm": 2.0497329566693945,
+      "learning_rate": 7.64610427760813e-06,
+      "loss": 0.9327,
+      "step": 5622
+    },
+    {
+      "epoch": 0.5884877027734171,
+      "grad_norm": 2.3462319809252232,
+      "learning_rate": 7.642809940141047e-06,
+      "loss": 1.0223,
+      "step": 5623
+    },
+    {
+      "epoch": 0.5885923600209314,
+      "grad_norm": 1.7385446984243167,
+      "learning_rate": 7.639515873519882e-06,
+      "loss": 0.8747,
+      "step": 5624
+    },
+    {
+      "epoch": 0.5886970172684458,
+      "grad_norm": 2.0670189309137323,
+      "learning_rate": 7.636222078123126e-06,
+      "loss": 0.9075,
+      "step": 5625
+    },
+    {
+      "epoch": 0.5888016745159602,
+      "grad_norm": 1.9507533341417447,
+      "learning_rate": 7.632928554329246e-06,
+      "loss": 0.9021,
+      "step": 5626
+    },
+    {
+      "epoch": 0.5889063317634746,
+      "grad_norm": 2.3873787735902074,
+      "learning_rate": 7.629635302516673e-06,
+      "loss": 1.0425,
+      "step": 5627
+    },
+    {
+      "epoch": 0.589010989010989,
+      "grad_norm": 2.252557905765437,
+      "learning_rate": 7.626342323063811e-06,
+      "loss": 0.9155,
+      "step": 5628
+    },
+    {
+      "epoch": 0.5891156462585034,
+      "grad_norm": 1.9671306821641084,
+      "learning_rate": 7.623049616349023e-06,
+      "loss": 0.8436,
+      "step": 5629
+    },
+    {
+      "epoch": 0.5892203035060178,
+      "grad_norm": 2.228355294848782,
+      "learning_rate": 7.619757182750655e-06,
+      "loss": 0.9464,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5893249607535321,
+      "grad_norm": 1.944122763671574,
+      "learning_rate": 7.616465022647015e-06,
+      "loss": 0.917,
+      "step": 5631
+    },
+    {
+      "epoch": 0.5894296180010465,
+      "grad_norm": 2.0093474365195956,
+      "learning_rate": 7.613173136416373e-06,
+      "loss": 0.9609,
+      "step": 5632
+    },
+    {
+      "epoch": 0.589534275248561,
+      "grad_norm": 2.3171664214346746,
+      "learning_rate": 7.6098815244369785e-06,
+      "loss": 0.9894,
+      "step": 5633
+    },
+    {
+      "epoch": 0.5896389324960754,
+      "grad_norm": 2.0398463213152893,
+      "learning_rate": 7.60659018708704e-06,
+      "loss": 0.9755,
+      "step": 5634
+    },
+    {
+      "epoch": 0.5897435897435898,
+      "grad_norm": 1.974866296695783,
+      "learning_rate": 7.603299124744743e-06,
+      "loss": 0.9829,
+      "step": 5635
+    },
+    {
+      "epoch": 0.5898482469911042,
+      "grad_norm": 1.8436780153736634,
+      "learning_rate": 7.600008337788233e-06,
+      "loss": 0.8847,
+      "step": 5636
+    },
+    {
+      "epoch": 0.5899529042386186,
+      "grad_norm": 1.7596168571097486,
+      "learning_rate": 7.596717826595627e-06,
+      "loss": 0.8302,
+      "step": 5637
+    },
+    {
+      "epoch": 0.5900575614861329,
+      "grad_norm": 2.286365759016598,
+      "learning_rate": 7.593427591545021e-06,
+      "loss": 0.9302,
+      "step": 5638
+    },
+    {
+      "epoch": 0.5901622187336473,
+      "grad_norm": 2.069205329083003,
+      "learning_rate": 7.5901376330144606e-06,
+      "loss": 0.8283,
+      "step": 5639
+    },
+    {
+      "epoch": 0.5902668759811617,
+      "grad_norm": 2.241407089031377,
+      "learning_rate": 7.586847951381973e-06,
+      "loss": 0.9715,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5903715332286761,
+      "grad_norm": 2.0834114666515267,
+      "learning_rate": 7.583558547025546e-06,
+      "loss": 0.9345,
+      "step": 5641
+    },
+    {
+      "epoch": 0.5904761904761905,
+      "grad_norm": 2.1953788167615262,
+      "learning_rate": 7.580269420323143e-06,
+      "loss": 1.0446,
+      "step": 5642
+    },
+    {
+      "epoch": 0.5905808477237049,
+      "grad_norm": 2.2480917530067375,
+      "learning_rate": 7.5769805716526876e-06,
+      "loss": 0.8577,
+      "step": 5643
+    },
+    {
+      "epoch": 0.5906855049712193,
+      "grad_norm": 2.4677953070932155,
+      "learning_rate": 7.5736920013920756e-06,
+      "loss": 0.9092,
+      "step": 5644
+    },
+    {
+      "epoch": 0.5907901622187336,
+      "grad_norm": 2.0662054886088934,
+      "learning_rate": 7.570403709919173e-06,
+      "loss": 0.7921,
+      "step": 5645
+    },
+    {
+      "epoch": 0.590894819466248,
+      "grad_norm": 2.1423059288297246,
+      "learning_rate": 7.5671156976118056e-06,
+      "loss": 0.9767,
+      "step": 5646
+    },
+    {
+      "epoch": 0.5909994767137624,
+      "grad_norm": 2.7938569718008646,
+      "learning_rate": 7.563827964847779e-06,
+      "loss": 0.9416,
+      "step": 5647
+    },
+    {
+      "epoch": 0.5911041339612768,
+      "grad_norm": 2.012361091935804,
+      "learning_rate": 7.560540512004859e-06,
+      "loss": 0.9601,
+      "step": 5648
+    },
+    {
+      "epoch": 0.5912087912087912,
+      "grad_norm": 2.025112150721346,
+      "learning_rate": 7.557253339460777e-06,
+      "loss": 1.0345,
+      "step": 5649
+    },
+    {
+      "epoch": 0.5913134484563056,
+      "grad_norm": 2.098963973383834,
+      "learning_rate": 7.5539664475932415e-06,
+      "loss": 0.953,
+      "step": 5650
+    },
+    {
+      "epoch": 0.59141810570382,
+      "grad_norm": 2.155862643447173,
+      "learning_rate": 7.550679836779917e-06,
+      "loss": 0.9,
+      "step": 5651
+    },
+    {
+      "epoch": 0.5915227629513344,
+      "grad_norm": 2.0776024089586214,
+      "learning_rate": 7.547393507398447e-06,
+      "loss": 0.8768,
+      "step": 5652
+    },
+    {
+      "epoch": 0.5916274201988487,
+      "grad_norm": 2.040547845106933,
+      "learning_rate": 7.544107459826433e-06,
+      "loss": 0.885,
+      "step": 5653
+    },
+    {
+      "epoch": 0.5917320774463631,
+      "grad_norm": 1.877629870992842,
+      "learning_rate": 7.5408216944414515e-06,
+      "loss": 0.7986,
+      "step": 5654
+    },
+    {
+      "epoch": 0.5918367346938775,
+      "grad_norm": 2.108404306286894,
+      "learning_rate": 7.537536211621039e-06,
+      "loss": 0.9868,
+      "step": 5655
+    },
+    {
+      "epoch": 0.591941391941392,
+      "grad_norm": 2.3003723833156635,
+      "learning_rate": 7.53425101174271e-06,
+      "loss": 0.854,
+      "step": 5656
+    },
+    {
+      "epoch": 0.5920460491889064,
+      "grad_norm": 2.14897580363839,
+      "learning_rate": 7.530966095183942e-06,
+      "loss": 0.8716,
+      "step": 5657
+    },
+    {
+      "epoch": 0.5921507064364208,
+      "grad_norm": 1.856710380053524,
+      "learning_rate": 7.5276814623221705e-06,
+      "loss": 0.9414,
+      "step": 5658
+    },
+    {
+      "epoch": 0.5922553636839352,
+      "grad_norm": 2.125965075529774,
+      "learning_rate": 7.5243971135348136e-06,
+      "loss": 0.8968,
+      "step": 5659
+    },
+    {
+      "epoch": 0.5923600209314495,
+      "grad_norm": 2.076781557575531,
+      "learning_rate": 7.521113049199244e-06,
+      "loss": 0.9287,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5924646781789639,
+      "grad_norm": 1.9002232671975265,
+      "learning_rate": 7.517829269692815e-06,
+      "loss": 1.0265,
+      "step": 5661
+    },
+    {
+      "epoch": 0.5925693354264783,
+      "grad_norm": 2.222162234459858,
+      "learning_rate": 7.514545775392833e-06,
+      "loss": 0.9957,
+      "step": 5662
+    },
+    {
+      "epoch": 0.5926739926739927,
+      "grad_norm": 2.124845865430392,
+      "learning_rate": 7.511262566676574e-06,
+      "loss": 1.0127,
+      "step": 5663
+    },
+    {
+      "epoch": 0.5927786499215071,
+      "grad_norm": 1.9891156119781586,
+      "learning_rate": 7.507979643921297e-06,
+      "loss": 0.8318,
+      "step": 5664
+    },
+    {
+      "epoch": 0.5928833071690215,
+      "grad_norm": 2.0909557900542133,
+      "learning_rate": 7.504697007504212e-06,
+      "loss": 0.8629,
+      "step": 5665
+    },
+    {
+      "epoch": 0.5929879644165359,
+      "grad_norm": 1.9403378716887565,
+      "learning_rate": 7.501414657802495e-06,
+      "loss": 0.9464,
+      "step": 5666
+    },
+    {
+      "epoch": 0.5930926216640502,
+      "grad_norm": 2.043310271462764,
+      "learning_rate": 7.498132595193304e-06,
+      "loss": 0.8681,
+      "step": 5667
+    },
+    {
+      "epoch": 0.5931972789115646,
+      "grad_norm": 1.9687560488818703,
+      "learning_rate": 7.494850820053745e-06,
+      "loss": 1.0069,
+      "step": 5668
+    },
+    {
+      "epoch": 0.593301936159079,
+      "grad_norm": 2.053403263724564,
+      "learning_rate": 7.491569332760909e-06,
+      "loss": 0.7221,
+      "step": 5669
+    },
+    {
+      "epoch": 0.5934065934065934,
+      "grad_norm": 2.046467258975427,
+      "learning_rate": 7.488288133691839e-06,
+      "loss": 0.9088,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5935112506541078,
+      "grad_norm": 1.9238393365949373,
+      "learning_rate": 7.485007223223556e-06,
+      "loss": 0.8819,
+      "step": 5671
+    },
+    {
+      "epoch": 0.5936159079016222,
+      "grad_norm": 1.6696389669150944,
+      "learning_rate": 7.481726601733037e-06,
+      "loss": 0.7529,
+      "step": 5672
+    },
+    {
+      "epoch": 0.5937205651491366,
+      "grad_norm": 2.59751100553255,
+      "learning_rate": 7.478446269597239e-06,
+      "loss": 0.7676,
+      "step": 5673
+    },
+    {
+      "epoch": 0.5938252223966509,
+      "grad_norm": 1.9990709348617675,
+      "learning_rate": 7.475166227193079e-06,
+      "loss": 0.889,
+      "step": 5674
+    },
+    {
+      "epoch": 0.5939298796441653,
+      "grad_norm": 1.910729649122657,
+      "learning_rate": 7.471886474897436e-06,
+      "loss": 0.9689,
+      "step": 5675
+    },
+    {
+      "epoch": 0.5940345368916797,
+      "grad_norm": 1.9841915683745623,
+      "learning_rate": 7.468607013087164e-06,
+      "loss": 0.9288,
+      "step": 5676
+    },
+    {
+      "epoch": 0.5941391941391941,
+      "grad_norm": 2.0782450749530095,
+      "learning_rate": 7.465327842139074e-06,
+      "loss": 0.8997,
+      "step": 5677
+    },
+    {
+      "epoch": 0.5942438513867085,
+      "grad_norm": 2.1382950383338013,
+      "learning_rate": 7.4620489624299576e-06,
+      "loss": 0.9772,
+      "step": 5678
+    },
+    {
+      "epoch": 0.594348508634223,
+      "grad_norm": 1.9347499610481047,
+      "learning_rate": 7.458770374336557e-06,
+      "loss": 0.8774,
+      "step": 5679
+    },
+    {
+      "epoch": 0.5944531658817374,
+      "grad_norm": 1.943431111602753,
+      "learning_rate": 7.455492078235596e-06,
+      "loss": 0.9336,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5945578231292517,
+      "grad_norm": 1.8838788238338073,
+      "learning_rate": 7.452214074503748e-06,
+      "loss": 0.8452,
+      "step": 5681
+    },
+    {
+      "epoch": 0.5946624803767661,
+      "grad_norm": 1.9609711265415661,
+      "learning_rate": 7.44893636351767e-06,
+      "loss": 0.91,
+      "step": 5682
+    },
+    {
+      "epoch": 0.5947671376242805,
+      "grad_norm": 2.2357068851381894,
+      "learning_rate": 7.445658945653975e-06,
+      "loss": 0.8955,
+      "step": 5683
+    },
+    {
+      "epoch": 0.5948717948717949,
+      "grad_norm": 2.129172955119575,
+      "learning_rate": 7.442381821289248e-06,
+      "loss": 0.8215,
+      "step": 5684
+    },
+    {
+      "epoch": 0.5949764521193093,
+      "grad_norm": 2.0015511967472412,
+      "learning_rate": 7.439104990800032e-06,
+      "loss": 0.9191,
+      "step": 5685
+    },
+    {
+      "epoch": 0.5950811093668237,
+      "grad_norm": 2.0354524088249515,
+      "learning_rate": 7.435828454562846e-06,
+      "loss": 0.8986,
+      "step": 5686
+    },
+    {
+      "epoch": 0.5951857666143381,
+      "grad_norm": 2.539924639282485,
+      "learning_rate": 7.432552212954167e-06,
+      "loss": 0.9566,
+      "step": 5687
+    },
+    {
+      "epoch": 0.5952904238618524,
+      "grad_norm": 2.6018174099109874,
+      "learning_rate": 7.429276266350445e-06,
+      "loss": 0.8285,
+      "step": 5688
+    },
+    {
+      "epoch": 0.5953950811093668,
+      "grad_norm": 2.0533166547957884,
+      "learning_rate": 7.426000615128087e-06,
+      "loss": 0.9661,
+      "step": 5689
+    },
+    {
+      "epoch": 0.5954997383568812,
+      "grad_norm": 1.7827839006990294,
+      "learning_rate": 7.422725259663479e-06,
+      "loss": 0.8243,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5956043956043956,
+      "grad_norm": 2.2481511288263603,
+      "learning_rate": 7.419450200332965e-06,
+      "loss": 0.9338,
+      "step": 5691
+    },
+    {
+      "epoch": 0.59570905285191,
+      "grad_norm": 1.9634491736223318,
+      "learning_rate": 7.416175437512854e-06,
+      "loss": 1.0209,
+      "step": 5692
+    },
+    {
+      "epoch": 0.5958137100994244,
+      "grad_norm": 1.9664186395780534,
+      "learning_rate": 7.412900971579422e-06,
+      "loss": 0.8375,
+      "step": 5693
+    },
+    {
+      "epoch": 0.5959183673469388,
+      "grad_norm": 2.002056184772738,
+      "learning_rate": 7.4096268029089136e-06,
+      "loss": 0.9765,
+      "step": 5694
+    },
+    {
+      "epoch": 0.5960230245944532,
+      "grad_norm": 2.0319582433132224,
+      "learning_rate": 7.406352931877538e-06,
+      "loss": 0.885,
+      "step": 5695
+    },
+    {
+      "epoch": 0.5961276818419675,
+      "grad_norm": 1.9584262931491643,
+      "learning_rate": 7.403079358861466e-06,
+      "loss": 1.0275,
+      "step": 5696
+    },
+    {
+      "epoch": 0.5962323390894819,
+      "grad_norm": 1.9530535654924333,
+      "learning_rate": 7.399806084236843e-06,
+      "loss": 0.7674,
+      "step": 5697
+    },
+    {
+      "epoch": 0.5963369963369963,
+      "grad_norm": 2.022147305775228,
+      "learning_rate": 7.396533108379767e-06,
+      "loss": 0.9294,
+      "step": 5698
+    },
+    {
+      "epoch": 0.5964416535845107,
+      "grad_norm": 2.0733270021944303,
+      "learning_rate": 7.393260431666317e-06,
+      "loss": 0.9985,
+      "step": 5699
+    },
+    {
+      "epoch": 0.5965463108320251,
+      "grad_norm": 2.057304250473014,
+      "learning_rate": 7.38998805447253e-06,
+      "loss": 1.0389,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5966509680795395,
+      "grad_norm": 2.465161068029499,
+      "learning_rate": 7.386715977174406e-06,
+      "loss": 0.9919,
+      "step": 5701
+    },
+    {
+      "epoch": 0.596755625327054,
+      "grad_norm": 2.3904950100409454,
+      "learning_rate": 7.383444200147914e-06,
+      "loss": 0.9943,
+      "step": 5702
+    },
+    {
+      "epoch": 0.5968602825745682,
+      "grad_norm": 2.083028232439757,
+      "learning_rate": 7.380172723768988e-06,
+      "loss": 0.9033,
+      "step": 5703
+    },
+    {
+      "epoch": 0.5969649398220827,
+      "grad_norm": 2.3016492387048535,
+      "learning_rate": 7.376901548413525e-06,
+      "loss": 1.0375,
+      "step": 5704
+    },
+    {
+      "epoch": 0.5970695970695971,
+      "grad_norm": 1.8832286691796583,
+      "learning_rate": 7.373630674457393e-06,
+      "loss": 0.8139,
+      "step": 5705
+    },
+    {
+      "epoch": 0.5971742543171115,
+      "grad_norm": 2.183575586002926,
+      "learning_rate": 7.370360102276419e-06,
+      "loss": 0.9699,
+      "step": 5706
+    },
+    {
+      "epoch": 0.5972789115646259,
+      "grad_norm": 2.067091529022829,
+      "learning_rate": 7.367089832246399e-06,
+      "loss": 0.8456,
+      "step": 5707
+    },
+    {
+      "epoch": 0.5973835688121403,
+      "grad_norm": 1.9426102438871136,
+      "learning_rate": 7.363819864743098e-06,
+      "loss": 0.9009,
+      "step": 5708
+    },
+    {
+      "epoch": 0.5974882260596547,
+      "grad_norm": 1.990935623039761,
+      "learning_rate": 7.360550200142238e-06,
+      "loss": 0.889,
+      "step": 5709
+    },
+    {
+      "epoch": 0.597592883307169,
+      "grad_norm": 1.7709079037998374,
+      "learning_rate": 7.357280838819512e-06,
+      "loss": 0.9062,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5976975405546834,
+      "grad_norm": 2.2877813592159004,
+      "learning_rate": 7.354011781150572e-06,
+      "loss": 1.0542,
+      "step": 5711
+    },
+    {
+      "epoch": 0.5978021978021978,
+      "grad_norm": 2.020894122061592,
+      "learning_rate": 7.350743027511045e-06,
+      "loss": 0.9956,
+      "step": 5712
+    },
+    {
+      "epoch": 0.5979068550497122,
+      "grad_norm": 1.9546500888309954,
+      "learning_rate": 7.3474745782765125e-06,
+      "loss": 1.0172,
+      "step": 5713
+    },
+    {
+      "epoch": 0.5980115122972266,
+      "grad_norm": 1.9265892498802972,
+      "learning_rate": 7.3442064338225315e-06,
+      "loss": 0.9464,
+      "step": 5714
+    },
+    {
+      "epoch": 0.598116169544741,
+      "grad_norm": 2.2740116881308605,
+      "learning_rate": 7.340938594524611e-06,
+      "loss": 0.935,
+      "step": 5715
+    },
+    {
+      "epoch": 0.5982208267922554,
+      "grad_norm": 2.3159963017309275,
+      "learning_rate": 7.337671060758236e-06,
+      "loss": 0.8863,
+      "step": 5716
+    },
+    {
+      "epoch": 0.5983254840397697,
+      "grad_norm": 2.1779592238981573,
+      "learning_rate": 7.334403832898856e-06,
+      "loss": 0.92,
+      "step": 5717
+    },
+    {
+      "epoch": 0.5984301412872841,
+      "grad_norm": 2.3970607095582284,
+      "learning_rate": 7.331136911321879e-06,
+      "loss": 0.9618,
+      "step": 5718
+    },
+    {
+      "epoch": 0.5985347985347985,
+      "grad_norm": 1.7681357135220832,
+      "learning_rate": 7.327870296402682e-06,
+      "loss": 0.8879,
+      "step": 5719
+    },
+    {
+      "epoch": 0.5986394557823129,
+      "grad_norm": 1.8481004018316487,
+      "learning_rate": 7.324603988516605e-06,
+      "loss": 0.9292,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5987441130298273,
+      "grad_norm": 2.0081982515082624,
+      "learning_rate": 7.321337988038952e-06,
+      "loss": 0.831,
+      "step": 5721
+    },
+    {
+      "epoch": 0.5988487702773417,
+      "grad_norm": 2.053737396236648,
+      "learning_rate": 7.318072295344996e-06,
+      "loss": 0.9926,
+      "step": 5722
+    },
+    {
+      "epoch": 0.5989534275248561,
+      "grad_norm": 1.932122093359507,
+      "learning_rate": 7.314806910809968e-06,
+      "loss": 0.8565,
+      "step": 5723
+    },
+    {
+      "epoch": 0.5990580847723704,
+      "grad_norm": 2.112245623049056,
+      "learning_rate": 7.311541834809066e-06,
+      "loss": 0.9997,
+      "step": 5724
+    },
+    {
+      "epoch": 0.5991627420198848,
+      "grad_norm": 2.2611995197979664,
+      "learning_rate": 7.308277067717464e-06,
+      "loss": 0.9893,
+      "step": 5725
+    },
+    {
+      "epoch": 0.5992673992673992,
+      "grad_norm": 2.1706750569520508,
+      "learning_rate": 7.30501260991028e-06,
+      "loss": 0.902,
+      "step": 5726
+    },
+    {
+      "epoch": 0.5993720565149137,
+      "grad_norm": 2.0515848466057873,
+      "learning_rate": 7.301748461762612e-06,
+      "loss": 0.918,
+      "step": 5727
+    },
+    {
+      "epoch": 0.5994767137624281,
+      "grad_norm": 2.0173947607217504,
+      "learning_rate": 7.298484623649515e-06,
+      "loss": 0.8884,
+      "step": 5728
+    },
+    {
+      "epoch": 0.5995813710099425,
+      "grad_norm": 2.538071049144709,
+      "learning_rate": 7.295221095946012e-06,
+      "loss": 0.9454,
+      "step": 5729
+    },
+    {
+      "epoch": 0.5996860282574569,
+      "grad_norm": 1.9533801934375317,
+      "learning_rate": 7.291957879027087e-06,
+      "loss": 0.9301,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5997906855049712,
+      "grad_norm": 2.25854110503167,
+      "learning_rate": 7.288694973267693e-06,
+      "loss": 1.059,
+      "step": 5731
+    },
+    {
+      "epoch": 0.5998953427524856,
+      "grad_norm": 1.9262169394873385,
+      "learning_rate": 7.2854323790427425e-06,
+      "loss": 0.8526,
+      "step": 5732
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 2.0501402686145234,
+      "learning_rate": 7.28217009672711e-06,
+      "loss": 0.8932,
+      "step": 5733
+    },
+    {
+      "epoch": 0.6001046572475144,
+      "grad_norm": 1.990109963355697,
+      "learning_rate": 7.2789081266956495e-06,
+      "loss": 0.9841,
+      "step": 5734
+    },
+    {
+      "epoch": 0.6002093144950288,
+      "grad_norm": 1.9142264065765036,
+      "learning_rate": 7.2756464693231585e-06,
+      "loss": 0.9663,
+      "step": 5735
+    },
+    {
+      "epoch": 0.6003139717425432,
+      "grad_norm": 2.1083186271489995,
+      "learning_rate": 7.272385124984414e-06,
+      "loss": 0.9541,
+      "step": 5736
+    },
+    {
+      "epoch": 0.6004186289900576,
+      "grad_norm": 2.438641624027379,
+      "learning_rate": 7.269124094054144e-06,
+      "loss": 0.9692,
+      "step": 5737
+    },
+    {
+      "epoch": 0.600523286237572,
+      "grad_norm": 1.7085030156404912,
+      "learning_rate": 7.265863376907054e-06,
+      "loss": 0.8373,
+      "step": 5738
+    },
+    {
+      "epoch": 0.6006279434850863,
+      "grad_norm": 2.175525308596229,
+      "learning_rate": 7.262602973917807e-06,
+      "loss": 0.8631,
+      "step": 5739
+    },
+    {
+      "epoch": 0.6007326007326007,
+      "grad_norm": 2.4383259594069577,
+      "learning_rate": 7.259342885461023e-06,
+      "loss": 0.9384,
+      "step": 5740
+    },
+    {
+      "epoch": 0.6008372579801151,
+      "grad_norm": 2.3558493908846363,
+      "learning_rate": 7.256083111911301e-06,
+      "loss": 0.8584,
+      "step": 5741
+    },
+    {
+      "epoch": 0.6009419152276295,
+      "grad_norm": 1.8065835519696472,
+      "learning_rate": 7.252823653643186e-06,
+      "loss": 0.938,
+      "step": 5742
+    },
+    {
+      "epoch": 0.6010465724751439,
+      "grad_norm": 2.2545538260720392,
+      "learning_rate": 7.249564511031206e-06,
+      "loss": 0.9175,
+      "step": 5743
+    },
+    {
+      "epoch": 0.6011512297226583,
+      "grad_norm": 1.8867293237713263,
+      "learning_rate": 7.246305684449842e-06,
+      "loss": 0.8941,
+      "step": 5744
+    },
+    {
+      "epoch": 0.6012558869701727,
+      "grad_norm": 2.091930423181412,
+      "learning_rate": 7.243047174273534e-06,
+      "loss": 1.0297,
+      "step": 5745
+    },
+    {
+      "epoch": 0.601360544217687,
+      "grad_norm": 1.9561529165386784,
+      "learning_rate": 7.239788980876696e-06,
+      "loss": 0.8983,
+      "step": 5746
+    },
+    {
+      "epoch": 0.6014652014652014,
+      "grad_norm": 2.088099544372334,
+      "learning_rate": 7.236531104633699e-06,
+      "loss": 0.9198,
+      "step": 5747
+    },
+    {
+      "epoch": 0.6015698587127158,
+      "grad_norm": 2.1188876407664066,
+      "learning_rate": 7.233273545918882e-06,
+      "loss": 0.8943,
+      "step": 5748
+    },
+    {
+      "epoch": 0.6016745159602302,
+      "grad_norm": 2.402914355301177,
+      "learning_rate": 7.230016305106539e-06,
+      "loss": 0.9966,
+      "step": 5749
+    },
+    {
+      "epoch": 0.6017791732077447,
+      "grad_norm": 2.157789260501364,
+      "learning_rate": 7.226759382570938e-06,
+      "loss": 1.0866,
+      "step": 5750
+    },
+    {
+      "epoch": 0.6018838304552591,
+      "grad_norm": 2.4322710098817746,
+      "learning_rate": 7.223502778686311e-06,
+      "loss": 0.9455,
+      "step": 5751
+    },
+    {
+      "epoch": 0.6019884877027735,
+      "grad_norm": 2.104737383056769,
+      "learning_rate": 7.220246493826839e-06,
+      "loss": 0.8708,
+      "step": 5752
+    },
+    {
+      "epoch": 0.6020931449502878,
+      "grad_norm": 1.9327478109090237,
+      "learning_rate": 7.216990528366683e-06,
+      "loss": 0.8164,
+      "step": 5753
+    },
+    {
+      "epoch": 0.6021978021978022,
+      "grad_norm": 1.9784215425645908,
+      "learning_rate": 7.213734882679955e-06,
+      "loss": 0.8658,
+      "step": 5754
+    },
+    {
+      "epoch": 0.6023024594453166,
+      "grad_norm": 2.0937291019563755,
+      "learning_rate": 7.21047955714074e-06,
+      "loss": 0.9194,
+      "step": 5755
+    },
+    {
+      "epoch": 0.602407116692831,
+      "grad_norm": 2.2283678360274823,
+      "learning_rate": 7.207224552123075e-06,
+      "loss": 1.0105,
+      "step": 5756
+    },
+    {
+      "epoch": 0.6025117739403454,
+      "grad_norm": 2.086487895813268,
+      "learning_rate": 7.203969868000975e-06,
+      "loss": 0.9044,
+      "step": 5757
+    },
+    {
+      "epoch": 0.6026164311878598,
+      "grad_norm": 2.0822701283652316,
+      "learning_rate": 7.200715505148404e-06,
+      "loss": 0.9292,
+      "step": 5758
+    },
+    {
+      "epoch": 0.6027210884353742,
+      "grad_norm": 2.51397980592967,
+      "learning_rate": 7.1974614639392915e-06,
+      "loss": 1.006,
+      "step": 5759
+    },
+    {
+      "epoch": 0.6028257456828885,
+      "grad_norm": 2.3703132970824274,
+      "learning_rate": 7.194207744747541e-06,
+      "loss": 0.907,
+      "step": 5760
+    },
+    {
+      "epoch": 0.6029304029304029,
+      "grad_norm": 2.063722563737003,
+      "learning_rate": 7.19095434794701e-06,
+      "loss": 0.8732,
+      "step": 5761
+    },
+    {
+      "epoch": 0.6030350601779173,
+      "grad_norm": 2.262710366424598,
+      "learning_rate": 7.187701273911518e-06,
+      "loss": 0.9704,
+      "step": 5762
+    },
+    {
+      "epoch": 0.6031397174254317,
+      "grad_norm": 2.2761471236819615,
+      "learning_rate": 7.184448523014851e-06,
+      "loss": 0.9213,
+      "step": 5763
+    },
+    {
+      "epoch": 0.6032443746729461,
+      "grad_norm": 2.252681155905622,
+      "learning_rate": 7.181196095630756e-06,
+      "loss": 1.0017,
+      "step": 5764
+    },
+    {
+      "epoch": 0.6033490319204605,
+      "grad_norm": 2.218841967236834,
+      "learning_rate": 7.177943992132945e-06,
+      "loss": 1.0078,
+      "step": 5765
+    },
+    {
+      "epoch": 0.6034536891679749,
+      "grad_norm": 2.371167922078118,
+      "learning_rate": 7.174692212895089e-06,
+      "loss": 1.0293,
+      "step": 5766
+    },
+    {
+      "epoch": 0.6035583464154892,
+      "grad_norm": 2.0568882242914177,
+      "learning_rate": 7.1714407582908265e-06,
+      "loss": 0.8454,
+      "step": 5767
+    },
+    {
+      "epoch": 0.6036630036630036,
+      "grad_norm": 1.691696703496195,
+      "learning_rate": 7.168189628693752e-06,
+      "loss": 0.8477,
+      "step": 5768
+    },
+    {
+      "epoch": 0.603767660910518,
+      "grad_norm": 1.8979982946907608,
+      "learning_rate": 7.16493882447743e-06,
+      "loss": 0.9209,
+      "step": 5769
+    },
+    {
+      "epoch": 0.6038723181580324,
+      "grad_norm": 2.5519832703990266,
+      "learning_rate": 7.1616883460153875e-06,
+      "loss": 0.8273,
+      "step": 5770
+    },
+    {
+      "epoch": 0.6039769754055468,
+      "grad_norm": 2.0415668920164434,
+      "learning_rate": 7.158438193681107e-06,
+      "loss": 0.861,
+      "step": 5771
+    },
+    {
+      "epoch": 0.6040816326530613,
+      "grad_norm": 2.0190515557119846,
+      "learning_rate": 7.155188367848038e-06,
+      "loss": 0.8852,
+      "step": 5772
+    },
+    {
+      "epoch": 0.6041862899005757,
+      "grad_norm": 2.018878210032007,
+      "learning_rate": 7.151938868889593e-06,
+      "loss": 0.9771,
+      "step": 5773
+    },
+    {
+      "epoch": 0.60429094714809,
+      "grad_norm": 2.3049004074694337,
+      "learning_rate": 7.148689697179147e-06,
+      "loss": 0.9558,
+      "step": 5774
+    },
+    {
+      "epoch": 0.6043956043956044,
+      "grad_norm": 1.9496529744525466,
+      "learning_rate": 7.145440853090033e-06,
+      "loss": 0.947,
+      "step": 5775
+    },
+    {
+      "epoch": 0.6045002616431188,
+      "grad_norm": 1.7724900507895518,
+      "learning_rate": 7.1421923369955485e-06,
+      "loss": 0.8816,
+      "step": 5776
+    },
+    {
+      "epoch": 0.6046049188906332,
+      "grad_norm": 2.3142724481991936,
+      "learning_rate": 7.138944149268963e-06,
+      "loss": 0.9659,
+      "step": 5777
+    },
+    {
+      "epoch": 0.6047095761381476,
+      "grad_norm": 2.1236464851755743,
+      "learning_rate": 7.1356962902834935e-06,
+      "loss": 0.9831,
+      "step": 5778
+    },
+    {
+      "epoch": 0.604814233385662,
+      "grad_norm": 2.2271103729867083,
+      "learning_rate": 7.132448760412326e-06,
+      "loss": 1.0104,
+      "step": 5779
+    },
+    {
+      "epoch": 0.6049188906331764,
+      "grad_norm": 2.0033029015172774,
+      "learning_rate": 7.129201560028611e-06,
+      "loss": 0.9995,
+      "step": 5780
+    },
+    {
+      "epoch": 0.6050235478806908,
+      "grad_norm": 2.5095302090496037,
+      "learning_rate": 7.1259546895054545e-06,
+      "loss": 0.979,
+      "step": 5781
+    },
+    {
+      "epoch": 0.6051282051282051,
+      "grad_norm": 1.8438079575156054,
+      "learning_rate": 7.122708149215931e-06,
+      "loss": 0.989,
+      "step": 5782
+    },
+    {
+      "epoch": 0.6052328623757195,
+      "grad_norm": 1.7506092086261658,
+      "learning_rate": 7.1194619395330746e-06,
+      "loss": 0.9699,
+      "step": 5783
+    },
+    {
+      "epoch": 0.6053375196232339,
+      "grad_norm": 1.8710351375548784,
+      "learning_rate": 7.116216060829881e-06,
+      "loss": 0.7888,
+      "step": 5784
+    },
+    {
+      "epoch": 0.6054421768707483,
+      "grad_norm": 2.253637141321553,
+      "learning_rate": 7.112970513479303e-06,
+      "loss": 1.0412,
+      "step": 5785
+    },
+    {
+      "epoch": 0.6055468341182627,
+      "grad_norm": 2.1113622507792926,
+      "learning_rate": 7.10972529785427e-06,
+      "loss": 0.9001,
+      "step": 5786
+    },
+    {
+      "epoch": 0.6056514913657771,
+      "grad_norm": 2.116957635415604,
+      "learning_rate": 7.106480414327661e-06,
+      "loss": 0.9627,
+      "step": 5787
+    },
+    {
+      "epoch": 0.6057561486132915,
+      "grad_norm": 1.9655834383321291,
+      "learning_rate": 7.103235863272315e-06,
+      "loss": 0.9217,
+      "step": 5788
+    },
+    {
+      "epoch": 0.6058608058608058,
+      "grad_norm": 1.9253182713618326,
+      "learning_rate": 7.099991645061044e-06,
+      "loss": 0.9115,
+      "step": 5789
+    },
+    {
+      "epoch": 0.6059654631083202,
+      "grad_norm": 1.9112762675976478,
+      "learning_rate": 7.096747760066609e-06,
+      "loss": 0.9361,
+      "step": 5790
+    },
+    {
+      "epoch": 0.6060701203558346,
+      "grad_norm": 1.9544948764185148,
+      "learning_rate": 7.0935042086617444e-06,
+      "loss": 0.8774,
+      "step": 5791
+    },
+    {
+      "epoch": 0.606174777603349,
+      "grad_norm": 2.1298758076771387,
+      "learning_rate": 7.090260991219135e-06,
+      "loss": 0.9291,
+      "step": 5792
+    },
+    {
+      "epoch": 0.6062794348508634,
+      "grad_norm": 1.8407570735350673,
+      "learning_rate": 7.087018108111441e-06,
+      "loss": 0.7909,
+      "step": 5793
+    },
+    {
+      "epoch": 0.6063840920983778,
+      "grad_norm": 2.0086101467681123,
+      "learning_rate": 7.083775559711265e-06,
+      "loss": 0.8417,
+      "step": 5794
+    },
+    {
+      "epoch": 0.6064887493458923,
+      "grad_norm": 2.1799980186333197,
+      "learning_rate": 7.080533346391192e-06,
+      "loss": 0.9765,
+      "step": 5795
+    },
+    {
+      "epoch": 0.6065934065934065,
+      "grad_norm": 2.045424197962744,
+      "learning_rate": 7.0772914685237585e-06,
+      "loss": 0.9536,
+      "step": 5796
+    },
+    {
+      "epoch": 0.606698063840921,
+      "grad_norm": 1.6592809893413245,
+      "learning_rate": 7.074049926481459e-06,
+      "loss": 0.846,
+      "step": 5797
+    },
+    {
+      "epoch": 0.6068027210884354,
+      "grad_norm": 2.6554474976351625,
+      "learning_rate": 7.070808720636754e-06,
+      "loss": 0.9715,
+      "step": 5798
+    },
+    {
+      "epoch": 0.6069073783359498,
+      "grad_norm": 2.0940703129068563,
+      "learning_rate": 7.067567851362066e-06,
+      "loss": 0.9867,
+      "step": 5799
+    },
+    {
+      "epoch": 0.6070120355834642,
+      "grad_norm": 2.1226712056254584,
+      "learning_rate": 7.064327319029775e-06,
+      "loss": 0.9931,
+      "step": 5800
+    },
+    {
+      "epoch": 0.6071166928309786,
+      "grad_norm": 1.7565753791576322,
+      "learning_rate": 7.061087124012227e-06,
+      "loss": 0.8574,
+      "step": 5801
+    },
+    {
+      "epoch": 0.607221350078493,
+      "grad_norm": 2.5046485756656036,
+      "learning_rate": 7.057847266681723e-06,
+      "loss": 0.9254,
+      "step": 5802
+    },
+    {
+      "epoch": 0.6073260073260073,
+      "grad_norm": 2.099170045438125,
+      "learning_rate": 7.054607747410535e-06,
+      "loss": 0.9769,
+      "step": 5803
+    },
+    {
+      "epoch": 0.6074306645735217,
+      "grad_norm": 2.422351961670763,
+      "learning_rate": 7.05136856657089e-06,
+      "loss": 0.9626,
+      "step": 5804
+    },
+    {
+      "epoch": 0.6075353218210361,
+      "grad_norm": 1.934073799795149,
+      "learning_rate": 7.048129724534971e-06,
+      "loss": 0.9296,
+      "step": 5805
+    },
+    {
+      "epoch": 0.6076399790685505,
+      "grad_norm": 1.9490333500220711,
+      "learning_rate": 7.044891221674934e-06,
+      "loss": 0.8619,
+      "step": 5806
+    },
+    {
+      "epoch": 0.6077446363160649,
+      "grad_norm": 2.0467303551021256,
+      "learning_rate": 7.041653058362883e-06,
+      "loss": 0.8767,
+      "step": 5807
+    },
+    {
+      "epoch": 0.6078492935635793,
+      "grad_norm": 2.1579085714167805,
+      "learning_rate": 7.038415234970896e-06,
+      "loss": 0.8894,
+      "step": 5808
+    },
+    {
+      "epoch": 0.6079539508110937,
+      "grad_norm": 1.94349634206974,
+      "learning_rate": 7.035177751870999e-06,
+      "loss": 0.9374,
+      "step": 5809
+    },
+    {
+      "epoch": 0.608058608058608,
+      "grad_norm": 2.0657703345239278,
+      "learning_rate": 7.031940609435192e-06,
+      "loss": 0.8923,
+      "step": 5810
+    },
+    {
+      "epoch": 0.6081632653061224,
+      "grad_norm": 2.070527663098741,
+      "learning_rate": 7.028703808035421e-06,
+      "loss": 0.8972,
+      "step": 5811
+    },
+    {
+      "epoch": 0.6082679225536368,
+      "grad_norm": 1.9175580806828474,
+      "learning_rate": 7.025467348043608e-06,
+      "loss": 0.947,
+      "step": 5812
+    },
+    {
+      "epoch": 0.6083725798011512,
+      "grad_norm": 2.106923878326725,
+      "learning_rate": 7.022231229831629e-06,
+      "loss": 0.9325,
+      "step": 5813
+    },
+    {
+      "epoch": 0.6084772370486656,
+      "grad_norm": 1.9730559307739923,
+      "learning_rate": 7.018995453771318e-06,
+      "loss": 0.8576,
+      "step": 5814
+    },
+    {
+      "epoch": 0.60858189429618,
+      "grad_norm": 2.0508153945924,
+      "learning_rate": 7.01576002023447e-06,
+      "loss": 0.8826,
+      "step": 5815
+    },
+    {
+      "epoch": 0.6086865515436944,
+      "grad_norm": 2.001780253253112,
+      "learning_rate": 7.012524929592848e-06,
+      "loss": 0.8283,
+      "step": 5816
+    },
+    {
+      "epoch": 0.6087912087912087,
+      "grad_norm": 1.8973036803507934,
+      "learning_rate": 7.009290182218166e-06,
+      "loss": 0.9091,
+      "step": 5817
+    },
+    {
+      "epoch": 0.6088958660387231,
+      "grad_norm": 1.9406056643325182,
+      "learning_rate": 7.006055778482106e-06,
+      "loss": 0.8443,
+      "step": 5818
+    },
+    {
+      "epoch": 0.6090005232862375,
+      "grad_norm": 2.111807902749534,
+      "learning_rate": 7.002821718756305e-06,
+      "loss": 0.8516,
+      "step": 5819
+    },
+    {
+      "epoch": 0.609105180533752,
+      "grad_norm": 2.0277118642001994,
+      "learning_rate": 6.999588003412363e-06,
+      "loss": 0.8915,
+      "step": 5820
+    },
+    {
+      "epoch": 0.6092098377812664,
+      "grad_norm": 1.8707711600366936,
+      "learning_rate": 6.996354632821845e-06,
+      "loss": 0.9782,
+      "step": 5821
+    },
+    {
+      "epoch": 0.6093144950287808,
+      "grad_norm": 1.826421423539411,
+      "learning_rate": 6.993121607356267e-06,
+      "loss": 0.8156,
+      "step": 5822
+    },
+    {
+      "epoch": 0.6094191522762952,
+      "grad_norm": 1.8648721842207836,
+      "learning_rate": 6.989888927387114e-06,
+      "loss": 0.8111,
+      "step": 5823
+    },
+    {
+      "epoch": 0.6095238095238096,
+      "grad_norm": 1.9578433557880752,
+      "learning_rate": 6.986656593285824e-06,
+      "loss": 0.7663,
+      "step": 5824
+    },
+    {
+      "epoch": 0.6096284667713239,
+      "grad_norm": 2.3788854431050046,
+      "learning_rate": 6.983424605423801e-06,
+      "loss": 0.8084,
+      "step": 5825
+    },
+    {
+      "epoch": 0.6097331240188383,
+      "grad_norm": 1.9702807385169419,
+      "learning_rate": 6.980192964172404e-06,
+      "loss": 0.9583,
+      "step": 5826
+    },
+    {
+      "epoch": 0.6098377812663527,
+      "grad_norm": 2.1748718305550687,
+      "learning_rate": 6.97696166990296e-06,
+      "loss": 0.9703,
+      "step": 5827
+    },
+    {
+      "epoch": 0.6099424385138671,
+      "grad_norm": 2.2580551673376763,
+      "learning_rate": 6.9737307229867435e-06,
+      "loss": 0.804,
+      "step": 5828
+    },
+    {
+      "epoch": 0.6100470957613815,
+      "grad_norm": 2.2895933354446316,
+      "learning_rate": 6.970500123795004e-06,
+      "loss": 0.9405,
+      "step": 5829
+    },
+    {
+      "epoch": 0.6101517530088959,
+      "grad_norm": 2.0595158690924906,
+      "learning_rate": 6.9672698726989455e-06,
+      "loss": 0.8,
+      "step": 5830
+    },
+    {
+      "epoch": 0.6102564102564103,
+      "grad_norm": 2.3271064177398637,
+      "learning_rate": 6.964039970069722e-06,
+      "loss": 0.9542,
+      "step": 5831
+    },
+    {
+      "epoch": 0.6103610675039246,
+      "grad_norm": 1.7796600145148627,
+      "learning_rate": 6.9608104162784655e-06,
+      "loss": 0.9259,
+      "step": 5832
+    },
+    {
+      "epoch": 0.610465724751439,
+      "grad_norm": 1.973121849768075,
+      "learning_rate": 6.957581211696251e-06,
+      "loss": 0.878,
+      "step": 5833
+    },
+    {
+      "epoch": 0.6105703819989534,
+      "grad_norm": 1.9358767168769708,
+      "learning_rate": 6.954352356694121e-06,
+      "loss": 0.9568,
+      "step": 5834
+    },
+    {
+      "epoch": 0.6106750392464678,
+      "grad_norm": 2.2637017462136377,
+      "learning_rate": 6.9511238516430814e-06,
+      "loss": 1.0218,
+      "step": 5835
+    },
+    {
+      "epoch": 0.6107796964939822,
+      "grad_norm": 1.679462688094085,
+      "learning_rate": 6.947895696914091e-06,
+      "loss": 0.7966,
+      "step": 5836
+    },
+    {
+      "epoch": 0.6108843537414966,
+      "grad_norm": 2.1125768150832873,
+      "learning_rate": 6.944667892878069e-06,
+      "loss": 0.7789,
+      "step": 5837
+    },
+    {
+      "epoch": 0.610989010989011,
+      "grad_norm": 2.1110418210605633,
+      "learning_rate": 6.941440439905904e-06,
+      "loss": 0.9139,
+      "step": 5838
+    },
+    {
+      "epoch": 0.6110936682365253,
+      "grad_norm": 2.4312531473703736,
+      "learning_rate": 6.938213338368431e-06,
+      "loss": 0.9509,
+      "step": 5839
+    },
+    {
+      "epoch": 0.6111983254840397,
+      "grad_norm": 2.215490332569269,
+      "learning_rate": 6.934986588636455e-06,
+      "loss": 0.8285,
+      "step": 5840
+    },
+    {
+      "epoch": 0.6113029827315541,
+      "grad_norm": 2.028628090171237,
+      "learning_rate": 6.931760191080732e-06,
+      "loss": 0.8616,
+      "step": 5841
+    },
+    {
+      "epoch": 0.6114076399790686,
+      "grad_norm": 2.1117389186178945,
+      "learning_rate": 6.928534146071983e-06,
+      "loss": 0.9494,
+      "step": 5842
+    },
+    {
+      "epoch": 0.611512297226583,
+      "grad_norm": 1.8477902899995793,
+      "learning_rate": 6.925308453980887e-06,
+      "loss": 0.8444,
+      "step": 5843
+    },
+    {
+      "epoch": 0.6116169544740974,
+      "grad_norm": 2.2634245815200615,
+      "learning_rate": 6.9220831151780845e-06,
+      "loss": 0.8982,
+      "step": 5844
+    },
+    {
+      "epoch": 0.6117216117216118,
+      "grad_norm": 1.8300503492297004,
+      "learning_rate": 6.918858130034167e-06,
+      "loss": 0.9627,
+      "step": 5845
+    },
+    {
+      "epoch": 0.6118262689691261,
+      "grad_norm": 2.0217643390425013,
+      "learning_rate": 6.915633498919698e-06,
+      "loss": 0.8742,
+      "step": 5846
+    },
+    {
+      "epoch": 0.6119309262166405,
+      "grad_norm": 2.1967587863707707,
+      "learning_rate": 6.9124092222051945e-06,
+      "loss": 0.9867,
+      "step": 5847
+    },
+    {
+      "epoch": 0.6120355834641549,
+      "grad_norm": 1.8375475239699013,
+      "learning_rate": 6.9091853002611296e-06,
+      "loss": 0.897,
+      "step": 5848
+    },
+    {
+      "epoch": 0.6121402407116693,
+      "grad_norm": 2.1183156451380487,
+      "learning_rate": 6.905961733457941e-06,
+      "loss": 0.8377,
+      "step": 5849
+    },
+    {
+      "epoch": 0.6122448979591837,
+      "grad_norm": 1.9783448138880826,
+      "learning_rate": 6.902738522166019e-06,
+      "loss": 0.9428,
+      "step": 5850
+    },
+    {
+      "epoch": 0.6123495552066981,
+      "grad_norm": 2.21998303732816,
+      "learning_rate": 6.899515666755722e-06,
+      "loss": 0.909,
+      "step": 5851
+    },
+    {
+      "epoch": 0.6124542124542125,
+      "grad_norm": 2.3831003494991063,
+      "learning_rate": 6.896293167597361e-06,
+      "loss": 0.9709,
+      "step": 5852
+    },
+    {
+      "epoch": 0.6125588697017268,
+      "grad_norm": 2.0207894472977608,
+      "learning_rate": 6.8930710250612044e-06,
+      "loss": 0.9789,
+      "step": 5853
+    },
+    {
+      "epoch": 0.6126635269492412,
+      "grad_norm": 1.9009358397826535,
+      "learning_rate": 6.889849239517483e-06,
+      "loss": 0.8241,
+      "step": 5854
+    },
+    {
+      "epoch": 0.6127681841967556,
+      "grad_norm": 1.8698070433483918,
+      "learning_rate": 6.886627811336393e-06,
+      "loss": 0.8008,
+      "step": 5855
+    },
+    {
+      "epoch": 0.61287284144427,
+      "grad_norm": 2.005435434602962,
+      "learning_rate": 6.883406740888077e-06,
+      "loss": 0.9513,
+      "step": 5856
+    },
+    {
+      "epoch": 0.6129774986917844,
+      "grad_norm": 1.8480823682591345,
+      "learning_rate": 6.880186028542647e-06,
+      "loss": 0.8963,
+      "step": 5857
+    },
+    {
+      "epoch": 0.6130821559392988,
+      "grad_norm": 1.8958475524506964,
+      "learning_rate": 6.876965674670164e-06,
+      "loss": 0.9899,
+      "step": 5858
+    },
+    {
+      "epoch": 0.6131868131868132,
+      "grad_norm": 2.2557462857033186,
+      "learning_rate": 6.8737456796406594e-06,
+      "loss": 0.9383,
+      "step": 5859
+    },
+    {
+      "epoch": 0.6132914704343275,
+      "grad_norm": 1.8928132503094905,
+      "learning_rate": 6.870526043824111e-06,
+      "loss": 0.8379,
+      "step": 5860
+    },
+    {
+      "epoch": 0.6133961276818419,
+      "grad_norm": 1.938497040014973,
+      "learning_rate": 6.867306767590466e-06,
+      "loss": 1.0145,
+      "step": 5861
+    },
+    {
+      "epoch": 0.6135007849293563,
+      "grad_norm": 1.8779981359486946,
+      "learning_rate": 6.864087851309622e-06,
+      "loss": 0.9063,
+      "step": 5862
+    },
+    {
+      "epoch": 0.6136054421768707,
+      "grad_norm": 1.834130321006479,
+      "learning_rate": 6.860869295351439e-06,
+      "loss": 0.8479,
+      "step": 5863
+    },
+    {
+      "epoch": 0.6137100994243851,
+      "grad_norm": 2.225164974831015,
+      "learning_rate": 6.857651100085742e-06,
+      "loss": 1.0329,
+      "step": 5864
+    },
+    {
+      "epoch": 0.6138147566718996,
+      "grad_norm": 2.2962326966980067,
+      "learning_rate": 6.854433265882301e-06,
+      "loss": 0.9488,
+      "step": 5865
+    },
+    {
+      "epoch": 0.613919413919414,
+      "grad_norm": 1.9342617679397862,
+      "learning_rate": 6.851215793110857e-06,
+      "loss": 0.9494,
+      "step": 5866
+    },
+    {
+      "epoch": 0.6140240711669284,
+      "grad_norm": 2.052176188364496,
+      "learning_rate": 6.8479986821411e-06,
+      "loss": 0.9748,
+      "step": 5867
+    },
+    {
+      "epoch": 0.6141287284144427,
+      "grad_norm": 2.1836996140380402,
+      "learning_rate": 6.844781933342687e-06,
+      "loss": 0.9203,
+      "step": 5868
+    },
+    {
+      "epoch": 0.6142333856619571,
+      "grad_norm": 2.1846424471266843,
+      "learning_rate": 6.841565547085223e-06,
+      "loss": 1.0104,
+      "step": 5869
+    },
+    {
+      "epoch": 0.6143380429094715,
+      "grad_norm": 2.063761945658502,
+      "learning_rate": 6.838349523738283e-06,
+      "loss": 1.0052,
+      "step": 5870
+    },
+    {
+      "epoch": 0.6144427001569859,
+      "grad_norm": 2.345754619202839,
+      "learning_rate": 6.83513386367139e-06,
+      "loss": 0.9294,
+      "step": 5871
+    },
+    {
+      "epoch": 0.6145473574045003,
+      "grad_norm": 1.8945517389181037,
+      "learning_rate": 6.83191856725403e-06,
+      "loss": 0.8727,
+      "step": 5872
+    },
+    {
+      "epoch": 0.6146520146520147,
+      "grad_norm": 2.042059497052054,
+      "learning_rate": 6.828703634855651e-06,
+      "loss": 0.8322,
+      "step": 5873
+    },
+    {
+      "epoch": 0.6147566718995291,
+      "grad_norm": 2.1770922500505354,
+      "learning_rate": 6.825489066845654e-06,
+      "loss": 0.8193,
+      "step": 5874
+    },
+    {
+      "epoch": 0.6148613291470434,
+      "grad_norm": 2.3497422750449126,
+      "learning_rate": 6.822274863593399e-06,
+      "loss": 0.9328,
+      "step": 5875
+    },
+    {
+      "epoch": 0.6149659863945578,
+      "grad_norm": 2.176112444967684,
+      "learning_rate": 6.8190610254682034e-06,
+      "loss": 0.9731,
+      "step": 5876
+    },
+    {
+      "epoch": 0.6150706436420722,
+      "grad_norm": 1.918343615201989,
+      "learning_rate": 6.815847552839343e-06,
+      "loss": 0.8911,
+      "step": 5877
+    },
+    {
+      "epoch": 0.6151753008895866,
+      "grad_norm": 2.01864048793102,
+      "learning_rate": 6.812634446076056e-06,
+      "loss": 1.0051,
+      "step": 5878
+    },
+    {
+      "epoch": 0.615279958137101,
+      "grad_norm": 2.1005849771125953,
+      "learning_rate": 6.809421705547532e-06,
+      "loss": 1.0042,
+      "step": 5879
+    },
+    {
+      "epoch": 0.6153846153846154,
+      "grad_norm": 2.1391206561805016,
+      "learning_rate": 6.806209331622919e-06,
+      "loss": 0.8773,
+      "step": 5880
+    },
+    {
+      "epoch": 0.6154892726321298,
+      "grad_norm": 2.1522010281571227,
+      "learning_rate": 6.802997324671332e-06,
+      "loss": 0.7671,
+      "step": 5881
+    },
+    {
+      "epoch": 0.6155939298796441,
+      "grad_norm": 1.888842162645125,
+      "learning_rate": 6.799785685061832e-06,
+      "loss": 0.8971,
+      "step": 5882
+    },
+    {
+      "epoch": 0.6156985871271585,
+      "grad_norm": 2.2616843959962356,
+      "learning_rate": 6.796574413163446e-06,
+      "loss": 0.7999,
+      "step": 5883
+    },
+    {
+      "epoch": 0.6158032443746729,
+      "grad_norm": 2.2496513510327247,
+      "learning_rate": 6.793363509345151e-06,
+      "loss": 1.0051,
+      "step": 5884
+    },
+    {
+      "epoch": 0.6159079016221873,
+      "grad_norm": 2.0525978260362687,
+      "learning_rate": 6.790152973975892e-06,
+      "loss": 0.9044,
+      "step": 5885
+    },
+    {
+      "epoch": 0.6160125588697017,
+      "grad_norm": 2.01635783084318,
+      "learning_rate": 6.786942807424561e-06,
+      "loss": 1.0252,
+      "step": 5886
+    },
+    {
+      "epoch": 0.6161172161172161,
+      "grad_norm": 2.153628252992972,
+      "learning_rate": 6.783733010060018e-06,
+      "loss": 0.8818,
+      "step": 5887
+    },
+    {
+      "epoch": 0.6162218733647306,
+      "grad_norm": 1.9635936104933305,
+      "learning_rate": 6.78052358225107e-06,
+      "loss": 0.9493,
+      "step": 5888
+    },
+    {
+      "epoch": 0.6163265306122448,
+      "grad_norm": 2.0898861954151733,
+      "learning_rate": 6.777314524366486e-06,
+      "loss": 0.89,
+      "step": 5889
+    },
+    {
+      "epoch": 0.6164311878597593,
+      "grad_norm": 1.9717149097514748,
+      "learning_rate": 6.774105836775e-06,
+      "loss": 0.8447,
+      "step": 5890
+    },
+    {
+      "epoch": 0.6165358451072737,
+      "grad_norm": 1.956228723574464,
+      "learning_rate": 6.770897519845294e-06,
+      "loss": 0.8423,
+      "step": 5891
+    },
+    {
+      "epoch": 0.6166405023547881,
+      "grad_norm": 1.9423670450685322,
+      "learning_rate": 6.767689573946008e-06,
+      "loss": 0.9756,
+      "step": 5892
+    },
+    {
+      "epoch": 0.6167451596023025,
+      "grad_norm": 2.0612156081235313,
+      "learning_rate": 6.7644819994457424e-06,
+      "loss": 0.8537,
+      "step": 5893
+    },
+    {
+      "epoch": 0.6168498168498169,
+      "grad_norm": 2.134024125429241,
+      "learning_rate": 6.761274796713053e-06,
+      "loss": 0.8747,
+      "step": 5894
+    },
+    {
+      "epoch": 0.6169544740973313,
+      "grad_norm": 2.1094468946194276,
+      "learning_rate": 6.7580679661164575e-06,
+      "loss": 0.9272,
+      "step": 5895
+    },
+    {
+      "epoch": 0.6170591313448456,
+      "grad_norm": 1.96616440967844,
+      "learning_rate": 6.754861508024422e-06,
+      "loss": 1.0121,
+      "step": 5896
+    },
+    {
+      "epoch": 0.61716378859236,
+      "grad_norm": 2.2360078591987667,
+      "learning_rate": 6.751655422805381e-06,
+      "loss": 0.9208,
+      "step": 5897
+    },
+    {
+      "epoch": 0.6172684458398744,
+      "grad_norm": 1.9125102631579798,
+      "learning_rate": 6.748449710827711e-06,
+      "loss": 0.9544,
+      "step": 5898
+    },
+    {
+      "epoch": 0.6173731030873888,
+      "grad_norm": 1.9824080317071273,
+      "learning_rate": 6.745244372459765e-06,
+      "loss": 0.9728,
+      "step": 5899
+    },
+    {
+      "epoch": 0.6174777603349032,
+      "grad_norm": 2.128988283347369,
+      "learning_rate": 6.7420394080698385e-06,
+      "loss": 0.8864,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6175824175824176,
+      "grad_norm": 1.8967381718005982,
+      "learning_rate": 6.738834818026187e-06,
+      "loss": 0.8736,
+      "step": 5901
+    },
+    {
+      "epoch": 0.617687074829932,
+      "grad_norm": 2.013465087057385,
+      "learning_rate": 6.7356306026970295e-06,
+      "loss": 1.0341,
+      "step": 5902
+    },
+    {
+      "epoch": 0.6177917320774463,
+      "grad_norm": 2.129706926698933,
+      "learning_rate": 6.7324267624505305e-06,
+      "loss": 0.8628,
+      "step": 5903
+    },
+    {
+      "epoch": 0.6178963893249607,
+      "grad_norm": 1.8225296515950085,
+      "learning_rate": 6.729223297654823e-06,
+      "loss": 0.8744,
+      "step": 5904
+    },
+    {
+      "epoch": 0.6180010465724751,
+      "grad_norm": 2.171135955598889,
+      "learning_rate": 6.726020208677986e-06,
+      "loss": 0.9598,
+      "step": 5905
+    },
+    {
+      "epoch": 0.6181057038199895,
+      "grad_norm": 1.9615288782179878,
+      "learning_rate": 6.722817495888063e-06,
+      "loss": 0.8919,
+      "step": 5906
+    },
+    {
+      "epoch": 0.6182103610675039,
+      "grad_norm": 2.3348478756088964,
+      "learning_rate": 6.719615159653057e-06,
+      "loss": 1.0335,
+      "step": 5907
+    },
+    {
+      "epoch": 0.6183150183150183,
+      "grad_norm": 1.8529500238451388,
+      "learning_rate": 6.716413200340917e-06,
+      "loss": 0.9656,
+      "step": 5908
+    },
+    {
+      "epoch": 0.6184196755625327,
+      "grad_norm": 2.101263167248366,
+      "learning_rate": 6.713211618319559e-06,
+      "loss": 0.8609,
+      "step": 5909
+    },
+    {
+      "epoch": 0.6185243328100472,
+      "grad_norm": 2.082342140819496,
+      "learning_rate": 6.710010413956849e-06,
+      "loss": 0.9781,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6186289900575614,
+      "grad_norm": 2.0995332207635595,
+      "learning_rate": 6.706809587620611e-06,
+      "loss": 0.9303,
+      "step": 5911
+    },
+    {
+      "epoch": 0.6187336473050759,
+      "grad_norm": 1.8736139083894665,
+      "learning_rate": 6.703609139678628e-06,
+      "loss": 0.7238,
+      "step": 5912
+    },
+    {
+      "epoch": 0.6188383045525903,
+      "grad_norm": 2.02936203552506,
+      "learning_rate": 6.700409070498636e-06,
+      "loss": 0.8549,
+      "step": 5913
+    },
+    {
+      "epoch": 0.6189429618001047,
+      "grad_norm": 2.1028471786177434,
+      "learning_rate": 6.697209380448333e-06,
+      "loss": 0.835,
+      "step": 5914
+    },
+    {
+      "epoch": 0.6190476190476191,
+      "grad_norm": 2.220094548261755,
+      "learning_rate": 6.694010069895363e-06,
+      "loss": 0.8769,
+      "step": 5915
+    },
+    {
+      "epoch": 0.6191522762951335,
+      "grad_norm": 1.7219940474883257,
+      "learning_rate": 6.6908111392073395e-06,
+      "loss": 0.7758,
+      "step": 5916
+    },
+    {
+      "epoch": 0.6192569335426479,
+      "grad_norm": 2.0998250638549525,
+      "learning_rate": 6.687612588751827e-06,
+      "loss": 0.9183,
+      "step": 5917
+    },
+    {
+      "epoch": 0.6193615907901622,
+      "grad_norm": 2.0505701485922643,
+      "learning_rate": 6.684414418896341e-06,
+      "loss": 0.8306,
+      "step": 5918
+    },
+    {
+      "epoch": 0.6194662480376766,
+      "grad_norm": 2.2514903555360717,
+      "learning_rate": 6.681216630008363e-06,
+      "loss": 1.044,
+      "step": 5919
+    },
+    {
+      "epoch": 0.619570905285191,
+      "grad_norm": 1.7740449558139142,
+      "learning_rate": 6.678019222455318e-06,
+      "loss": 0.9587,
+      "step": 5920
+    },
+    {
+      "epoch": 0.6196755625327054,
+      "grad_norm": 1.9435874998522282,
+      "learning_rate": 6.674822196604601e-06,
+      "loss": 0.8681,
+      "step": 5921
+    },
+    {
+      "epoch": 0.6197802197802198,
+      "grad_norm": 2.136025694684406,
+      "learning_rate": 6.671625552823553e-06,
+      "loss": 0.9398,
+      "step": 5922
+    },
+    {
+      "epoch": 0.6198848770277342,
+      "grad_norm": 1.923864776761002,
+      "learning_rate": 6.668429291479479e-06,
+      "loss": 0.9224,
+      "step": 5923
+    },
+    {
+      "epoch": 0.6199895342752486,
+      "grad_norm": 2.163924861536042,
+      "learning_rate": 6.665233412939628e-06,
+      "loss": 0.8994,
+      "step": 5924
+    },
+    {
+      "epoch": 0.6200941915227629,
+      "grad_norm": 2.4204765320256647,
+      "learning_rate": 6.6620379175712205e-06,
+      "loss": 1.0021,
+      "step": 5925
+    },
+    {
+      "epoch": 0.6201988487702773,
+      "grad_norm": 2.06754965058645,
+      "learning_rate": 6.6588428057414256e-06,
+      "loss": 0.9148,
+      "step": 5926
+    },
+    {
+      "epoch": 0.6203035060177917,
+      "grad_norm": 2.337630297175959,
+      "learning_rate": 6.655648077817362e-06,
+      "loss": 0.9678,
+      "step": 5927
+    },
+    {
+      "epoch": 0.6204081632653061,
+      "grad_norm": 2.040465726503583,
+      "learning_rate": 6.652453734166118e-06,
+      "loss": 0.9536,
+      "step": 5928
+    },
+    {
+      "epoch": 0.6205128205128205,
+      "grad_norm": 2.092903427609399,
+      "learning_rate": 6.649259775154725e-06,
+      "loss": 0.9262,
+      "step": 5929
+    },
+    {
+      "epoch": 0.6206174777603349,
+      "grad_norm": 1.9715070313368872,
+      "learning_rate": 6.646066201150174e-06,
+      "loss": 1.048,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6207221350078493,
+      "grad_norm": 1.8718505714062148,
+      "learning_rate": 6.6428730125194196e-06,
+      "loss": 0.9048,
+      "step": 5931
+    },
+    {
+      "epoch": 0.6208267922553636,
+      "grad_norm": 1.8924153254665501,
+      "learning_rate": 6.6396802096293555e-06,
+      "loss": 0.951,
+      "step": 5932
+    },
+    {
+      "epoch": 0.620931449502878,
+      "grad_norm": 2.242893217335704,
+      "learning_rate": 6.6364877928468505e-06,
+      "loss": 0.9777,
+      "step": 5933
+    },
+    {
+      "epoch": 0.6210361067503924,
+      "grad_norm": 2.0854393025410247,
+      "learning_rate": 6.633295762538718e-06,
+      "loss": 0.9969,
+      "step": 5934
+    },
+    {
+      "epoch": 0.6211407639979069,
+      "grad_norm": 1.8436423025912454,
+      "learning_rate": 6.6301041190717255e-06,
+      "loss": 0.964,
+      "step": 5935
+    },
+    {
+      "epoch": 0.6212454212454213,
+      "grad_norm": 1.8863362444224359,
+      "learning_rate": 6.626912862812605e-06,
+      "loss": 0.8836,
+      "step": 5936
+    },
+    {
+      "epoch": 0.6213500784929357,
+      "grad_norm": 2.262054897692573,
+      "learning_rate": 6.62372199412803e-06,
+      "loss": 0.8957,
+      "step": 5937
+    },
+    {
+      "epoch": 0.6214547357404501,
+      "grad_norm": 2.201402572775819,
+      "learning_rate": 6.620531513384645e-06,
+      "loss": 0.9112,
+      "step": 5938
+    },
+    {
+      "epoch": 0.6215593929879644,
+      "grad_norm": 2.2199717929386664,
+      "learning_rate": 6.6173414209490395e-06,
+      "loss": 1.1,
+      "step": 5939
+    },
+    {
+      "epoch": 0.6216640502354788,
+      "grad_norm": 2.3970536946249505,
+      "learning_rate": 6.614151717187762e-06,
+      "loss": 0.935,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6217687074829932,
+      "grad_norm": 1.9767721037140733,
+      "learning_rate": 6.610962402467314e-06,
+      "loss": 0.9553,
+      "step": 5941
+    },
+    {
+      "epoch": 0.6218733647305076,
+      "grad_norm": 2.205710848436116,
+      "learning_rate": 6.607773477154156e-06,
+      "loss": 0.8938,
+      "step": 5942
+    },
+    {
+      "epoch": 0.621978021978022,
+      "grad_norm": 1.9654480260758713,
+      "learning_rate": 6.604584941614706e-06,
+      "loss": 0.9242,
+      "step": 5943
+    },
+    {
+      "epoch": 0.6220826792255364,
+      "grad_norm": 1.750180216403602,
+      "learning_rate": 6.601396796215327e-06,
+      "loss": 0.8576,
+      "step": 5944
+    },
+    {
+      "epoch": 0.6221873364730508,
+      "grad_norm": 2.236278725054879,
+      "learning_rate": 6.5982090413223476e-06,
+      "loss": 1.0135,
+      "step": 5945
+    },
+    {
+      "epoch": 0.6222919937205651,
+      "grad_norm": 2.027612274915688,
+      "learning_rate": 6.595021677302044e-06,
+      "loss": 0.939,
+      "step": 5946
+    },
+    {
+      "epoch": 0.6223966509680795,
+      "grad_norm": 2.244750079445713,
+      "learning_rate": 6.591834704520653e-06,
+      "loss": 0.8214,
+      "step": 5947
+    },
+    {
+      "epoch": 0.6225013082155939,
+      "grad_norm": 2.4590294267920476,
+      "learning_rate": 6.5886481233443656e-06,
+      "loss": 0.9864,
+      "step": 5948
+    },
+    {
+      "epoch": 0.6226059654631083,
+      "grad_norm": 2.030725863909143,
+      "learning_rate": 6.585461934139321e-06,
+      "loss": 0.9592,
+      "step": 5949
+    },
+    {
+      "epoch": 0.6227106227106227,
+      "grad_norm": 1.988399104391876,
+      "learning_rate": 6.5822761372716195e-06,
+      "loss": 0.9493,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6228152799581371,
+      "grad_norm": 1.954025794060076,
+      "learning_rate": 6.579090733107323e-06,
+      "loss": 0.9747,
+      "step": 5951
+    },
+    {
+      "epoch": 0.6229199372056515,
+      "grad_norm": 2.1922622101716223,
+      "learning_rate": 6.575905722012435e-06,
+      "loss": 0.9192,
+      "step": 5952
+    },
+    {
+      "epoch": 0.6230245944531659,
+      "grad_norm": 2.1750494542346788,
+      "learning_rate": 6.5727211043529214e-06,
+      "loss": 0.9707,
+      "step": 5953
+    },
+    {
+      "epoch": 0.6231292517006802,
+      "grad_norm": 2.2279016358653125,
+      "learning_rate": 6.5695368804947e-06,
+      "loss": 0.8988,
+      "step": 5954
+    },
+    {
+      "epoch": 0.6232339089481946,
+      "grad_norm": 2.3809272214435104,
+      "learning_rate": 6.566353050803646e-06,
+      "loss": 1.0132,
+      "step": 5955
+    },
+    {
+      "epoch": 0.623338566195709,
+      "grad_norm": 2.3706566423677065,
+      "learning_rate": 6.563169615645586e-06,
+      "loss": 1.0054,
+      "step": 5956
+    },
+    {
+      "epoch": 0.6234432234432234,
+      "grad_norm": 1.7752367380253748,
+      "learning_rate": 6.559986575386307e-06,
+      "loss": 0.862,
+      "step": 5957
+    },
+    {
+      "epoch": 0.6235478806907379,
+      "grad_norm": 1.8311258992017134,
+      "learning_rate": 6.556803930391542e-06,
+      "loss": 0.7931,
+      "step": 5958
+    },
+    {
+      "epoch": 0.6236525379382523,
+      "grad_norm": 2.005969903941888,
+      "learning_rate": 6.553621681026983e-06,
+      "loss": 0.8692,
+      "step": 5959
+    },
+    {
+      "epoch": 0.6237571951857667,
+      "grad_norm": 2.4536539179443575,
+      "learning_rate": 6.550439827658282e-06,
+      "loss": 1.0995,
+      "step": 5960
+    },
+    {
+      "epoch": 0.623861852433281,
+      "grad_norm": 1.829663168954961,
+      "learning_rate": 6.5472583706510375e-06,
+      "loss": 0.9042,
+      "step": 5961
+    },
+    {
+      "epoch": 0.6239665096807954,
+      "grad_norm": 1.8713774810103647,
+      "learning_rate": 6.544077310370806e-06,
+      "loss": 0.8337,
+      "step": 5962
+    },
+    {
+      "epoch": 0.6240711669283098,
+      "grad_norm": 1.7943958864399616,
+      "learning_rate": 6.540896647183098e-06,
+      "loss": 0.8306,
+      "step": 5963
+    },
+    {
+      "epoch": 0.6241758241758242,
+      "grad_norm": 2.142987115852985,
+      "learning_rate": 6.537716381453379e-06,
+      "loss": 0.9367,
+      "step": 5964
+    },
+    {
+      "epoch": 0.6242804814233386,
+      "grad_norm": 2.0012001401459862,
+      "learning_rate": 6.534536513547065e-06,
+      "loss": 0.8729,
+      "step": 5965
+    },
+    {
+      "epoch": 0.624385138670853,
+      "grad_norm": 1.976227019935359,
+      "learning_rate": 6.531357043829529e-06,
+      "loss": 0.8953,
+      "step": 5966
+    },
+    {
+      "epoch": 0.6244897959183674,
+      "grad_norm": 2.044153940630976,
+      "learning_rate": 6.5281779726661e-06,
+      "loss": 0.9151,
+      "step": 5967
+    },
+    {
+      "epoch": 0.6245944531658817,
+      "grad_norm": 2.5746730325720395,
+      "learning_rate": 6.524999300422064e-06,
+      "loss": 0.8603,
+      "step": 5968
+    },
+    {
+      "epoch": 0.6246991104133961,
+      "grad_norm": 1.8639820031464824,
+      "learning_rate": 6.5218210274626495e-06,
+      "loss": 0.868,
+      "step": 5969
+    },
+    {
+      "epoch": 0.6248037676609105,
+      "grad_norm": 2.294972502505636,
+      "learning_rate": 6.5186431541530514e-06,
+      "loss": 0.9509,
+      "step": 5970
+    },
+    {
+      "epoch": 0.6249084249084249,
+      "grad_norm": 2.251140816785971,
+      "learning_rate": 6.515465680858412e-06,
+      "loss": 0.8821,
+      "step": 5971
+    },
+    {
+      "epoch": 0.6250130821559393,
+      "grad_norm": 1.8794029352605075,
+      "learning_rate": 6.512288607943831e-06,
+      "loss": 0.7838,
+      "step": 5972
+    },
+    {
+      "epoch": 0.6251177394034537,
+      "grad_norm": 1.9880704609132205,
+      "learning_rate": 6.509111935774357e-06,
+      "loss": 0.9442,
+      "step": 5973
+    },
+    {
+      "epoch": 0.6252223966509681,
+      "grad_norm": 2.1245881624662153,
+      "learning_rate": 6.5059356647150004e-06,
+      "loss": 0.9828,
+      "step": 5974
+    },
+    {
+      "epoch": 0.6253270538984824,
+      "grad_norm": 2.241601235510337,
+      "learning_rate": 6.502759795130717e-06,
+      "loss": 1.0231,
+      "step": 5975
+    },
+    {
+      "epoch": 0.6254317111459968,
+      "grad_norm": 2.075450945502386,
+      "learning_rate": 6.499584327386422e-06,
+      "loss": 0.9298,
+      "step": 5976
+    },
+    {
+      "epoch": 0.6255363683935112,
+      "grad_norm": 1.9389873315576378,
+      "learning_rate": 6.496409261846988e-06,
+      "loss": 0.7791,
+      "step": 5977
+    },
+    {
+      "epoch": 0.6256410256410256,
+      "grad_norm": 2.1343252505964387,
+      "learning_rate": 6.493234598877229e-06,
+      "loss": 1.1021,
+      "step": 5978
+    },
+    {
+      "epoch": 0.62574568288854,
+      "grad_norm": 1.7296284100239168,
+      "learning_rate": 6.490060338841926e-06,
+      "loss": 0.8877,
+      "step": 5979
+    },
+    {
+      "epoch": 0.6258503401360545,
+      "grad_norm": 2.6208758089241075,
+      "learning_rate": 6.486886482105803e-06,
+      "loss": 1.0258,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6259549973835689,
+      "grad_norm": 1.6310737999103246,
+      "learning_rate": 6.483713029033547e-06,
+      "loss": 0.8419,
+      "step": 5981
+    },
+    {
+      "epoch": 0.6260596546310832,
+      "grad_norm": 1.9090171828142979,
+      "learning_rate": 6.4805399799897905e-06,
+      "loss": 0.9513,
+      "step": 5982
+    },
+    {
+      "epoch": 0.6261643118785976,
+      "grad_norm": 2.0317937367699104,
+      "learning_rate": 6.477367335339129e-06,
+      "loss": 0.8979,
+      "step": 5983
+    },
+    {
+      "epoch": 0.626268969126112,
+      "grad_norm": 1.9744350469086347,
+      "learning_rate": 6.474195095446099e-06,
+      "loss": 0.9687,
+      "step": 5984
+    },
+    {
+      "epoch": 0.6263736263736264,
+      "grad_norm": 2.0167761714344996,
+      "learning_rate": 6.471023260675196e-06,
+      "loss": 0.9428,
+      "step": 5985
+    },
+    {
+      "epoch": 0.6264782836211408,
+      "grad_norm": 1.944143991441491,
+      "learning_rate": 6.467851831390879e-06,
+      "loss": 1.0335,
+      "step": 5986
+    },
+    {
+      "epoch": 0.6265829408686552,
+      "grad_norm": 2.037332698815376,
+      "learning_rate": 6.464680807957547e-06,
+      "loss": 0.9471,
+      "step": 5987
+    },
+    {
+      "epoch": 0.6266875981161696,
+      "grad_norm": 2.1505226752128794,
+      "learning_rate": 6.461510190739555e-06,
+      "loss": 0.9963,
+      "step": 5988
+    },
+    {
+      "epoch": 0.6267922553636839,
+      "grad_norm": 1.970070926677773,
+      "learning_rate": 6.458339980101218e-06,
+      "loss": 0.9675,
+      "step": 5989
+    },
+    {
+      "epoch": 0.6268969126111983,
+      "grad_norm": 1.8959906001986315,
+      "learning_rate": 6.455170176406795e-06,
+      "loss": 0.8347,
+      "step": 5990
+    },
+    {
+      "epoch": 0.6270015698587127,
+      "grad_norm": 1.8273443555041868,
+      "learning_rate": 6.452000780020507e-06,
+      "loss": 0.9721,
+      "step": 5991
+    },
+    {
+      "epoch": 0.6271062271062271,
+      "grad_norm": 2.1115215989855187,
+      "learning_rate": 6.448831791306518e-06,
+      "loss": 0.9706,
+      "step": 5992
+    },
+    {
+      "epoch": 0.6272108843537415,
+      "grad_norm": 1.9586483418053429,
+      "learning_rate": 6.445663210628955e-06,
+      "loss": 0.8123,
+      "step": 5993
+    },
+    {
+      "epoch": 0.6273155416012559,
+      "grad_norm": 2.2343184089780768,
+      "learning_rate": 6.442495038351898e-06,
+      "loss": 0.9436,
+      "step": 5994
+    },
+    {
+      "epoch": 0.6274201988487703,
+      "grad_norm": 2.0306223291943546,
+      "learning_rate": 6.439327274839371e-06,
+      "loss": 0.8866,
+      "step": 5995
+    },
+    {
+      "epoch": 0.6275248560962847,
+      "grad_norm": 2.2375207468319482,
+      "learning_rate": 6.4361599204553604e-06,
+      "loss": 0.9457,
+      "step": 5996
+    },
+    {
+      "epoch": 0.627629513343799,
+      "grad_norm": 1.8709896809018647,
+      "learning_rate": 6.4329929755637975e-06,
+      "loss": 0.9759,
+      "step": 5997
+    },
+    {
+      "epoch": 0.6277341705913134,
+      "grad_norm": 2.0216688575621453,
+      "learning_rate": 6.429826440528575e-06,
+      "loss": 0.888,
+      "step": 5998
+    },
+    {
+      "epoch": 0.6278388278388278,
+      "grad_norm": 1.8399467700878034,
+      "learning_rate": 6.4266603157135295e-06,
+      "loss": 0.8394,
+      "step": 5999
+    },
+    {
+      "epoch": 0.6279434850863422,
+      "grad_norm": 2.1376344086932373,
+      "learning_rate": 6.4234946014824604e-06,
+      "loss": 0.9686,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6280481423338566,
+      "grad_norm": 1.9022841655578102,
+      "learning_rate": 6.42032929819911e-06,
+      "loss": 0.864,
+      "step": 6001
+    },
+    {
+      "epoch": 0.628152799581371,
+      "grad_norm": 2.083437132779924,
+      "learning_rate": 6.417164406227177e-06,
+      "loss": 0.9067,
+      "step": 6002
+    },
+    {
+      "epoch": 0.6282574568288855,
+      "grad_norm": 2.2018711885514612,
+      "learning_rate": 6.413999925930321e-06,
+      "loss": 0.9072,
+      "step": 6003
+    },
+    {
+      "epoch": 0.6283621140763997,
+      "grad_norm": 1.9021036398571836,
+      "learning_rate": 6.410835857672142e-06,
+      "loss": 0.9087,
+      "step": 6004
+    },
+    {
+      "epoch": 0.6284667713239142,
+      "grad_norm": 1.935354772806226,
+      "learning_rate": 6.4076722018162e-06,
+      "loss": 0.9119,
+      "step": 6005
+    },
+    {
+      "epoch": 0.6285714285714286,
+      "grad_norm": 1.9033643377647045,
+      "learning_rate": 6.4045089587260054e-06,
+      "loss": 0.8665,
+      "step": 6006
+    },
+    {
+      "epoch": 0.628676085818943,
+      "grad_norm": 2.3015121040796434,
+      "learning_rate": 6.401346128765018e-06,
+      "loss": 0.9669,
+      "step": 6007
+    },
+    {
+      "epoch": 0.6287807430664574,
+      "grad_norm": 2.203066783924028,
+      "learning_rate": 6.398183712296657e-06,
+      "loss": 0.8435,
+      "step": 6008
+    },
+    {
+      "epoch": 0.6288854003139718,
+      "grad_norm": 1.833007785001033,
+      "learning_rate": 6.395021709684288e-06,
+      "loss": 0.9143,
+      "step": 6009
+    },
+    {
+      "epoch": 0.6289900575614862,
+      "grad_norm": 2.3499962012268814,
+      "learning_rate": 6.391860121291233e-06,
+      "loss": 0.8905,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6290947148090005,
+      "grad_norm": 2.1025471869532764,
+      "learning_rate": 6.388698947480762e-06,
+      "loss": 0.8267,
+      "step": 6011
+    },
+    {
+      "epoch": 0.6291993720565149,
+      "grad_norm": 1.8540577312675692,
+      "learning_rate": 6.385538188616105e-06,
+      "loss": 0.8396,
+      "step": 6012
+    },
+    {
+      "epoch": 0.6293040293040293,
+      "grad_norm": 1.8832622109992163,
+      "learning_rate": 6.3823778450604376e-06,
+      "loss": 0.8341,
+      "step": 6013
+    },
+    {
+      "epoch": 0.6294086865515437,
+      "grad_norm": 2.404866254697502,
+      "learning_rate": 6.37921791717689e-06,
+      "loss": 0.9017,
+      "step": 6014
+    },
+    {
+      "epoch": 0.6295133437990581,
+      "grad_norm": 1.940080093100944,
+      "learning_rate": 6.376058405328543e-06,
+      "loss": 0.964,
+      "step": 6015
+    },
+    {
+      "epoch": 0.6296180010465725,
+      "grad_norm": 2.045127466425197,
+      "learning_rate": 6.372899309878432e-06,
+      "loss": 0.9815,
+      "step": 6016
+    },
+    {
+      "epoch": 0.6297226582940869,
+      "grad_norm": 2.1577986035955505,
+      "learning_rate": 6.369740631189545e-06,
+      "loss": 1.0398,
+      "step": 6017
+    },
+    {
+      "epoch": 0.6298273155416012,
+      "grad_norm": 1.749672091610911,
+      "learning_rate": 6.366582369624816e-06,
+      "loss": 0.7923,
+      "step": 6018
+    },
+    {
+      "epoch": 0.6299319727891156,
+      "grad_norm": 2.109385169681741,
+      "learning_rate": 6.363424525547136e-06,
+      "loss": 1.0015,
+      "step": 6019
+    },
+    {
+      "epoch": 0.63003663003663,
+      "grad_norm": 2.4515134604996156,
+      "learning_rate": 6.360267099319355e-06,
+      "loss": 1.0339,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6301412872841444,
+      "grad_norm": 1.8551476559676408,
+      "learning_rate": 6.357110091304259e-06,
+      "loss": 0.8783,
+      "step": 6021
+    },
+    {
+      "epoch": 0.6302459445316588,
+      "grad_norm": 2.0924036270338786,
+      "learning_rate": 6.353953501864602e-06,
+      "loss": 0.9481,
+      "step": 6022
+    },
+    {
+      "epoch": 0.6303506017791732,
+      "grad_norm": 1.895267228752824,
+      "learning_rate": 6.35079733136308e-06,
+      "loss": 0.9792,
+      "step": 6023
+    },
+    {
+      "epoch": 0.6304552590266876,
+      "grad_norm": 1.8118380058338375,
+      "learning_rate": 6.3476415801623394e-06,
+      "loss": 0.8829,
+      "step": 6024
+    },
+    {
+      "epoch": 0.6305599162742019,
+      "grad_norm": 1.8405134542145882,
+      "learning_rate": 6.344486248624987e-06,
+      "loss": 0.7713,
+      "step": 6025
+    },
+    {
+      "epoch": 0.6306645735217163,
+      "grad_norm": 2.218547626688002,
+      "learning_rate": 6.341331337113573e-06,
+      "loss": 0.9786,
+      "step": 6026
+    },
+    {
+      "epoch": 0.6307692307692307,
+      "grad_norm": 1.882913448926194,
+      "learning_rate": 6.338176845990608e-06,
+      "loss": 0.9793,
+      "step": 6027
+    },
+    {
+      "epoch": 0.6308738880167452,
+      "grad_norm": 1.9751456320726135,
+      "learning_rate": 6.335022775618542e-06,
+      "loss": 0.8445,
+      "step": 6028
+    },
+    {
+      "epoch": 0.6309785452642596,
+      "grad_norm": 2.1125744499378496,
+      "learning_rate": 6.331869126359791e-06,
+      "loss": 0.9355,
+      "step": 6029
+    },
+    {
+      "epoch": 0.631083202511774,
+      "grad_norm": 1.6815396479096147,
+      "learning_rate": 6.328715898576716e-06,
+      "loss": 0.7911,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6311878597592884,
+      "grad_norm": 1.7816434864441257,
+      "learning_rate": 6.3255630926316236e-06,
+      "loss": 0.8058,
+      "step": 6031
+    },
+    {
+      "epoch": 0.6312925170068027,
+      "grad_norm": 2.075892498540421,
+      "learning_rate": 6.3224107088867835e-06,
+      "loss": 0.9163,
+      "step": 6032
+    },
+    {
+      "epoch": 0.6313971742543171,
+      "grad_norm": 1.7175221205045825,
+      "learning_rate": 6.319258747704406e-06,
+      "loss": 0.7745,
+      "step": 6033
+    },
+    {
+      "epoch": 0.6315018315018315,
+      "grad_norm": 1.90865898477721,
+      "learning_rate": 6.316107209446662e-06,
+      "loss": 0.8428,
+      "step": 6034
+    },
+    {
+      "epoch": 0.6316064887493459,
+      "grad_norm": 2.011691584165178,
+      "learning_rate": 6.312956094475666e-06,
+      "loss": 0.9125,
+      "step": 6035
+    },
+    {
+      "epoch": 0.6317111459968603,
+      "grad_norm": 2.117883184326949,
+      "learning_rate": 6.30980540315349e-06,
+      "loss": 1.0146,
+      "step": 6036
+    },
+    {
+      "epoch": 0.6318158032443747,
+      "grad_norm": 1.9763590631964858,
+      "learning_rate": 6.306655135842151e-06,
+      "loss": 0.8664,
+      "step": 6037
+    },
+    {
+      "epoch": 0.6319204604918891,
+      "grad_norm": 1.9200563612501522,
+      "learning_rate": 6.303505292903627e-06,
+      "loss": 0.865,
+      "step": 6038
+    },
+    {
+      "epoch": 0.6320251177394035,
+      "grad_norm": 1.95702742334413,
+      "learning_rate": 6.300355874699839e-06,
+      "loss": 0.984,
+      "step": 6039
+    },
+    {
+      "epoch": 0.6321297749869178,
+      "grad_norm": 1.9835884003747482,
+      "learning_rate": 6.297206881592659e-06,
+      "loss": 0.8374,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6322344322344322,
+      "grad_norm": 1.9891775252939559,
+      "learning_rate": 6.294058313943916e-06,
+      "loss": 0.8428,
+      "step": 6041
+    },
+    {
+      "epoch": 0.6323390894819466,
+      "grad_norm": 2.2448267920259655,
+      "learning_rate": 6.290910172115387e-06,
+      "loss": 0.8765,
+      "step": 6042
+    },
+    {
+      "epoch": 0.632443746729461,
+      "grad_norm": 2.0880819996965405,
+      "learning_rate": 6.287762456468796e-06,
+      "loss": 1.0132,
+      "step": 6043
+    },
+    {
+      "epoch": 0.6325484039769754,
+      "grad_norm": 2.321191163987611,
+      "learning_rate": 6.284615167365826e-06,
+      "loss": 0.9126,
+      "step": 6044
+    },
+    {
+      "epoch": 0.6326530612244898,
+      "grad_norm": 2.157686318063533,
+      "learning_rate": 6.281468305168102e-06,
+      "loss": 0.9658,
+      "step": 6045
+    },
+    {
+      "epoch": 0.6327577184720042,
+      "grad_norm": 1.8521424353333522,
+      "learning_rate": 6.278321870237207e-06,
+      "loss": 0.9874,
+      "step": 6046
+    },
+    {
+      "epoch": 0.6328623757195185,
+      "grad_norm": 1.985600041977417,
+      "learning_rate": 6.275175862934679e-06,
+      "loss": 0.8763,
+      "step": 6047
+    },
+    {
+      "epoch": 0.6329670329670329,
+      "grad_norm": 2.167359949334803,
+      "learning_rate": 6.2720302836219925e-06,
+      "loss": 0.8538,
+      "step": 6048
+    },
+    {
+      "epoch": 0.6330716902145473,
+      "grad_norm": 1.892338682422864,
+      "learning_rate": 6.268885132660585e-06,
+      "loss": 0.857,
+      "step": 6049
+    },
+    {
+      "epoch": 0.6331763474620618,
+      "grad_norm": 2.074732815371282,
+      "learning_rate": 6.265740410411838e-06,
+      "loss": 0.9257,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6332810047095762,
+      "grad_norm": 1.9538306094870916,
+      "learning_rate": 6.26259611723709e-06,
+      "loss": 0.9022,
+      "step": 6051
+    },
+    {
+      "epoch": 0.6333856619570906,
+      "grad_norm": 2.052293995501398,
+      "learning_rate": 6.259452253497622e-06,
+      "loss": 0.9688,
+      "step": 6052
+    },
+    {
+      "epoch": 0.633490319204605,
+      "grad_norm": 1.9040390605918567,
+      "learning_rate": 6.256308819554676e-06,
+      "loss": 0.9216,
+      "step": 6053
+    },
+    {
+      "epoch": 0.6335949764521193,
+      "grad_norm": 1.887491395994407,
+      "learning_rate": 6.253165815769431e-06,
+      "loss": 0.9326,
+      "step": 6054
+    },
+    {
+      "epoch": 0.6336996336996337,
+      "grad_norm": 2.1059814291825054,
+      "learning_rate": 6.250023242503031e-06,
+      "loss": 0.9491,
+      "step": 6055
+    },
+    {
+      "epoch": 0.6338042909471481,
+      "grad_norm": 1.949072153051419,
+      "learning_rate": 6.246881100116566e-06,
+      "loss": 0.7935,
+      "step": 6056
+    },
+    {
+      "epoch": 0.6339089481946625,
+      "grad_norm": 2.0739299147239607,
+      "learning_rate": 6.243739388971068e-06,
+      "loss": 0.959,
+      "step": 6057
+    },
+    {
+      "epoch": 0.6340136054421769,
+      "grad_norm": 2.001653115479742,
+      "learning_rate": 6.240598109427531e-06,
+      "loss": 0.9694,
+      "step": 6058
+    },
+    {
+      "epoch": 0.6341182626896913,
+      "grad_norm": 2.2062451919821355,
+      "learning_rate": 6.237457261846891e-06,
+      "loss": 0.9713,
+      "step": 6059
+    },
+    {
+      "epoch": 0.6342229199372057,
+      "grad_norm": 2.0518558414861845,
+      "learning_rate": 6.234316846590041e-06,
+      "loss": 0.9721,
+      "step": 6060
+    },
+    {
+      "epoch": 0.63432757718472,
+      "grad_norm": 2.0231758494324352,
+      "learning_rate": 6.231176864017819e-06,
+      "loss": 1.0087,
+      "step": 6061
+    },
+    {
+      "epoch": 0.6344322344322344,
+      "grad_norm": 2.315599702222333,
+      "learning_rate": 6.228037314491013e-06,
+      "loss": 0.9439,
+      "step": 6062
+    },
+    {
+      "epoch": 0.6345368916797488,
+      "grad_norm": 2.0500400828302956,
+      "learning_rate": 6.224898198370363e-06,
+      "loss": 0.9502,
+      "step": 6063
+    },
+    {
+      "epoch": 0.6346415489272632,
+      "grad_norm": 1.9582030992702546,
+      "learning_rate": 6.22175951601657e-06,
+      "loss": 0.9388,
+      "step": 6064
+    },
+    {
+      "epoch": 0.6347462061747776,
+      "grad_norm": 1.8671881803341661,
+      "learning_rate": 6.218621267790263e-06,
+      "loss": 0.9292,
+      "step": 6065
+    },
+    {
+      "epoch": 0.634850863422292,
+      "grad_norm": 1.9018283992660436,
+      "learning_rate": 6.215483454052043e-06,
+      "loss": 0.8481,
+      "step": 6066
+    },
+    {
+      "epoch": 0.6349555206698064,
+      "grad_norm": 2.1228197696527915,
+      "learning_rate": 6.2123460751624425e-06,
+      "loss": 0.9088,
+      "step": 6067
+    },
+    {
+      "epoch": 0.6350601779173207,
+      "grad_norm": 2.0541407952110817,
+      "learning_rate": 6.209209131481957e-06,
+      "loss": 0.8942,
+      "step": 6068
+    },
+    {
+      "epoch": 0.6351648351648351,
+      "grad_norm": 2.4812046684098146,
+      "learning_rate": 6.206072623371027e-06,
+      "loss": 0.9851,
+      "step": 6069
+    },
+    {
+      "epoch": 0.6352694924123495,
+      "grad_norm": 2.466479670275315,
+      "learning_rate": 6.202936551190044e-06,
+      "loss": 0.9819,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6353741496598639,
+      "grad_norm": 1.928191042129914,
+      "learning_rate": 6.199800915299345e-06,
+      "loss": 0.8792,
+      "step": 6071
+    },
+    {
+      "epoch": 0.6354788069073783,
+      "grad_norm": 1.638080208345466,
+      "learning_rate": 6.196665716059229e-06,
+      "loss": 0.7502,
+      "step": 6072
+    },
+    {
+      "epoch": 0.6355834641548928,
+      "grad_norm": 1.935024991236978,
+      "learning_rate": 6.193530953829931e-06,
+      "loss": 0.9067,
+      "step": 6073
+    },
+    {
+      "epoch": 0.6356881214024072,
+      "grad_norm": 1.9042984524406725,
+      "learning_rate": 6.190396628971644e-06,
+      "loss": 1.0139,
+      "step": 6074
+    },
+    {
+      "epoch": 0.6357927786499215,
+      "grad_norm": 2.096217219665164,
+      "learning_rate": 6.187262741844507e-06,
+      "loss": 0.9256,
+      "step": 6075
+    },
+    {
+      "epoch": 0.6358974358974359,
+      "grad_norm": 2.1629109423290602,
+      "learning_rate": 6.18412929280861e-06,
+      "loss": 1.1161,
+      "step": 6076
+    },
+    {
+      "epoch": 0.6360020931449503,
+      "grad_norm": 2.4819953770244543,
+      "learning_rate": 6.180996282223993e-06,
+      "loss": 1.015,
+      "step": 6077
+    },
+    {
+      "epoch": 0.6361067503924647,
+      "grad_norm": 1.973489568225318,
+      "learning_rate": 6.177863710450645e-06,
+      "loss": 0.8687,
+      "step": 6078
+    },
+    {
+      "epoch": 0.6362114076399791,
+      "grad_norm": 2.0537245713359455,
+      "learning_rate": 6.174731577848505e-06,
+      "loss": 1.0115,
+      "step": 6079
+    },
+    {
+      "epoch": 0.6363160648874935,
+      "grad_norm": 2.1026542763075056,
+      "learning_rate": 6.171599884777458e-06,
+      "loss": 0.8719,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6364207221350079,
+      "grad_norm": 1.9810644360085412,
+      "learning_rate": 6.168468631597349e-06,
+      "loss": 0.9621,
+      "step": 6081
+    },
+    {
+      "epoch": 0.6365253793825223,
+      "grad_norm": 1.8498313936174973,
+      "learning_rate": 6.16533781866796e-06,
+      "loss": 0.8684,
+      "step": 6082
+    },
+    {
+      "epoch": 0.6366300366300366,
+      "grad_norm": 1.882513401890856,
+      "learning_rate": 6.162207446349031e-06,
+      "loss": 0.8804,
+      "step": 6083
+    },
+    {
+      "epoch": 0.636734693877551,
+      "grad_norm": 2.1454199707720045,
+      "learning_rate": 6.159077515000245e-06,
+      "loss": 0.8786,
+      "step": 6084
+    },
+    {
+      "epoch": 0.6368393511250654,
+      "grad_norm": 2.0348014676633848,
+      "learning_rate": 6.155948024981241e-06,
+      "loss": 0.9031,
+      "step": 6085
+    },
+    {
+      "epoch": 0.6369440083725798,
+      "grad_norm": 2.0035495835149786,
+      "learning_rate": 6.1528189766515996e-06,
+      "loss": 0.965,
+      "step": 6086
+    },
+    {
+      "epoch": 0.6370486656200942,
+      "grad_norm": 2.180282601428462,
+      "learning_rate": 6.1496903703708575e-06,
+      "loss": 1.0432,
+      "step": 6087
+    },
+    {
+      "epoch": 0.6371533228676086,
+      "grad_norm": 2.1368862326905043,
+      "learning_rate": 6.146562206498497e-06,
+      "loss": 0.9117,
+      "step": 6088
+    },
+    {
+      "epoch": 0.637257980115123,
+      "grad_norm": 2.043077967035869,
+      "learning_rate": 6.143434485393947e-06,
+      "loss": 0.9375,
+      "step": 6089
+    },
+    {
+      "epoch": 0.6373626373626373,
+      "grad_norm": 2.133413981464569,
+      "learning_rate": 6.140307207416598e-06,
+      "loss": 0.9432,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6374672946101517,
+      "grad_norm": 1.8874105602733608,
+      "learning_rate": 6.137180372925773e-06,
+      "loss": 0.9215,
+      "step": 6091
+    },
+    {
+      "epoch": 0.6375719518576661,
+      "grad_norm": 2.1540406495422624,
+      "learning_rate": 6.134053982280756e-06,
+      "loss": 0.8921,
+      "step": 6092
+    },
+    {
+      "epoch": 0.6376766091051805,
+      "grad_norm": 2.267611660094676,
+      "learning_rate": 6.130928035840771e-06,
+      "loss": 0.9158,
+      "step": 6093
+    },
+    {
+      "epoch": 0.6377812663526949,
+      "grad_norm": 2.132196533283258,
+      "learning_rate": 6.127802533965001e-06,
+      "loss": 0.9576,
+      "step": 6094
+    },
+    {
+      "epoch": 0.6378859236002093,
+      "grad_norm": 1.7952285637216865,
+      "learning_rate": 6.124677477012566e-06,
+      "loss": 0.8652,
+      "step": 6095
+    },
+    {
+      "epoch": 0.6379905808477238,
+      "grad_norm": 2.268355822133597,
+      "learning_rate": 6.121552865342548e-06,
+      "loss": 0.9496,
+      "step": 6096
+    },
+    {
+      "epoch": 0.638095238095238,
+      "grad_norm": 2.480468965787619,
+      "learning_rate": 6.118428699313965e-06,
+      "loss": 0.9992,
+      "step": 6097
+    },
+    {
+      "epoch": 0.6381998953427525,
+      "grad_norm": 2.284899377257853,
+      "learning_rate": 6.115304979285794e-06,
+      "loss": 1.0189,
+      "step": 6098
+    },
+    {
+      "epoch": 0.6383045525902669,
+      "grad_norm": 2.2705419138837386,
+      "learning_rate": 6.112181705616958e-06,
+      "loss": 0.7452,
+      "step": 6099
+    },
+    {
+      "epoch": 0.6384092098377813,
+      "grad_norm": 1.9550159382736858,
+      "learning_rate": 6.109058878666326e-06,
+      "loss": 0.934,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6385138670852957,
+      "grad_norm": 2.380989502303415,
+      "learning_rate": 6.105936498792715e-06,
+      "loss": 0.8759,
+      "step": 6101
+    },
+    {
+      "epoch": 0.6386185243328101,
+      "grad_norm": 2.3521973407044827,
+      "learning_rate": 6.102814566354896e-06,
+      "loss": 0.9418,
+      "step": 6102
+    },
+    {
+      "epoch": 0.6387231815803245,
+      "grad_norm": 2.1620835017709092,
+      "learning_rate": 6.099693081711582e-06,
+      "loss": 0.9934,
+      "step": 6103
+    },
+    {
+      "epoch": 0.6388278388278388,
+      "grad_norm": 2.304478296902805,
+      "learning_rate": 6.09657204522144e-06,
+      "loss": 0.9824,
+      "step": 6104
+    },
+    {
+      "epoch": 0.6389324960753532,
+      "grad_norm": 2.124733642865628,
+      "learning_rate": 6.093451457243082e-06,
+      "loss": 0.8826,
+      "step": 6105
+    },
+    {
+      "epoch": 0.6390371533228676,
+      "grad_norm": 2.2586347480935216,
+      "learning_rate": 6.09033131813507e-06,
+      "loss": 1.0353,
+      "step": 6106
+    },
+    {
+      "epoch": 0.639141810570382,
+      "grad_norm": 2.183731475063431,
+      "learning_rate": 6.087211628255918e-06,
+      "loss": 0.9315,
+      "step": 6107
+    },
+    {
+      "epoch": 0.6392464678178964,
+      "grad_norm": 2.1727879672656063,
+      "learning_rate": 6.08409238796408e-06,
+      "loss": 1.017,
+      "step": 6108
+    },
+    {
+      "epoch": 0.6393511250654108,
+      "grad_norm": 1.8520184743679367,
+      "learning_rate": 6.080973597617969e-06,
+      "loss": 0.7484,
+      "step": 6109
+    },
+    {
+      "epoch": 0.6394557823129252,
+      "grad_norm": 2.1309335686211384,
+      "learning_rate": 6.077855257575932e-06,
+      "loss": 0.9459,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6395604395604395,
+      "grad_norm": 2.112146194809444,
+      "learning_rate": 6.074737368196279e-06,
+      "loss": 0.9341,
+      "step": 6111
+    },
+    {
+      "epoch": 0.6396650968079539,
+      "grad_norm": 2.2818082480682373,
+      "learning_rate": 6.071619929837259e-06,
+      "loss": 0.9016,
+      "step": 6112
+    },
+    {
+      "epoch": 0.6397697540554683,
+      "grad_norm": 1.9751366903614145,
+      "learning_rate": 6.068502942857075e-06,
+      "loss": 0.7711,
+      "step": 6113
+    },
+    {
+      "epoch": 0.6398744113029827,
+      "grad_norm": 2.0611570942571813,
+      "learning_rate": 6.065386407613869e-06,
+      "loss": 0.9724,
+      "step": 6114
+    },
+    {
+      "epoch": 0.6399790685504971,
+      "grad_norm": 2.0283236031744316,
+      "learning_rate": 6.06227032446574e-06,
+      "loss": 0.8681,
+      "step": 6115
+    },
+    {
+      "epoch": 0.6400837257980115,
+      "grad_norm": 1.840467172717708,
+      "learning_rate": 6.0591546937707375e-06,
+      "loss": 0.8617,
+      "step": 6116
+    },
+    {
+      "epoch": 0.6401883830455259,
+      "grad_norm": 2.216732313893134,
+      "learning_rate": 6.056039515886848e-06,
+      "loss": 0.9226,
+      "step": 6117
+    },
+    {
+      "epoch": 0.6402930402930402,
+      "grad_norm": 2.0576439170006062,
+      "learning_rate": 6.0529247911720145e-06,
+      "loss": 0.9738,
+      "step": 6118
+    },
+    {
+      "epoch": 0.6403976975405546,
+      "grad_norm": 2.004555679033776,
+      "learning_rate": 6.049810519984125e-06,
+      "loss": 0.9752,
+      "step": 6119
+    },
+    {
+      "epoch": 0.640502354788069,
+      "grad_norm": 2.121801609117083,
+      "learning_rate": 6.046696702681012e-06,
+      "loss": 0.9376,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6406070120355835,
+      "grad_norm": 2.0557921754943638,
+      "learning_rate": 6.043583339620465e-06,
+      "loss": 0.8175,
+      "step": 6121
+    },
+    {
+      "epoch": 0.6407116692830979,
+      "grad_norm": 2.117196486653035,
+      "learning_rate": 6.0404704311602095e-06,
+      "loss": 0.914,
+      "step": 6122
+    },
+    {
+      "epoch": 0.6408163265306123,
+      "grad_norm": 2.454371990920559,
+      "learning_rate": 6.037357977657926e-06,
+      "loss": 0.8974,
+      "step": 6123
+    },
+    {
+      "epoch": 0.6409209837781267,
+      "grad_norm": 1.779633370179486,
+      "learning_rate": 6.034245979471249e-06,
+      "loss": 0.8059,
+      "step": 6124
+    },
+    {
+      "epoch": 0.6410256410256411,
+      "grad_norm": 2.224227729668064,
+      "learning_rate": 6.031134436957747e-06,
+      "loss": 0.9438,
+      "step": 6125
+    },
+    {
+      "epoch": 0.6411302982731554,
+      "grad_norm": 1.8793850100294136,
+      "learning_rate": 6.028023350474943e-06,
+      "loss": 0.8463,
+      "step": 6126
+    },
+    {
+      "epoch": 0.6412349555206698,
+      "grad_norm": 2.0192489810899823,
+      "learning_rate": 6.024912720380309e-06,
+      "loss": 0.8065,
+      "step": 6127
+    },
+    {
+      "epoch": 0.6413396127681842,
+      "grad_norm": 2.086275873749281,
+      "learning_rate": 6.021802547031263e-06,
+      "loss": 0.9134,
+      "step": 6128
+    },
+    {
+      "epoch": 0.6414442700156986,
+      "grad_norm": 2.239582460843957,
+      "learning_rate": 6.018692830785167e-06,
+      "loss": 0.8925,
+      "step": 6129
+    },
+    {
+      "epoch": 0.641548927263213,
+      "grad_norm": 1.9036697749177023,
+      "learning_rate": 6.015583571999337e-06,
+      "loss": 0.9513,
+      "step": 6130
+    },
+    {
+      "epoch": 0.6416535845107274,
+      "grad_norm": 1.967430877780038,
+      "learning_rate": 6.012474771031029e-06,
+      "loss": 0.9026,
+      "step": 6131
+    },
+    {
+      "epoch": 0.6417582417582418,
+      "grad_norm": 1.9634313043742981,
+      "learning_rate": 6.009366428237453e-06,
+      "loss": 0.9432,
+      "step": 6132
+    },
+    {
+      "epoch": 0.6418628990057561,
+      "grad_norm": 1.907415430341769,
+      "learning_rate": 6.006258543975769e-06,
+      "loss": 1.0205,
+      "step": 6133
+    },
+    {
+      "epoch": 0.6419675562532705,
+      "grad_norm": 1.9958836925353802,
+      "learning_rate": 6.003151118603071e-06,
+      "loss": 0.9038,
+      "step": 6134
+    },
+    {
+      "epoch": 0.6420722135007849,
+      "grad_norm": 1.974573234104124,
+      "learning_rate": 6.000044152476414e-06,
+      "loss": 0.8984,
+      "step": 6135
+    },
+    {
+      "epoch": 0.6421768707482993,
+      "grad_norm": 1.9662322439234587,
+      "learning_rate": 5.996937645952792e-06,
+      "loss": 0.8528,
+      "step": 6136
+    },
+    {
+      "epoch": 0.6422815279958137,
+      "grad_norm": 2.049330183711433,
+      "learning_rate": 5.993831599389149e-06,
+      "loss": 0.8704,
+      "step": 6137
+    },
+    {
+      "epoch": 0.6423861852433281,
+      "grad_norm": 1.8386217231992765,
+      "learning_rate": 5.990726013142378e-06,
+      "loss": 0.8267,
+      "step": 6138
+    },
+    {
+      "epoch": 0.6424908424908425,
+      "grad_norm": 1.9591896043802715,
+      "learning_rate": 5.987620887569314e-06,
+      "loss": 0.8269,
+      "step": 6139
+    },
+    {
+      "epoch": 0.6425954997383568,
+      "grad_norm": 2.0298076454052065,
+      "learning_rate": 5.984516223026746e-06,
+      "loss": 0.9347,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6427001569858712,
+      "grad_norm": 2.0195419480669683,
+      "learning_rate": 5.9814120198714e-06,
+      "loss": 0.9478,
+      "step": 6141
+    },
+    {
+      "epoch": 0.6428048142333856,
+      "grad_norm": 2.0883419789864335,
+      "learning_rate": 5.9783082784599615e-06,
+      "loss": 0.8287,
+      "step": 6142
+    },
+    {
+      "epoch": 0.6429094714809,
+      "grad_norm": 2.1269657774275004,
+      "learning_rate": 5.975204999149056e-06,
+      "loss": 0.9302,
+      "step": 6143
+    },
+    {
+      "epoch": 0.6430141287284145,
+      "grad_norm": 2.016828126075074,
+      "learning_rate": 5.972102182295254e-06,
+      "loss": 1.0265,
+      "step": 6144
+    },
+    {
+      "epoch": 0.6431187859759289,
+      "grad_norm": 1.6917299157653185,
+      "learning_rate": 5.968999828255079e-06,
+      "loss": 0.7979,
+      "step": 6145
+    },
+    {
+      "epoch": 0.6432234432234433,
+      "grad_norm": 1.8616864521421854,
+      "learning_rate": 5.965897937384992e-06,
+      "loss": 0.7963,
+      "step": 6146
+    },
+    {
+      "epoch": 0.6433281004709576,
+      "grad_norm": 2.341785220873819,
+      "learning_rate": 5.962796510041413e-06,
+      "loss": 0.9599,
+      "step": 6147
+    },
+    {
+      "epoch": 0.643432757718472,
+      "grad_norm": 1.9957063834445605,
+      "learning_rate": 5.9596955465806974e-06,
+      "loss": 1.0314,
+      "step": 6148
+    },
+    {
+      "epoch": 0.6435374149659864,
+      "grad_norm": 2.08882364485049,
+      "learning_rate": 5.956595047359151e-06,
+      "loss": 0.8416,
+      "step": 6149
+    },
+    {
+      "epoch": 0.6436420722135008,
+      "grad_norm": 1.8725034362012762,
+      "learning_rate": 5.953495012733035e-06,
+      "loss": 0.8405,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6437467294610152,
+      "grad_norm": 1.9625445407363562,
+      "learning_rate": 5.950395443058543e-06,
+      "loss": 0.9756,
+      "step": 6151
+    },
+    {
+      "epoch": 0.6438513867085296,
+      "grad_norm": 2.0630976658247766,
+      "learning_rate": 5.947296338691827e-06,
+      "loss": 0.9711,
+      "step": 6152
+    },
+    {
+      "epoch": 0.643956043956044,
+      "grad_norm": 1.9216746094429082,
+      "learning_rate": 5.944197699988975e-06,
+      "loss": 0.9195,
+      "step": 6153
+    },
+    {
+      "epoch": 0.6440607012035583,
+      "grad_norm": 2.2677133214478324,
+      "learning_rate": 5.9410995273060316e-06,
+      "loss": 0.9668,
+      "step": 6154
+    },
+    {
+      "epoch": 0.6441653584510727,
+      "grad_norm": 2.026665277815765,
+      "learning_rate": 5.93800182099898e-06,
+      "loss": 1.006,
+      "step": 6155
+    },
+    {
+      "epoch": 0.6442700156985871,
+      "grad_norm": 2.309012027737432,
+      "learning_rate": 5.93490458142375e-06,
+      "loss": 0.8949,
+      "step": 6156
+    },
+    {
+      "epoch": 0.6443746729461015,
+      "grad_norm": 1.8376028280551246,
+      "learning_rate": 5.93180780893623e-06,
+      "loss": 0.9266,
+      "step": 6157
+    },
+    {
+      "epoch": 0.6444793301936159,
+      "grad_norm": 1.8687096935531817,
+      "learning_rate": 5.928711503892233e-06,
+      "loss": 0.7169,
+      "step": 6158
+    },
+    {
+      "epoch": 0.6445839874411303,
+      "grad_norm": 1.7565771976435207,
+      "learning_rate": 5.925615666647541e-06,
+      "loss": 0.8147,
+      "step": 6159
+    },
+    {
+      "epoch": 0.6446886446886447,
+      "grad_norm": 2.2216686274696182,
+      "learning_rate": 5.92252029755787e-06,
+      "loss": 0.9832,
+      "step": 6160
+    },
+    {
+      "epoch": 0.644793301936159,
+      "grad_norm": 1.9664584115913126,
+      "learning_rate": 5.91942539697888e-06,
+      "loss": 0.9132,
+      "step": 6161
+    },
+    {
+      "epoch": 0.6448979591836734,
+      "grad_norm": 2.2259482681346325,
+      "learning_rate": 5.9163309652661886e-06,
+      "loss": 1.0205,
+      "step": 6162
+    },
+    {
+      "epoch": 0.6450026164311878,
+      "grad_norm": 1.974310097833552,
+      "learning_rate": 5.913237002775342e-06,
+      "loss": 0.8725,
+      "step": 6163
+    },
+    {
+      "epoch": 0.6451072736787022,
+      "grad_norm": 1.9361555583674028,
+      "learning_rate": 5.910143509861854e-06,
+      "loss": 0.8765,
+      "step": 6164
+    },
+    {
+      "epoch": 0.6452119309262166,
+      "grad_norm": 2.3259766646067805,
+      "learning_rate": 5.907050486881164e-06,
+      "loss": 0.8672,
+      "step": 6165
+    },
+    {
+      "epoch": 0.645316588173731,
+      "grad_norm": 2.0520065137688976,
+      "learning_rate": 5.903957934188673e-06,
+      "loss": 0.942,
+      "step": 6166
+    },
+    {
+      "epoch": 0.6454212454212455,
+      "grad_norm": 2.308496686545806,
+      "learning_rate": 5.900865852139715e-06,
+      "loss": 1.0235,
+      "step": 6167
+    },
+    {
+      "epoch": 0.6455259026687599,
+      "grad_norm": 1.9275377714588016,
+      "learning_rate": 5.897774241089581e-06,
+      "loss": 0.9101,
+      "step": 6168
+    },
+    {
+      "epoch": 0.6456305599162742,
+      "grad_norm": 1.7246090565713368,
+      "learning_rate": 5.894683101393505e-06,
+      "loss": 0.813,
+      "step": 6169
+    },
+    {
+      "epoch": 0.6457352171637886,
+      "grad_norm": 2.090211706193853,
+      "learning_rate": 5.8915924334066635e-06,
+      "loss": 0.9071,
+      "step": 6170
+    },
+    {
+      "epoch": 0.645839874411303,
+      "grad_norm": 2.225562586570551,
+      "learning_rate": 5.888502237484179e-06,
+      "loss": 0.89,
+      "step": 6171
+    },
+    {
+      "epoch": 0.6459445316588174,
+      "grad_norm": 2.1959182655755507,
+      "learning_rate": 5.885412513981121e-06,
+      "loss": 0.9924,
+      "step": 6172
+    },
+    {
+      "epoch": 0.6460491889063318,
+      "grad_norm": 2.0400315285376918,
+      "learning_rate": 5.88232326325251e-06,
+      "loss": 0.9976,
+      "step": 6173
+    },
+    {
+      "epoch": 0.6461538461538462,
+      "grad_norm": 2.2524957844292084,
+      "learning_rate": 5.879234485653302e-06,
+      "loss": 0.9471,
+      "step": 6174
+    },
+    {
+      "epoch": 0.6462585034013606,
+      "grad_norm": 1.9865407512804576,
+      "learning_rate": 5.8761461815384005e-06,
+      "loss": 1.007,
+      "step": 6175
+    },
+    {
+      "epoch": 0.6463631606488749,
+      "grad_norm": 1.8960367283948207,
+      "learning_rate": 5.873058351262666e-06,
+      "loss": 0.8504,
+      "step": 6176
+    },
+    {
+      "epoch": 0.6464678178963893,
+      "grad_norm": 1.7594307610018052,
+      "learning_rate": 5.869970995180896e-06,
+      "loss": 0.9622,
+      "step": 6177
+    },
+    {
+      "epoch": 0.6465724751439037,
+      "grad_norm": 2.262308058632982,
+      "learning_rate": 5.866884113647827e-06,
+      "loss": 1.0252,
+      "step": 6178
+    },
+    {
+      "epoch": 0.6466771323914181,
+      "grad_norm": 1.9209190231553315,
+      "learning_rate": 5.863797707018155e-06,
+      "loss": 0.8007,
+      "step": 6179
+    },
+    {
+      "epoch": 0.6467817896389325,
+      "grad_norm": 1.5447458009603428,
+      "learning_rate": 5.86071177564651e-06,
+      "loss": 0.778,
+      "step": 6180
+    },
+    {
+      "epoch": 0.6468864468864469,
+      "grad_norm": 1.8092557053926288,
+      "learning_rate": 5.857626319887475e-06,
+      "loss": 0.9498,
+      "step": 6181
+    },
+    {
+      "epoch": 0.6469911041339613,
+      "grad_norm": 2.2187668350942693,
+      "learning_rate": 5.85454134009557e-06,
+      "loss": 0.8878,
+      "step": 6182
+    },
+    {
+      "epoch": 0.6470957613814756,
+      "grad_norm": 2.0160589952562384,
+      "learning_rate": 5.851456836625271e-06,
+      "loss": 0.8794,
+      "step": 6183
+    },
+    {
+      "epoch": 0.64720041862899,
+      "grad_norm": 1.8692365583128616,
+      "learning_rate": 5.848372809830989e-06,
+      "loss": 0.9669,
+      "step": 6184
+    },
+    {
+      "epoch": 0.6473050758765044,
+      "grad_norm": 2.127493895012381,
+      "learning_rate": 5.845289260067089e-06,
+      "loss": 0.92,
+      "step": 6185
+    },
+    {
+      "epoch": 0.6474097331240188,
+      "grad_norm": 2.2617699972750125,
+      "learning_rate": 5.842206187687876e-06,
+      "loss": 0.889,
+      "step": 6186
+    },
+    {
+      "epoch": 0.6475143903715332,
+      "grad_norm": 2.134441719505655,
+      "learning_rate": 5.8391235930476e-06,
+      "loss": 0.8696,
+      "step": 6187
+    },
+    {
+      "epoch": 0.6476190476190476,
+      "grad_norm": 2.6083578482447947,
+      "learning_rate": 5.836041476500458e-06,
+      "loss": 0.9542,
+      "step": 6188
+    },
+    {
+      "epoch": 0.6477237048665621,
+      "grad_norm": 2.3257712734015388,
+      "learning_rate": 5.832959838400593e-06,
+      "loss": 0.8224,
+      "step": 6189
+    },
+    {
+      "epoch": 0.6478283621140764,
+      "grad_norm": 2.3044088351111256,
+      "learning_rate": 5.82987867910209e-06,
+      "loss": 0.8298,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6479330193615908,
+      "grad_norm": 2.026561166745952,
+      "learning_rate": 5.8267979989589815e-06,
+      "loss": 0.82,
+      "step": 6191
+    },
+    {
+      "epoch": 0.6480376766091052,
+      "grad_norm": 1.965006660005929,
+      "learning_rate": 5.823717798325238e-06,
+      "loss": 0.8973,
+      "step": 6192
+    },
+    {
+      "epoch": 0.6481423338566196,
+      "grad_norm": 1.9771041937813247,
+      "learning_rate": 5.820638077554785e-06,
+      "loss": 0.9496,
+      "step": 6193
+    },
+    {
+      "epoch": 0.648246991104134,
+      "grad_norm": 2.1258967665098134,
+      "learning_rate": 5.8175588370014955e-06,
+      "loss": 0.9364,
+      "step": 6194
+    },
+    {
+      "epoch": 0.6483516483516484,
+      "grad_norm": 2.1439985227923644,
+      "learning_rate": 5.814480077019173e-06,
+      "loss": 0.988,
+      "step": 6195
+    },
+    {
+      "epoch": 0.6484563055991628,
+      "grad_norm": 2.133683906939611,
+      "learning_rate": 5.811401797961576e-06,
+      "loss": 0.7218,
+      "step": 6196
+    },
+    {
+      "epoch": 0.6485609628466771,
+      "grad_norm": 2.1684367284425567,
+      "learning_rate": 5.8083240001824015e-06,
+      "loss": 0.9307,
+      "step": 6197
+    },
+    {
+      "epoch": 0.6486656200941915,
+      "grad_norm": 2.1509821140259078,
+      "learning_rate": 5.805246684035293e-06,
+      "loss": 0.9419,
+      "step": 6198
+    },
+    {
+      "epoch": 0.6487702773417059,
+      "grad_norm": 2.40745342036477,
+      "learning_rate": 5.802169849873849e-06,
+      "loss": 0.971,
+      "step": 6199
+    },
+    {
+      "epoch": 0.6488749345892203,
+      "grad_norm": 1.8193600056832606,
+      "learning_rate": 5.799093498051599e-06,
+      "loss": 0.948,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6489795918367347,
+      "grad_norm": 1.7020389159483353,
+      "learning_rate": 5.796017628922018e-06,
+      "loss": 0.7398,
+      "step": 6201
+    },
+    {
+      "epoch": 0.6490842490842491,
+      "grad_norm": 2.04512342282411,
+      "learning_rate": 5.792942242838537e-06,
+      "loss": 0.8347,
+      "step": 6202
+    },
+    {
+      "epoch": 0.6491889063317635,
+      "grad_norm": 2.5262087123815844,
+      "learning_rate": 5.789867340154518e-06,
+      "loss": 0.8673,
+      "step": 6203
+    },
+    {
+      "epoch": 0.6492935635792778,
+      "grad_norm": 1.7089620050014493,
+      "learning_rate": 5.786792921223281e-06,
+      "loss": 0.7972,
+      "step": 6204
+    },
+    {
+      "epoch": 0.6493982208267922,
+      "grad_norm": 2.1423319228812825,
+      "learning_rate": 5.783718986398077e-06,
+      "loss": 0.84,
+      "step": 6205
+    },
+    {
+      "epoch": 0.6495028780743066,
+      "grad_norm": 2.095804353523492,
+      "learning_rate": 5.78064553603211e-06,
+      "loss": 0.8465,
+      "step": 6206
+    },
+    {
+      "epoch": 0.649607535321821,
+      "grad_norm": 1.6986750357778488,
+      "learning_rate": 5.77757257047852e-06,
+      "loss": 0.8179,
+      "step": 6207
+    },
+    {
+      "epoch": 0.6497121925693354,
+      "grad_norm": 1.9991041419560274,
+      "learning_rate": 5.774500090090404e-06,
+      "loss": 0.8437,
+      "step": 6208
+    },
+    {
+      "epoch": 0.6498168498168498,
+      "grad_norm": 2.5476008805446435,
+      "learning_rate": 5.7714280952207955e-06,
+      "loss": 0.894,
+      "step": 6209
+    },
+    {
+      "epoch": 0.6499215070643642,
+      "grad_norm": 1.840271331808465,
+      "learning_rate": 5.7683565862226676e-06,
+      "loss": 0.9403,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6500261643118787,
+      "grad_norm": 2.5171778195678955,
+      "learning_rate": 5.765285563448948e-06,
+      "loss": 0.8461,
+      "step": 6211
+    },
+    {
+      "epoch": 0.650130821559393,
+      "grad_norm": 1.7716194503918323,
+      "learning_rate": 5.7622150272525e-06,
+      "loss": 0.9805,
+      "step": 6212
+    },
+    {
+      "epoch": 0.6502354788069074,
+      "grad_norm": 1.8377563447111556,
+      "learning_rate": 5.759144977986138e-06,
+      "loss": 0.8581,
+      "step": 6213
+    },
+    {
+      "epoch": 0.6503401360544218,
+      "grad_norm": 1.8723011718069902,
+      "learning_rate": 5.756075416002618e-06,
+      "loss": 0.9373,
+      "step": 6214
+    },
+    {
+      "epoch": 0.6504447933019362,
+      "grad_norm": 2.069818431579889,
+      "learning_rate": 5.753006341654634e-06,
+      "loss": 0.9817,
+      "step": 6215
+    },
+    {
+      "epoch": 0.6505494505494506,
+      "grad_norm": 2.1304030948159105,
+      "learning_rate": 5.749937755294831e-06,
+      "loss": 0.8727,
+      "step": 6216
+    },
+    {
+      "epoch": 0.650654107796965,
+      "grad_norm": 1.8379536286843263,
+      "learning_rate": 5.746869657275792e-06,
+      "loss": 0.8884,
+      "step": 6217
+    },
+    {
+      "epoch": 0.6507587650444794,
+      "grad_norm": 2.193473187783517,
+      "learning_rate": 5.743802047950055e-06,
+      "loss": 0.9205,
+      "step": 6218
+    },
+    {
+      "epoch": 0.6508634222919937,
+      "grad_norm": 2.122540449060093,
+      "learning_rate": 5.740734927670089e-06,
+      "loss": 0.8567,
+      "step": 6219
+    },
+    {
+      "epoch": 0.6509680795395081,
+      "grad_norm": 1.9639756156046886,
+      "learning_rate": 5.737668296788314e-06,
+      "loss": 0.9146,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6510727367870225,
+      "grad_norm": 1.952699246508381,
+      "learning_rate": 5.734602155657096e-06,
+      "loss": 0.8395,
+      "step": 6221
+    },
+    {
+      "epoch": 0.6511773940345369,
+      "grad_norm": 1.8418042050457508,
+      "learning_rate": 5.731536504628732e-06,
+      "loss": 0.8573,
+      "step": 6222
+    },
+    {
+      "epoch": 0.6512820512820513,
+      "grad_norm": 1.8770950809548828,
+      "learning_rate": 5.728471344055482e-06,
+      "loss": 0.9337,
+      "step": 6223
+    },
+    {
+      "epoch": 0.6513867085295657,
+      "grad_norm": 1.8578509595752808,
+      "learning_rate": 5.725406674289532e-06,
+      "loss": 0.8205,
+      "step": 6224
+    },
+    {
+      "epoch": 0.6514913657770801,
+      "grad_norm": 2.250049618236253,
+      "learning_rate": 5.722342495683021e-06,
+      "loss": 0.9339,
+      "step": 6225
+    },
+    {
+      "epoch": 0.6515960230245944,
+      "grad_norm": 2.075268254460256,
+      "learning_rate": 5.7192788085880245e-06,
+      "loss": 0.9066,
+      "step": 6226
+    },
+    {
+      "epoch": 0.6517006802721088,
+      "grad_norm": 2.0103126248212058,
+      "learning_rate": 5.716215613356577e-06,
+      "loss": 0.8782,
+      "step": 6227
+    },
+    {
+      "epoch": 0.6518053375196232,
+      "grad_norm": 2.170160364779897,
+      "learning_rate": 5.713152910340634e-06,
+      "loss": 0.9634,
+      "step": 6228
+    },
+    {
+      "epoch": 0.6519099947671376,
+      "grad_norm": 1.8429017442207223,
+      "learning_rate": 5.7100906998921154e-06,
+      "loss": 0.8229,
+      "step": 6229
+    },
+    {
+      "epoch": 0.652014652014652,
+      "grad_norm": 1.9662643778222724,
+      "learning_rate": 5.707028982362873e-06,
+      "loss": 0.9536,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6521193092621664,
+      "grad_norm": 2.3974867995173166,
+      "learning_rate": 5.7039677581046984e-06,
+      "loss": 0.9297,
+      "step": 6231
+    },
+    {
+      "epoch": 0.6522239665096808,
+      "grad_norm": 2.0737121782229746,
+      "learning_rate": 5.700907027469342e-06,
+      "loss": 0.9138,
+      "step": 6232
+    },
+    {
+      "epoch": 0.6523286237571951,
+      "grad_norm": 1.9226135804608533,
+      "learning_rate": 5.697846790808483e-06,
+      "loss": 0.7444,
+      "step": 6233
+    },
+    {
+      "epoch": 0.6524332810047095,
+      "grad_norm": 2.0066517801777994,
+      "learning_rate": 5.69478704847375e-06,
+      "loss": 0.82,
+      "step": 6234
+    },
+    {
+      "epoch": 0.652537938252224,
+      "grad_norm": 1.891347237478382,
+      "learning_rate": 5.691727800816712e-06,
+      "loss": 0.9539,
+      "step": 6235
+    },
+    {
+      "epoch": 0.6526425954997384,
+      "grad_norm": 1.9297427031420433,
+      "learning_rate": 5.68866904818888e-06,
+      "loss": 0.8912,
+      "step": 6236
+    },
+    {
+      "epoch": 0.6527472527472528,
+      "grad_norm": 1.9332020266170267,
+      "learning_rate": 5.685610790941713e-06,
+      "loss": 0.8453,
+      "step": 6237
+    },
+    {
+      "epoch": 0.6528519099947672,
+      "grad_norm": 2.0549224122607925,
+      "learning_rate": 5.6825530294266185e-06,
+      "loss": 0.9198,
+      "step": 6238
+    },
+    {
+      "epoch": 0.6529565672422816,
+      "grad_norm": 1.7906027105209696,
+      "learning_rate": 5.679495763994931e-06,
+      "loss": 0.9266,
+      "step": 6239
+    },
+    {
+      "epoch": 0.6530612244897959,
+      "grad_norm": 1.6523611298603982,
+      "learning_rate": 5.67643899499794e-06,
+      "loss": 0.8616,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6531658817373103,
+      "grad_norm": 2.1388297304679718,
+      "learning_rate": 5.673382722786869e-06,
+      "loss": 0.8582,
+      "step": 6241
+    },
+    {
+      "epoch": 0.6532705389848247,
+      "grad_norm": 1.918250281902123,
+      "learning_rate": 5.670326947712899e-06,
+      "loss": 0.9843,
+      "step": 6242
+    },
+    {
+      "epoch": 0.6533751962323391,
+      "grad_norm": 1.990471992424265,
+      "learning_rate": 5.667271670127138e-06,
+      "loss": 0.8863,
+      "step": 6243
+    },
+    {
+      "epoch": 0.6534798534798535,
+      "grad_norm": 2.053395723942595,
+      "learning_rate": 5.664216890380647e-06,
+      "loss": 1.0306,
+      "step": 6244
+    },
+    {
+      "epoch": 0.6535845107273679,
+      "grad_norm": 1.9787757140031736,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.8954,
+      "step": 6245
+    },
+    {
+      "epoch": 0.6536891679748823,
+      "grad_norm": 2.1654076339913457,
+      "learning_rate": 5.6581088258094054e-06,
+      "loss": 0.9597,
+      "step": 6246
+    },
+    {
+      "epoch": 0.6537938252223966,
+      "grad_norm": 2.18172363279797,
+      "learning_rate": 5.655055541686491e-06,
+      "loss": 1.008,
+      "step": 6247
+    },
+    {
+      "epoch": 0.653898482469911,
+      "grad_norm": 1.9685000331248872,
+      "learning_rate": 5.652002756806506e-06,
+      "loss": 0.8069,
+      "step": 6248
+    },
+    {
+      "epoch": 0.6540031397174254,
+      "grad_norm": 2.024974383790366,
+      "learning_rate": 5.648950471520217e-06,
+      "loss": 0.888,
+      "step": 6249
+    },
+    {
+      "epoch": 0.6541077969649398,
+      "grad_norm": 1.9484380553012453,
+      "learning_rate": 5.645898686178335e-06,
+      "loss": 0.944,
+      "step": 6250
+    },
+    {
+      "epoch": 0.6542124542124542,
+      "grad_norm": 2.231538349975824,
+      "learning_rate": 5.642847401131526e-06,
+      "loss": 0.8687,
+      "step": 6251
+    },
+    {
+      "epoch": 0.6543171114599686,
+      "grad_norm": 1.9993737936439624,
+      "learning_rate": 5.639796616730382e-06,
+      "loss": 0.9617,
+      "step": 6252
+    },
+    {
+      "epoch": 0.654421768707483,
+      "grad_norm": 2.1435053716762376,
+      "learning_rate": 5.636746333325447e-06,
+      "loss": 1.0731,
+      "step": 6253
+    },
+    {
+      "epoch": 0.6545264259549974,
+      "grad_norm": 1.9548403870891942,
+      "learning_rate": 5.633696551267198e-06,
+      "loss": 0.9179,
+      "step": 6254
+    },
+    {
+      "epoch": 0.6546310832025117,
+      "grad_norm": 1.910022534905838,
+      "learning_rate": 5.630647270906071e-06,
+      "loss": 0.9102,
+      "step": 6255
+    },
+    {
+      "epoch": 0.6547357404500261,
+      "grad_norm": 1.8907963804653292,
+      "learning_rate": 5.627598492592428e-06,
+      "loss": 0.922,
+      "step": 6256
+    },
+    {
+      "epoch": 0.6548403976975405,
+      "grad_norm": 1.954572343884513,
+      "learning_rate": 5.624550216676584e-06,
+      "loss": 0.9166,
+      "step": 6257
+    },
+    {
+      "epoch": 0.654945054945055,
+      "grad_norm": 1.8353655197423198,
+      "learning_rate": 5.621502443508791e-06,
+      "loss": 0.9632,
+      "step": 6258
+    },
+    {
+      "epoch": 0.6550497121925694,
+      "grad_norm": 2.18435647821634,
+      "learning_rate": 5.618455173439244e-06,
+      "loss": 0.8807,
+      "step": 6259
+    },
+    {
+      "epoch": 0.6551543694400838,
+      "grad_norm": 2.527165737623044,
+      "learning_rate": 5.615408406818074e-06,
+      "loss": 1.04,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6552590266875982,
+      "grad_norm": 2.1620951304463962,
+      "learning_rate": 5.6123621439953715e-06,
+      "loss": 0.8673,
+      "step": 6261
+    },
+    {
+      "epoch": 0.6553636839351125,
+      "grad_norm": 1.9064074958893684,
+      "learning_rate": 5.609316385321149e-06,
+      "loss": 0.8773,
+      "step": 6262
+    },
+    {
+      "epoch": 0.6554683411826269,
+      "grad_norm": 2.240290365862607,
+      "learning_rate": 5.60627113114538e-06,
+      "loss": 0.8222,
+      "step": 6263
+    },
+    {
+      "epoch": 0.6555729984301413,
+      "grad_norm": 2.0718457091136147,
+      "learning_rate": 5.6032263818179635e-06,
+      "loss": 0.8648,
+      "step": 6264
+    },
+    {
+      "epoch": 0.6556776556776557,
+      "grad_norm": 2.0971279832921965,
+      "learning_rate": 5.6001821376887454e-06,
+      "loss": 0.9237,
+      "step": 6265
+    },
+    {
+      "epoch": 0.6557823129251701,
+      "grad_norm": 1.9785023614732484,
+      "learning_rate": 5.5971383991075234e-06,
+      "loss": 0.7651,
+      "step": 6266
+    },
+    {
+      "epoch": 0.6558869701726845,
+      "grad_norm": 1.9909685665336019,
+      "learning_rate": 5.594095166424025e-06,
+      "loss": 0.8833,
+      "step": 6267
+    },
+    {
+      "epoch": 0.6559916274201989,
+      "grad_norm": 2.105567126725272,
+      "learning_rate": 5.591052439987923e-06,
+      "loss": 0.8752,
+      "step": 6268
+    },
+    {
+      "epoch": 0.6560962846677132,
+      "grad_norm": 1.8400676598198988,
+      "learning_rate": 5.588010220148834e-06,
+      "loss": 0.8795,
+      "step": 6269
+    },
+    {
+      "epoch": 0.6562009419152276,
+      "grad_norm": 1.8357939138695751,
+      "learning_rate": 5.58496850725631e-06,
+      "loss": 0.7996,
+      "step": 6270
+    },
+    {
+      "epoch": 0.656305599162742,
+      "grad_norm": 2.0900322388751653,
+      "learning_rate": 5.581927301659855e-06,
+      "loss": 1.0018,
+      "step": 6271
+    },
+    {
+      "epoch": 0.6564102564102564,
+      "grad_norm": 1.9794563181186657,
+      "learning_rate": 5.578886603708914e-06,
+      "loss": 0.9069,
+      "step": 6272
+    },
+    {
+      "epoch": 0.6565149136577708,
+      "grad_norm": 1.9823398345613306,
+      "learning_rate": 5.575846413752865e-06,
+      "loss": 0.8751,
+      "step": 6273
+    },
+    {
+      "epoch": 0.6566195709052852,
+      "grad_norm": 2.443483126872738,
+      "learning_rate": 5.572806732141031e-06,
+      "loss": 0.891,
+      "step": 6274
+    },
+    {
+      "epoch": 0.6567242281527996,
+      "grad_norm": 1.952400398515589,
+      "learning_rate": 5.569767559222674e-06,
+      "loss": 0.8864,
+      "step": 6275
+    },
+    {
+      "epoch": 0.6568288854003139,
+      "grad_norm": 1.8830733377298436,
+      "learning_rate": 5.5667288953470115e-06,
+      "loss": 0.9083,
+      "step": 6276
+    },
+    {
+      "epoch": 0.6569335426478283,
+      "grad_norm": 1.9839437867503655,
+      "learning_rate": 5.563690740863184e-06,
+      "loss": 0.8745,
+      "step": 6277
+    },
+    {
+      "epoch": 0.6570381998953427,
+      "grad_norm": 1.8975928814696994,
+      "learning_rate": 5.560653096120283e-06,
+      "loss": 0.9024,
+      "step": 6278
+    },
+    {
+      "epoch": 0.6571428571428571,
+      "grad_norm": 2.147697717951089,
+      "learning_rate": 5.557615961467338e-06,
+      "loss": 0.8365,
+      "step": 6279
+    },
+    {
+      "epoch": 0.6572475143903715,
+      "grad_norm": 1.7757224649611865,
+      "learning_rate": 5.554579337253326e-06,
+      "loss": 0.827,
+      "step": 6280
+    },
+    {
+      "epoch": 0.657352171637886,
+      "grad_norm": 2.306267658791178,
+      "learning_rate": 5.551543223827162e-06,
+      "loss": 0.8783,
+      "step": 6281
+    },
+    {
+      "epoch": 0.6574568288854004,
+      "grad_norm": 1.938353410885278,
+      "learning_rate": 5.5485076215377e-06,
+      "loss": 0.8631,
+      "step": 6282
+    },
+    {
+      "epoch": 0.6575614861329147,
+      "grad_norm": 3.588827654123888,
+      "learning_rate": 5.545472530733738e-06,
+      "loss": 0.922,
+      "step": 6283
+    },
+    {
+      "epoch": 0.6576661433804291,
+      "grad_norm": 2.3146863860542197,
+      "learning_rate": 5.542437951764006e-06,
+      "loss": 0.9648,
+      "step": 6284
+    },
+    {
+      "epoch": 0.6577708006279435,
+      "grad_norm": 1.8737153115128953,
+      "learning_rate": 5.539403884977195e-06,
+      "loss": 0.8414,
+      "step": 6285
+    },
+    {
+      "epoch": 0.6578754578754579,
+      "grad_norm": 1.92712911224643,
+      "learning_rate": 5.536370330721921e-06,
+      "loss": 0.9653,
+      "step": 6286
+    },
+    {
+      "epoch": 0.6579801151229723,
+      "grad_norm": 1.7867220203896153,
+      "learning_rate": 5.533337289346743e-06,
+      "loss": 0.9625,
+      "step": 6287
+    },
+    {
+      "epoch": 0.6580847723704867,
+      "grad_norm": 2.618706543055954,
+      "learning_rate": 5.530304761200162e-06,
+      "loss": 1.0739,
+      "step": 6288
+    },
+    {
+      "epoch": 0.6581894296180011,
+      "grad_norm": 1.9789921876753085,
+      "learning_rate": 5.527272746630625e-06,
+      "loss": 0.8933,
+      "step": 6289
+    },
+    {
+      "epoch": 0.6582940868655154,
+      "grad_norm": 1.9939197083721731,
+      "learning_rate": 5.5242412459865215e-06,
+      "loss": 1.0279,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6583987441130298,
+      "grad_norm": 1.911672548685458,
+      "learning_rate": 5.521210259616171e-06,
+      "loss": 0.9188,
+      "step": 6291
+    },
+    {
+      "epoch": 0.6585034013605442,
+      "grad_norm": 2.081526787157685,
+      "learning_rate": 5.518179787867841e-06,
+      "loss": 0.8945,
+      "step": 6292
+    },
+    {
+      "epoch": 0.6586080586080586,
+      "grad_norm": 2.102166057104072,
+      "learning_rate": 5.515149831089739e-06,
+      "loss": 0.901,
+      "step": 6293
+    },
+    {
+      "epoch": 0.658712715855573,
+      "grad_norm": 1.7756557962954296,
+      "learning_rate": 5.51212038963001e-06,
+      "loss": 0.933,
+      "step": 6294
+    },
+    {
+      "epoch": 0.6588173731030874,
+      "grad_norm": 1.8389100423069065,
+      "learning_rate": 5.5090914638367495e-06,
+      "loss": 0.9373,
+      "step": 6295
+    },
+    {
+      "epoch": 0.6589220303506018,
+      "grad_norm": 2.112443963445209,
+      "learning_rate": 5.506063054057984e-06,
+      "loss": 0.9985,
+      "step": 6296
+    },
+    {
+      "epoch": 0.6590266875981162,
+      "grad_norm": 1.9306029603381285,
+      "learning_rate": 5.50303516064168e-06,
+      "loss": 0.7859,
+      "step": 6297
+    },
+    {
+      "epoch": 0.6591313448456305,
+      "grad_norm": 1.9275367700602777,
+      "learning_rate": 5.500007783935757e-06,
+      "loss": 0.9392,
+      "step": 6298
+    },
+    {
+      "epoch": 0.6592360020931449,
+      "grad_norm": 1.9953473847028906,
+      "learning_rate": 5.4969809242880555e-06,
+      "loss": 1.0136,
+      "step": 6299
+    },
+    {
+      "epoch": 0.6593406593406593,
+      "grad_norm": 2.0576053965813808,
+      "learning_rate": 5.493954582046381e-06,
+      "loss": 1.006,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6594453165881737,
+      "grad_norm": 1.9656517014244708,
+      "learning_rate": 5.490928757558458e-06,
+      "loss": 0.8898,
+      "step": 6301
+    },
+    {
+      "epoch": 0.6595499738356881,
+      "grad_norm": 2.328622538340134,
+      "learning_rate": 5.487903451171963e-06,
+      "loss": 0.9392,
+      "step": 6302
+    },
+    {
+      "epoch": 0.6596546310832025,
+      "grad_norm": 2.024875566385732,
+      "learning_rate": 5.484878663234503e-06,
+      "loss": 0.9248,
+      "step": 6303
+    },
+    {
+      "epoch": 0.659759288330717,
+      "grad_norm": 2.151355182140815,
+      "learning_rate": 5.481854394093643e-06,
+      "loss": 0.879,
+      "step": 6304
+    },
+    {
+      "epoch": 0.6598639455782312,
+      "grad_norm": 2.190977467557319,
+      "learning_rate": 5.478830644096872e-06,
+      "loss": 1.0399,
+      "step": 6305
+    },
+    {
+      "epoch": 0.6599686028257457,
+      "grad_norm": 1.7250845041049725,
+      "learning_rate": 5.475807413591621e-06,
+      "loss": 0.8283,
+      "step": 6306
+    },
+    {
+      "epoch": 0.6600732600732601,
+      "grad_norm": 2.0039747052514776,
+      "learning_rate": 5.4727847029252735e-06,
+      "loss": 0.8717,
+      "step": 6307
+    },
+    {
+      "epoch": 0.6601779173207745,
+      "grad_norm": 2.2311307267829874,
+      "learning_rate": 5.469762512445142e-06,
+      "loss": 0.8769,
+      "step": 6308
+    },
+    {
+      "epoch": 0.6602825745682889,
+      "grad_norm": 1.992690818121849,
+      "learning_rate": 5.466740842498477e-06,
+      "loss": 0.8671,
+      "step": 6309
+    },
+    {
+      "epoch": 0.6603872318158033,
+      "grad_norm": 2.2879738880594176,
+      "learning_rate": 5.463719693432483e-06,
+      "loss": 0.9096,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6604918890633177,
+      "grad_norm": 2.0374811856000465,
+      "learning_rate": 5.460699065594292e-06,
+      "loss": 0.9504,
+      "step": 6311
+    },
+    {
+      "epoch": 0.660596546310832,
+      "grad_norm": 1.8719692965200658,
+      "learning_rate": 5.4576789593309805e-06,
+      "loss": 0.9048,
+      "step": 6312
+    },
+    {
+      "epoch": 0.6607012035583464,
+      "grad_norm": 2.2231971812121794,
+      "learning_rate": 5.454659374989563e-06,
+      "loss": 0.9228,
+      "step": 6313
+    },
+    {
+      "epoch": 0.6608058608058608,
+      "grad_norm": 1.965484690948648,
+      "learning_rate": 5.451640312916995e-06,
+      "loss": 0.9135,
+      "step": 6314
+    },
+    {
+      "epoch": 0.6609105180533752,
+      "grad_norm": 2.1030519351417505,
+      "learning_rate": 5.448621773460181e-06,
+      "loss": 0.9864,
+      "step": 6315
+    },
+    {
+      "epoch": 0.6610151753008896,
+      "grad_norm": 1.9317968866735937,
+      "learning_rate": 5.445603756965952e-06,
+      "loss": 0.9095,
+      "step": 6316
+    },
+    {
+      "epoch": 0.661119832548404,
+      "grad_norm": 2.0771595808477943,
+      "learning_rate": 5.442586263781082e-06,
+      "loss": 0.979,
+      "step": 6317
+    },
+    {
+      "epoch": 0.6612244897959184,
+      "grad_norm": 2.108827558618817,
+      "learning_rate": 5.439569294252287e-06,
+      "loss": 0.9105,
+      "step": 6318
+    },
+    {
+      "epoch": 0.6613291470434327,
+      "grad_norm": 2.0983819392455314,
+      "learning_rate": 5.436552848726229e-06,
+      "loss": 0.9923,
+      "step": 6319
+    },
+    {
+      "epoch": 0.6614338042909471,
+      "grad_norm": 2.473425853943777,
+      "learning_rate": 5.433536927549499e-06,
+      "loss": 1.0007,
+      "step": 6320
+    },
+    {
+      "epoch": 0.6615384615384615,
+      "grad_norm": 1.9633903506211725,
+      "learning_rate": 5.430521531068634e-06,
+      "loss": 0.8951,
+      "step": 6321
+    },
+    {
+      "epoch": 0.6616431187859759,
+      "grad_norm": 2.100683110676726,
+      "learning_rate": 5.427506659630104e-06,
+      "loss": 0.9602,
+      "step": 6322
+    },
+    {
+      "epoch": 0.6617477760334903,
+      "grad_norm": 2.266215087701523,
+      "learning_rate": 5.4244923135803275e-06,
+      "loss": 0.9114,
+      "step": 6323
+    },
+    {
+      "epoch": 0.6618524332810047,
+      "grad_norm": 1.8670218973616595,
+      "learning_rate": 5.421478493265664e-06,
+      "loss": 0.9076,
+      "step": 6324
+    },
+    {
+      "epoch": 0.6619570905285191,
+      "grad_norm": 2.0326379729684545,
+      "learning_rate": 5.418465199032404e-06,
+      "loss": 0.849,
+      "step": 6325
+    },
+    {
+      "epoch": 0.6620617477760334,
+      "grad_norm": 1.9942841480440203,
+      "learning_rate": 5.4154524312267795e-06,
+      "loss": 0.886,
+      "step": 6326
+    },
+    {
+      "epoch": 0.6621664050235478,
+      "grad_norm": 2.0675772411489386,
+      "learning_rate": 5.412440190194965e-06,
+      "loss": 0.8767,
+      "step": 6327
+    },
+    {
+      "epoch": 0.6622710622710622,
+      "grad_norm": 2.216673868421644,
+      "learning_rate": 5.409428476283068e-06,
+      "loss": 0.8531,
+      "step": 6328
+    },
+    {
+      "epoch": 0.6623757195185767,
+      "grad_norm": 2.114144277955494,
+      "learning_rate": 5.4064172898371506e-06,
+      "loss": 0.8506,
+      "step": 6329
+    },
+    {
+      "epoch": 0.6624803767660911,
+      "grad_norm": 2.11736267037912,
+      "learning_rate": 5.403406631203197e-06,
+      "loss": 0.8308,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6625850340136055,
+      "grad_norm": 1.8344929417267373,
+      "learning_rate": 5.400396500727141e-06,
+      "loss": 0.885,
+      "step": 6331
+    },
+    {
+      "epoch": 0.6626896912611199,
+      "grad_norm": 2.005302325582287,
+      "learning_rate": 5.397386898754847e-06,
+      "loss": 0.7028,
+      "step": 6332
+    },
+    {
+      "epoch": 0.6627943485086342,
+      "grad_norm": 2.007310766798443,
+      "learning_rate": 5.394377825632129e-06,
+      "loss": 1.0009,
+      "step": 6333
+    },
+    {
+      "epoch": 0.6628990057561486,
+      "grad_norm": 2.298430147972449,
+      "learning_rate": 5.3913692817047395e-06,
+      "loss": 0.7607,
+      "step": 6334
+    },
+    {
+      "epoch": 0.663003663003663,
+      "grad_norm": 2.0426983676977746,
+      "learning_rate": 5.388361267318362e-06,
+      "loss": 1.0074,
+      "step": 6335
+    },
+    {
+      "epoch": 0.6631083202511774,
+      "grad_norm": 2.36635698834763,
+      "learning_rate": 5.385353782818623e-06,
+      "loss": 0.8771,
+      "step": 6336
+    },
+    {
+      "epoch": 0.6632129774986918,
+      "grad_norm": 1.89057291548456,
+      "learning_rate": 5.382346828551086e-06,
+      "loss": 0.9702,
+      "step": 6337
+    },
+    {
+      "epoch": 0.6633176347462062,
+      "grad_norm": 2.1549128599478347,
+      "learning_rate": 5.379340404861263e-06,
+      "loss": 0.9578,
+      "step": 6338
+    },
+    {
+      "epoch": 0.6634222919937206,
+      "grad_norm": 2.264597570417053,
+      "learning_rate": 5.3763345120945944e-06,
+      "loss": 0.9258,
+      "step": 6339
+    },
+    {
+      "epoch": 0.663526949241235,
+      "grad_norm": 1.9983468153168773,
+      "learning_rate": 5.37332915059646e-06,
+      "loss": 0.8108,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6636316064887493,
+      "grad_norm": 2.199629599477845,
+      "learning_rate": 5.370324320712189e-06,
+      "loss": 0.7242,
+      "step": 6341
+    },
+    {
+      "epoch": 0.6637362637362637,
+      "grad_norm": 2.037900821563259,
+      "learning_rate": 5.3673200227870345e-06,
+      "loss": 0.856,
+      "step": 6342
+    },
+    {
+      "epoch": 0.6638409209837781,
+      "grad_norm": 1.9505192747155962,
+      "learning_rate": 5.364316257166207e-06,
+      "loss": 0.8252,
+      "step": 6343
+    },
+    {
+      "epoch": 0.6639455782312925,
+      "grad_norm": 2.092352950973099,
+      "learning_rate": 5.361313024194837e-06,
+      "loss": 0.9537,
+      "step": 6344
+    },
+    {
+      "epoch": 0.6640502354788069,
+      "grad_norm": 2.284549338180081,
+      "learning_rate": 5.358310324218003e-06,
+      "loss": 0.7024,
+      "step": 6345
+    },
+    {
+      "epoch": 0.6641548927263213,
+      "grad_norm": 2.113760328686243,
+      "learning_rate": 5.355308157580725e-06,
+      "loss": 0.8261,
+      "step": 6346
+    },
+    {
+      "epoch": 0.6642595499738357,
+      "grad_norm": 2.256190025263873,
+      "learning_rate": 5.35230652462795e-06,
+      "loss": 0.9598,
+      "step": 6347
+    },
+    {
+      "epoch": 0.66436420722135,
+      "grad_norm": 1.9207259907187089,
+      "learning_rate": 5.349305425704583e-06,
+      "loss": 0.9572,
+      "step": 6348
+    },
+    {
+      "epoch": 0.6644688644688644,
+      "grad_norm": 2.3515307832470334,
+      "learning_rate": 5.346304861155445e-06,
+      "loss": 0.9013,
+      "step": 6349
+    },
+    {
+      "epoch": 0.6645735217163788,
+      "grad_norm": 1.988489105543637,
+      "learning_rate": 5.343304831325315e-06,
+      "loss": 0.8901,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6646781789638933,
+      "grad_norm": 2.1734081104477965,
+      "learning_rate": 5.340305336558902e-06,
+      "loss": 1.0388,
+      "step": 6351
+    },
+    {
+      "epoch": 0.6647828362114077,
+      "grad_norm": 2.0143701834494263,
+      "learning_rate": 5.337306377200848e-06,
+      "loss": 0.8924,
+      "step": 6352
+    },
+    {
+      "epoch": 0.6648874934589221,
+      "grad_norm": 1.9233691798259414,
+      "learning_rate": 5.334307953595747e-06,
+      "loss": 0.9476,
+      "step": 6353
+    },
+    {
+      "epoch": 0.6649921507064365,
+      "grad_norm": 2.130187551657956,
+      "learning_rate": 5.331310066088121e-06,
+      "loss": 0.8532,
+      "step": 6354
+    },
+    {
+      "epoch": 0.6650968079539508,
+      "grad_norm": 1.9538394918636528,
+      "learning_rate": 5.328312715022432e-06,
+      "loss": 0.9532,
+      "step": 6355
+    },
+    {
+      "epoch": 0.6652014652014652,
+      "grad_norm": 2.3831371289530074,
+      "learning_rate": 5.32531590074308e-06,
+      "loss": 0.9532,
+      "step": 6356
+    },
+    {
+      "epoch": 0.6653061224489796,
+      "grad_norm": 2.4969974074294585,
+      "learning_rate": 5.32231962359441e-06,
+      "loss": 0.9987,
+      "step": 6357
+    },
+    {
+      "epoch": 0.665410779696494,
+      "grad_norm": 2.176103875459444,
+      "learning_rate": 5.319323883920695e-06,
+      "loss": 0.988,
+      "step": 6358
+    },
+    {
+      "epoch": 0.6655154369440084,
+      "grad_norm": 1.6925971627256475,
+      "learning_rate": 5.316328682066157e-06,
+      "loss": 0.9428,
+      "step": 6359
+    },
+    {
+      "epoch": 0.6656200941915228,
+      "grad_norm": 2.0471516717225784,
+      "learning_rate": 5.313334018374949e-06,
+      "loss": 0.921,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6657247514390372,
+      "grad_norm": 1.9398078548541235,
+      "learning_rate": 5.310339893191161e-06,
+      "loss": 0.8818,
+      "step": 6361
+    },
+    {
+      "epoch": 0.6658294086865515,
+      "grad_norm": 1.9679163945822484,
+      "learning_rate": 5.307346306858827e-06,
+      "loss": 0.9378,
+      "step": 6362
+    },
+    {
+      "epoch": 0.6659340659340659,
+      "grad_norm": 1.9372496636716319,
+      "learning_rate": 5.304353259721917e-06,
+      "loss": 0.8058,
+      "step": 6363
+    },
+    {
+      "epoch": 0.6660387231815803,
+      "grad_norm": 1.943843808421226,
+      "learning_rate": 5.301360752124337e-06,
+      "loss": 0.8519,
+      "step": 6364
+    },
+    {
+      "epoch": 0.6661433804290947,
+      "grad_norm": 2.0411186947239552,
+      "learning_rate": 5.298368784409931e-06,
+      "loss": 0.8791,
+      "step": 6365
+    },
+    {
+      "epoch": 0.6662480376766091,
+      "grad_norm": 2.2814426931295793,
+      "learning_rate": 5.29537735692248e-06,
+      "loss": 1.0396,
+      "step": 6366
+    },
+    {
+      "epoch": 0.6663526949241235,
+      "grad_norm": 2.0786777164691888,
+      "learning_rate": 5.292386470005706e-06,
+      "loss": 0.844,
+      "step": 6367
+    },
+    {
+      "epoch": 0.6664573521716379,
+      "grad_norm": 2.04748550942362,
+      "learning_rate": 5.289396124003274e-06,
+      "loss": 1.0169,
+      "step": 6368
+    },
+    {
+      "epoch": 0.6665620094191522,
+      "grad_norm": 2.123957153054228,
+      "learning_rate": 5.286406319258779e-06,
+      "loss": 0.9936,
+      "step": 6369
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 2.0189091508075583,
+      "learning_rate": 5.2834170561157514e-06,
+      "loss": 0.8023,
+      "step": 6370
+    },
+    {
+      "epoch": 0.666771323914181,
+      "grad_norm": 1.9067663948673013,
+      "learning_rate": 5.280428334917662e-06,
+      "loss": 0.8841,
+      "step": 6371
+    },
+    {
+      "epoch": 0.6668759811616954,
+      "grad_norm": 2.0191663801552426,
+      "learning_rate": 5.277440156007929e-06,
+      "loss": 0.8464,
+      "step": 6372
+    },
+    {
+      "epoch": 0.6669806384092098,
+      "grad_norm": 1.9211457524771827,
+      "learning_rate": 5.274452519729895e-06,
+      "loss": 0.8674,
+      "step": 6373
+    },
+    {
+      "epoch": 0.6670852956567243,
+      "grad_norm": 1.9382829851230479,
+      "learning_rate": 5.271465426426847e-06,
+      "loss": 1.0491,
+      "step": 6374
+    },
+    {
+      "epoch": 0.6671899529042387,
+      "grad_norm": 1.8888165263402947,
+      "learning_rate": 5.268478876442003e-06,
+      "loss": 0.8538,
+      "step": 6375
+    },
+    {
+      "epoch": 0.667294610151753,
+      "grad_norm": 2.091423517591985,
+      "learning_rate": 5.2654928701185274e-06,
+      "loss": 0.8253,
+      "step": 6376
+    },
+    {
+      "epoch": 0.6673992673992674,
+      "grad_norm": 2.2768215094704822,
+      "learning_rate": 5.262507407799522e-06,
+      "loss": 0.9538,
+      "step": 6377
+    },
+    {
+      "epoch": 0.6675039246467818,
+      "grad_norm": 2.1494565343878533,
+      "learning_rate": 5.259522489828022e-06,
+      "loss": 0.9563,
+      "step": 6378
+    },
+    {
+      "epoch": 0.6676085818942962,
+      "grad_norm": 2.189850705311329,
+      "learning_rate": 5.2565381165469954e-06,
+      "loss": 0.8845,
+      "step": 6379
+    },
+    {
+      "epoch": 0.6677132391418106,
+      "grad_norm": 2.3476427995274882,
+      "learning_rate": 5.253554288299352e-06,
+      "loss": 0.893,
+      "step": 6380
+    },
+    {
+      "epoch": 0.667817896389325,
+      "grad_norm": 1.92825991613659,
+      "learning_rate": 5.250571005427947e-06,
+      "loss": 0.8203,
+      "step": 6381
+    },
+    {
+      "epoch": 0.6679225536368394,
+      "grad_norm": 2.031488675623292,
+      "learning_rate": 5.247588268275561e-06,
+      "loss": 0.8845,
+      "step": 6382
+    },
+    {
+      "epoch": 0.6680272108843538,
+      "grad_norm": 1.7586373295933992,
+      "learning_rate": 5.244606077184918e-06,
+      "loss": 0.7734,
+      "step": 6383
+    },
+    {
+      "epoch": 0.6681318681318681,
+      "grad_norm": 2.394679003738517,
+      "learning_rate": 5.241624432498673e-06,
+      "loss": 0.9904,
+      "step": 6384
+    },
+    {
+      "epoch": 0.6682365253793825,
+      "grad_norm": 2.2617940928572047,
+      "learning_rate": 5.23864333455943e-06,
+      "loss": 0.9123,
+      "step": 6385
+    },
+    {
+      "epoch": 0.6683411826268969,
+      "grad_norm": 6.141677769409521,
+      "learning_rate": 5.235662783709717e-06,
+      "loss": 0.8455,
+      "step": 6386
+    },
+    {
+      "epoch": 0.6684458398744113,
+      "grad_norm": 1.8536950099015304,
+      "learning_rate": 5.232682780292012e-06,
+      "loss": 0.872,
+      "step": 6387
+    },
+    {
+      "epoch": 0.6685504971219257,
+      "grad_norm": 1.9152603791182172,
+      "learning_rate": 5.22970332464872e-06,
+      "loss": 0.9247,
+      "step": 6388
+    },
+    {
+      "epoch": 0.6686551543694401,
+      "grad_norm": 2.1906359883969166,
+      "learning_rate": 5.2267244171221864e-06,
+      "loss": 0.9432,
+      "step": 6389
+    },
+    {
+      "epoch": 0.6687598116169545,
+      "grad_norm": 1.8408546196814852,
+      "learning_rate": 5.223746058054691e-06,
+      "loss": 0.9843,
+      "step": 6390
+    },
+    {
+      "epoch": 0.6688644688644688,
+      "grad_norm": 1.977899230941687,
+      "learning_rate": 5.220768247788458e-06,
+      "loss": 0.8972,
+      "step": 6391
+    },
+    {
+      "epoch": 0.6689691261119832,
+      "grad_norm": 2.0400496145208225,
+      "learning_rate": 5.217790986665639e-06,
+      "loss": 0.8671,
+      "step": 6392
+    },
+    {
+      "epoch": 0.6690737833594976,
+      "grad_norm": 2.7116537319504066,
+      "learning_rate": 5.214814275028334e-06,
+      "loss": 1.0178,
+      "step": 6393
+    },
+    {
+      "epoch": 0.669178440607012,
+      "grad_norm": 2.073375860762836,
+      "learning_rate": 5.211838113218568e-06,
+      "loss": 0.9932,
+      "step": 6394
+    },
+    {
+      "epoch": 0.6692830978545264,
+      "grad_norm": 2.096782968232844,
+      "learning_rate": 5.208862501578307e-06,
+      "loss": 0.7974,
+      "step": 6395
+    },
+    {
+      "epoch": 0.6693877551020408,
+      "grad_norm": 1.8056629099927985,
+      "learning_rate": 5.205887440449462e-06,
+      "loss": 0.8922,
+      "step": 6396
+    },
+    {
+      "epoch": 0.6694924123495553,
+      "grad_norm": 2.1562515150078734,
+      "learning_rate": 5.202912930173867e-06,
+      "loss": 0.9763,
+      "step": 6397
+    },
+    {
+      "epoch": 0.6695970695970695,
+      "grad_norm": 1.964969717746662,
+      "learning_rate": 5.1999389710933015e-06,
+      "loss": 0.7849,
+      "step": 6398
+    },
+    {
+      "epoch": 0.669701726844584,
+      "grad_norm": 2.1575433456324498,
+      "learning_rate": 5.196965563549475e-06,
+      "loss": 0.8537,
+      "step": 6399
+    },
+    {
+      "epoch": 0.6698063840920984,
+      "grad_norm": 2.2390395785938106,
+      "learning_rate": 5.193992707884045e-06,
+      "loss": 0.9165,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6699110413396128,
+      "grad_norm": 2.0512664839575727,
+      "learning_rate": 5.1910204044385935e-06,
+      "loss": 0.8743,
+      "step": 6401
+    },
+    {
+      "epoch": 0.6700156985871272,
+      "grad_norm": 1.832927563252606,
+      "learning_rate": 5.188048653554649e-06,
+      "loss": 0.9634,
+      "step": 6402
+    },
+    {
+      "epoch": 0.6701203558346416,
+      "grad_norm": 2.163369405212209,
+      "learning_rate": 5.185077455573671e-06,
+      "loss": 0.8162,
+      "step": 6403
+    },
+    {
+      "epoch": 0.670225013082156,
+      "grad_norm": 2.096116535555188,
+      "learning_rate": 5.182106810837053e-06,
+      "loss": 0.8113,
+      "step": 6404
+    },
+    {
+      "epoch": 0.6703296703296703,
+      "grad_norm": 2.2187096652736225,
+      "learning_rate": 5.179136719686124e-06,
+      "loss": 0.7653,
+      "step": 6405
+    },
+    {
+      "epoch": 0.6704343275771847,
+      "grad_norm": 1.7426313844832375,
+      "learning_rate": 5.176167182462164e-06,
+      "loss": 0.7568,
+      "step": 6406
+    },
+    {
+      "epoch": 0.6705389848246991,
+      "grad_norm": 2.610383984694903,
+      "learning_rate": 5.173198199506375e-06,
+      "loss": 1.0115,
+      "step": 6407
+    },
+    {
+      "epoch": 0.6706436420722135,
+      "grad_norm": 1.7869109945081258,
+      "learning_rate": 5.170229771159896e-06,
+      "loss": 0.8499,
+      "step": 6408
+    },
+    {
+      "epoch": 0.6707482993197279,
+      "grad_norm": 1.8327398454588113,
+      "learning_rate": 5.167261897763804e-06,
+      "loss": 0.9006,
+      "step": 6409
+    },
+    {
+      "epoch": 0.6708529565672423,
+      "grad_norm": 1.7464546014687978,
+      "learning_rate": 5.164294579659117e-06,
+      "loss": 0.8632,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6709576138147567,
+      "grad_norm": 2.0024420499608797,
+      "learning_rate": 5.161327817186789e-06,
+      "loss": 0.8386,
+      "step": 6411
+    },
+    {
+      "epoch": 0.671062271062271,
+      "grad_norm": 3.683976549889003,
+      "learning_rate": 5.158361610687704e-06,
+      "loss": 0.9202,
+      "step": 6412
+    },
+    {
+      "epoch": 0.6711669283097854,
+      "grad_norm": 2.418966089950013,
+      "learning_rate": 5.155395960502685e-06,
+      "loss": 0.8548,
+      "step": 6413
+    },
+    {
+      "epoch": 0.6712715855572998,
+      "grad_norm": 1.8796425565869193,
+      "learning_rate": 5.152430866972489e-06,
+      "loss": 0.8549,
+      "step": 6414
+    },
+    {
+      "epoch": 0.6713762428048142,
+      "grad_norm": 2.208142074213644,
+      "learning_rate": 5.1494663304378144e-06,
+      "loss": 0.9177,
+      "step": 6415
+    },
+    {
+      "epoch": 0.6714809000523286,
+      "grad_norm": 2.2446841553324903,
+      "learning_rate": 5.146502351239293e-06,
+      "loss": 0.8943,
+      "step": 6416
+    },
+    {
+      "epoch": 0.671585557299843,
+      "grad_norm": 1.8751432705355553,
+      "learning_rate": 5.143538929717491e-06,
+      "loss": 0.7813,
+      "step": 6417
+    },
+    {
+      "epoch": 0.6716902145473574,
+      "grad_norm": 1.9930880359584375,
+      "learning_rate": 5.140576066212906e-06,
+      "loss": 0.8551,
+      "step": 6418
+    },
+    {
+      "epoch": 0.6717948717948717,
+      "grad_norm": 2.4278638629645815,
+      "learning_rate": 5.137613761065983e-06,
+      "loss": 0.9474,
+      "step": 6419
+    },
+    {
+      "epoch": 0.6718995290423861,
+      "grad_norm": 2.14429691352793,
+      "learning_rate": 5.134652014617099e-06,
+      "loss": 0.9384,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6720041862899006,
+      "grad_norm": 1.8951332364379743,
+      "learning_rate": 5.131690827206563e-06,
+      "loss": 0.9869,
+      "step": 6421
+    },
+    {
+      "epoch": 0.672108843537415,
+      "grad_norm": 1.9516044400568116,
+      "learning_rate": 5.128730199174619e-06,
+      "loss": 0.885,
+      "step": 6422
+    },
+    {
+      "epoch": 0.6722135007849294,
+      "grad_norm": 2.1976500792646414,
+      "learning_rate": 5.125770130861449e-06,
+      "loss": 1.0194,
+      "step": 6423
+    },
+    {
+      "epoch": 0.6723181580324438,
+      "grad_norm": 2.0018038250643317,
+      "learning_rate": 5.12281062260717e-06,
+      "loss": 0.9788,
+      "step": 6424
+    },
+    {
+      "epoch": 0.6724228152799582,
+      "grad_norm": 2.1270551713262744,
+      "learning_rate": 5.119851674751841e-06,
+      "loss": 0.9429,
+      "step": 6425
+    },
+    {
+      "epoch": 0.6725274725274726,
+      "grad_norm": 2.025799543591311,
+      "learning_rate": 5.116893287635448e-06,
+      "loss": 0.7762,
+      "step": 6426
+    },
+    {
+      "epoch": 0.6726321297749869,
+      "grad_norm": 1.8699078179658104,
+      "learning_rate": 5.1139354615979105e-06,
+      "loss": 0.8726,
+      "step": 6427
+    },
+    {
+      "epoch": 0.6727367870225013,
+      "grad_norm": 1.9041060855902985,
+      "learning_rate": 5.110978196979098e-06,
+      "loss": 0.84,
+      "step": 6428
+    },
+    {
+      "epoch": 0.6728414442700157,
+      "grad_norm": 2.1433565757666115,
+      "learning_rate": 5.1080214941187975e-06,
+      "loss": 0.954,
+      "step": 6429
+    },
+    {
+      "epoch": 0.6729461015175301,
+      "grad_norm": 2.046061335552731,
+      "learning_rate": 5.1050653533567496e-06,
+      "loss": 0.8755,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6730507587650445,
+      "grad_norm": 1.8884753453862546,
+      "learning_rate": 5.102109775032615e-06,
+      "loss": 0.7611,
+      "step": 6431
+    },
+    {
+      "epoch": 0.6731554160125589,
+      "grad_norm": 2.033544305804191,
+      "learning_rate": 5.0991547594859955e-06,
+      "loss": 0.908,
+      "step": 6432
+    },
+    {
+      "epoch": 0.6732600732600733,
+      "grad_norm": 1.8338267562653663,
+      "learning_rate": 5.096200307056426e-06,
+      "loss": 1.005,
+      "step": 6433
+    },
+    {
+      "epoch": 0.6733647305075876,
+      "grad_norm": 2.108484710554183,
+      "learning_rate": 5.093246418083386e-06,
+      "loss": 0.9521,
+      "step": 6434
+    },
+    {
+      "epoch": 0.673469387755102,
+      "grad_norm": 2.085184894766022,
+      "learning_rate": 5.09029309290628e-06,
+      "loss": 0.8918,
+      "step": 6435
+    },
+    {
+      "epoch": 0.6735740450026164,
+      "grad_norm": 2.1541212873096636,
+      "learning_rate": 5.087340331864446e-06,
+      "loss": 0.7858,
+      "step": 6436
+    },
+    {
+      "epoch": 0.6736787022501308,
+      "grad_norm": 2.1748331560744893,
+      "learning_rate": 5.084388135297171e-06,
+      "loss": 0.9436,
+      "step": 6437
+    },
+    {
+      "epoch": 0.6737833594976452,
+      "grad_norm": 2.213867089819224,
+      "learning_rate": 5.08143650354366e-06,
+      "loss": 0.9051,
+      "step": 6438
+    },
+    {
+      "epoch": 0.6738880167451596,
+      "grad_norm": 2.276279627419914,
+      "learning_rate": 5.07848543694307e-06,
+      "loss": 0.8651,
+      "step": 6439
+    },
+    {
+      "epoch": 0.673992673992674,
+      "grad_norm": 1.7847556728333231,
+      "learning_rate": 5.075534935834481e-06,
+      "loss": 0.7922,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6740973312401883,
+      "grad_norm": 1.9203154340553588,
+      "learning_rate": 5.07258500055691e-06,
+      "loss": 0.8485,
+      "step": 6441
+    },
+    {
+      "epoch": 0.6742019884877027,
+      "grad_norm": 2.033181016154668,
+      "learning_rate": 5.069635631449311e-06,
+      "loss": 0.9363,
+      "step": 6442
+    },
+    {
+      "epoch": 0.6743066457352171,
+      "grad_norm": 1.953116403946782,
+      "learning_rate": 5.066686828850569e-06,
+      "loss": 0.9573,
+      "step": 6443
+    },
+    {
+      "epoch": 0.6744113029827316,
+      "grad_norm": 2.146061199630762,
+      "learning_rate": 5.063738593099512e-06,
+      "loss": 0.883,
+      "step": 6444
+    },
+    {
+      "epoch": 0.674515960230246,
+      "grad_norm": 2.1400402168097457,
+      "learning_rate": 5.060790924534902e-06,
+      "loss": 0.8179,
+      "step": 6445
+    },
+    {
+      "epoch": 0.6746206174777604,
+      "grad_norm": 1.7980728816836642,
+      "learning_rate": 5.057843823495425e-06,
+      "loss": 0.8783,
+      "step": 6446
+    },
+    {
+      "epoch": 0.6747252747252748,
+      "grad_norm": 2.283729906739794,
+      "learning_rate": 5.054897290319713e-06,
+      "loss": 0.8026,
+      "step": 6447
+    },
+    {
+      "epoch": 0.6748299319727891,
+      "grad_norm": 2.037589904474863,
+      "learning_rate": 5.051951325346325e-06,
+      "loss": 0.9345,
+      "step": 6448
+    },
+    {
+      "epoch": 0.6749345892203035,
+      "grad_norm": 2.1144335870638185,
+      "learning_rate": 5.049005928913762e-06,
+      "loss": 0.9946,
+      "step": 6449
+    },
+    {
+      "epoch": 0.6750392464678179,
+      "grad_norm": 1.936401258521119,
+      "learning_rate": 5.046061101360456e-06,
+      "loss": 0.8188,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6751439037153323,
+      "grad_norm": 2.2507533396021926,
+      "learning_rate": 5.04311684302477e-06,
+      "loss": 0.7269,
+      "step": 6451
+    },
+    {
+      "epoch": 0.6752485609628467,
+      "grad_norm": 1.963538133706413,
+      "learning_rate": 5.0401731542450046e-06,
+      "loss": 0.8515,
+      "step": 6452
+    },
+    {
+      "epoch": 0.6753532182103611,
+      "grad_norm": 1.9890165032492324,
+      "learning_rate": 5.037230035359398e-06,
+      "loss": 0.8833,
+      "step": 6453
+    },
+    {
+      "epoch": 0.6754578754578755,
+      "grad_norm": 1.9566124199728854,
+      "learning_rate": 5.034287486706126e-06,
+      "loss": 0.8651,
+      "step": 6454
+    },
+    {
+      "epoch": 0.6755625327053898,
+      "grad_norm": 1.9567451091759998,
+      "learning_rate": 5.031345508623287e-06,
+      "loss": 0.8554,
+      "step": 6455
+    },
+    {
+      "epoch": 0.6756671899529042,
+      "grad_norm": 2.0495146610167954,
+      "learning_rate": 5.028404101448923e-06,
+      "loss": 0.9391,
+      "step": 6456
+    },
+    {
+      "epoch": 0.6757718472004186,
+      "grad_norm": 1.9180912820360527,
+      "learning_rate": 5.025463265521001e-06,
+      "loss": 0.8958,
+      "step": 6457
+    },
+    {
+      "epoch": 0.675876504447933,
+      "grad_norm": 2.108925051546432,
+      "learning_rate": 5.0225230011774395e-06,
+      "loss": 0.9158,
+      "step": 6458
+    },
+    {
+      "epoch": 0.6759811616954474,
+      "grad_norm": 2.0210637600732655,
+      "learning_rate": 5.0195833087560745e-06,
+      "loss": 0.8969,
+      "step": 6459
+    },
+    {
+      "epoch": 0.6760858189429618,
+      "grad_norm": 1.978491474863179,
+      "learning_rate": 5.016644188594683e-06,
+      "loss": 0.9462,
+      "step": 6460
+    },
+    {
+      "epoch": 0.6761904761904762,
+      "grad_norm": 1.7918602559844456,
+      "learning_rate": 5.013705641030978e-06,
+      "loss": 0.916,
+      "step": 6461
+    },
+    {
+      "epoch": 0.6762951334379905,
+      "grad_norm": 2.1981945869138233,
+      "learning_rate": 5.010767666402599e-06,
+      "loss": 0.9247,
+      "step": 6462
+    },
+    {
+      "epoch": 0.6763997906855049,
+      "grad_norm": 2.1851426012383866,
+      "learning_rate": 5.007830265047129e-06,
+      "loss": 0.9502,
+      "step": 6463
+    },
+    {
+      "epoch": 0.6765044479330193,
+      "grad_norm": 2.2216190305636596,
+      "learning_rate": 5.004893437302085e-06,
+      "loss": 1.0375,
+      "step": 6464
+    },
+    {
+      "epoch": 0.6766091051805337,
+      "grad_norm": 1.815266465146527,
+      "learning_rate": 5.001957183504913e-06,
+      "loss": 0.9195,
+      "step": 6465
+    },
+    {
+      "epoch": 0.6767137624280481,
+      "grad_norm": 2.0023594949351593,
+      "learning_rate": 4.999021503992992e-06,
+      "loss": 0.7293,
+      "step": 6466
+    },
+    {
+      "epoch": 0.6768184196755626,
+      "grad_norm": 2.5183477674119086,
+      "learning_rate": 4.996086399103633e-06,
+      "loss": 0.9346,
+      "step": 6467
+    },
+    {
+      "epoch": 0.676923076923077,
+      "grad_norm": 2.98805111119296,
+      "learning_rate": 4.9931518691740954e-06,
+      "loss": 0.8498,
+      "step": 6468
+    },
+    {
+      "epoch": 0.6770277341705914,
+      "grad_norm": 1.9793193853571587,
+      "learning_rate": 4.990217914541559e-06,
+      "loss": 0.8217,
+      "step": 6469
+    },
+    {
+      "epoch": 0.6771323914181057,
+      "grad_norm": 2.2564143430387302,
+      "learning_rate": 4.987284535543139e-06,
+      "loss": 0.8779,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6772370486656201,
+      "grad_norm": 1.671524269897445,
+      "learning_rate": 4.984351732515883e-06,
+      "loss": 0.8147,
+      "step": 6471
+    },
+    {
+      "epoch": 0.6773417059131345,
+      "grad_norm": 2.255727997330643,
+      "learning_rate": 4.981419505796782e-06,
+      "loss": 0.9483,
+      "step": 6472
+    },
+    {
+      "epoch": 0.6774463631606489,
+      "grad_norm": 2.0870637943921917,
+      "learning_rate": 4.978487855722757e-06,
+      "loss": 0.9186,
+      "step": 6473
+    },
+    {
+      "epoch": 0.6775510204081633,
+      "grad_norm": 2.061420193131777,
+      "learning_rate": 4.975556782630657e-06,
+      "loss": 0.8609,
+      "step": 6474
+    },
+    {
+      "epoch": 0.6776556776556777,
+      "grad_norm": 2.1025694958037966,
+      "learning_rate": 4.972626286857268e-06,
+      "loss": 0.9063,
+      "step": 6475
+    },
+    {
+      "epoch": 0.6777603349031921,
+      "grad_norm": 2.0926436744834533,
+      "learning_rate": 4.969696368739308e-06,
+      "loss": 0.9524,
+      "step": 6476
+    },
+    {
+      "epoch": 0.6778649921507064,
+      "grad_norm": 2.4314254518234386,
+      "learning_rate": 4.966767028613435e-06,
+      "loss": 0.965,
+      "step": 6477
+    },
+    {
+      "epoch": 0.6779696493982208,
+      "grad_norm": 2.473343295432155,
+      "learning_rate": 4.963838266816234e-06,
+      "loss": 0.7574,
+      "step": 6478
+    },
+    {
+      "epoch": 0.6780743066457352,
+      "grad_norm": 2.163717925184423,
+      "learning_rate": 4.960910083684222e-06,
+      "loss": 0.9157,
+      "step": 6479
+    },
+    {
+      "epoch": 0.6781789638932496,
+      "grad_norm": 1.9819090705808335,
+      "learning_rate": 4.95798247955386e-06,
+      "loss": 0.7951,
+      "step": 6480
+    },
+    {
+      "epoch": 0.678283621140764,
+      "grad_norm": 2.306299973255293,
+      "learning_rate": 4.955055454761532e-06,
+      "loss": 0.8959,
+      "step": 6481
+    },
+    {
+      "epoch": 0.6783882783882784,
+      "grad_norm": 1.8508763399022372,
+      "learning_rate": 4.952129009643557e-06,
+      "loss": 0.8809,
+      "step": 6482
+    },
+    {
+      "epoch": 0.6784929356357928,
+      "grad_norm": 2.1808393788781903,
+      "learning_rate": 4.949203144536195e-06,
+      "loss": 0.8524,
+      "step": 6483
+    },
+    {
+      "epoch": 0.6785975928833071,
+      "grad_norm": 2.6263905356474657,
+      "learning_rate": 4.946277859775631e-06,
+      "loss": 0.9204,
+      "step": 6484
+    },
+    {
+      "epoch": 0.6787022501308215,
+      "grad_norm": 1.9970126875729537,
+      "learning_rate": 4.943353155697985e-06,
+      "loss": 0.8717,
+      "step": 6485
+    },
+    {
+      "epoch": 0.6788069073783359,
+      "grad_norm": 2.0984787953717556,
+      "learning_rate": 4.94042903263931e-06,
+      "loss": 0.9325,
+      "step": 6486
+    },
+    {
+      "epoch": 0.6789115646258503,
+      "grad_norm": 2.00008151496788,
+      "learning_rate": 4.937505490935599e-06,
+      "loss": 0.9307,
+      "step": 6487
+    },
+    {
+      "epoch": 0.6790162218733647,
+      "grad_norm": 1.9504354659273084,
+      "learning_rate": 4.934582530922765e-06,
+      "loss": 0.889,
+      "step": 6488
+    },
+    {
+      "epoch": 0.6791208791208792,
+      "grad_norm": 1.9738078584785779,
+      "learning_rate": 4.931660152936673e-06,
+      "loss": 0.829,
+      "step": 6489
+    },
+    {
+      "epoch": 0.6792255363683936,
+      "grad_norm": 2.258801843221124,
+      "learning_rate": 4.928738357313102e-06,
+      "loss": 0.865,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6793301936159079,
+      "grad_norm": 1.7536804465963125,
+      "learning_rate": 4.9258171443877715e-06,
+      "loss": 0.9223,
+      "step": 6491
+    },
+    {
+      "epoch": 0.6794348508634223,
+      "grad_norm": 1.8798393029654916,
+      "learning_rate": 4.922896514496341e-06,
+      "loss": 0.8812,
+      "step": 6492
+    },
+    {
+      "epoch": 0.6795395081109367,
+      "grad_norm": 1.9041463209466114,
+      "learning_rate": 4.919976467974393e-06,
+      "loss": 0.8996,
+      "step": 6493
+    },
+    {
+      "epoch": 0.6796441653584511,
+      "grad_norm": 2.2063895357599925,
+      "learning_rate": 4.917057005157447e-06,
+      "loss": 0.8618,
+      "step": 6494
+    },
+    {
+      "epoch": 0.6797488226059655,
+      "grad_norm": 2.068532864168024,
+      "learning_rate": 4.914138126380952e-06,
+      "loss": 0.9522,
+      "step": 6495
+    },
+    {
+      "epoch": 0.6798534798534799,
+      "grad_norm": 1.870324584594993,
+      "learning_rate": 4.911219831980299e-06,
+      "loss": 0.9341,
+      "step": 6496
+    },
+    {
+      "epoch": 0.6799581371009943,
+      "grad_norm": 2.014330793845203,
+      "learning_rate": 4.908302122290801e-06,
+      "loss": 0.8728,
+      "step": 6497
+    },
+    {
+      "epoch": 0.6800627943485086,
+      "grad_norm": 2.038100868952104,
+      "learning_rate": 4.905384997647715e-06,
+      "loss": 0.8701,
+      "step": 6498
+    },
+    {
+      "epoch": 0.680167451596023,
+      "grad_norm": 2.1455491337861843,
+      "learning_rate": 4.9024684583862206e-06,
+      "loss": 0.8911,
+      "step": 6499
+    },
+    {
+      "epoch": 0.6802721088435374,
+      "grad_norm": 1.903568909831072,
+      "learning_rate": 4.8995525048414325e-06,
+      "loss": 0.7694,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6803767660910518,
+      "grad_norm": 1.6985342718111633,
+      "learning_rate": 4.896637137348399e-06,
+      "loss": 0.8418,
+      "step": 6501
+    },
+    {
+      "epoch": 0.6804814233385662,
+      "grad_norm": 2.2017283202638773,
+      "learning_rate": 4.893722356242108e-06,
+      "loss": 0.9125,
+      "step": 6502
+    },
+    {
+      "epoch": 0.6805860805860806,
+      "grad_norm": 2.2531458138902702,
+      "learning_rate": 4.8908081618574685e-06,
+      "loss": 0.8669,
+      "step": 6503
+    },
+    {
+      "epoch": 0.680690737833595,
+      "grad_norm": 1.853444123308966,
+      "learning_rate": 4.88789455452933e-06,
+      "loss": 0.8773,
+      "step": 6504
+    },
+    {
+      "epoch": 0.6807953950811093,
+      "grad_norm": 2.3305108211749928,
+      "learning_rate": 4.884981534592466e-06,
+      "loss": 0.9728,
+      "step": 6505
+    },
+    {
+      "epoch": 0.6809000523286237,
+      "grad_norm": 2.424972902046859,
+      "learning_rate": 4.882069102381593e-06,
+      "loss": 0.9775,
+      "step": 6506
+    },
+    {
+      "epoch": 0.6810047095761381,
+      "grad_norm": 2.5274554940365643,
+      "learning_rate": 4.879157258231358e-06,
+      "loss": 0.8468,
+      "step": 6507
+    },
+    {
+      "epoch": 0.6811093668236525,
+      "grad_norm": 2.0276231050948446,
+      "learning_rate": 4.876246002476337e-06,
+      "loss": 0.933,
+      "step": 6508
+    },
+    {
+      "epoch": 0.6812140240711669,
+      "grad_norm": 1.9937132552470647,
+      "learning_rate": 4.873335335451036e-06,
+      "loss": 1.0518,
+      "step": 6509
+    },
+    {
+      "epoch": 0.6813186813186813,
+      "grad_norm": 1.920197673037133,
+      "learning_rate": 4.870425257489895e-06,
+      "loss": 0.8334,
+      "step": 6510
+    },
+    {
+      "epoch": 0.6814233385661957,
+      "grad_norm": 1.7724221748359235,
+      "learning_rate": 4.8675157689272936e-06,
+      "loss": 0.8897,
+      "step": 6511
+    },
+    {
+      "epoch": 0.6815279958137102,
+      "grad_norm": 2.30963352647495,
+      "learning_rate": 4.864606870097535e-06,
+      "loss": 0.867,
+      "step": 6512
+    },
+    {
+      "epoch": 0.6816326530612244,
+      "grad_norm": 1.9385477533205635,
+      "learning_rate": 4.861698561334858e-06,
+      "loss": 0.842,
+      "step": 6513
+    },
+    {
+      "epoch": 0.6817373103087389,
+      "grad_norm": 2.362323418390193,
+      "learning_rate": 4.858790842973428e-06,
+      "loss": 1.0088,
+      "step": 6514
+    },
+    {
+      "epoch": 0.6818419675562533,
+      "grad_norm": 1.9262422529795675,
+      "learning_rate": 4.855883715347353e-06,
+      "loss": 0.8423,
+      "step": 6515
+    },
+    {
+      "epoch": 0.6819466248037677,
+      "grad_norm": 1.9990004709172928,
+      "learning_rate": 4.852977178790671e-06,
+      "loss": 0.9374,
+      "step": 6516
+    },
+    {
+      "epoch": 0.6820512820512821,
+      "grad_norm": 2.3761732334999084,
+      "learning_rate": 4.850071233637345e-06,
+      "loss": 0.9739,
+      "step": 6517
+    },
+    {
+      "epoch": 0.6821559392987965,
+      "grad_norm": 1.9130503375804242,
+      "learning_rate": 4.847165880221275e-06,
+      "loss": 0.9095,
+      "step": 6518
+    },
+    {
+      "epoch": 0.6822605965463109,
+      "grad_norm": 1.8654272704894526,
+      "learning_rate": 4.844261118876291e-06,
+      "loss": 0.8022,
+      "step": 6519
+    },
+    {
+      "epoch": 0.6823652537938252,
+      "grad_norm": 2.2423240527964166,
+      "learning_rate": 4.841356949936152e-06,
+      "loss": 1.0298,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6824699110413396,
+      "grad_norm": 2.0407588030898745,
+      "learning_rate": 4.838453373734562e-06,
+      "loss": 0.947,
+      "step": 6521
+    },
+    {
+      "epoch": 0.682574568288854,
+      "grad_norm": 2.273219656507149,
+      "learning_rate": 4.835550390605141e-06,
+      "loss": 0.8439,
+      "step": 6522
+    },
+    {
+      "epoch": 0.6826792255363684,
+      "grad_norm": 2.3321959685240374,
+      "learning_rate": 4.832648000881448e-06,
+      "loss": 0.9268,
+      "step": 6523
+    },
+    {
+      "epoch": 0.6827838827838828,
+      "grad_norm": 1.922263615943614,
+      "learning_rate": 4.829746204896978e-06,
+      "loss": 0.953,
+      "step": 6524
+    },
+    {
+      "epoch": 0.6828885400313972,
+      "grad_norm": 1.9490059628377225,
+      "learning_rate": 4.8268450029851456e-06,
+      "loss": 0.8965,
+      "step": 6525
+    },
+    {
+      "epoch": 0.6829931972789116,
+      "grad_norm": 2.0239108880099144,
+      "learning_rate": 4.823944395479314e-06,
+      "loss": 0.9316,
+      "step": 6526
+    },
+    {
+      "epoch": 0.6830978545264259,
+      "grad_norm": 2.019795192924344,
+      "learning_rate": 4.821044382712764e-06,
+      "loss": 0.9969,
+      "step": 6527
+    },
+    {
+      "epoch": 0.6832025117739403,
+      "grad_norm": 2.066141054067395,
+      "learning_rate": 4.818144965018714e-06,
+      "loss": 0.911,
+      "step": 6528
+    },
+    {
+      "epoch": 0.6833071690214547,
+      "grad_norm": 2.085608618994838,
+      "learning_rate": 4.8152461427303075e-06,
+      "loss": 0.8309,
+      "step": 6529
+    },
+    {
+      "epoch": 0.6834118262689691,
+      "grad_norm": 2.546699608829966,
+      "learning_rate": 4.812347916180634e-06,
+      "loss": 0.9144,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6835164835164835,
+      "grad_norm": 2.214180533685667,
+      "learning_rate": 4.809450285702697e-06,
+      "loss": 0.8954,
+      "step": 6531
+    },
+    {
+      "epoch": 0.6836211407639979,
+      "grad_norm": 2.308704951945097,
+      "learning_rate": 4.806553251629449e-06,
+      "loss": 0.7669,
+      "step": 6532
+    },
+    {
+      "epoch": 0.6837257980115123,
+      "grad_norm": 1.9800008001673324,
+      "learning_rate": 4.803656814293761e-06,
+      "loss": 0.8632,
+      "step": 6533
+    },
+    {
+      "epoch": 0.6838304552590266,
+      "grad_norm": 1.6800482696957828,
+      "learning_rate": 4.800760974028435e-06,
+      "loss": 0.816,
+      "step": 6534
+    },
+    {
+      "epoch": 0.683935112506541,
+      "grad_norm": 2.1903832073879825,
+      "learning_rate": 4.7978657311662155e-06,
+      "loss": 0.9341,
+      "step": 6535
+    },
+    {
+      "epoch": 0.6840397697540554,
+      "grad_norm": 2.1866955766796607,
+      "learning_rate": 4.794971086039771e-06,
+      "loss": 0.9187,
+      "step": 6536
+    },
+    {
+      "epoch": 0.6841444270015699,
+      "grad_norm": 2.171235833925488,
+      "learning_rate": 4.7920770389816995e-06,
+      "loss": 0.9389,
+      "step": 6537
+    },
+    {
+      "epoch": 0.6842490842490843,
+      "grad_norm": 2.133592554033059,
+      "learning_rate": 4.7891835903245345e-06,
+      "loss": 0.8337,
+      "step": 6538
+    },
+    {
+      "epoch": 0.6843537414965987,
+      "grad_norm": 2.164792966032931,
+      "learning_rate": 4.786290740400734e-06,
+      "loss": 0.8745,
+      "step": 6539
+    },
+    {
+      "epoch": 0.6844583987441131,
+      "grad_norm": 2.566602076396997,
+      "learning_rate": 4.783398489542696e-06,
+      "loss": 1.0129,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6845630559916274,
+      "grad_norm": 2.1672816313961385,
+      "learning_rate": 4.780506838082752e-06,
+      "loss": 0.91,
+      "step": 6541
+    },
+    {
+      "epoch": 0.6846677132391418,
+      "grad_norm": 2.5299316051254173,
+      "learning_rate": 4.7776157863531535e-06,
+      "loss": 0.8804,
+      "step": 6542
+    },
+    {
+      "epoch": 0.6847723704866562,
+      "grad_norm": 2.050095970695568,
+      "learning_rate": 4.7747253346860865e-06,
+      "loss": 0.9931,
+      "step": 6543
+    },
+    {
+      "epoch": 0.6848770277341706,
+      "grad_norm": 1.950038717236096,
+      "learning_rate": 4.771835483413668e-06,
+      "loss": 0.8983,
+      "step": 6544
+    },
+    {
+      "epoch": 0.684981684981685,
+      "grad_norm": 1.7294746393168354,
+      "learning_rate": 4.768946232867956e-06,
+      "loss": 0.7711,
+      "step": 6545
+    },
+    {
+      "epoch": 0.6850863422291994,
+      "grad_norm": 2.1177999258178346,
+      "learning_rate": 4.766057583380925e-06,
+      "loss": 0.9021,
+      "step": 6546
+    },
+    {
+      "epoch": 0.6851909994767138,
+      "grad_norm": 2.1071905114151597,
+      "learning_rate": 4.763169535284488e-06,
+      "loss": 0.923,
+      "step": 6547
+    },
+    {
+      "epoch": 0.6852956567242282,
+      "grad_norm": 2.186034990894795,
+      "learning_rate": 4.760282088910485e-06,
+      "loss": 0.9689,
+      "step": 6548
+    },
+    {
+      "epoch": 0.6854003139717425,
+      "grad_norm": 2.0749256838860552,
+      "learning_rate": 4.757395244590692e-06,
+      "loss": 0.9672,
+      "step": 6549
+    },
+    {
+      "epoch": 0.6855049712192569,
+      "grad_norm": 1.9694250708561203,
+      "learning_rate": 4.754509002656815e-06,
+      "loss": 0.936,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6856096284667713,
+      "grad_norm": 1.698078542210978,
+      "learning_rate": 4.751623363440488e-06,
+      "loss": 0.8718,
+      "step": 6551
+    },
+    {
+      "epoch": 0.6857142857142857,
+      "grad_norm": 2.0573301693561583,
+      "learning_rate": 4.748738327273277e-06,
+      "loss": 0.93,
+      "step": 6552
+    },
+    {
+      "epoch": 0.6858189429618001,
+      "grad_norm": 2.2826375431191774,
+      "learning_rate": 4.745853894486674e-06,
+      "loss": 1.0443,
+      "step": 6553
+    },
+    {
+      "epoch": 0.6859236002093145,
+      "grad_norm": 1.8209812613407657,
+      "learning_rate": 4.742970065412112e-06,
+      "loss": 0.8636,
+      "step": 6554
+    },
+    {
+      "epoch": 0.6860282574568289,
+      "grad_norm": 2.050736728583209,
+      "learning_rate": 4.740086840380948e-06,
+      "loss": 0.9672,
+      "step": 6555
+    },
+    {
+      "epoch": 0.6861329147043432,
+      "grad_norm": 1.9397854484778958,
+      "learning_rate": 4.7372042197244684e-06,
+      "loss": 0.9258,
+      "step": 6556
+    },
+    {
+      "epoch": 0.6862375719518576,
+      "grad_norm": 2.086617574002967,
+      "learning_rate": 4.734322203773889e-06,
+      "loss": 0.9393,
+      "step": 6557
+    },
+    {
+      "epoch": 0.686342229199372,
+      "grad_norm": 2.012325233074569,
+      "learning_rate": 4.731440792860368e-06,
+      "loss": 0.8641,
+      "step": 6558
+    },
+    {
+      "epoch": 0.6864468864468865,
+      "grad_norm": 2.1204083885533227,
+      "learning_rate": 4.728559987314975e-06,
+      "loss": 0.9989,
+      "step": 6559
+    },
+    {
+      "epoch": 0.6865515436944009,
+      "grad_norm": 2.148258411085086,
+      "learning_rate": 4.72567978746873e-06,
+      "loss": 0.8573,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6866562009419153,
+      "grad_norm": 2.079331381933887,
+      "learning_rate": 4.72280019365257e-06,
+      "loss": 0.8727,
+      "step": 6561
+    },
+    {
+      "epoch": 0.6867608581894297,
+      "grad_norm": 1.8462912023639682,
+      "learning_rate": 4.719921206197365e-06,
+      "loss": 0.8153,
+      "step": 6562
+    },
+    {
+      "epoch": 0.686865515436944,
+      "grad_norm": 2.1589001836310433,
+      "learning_rate": 4.717042825433914e-06,
+      "loss": 0.9984,
+      "step": 6563
+    },
+    {
+      "epoch": 0.6869701726844584,
+      "grad_norm": 2.0590111114309293,
+      "learning_rate": 4.714165051692956e-06,
+      "loss": 0.9904,
+      "step": 6564
+    },
+    {
+      "epoch": 0.6870748299319728,
+      "grad_norm": 2.0472445373680217,
+      "learning_rate": 4.711287885305149e-06,
+      "loss": 0.8908,
+      "step": 6565
+    },
+    {
+      "epoch": 0.6871794871794872,
+      "grad_norm": 2.1696081831633967,
+      "learning_rate": 4.708411326601081e-06,
+      "loss": 0.9935,
+      "step": 6566
+    },
+    {
+      "epoch": 0.6872841444270016,
+      "grad_norm": 2.7641764279513628,
+      "learning_rate": 4.705535375911283e-06,
+      "loss": 0.9877,
+      "step": 6567
+    },
+    {
+      "epoch": 0.687388801674516,
+      "grad_norm": 2.131915916399177,
+      "learning_rate": 4.7026600335662e-06,
+      "loss": 0.9278,
+      "step": 6568
+    },
+    {
+      "epoch": 0.6874934589220304,
+      "grad_norm": 1.8259729970596064,
+      "learning_rate": 4.69978529989622e-06,
+      "loss": 0.7988,
+      "step": 6569
+    },
+    {
+      "epoch": 0.6875981161695447,
+      "grad_norm": 2.1246313055880135,
+      "learning_rate": 4.696911175231655e-06,
+      "loss": 0.9864,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6877027734170591,
+      "grad_norm": 2.028298179481332,
+      "learning_rate": 4.694037659902747e-06,
+      "loss": 0.9062,
+      "step": 6571
+    },
+    {
+      "epoch": 0.6878074306645735,
+      "grad_norm": 1.6089562184719646,
+      "learning_rate": 4.691164754239663e-06,
+      "loss": 0.7789,
+      "step": 6572
+    },
+    {
+      "epoch": 0.6879120879120879,
+      "grad_norm": 1.8248486409728757,
+      "learning_rate": 4.6882924585725155e-06,
+      "loss": 0.8198,
+      "step": 6573
+    },
+    {
+      "epoch": 0.6880167451596023,
+      "grad_norm": 2.095291077164802,
+      "learning_rate": 4.685420773231333e-06,
+      "loss": 0.9641,
+      "step": 6574
+    },
+    {
+      "epoch": 0.6881214024071167,
+      "grad_norm": 2.2450995973799013,
+      "learning_rate": 4.682549698546073e-06,
+      "loss": 0.9317,
+      "step": 6575
+    },
+    {
+      "epoch": 0.6882260596546311,
+      "grad_norm": 2.0526149027810283,
+      "learning_rate": 4.679679234846636e-06,
+      "loss": 0.9398,
+      "step": 6576
+    },
+    {
+      "epoch": 0.6883307169021454,
+      "grad_norm": 1.7534041188486345,
+      "learning_rate": 4.67680938246284e-06,
+      "loss": 0.8108,
+      "step": 6577
+    },
+    {
+      "epoch": 0.6884353741496598,
+      "grad_norm": 2.0637109461898553,
+      "learning_rate": 4.6739401417244335e-06,
+      "loss": 0.9386,
+      "step": 6578
+    },
+    {
+      "epoch": 0.6885400313971742,
+      "grad_norm": 1.954071200949578,
+      "learning_rate": 4.671071512961106e-06,
+      "loss": 0.8725,
+      "step": 6579
+    },
+    {
+      "epoch": 0.6886446886446886,
+      "grad_norm": 2.128303103825469,
+      "learning_rate": 4.668203496502464e-06,
+      "loss": 0.9578,
+      "step": 6580
+    },
+    {
+      "epoch": 0.688749345892203,
+      "grad_norm": 1.8201644126560876,
+      "learning_rate": 4.665336092678049e-06,
+      "loss": 0.903,
+      "step": 6581
+    },
+    {
+      "epoch": 0.6888540031397175,
+      "grad_norm": 2.1543435935120763,
+      "learning_rate": 4.6624693018173285e-06,
+      "loss": 0.9748,
+      "step": 6582
+    },
+    {
+      "epoch": 0.6889586603872319,
+      "grad_norm": 2.1608061059358765,
+      "learning_rate": 4.659603124249704e-06,
+      "loss": 0.9338,
+      "step": 6583
+    },
+    {
+      "epoch": 0.6890633176347462,
+      "grad_norm": 2.260571417991753,
+      "learning_rate": 4.6567375603045114e-06,
+      "loss": 0.8783,
+      "step": 6584
+    },
+    {
+      "epoch": 0.6891679748822606,
+      "grad_norm": 2.0583659884788923,
+      "learning_rate": 4.653872610311005e-06,
+      "loss": 0.922,
+      "step": 6585
+    },
+    {
+      "epoch": 0.689272632129775,
+      "grad_norm": 1.7393344357365446,
+      "learning_rate": 4.651008274598373e-06,
+      "loss": 0.8636,
+      "step": 6586
+    },
+    {
+      "epoch": 0.6893772893772894,
+      "grad_norm": 2.1302121690545084,
+      "learning_rate": 4.648144553495732e-06,
+      "loss": 0.895,
+      "step": 6587
+    },
+    {
+      "epoch": 0.6894819466248038,
+      "grad_norm": 1.870330710588847,
+      "learning_rate": 4.645281447332133e-06,
+      "loss": 0.9744,
+      "step": 6588
+    },
+    {
+      "epoch": 0.6895866038723182,
+      "grad_norm": 2.1942089129705358,
+      "learning_rate": 4.642418956436551e-06,
+      "loss": 0.8749,
+      "step": 6589
+    },
+    {
+      "epoch": 0.6896912611198326,
+      "grad_norm": 1.7509480014685161,
+      "learning_rate": 4.639557081137891e-06,
+      "loss": 0.8509,
+      "step": 6590
+    },
+    {
+      "epoch": 0.689795918367347,
+      "grad_norm": 2.0185777788794015,
+      "learning_rate": 4.636695821764987e-06,
+      "loss": 0.9448,
+      "step": 6591
+    },
+    {
+      "epoch": 0.6899005756148613,
+      "grad_norm": 2.0964031784253674,
+      "learning_rate": 4.633835178646605e-06,
+      "loss": 0.9755,
+      "step": 6592
+    },
+    {
+      "epoch": 0.6900052328623757,
+      "grad_norm": 2.1491220196897958,
+      "learning_rate": 4.630975152111443e-06,
+      "loss": 0.9519,
+      "step": 6593
+    },
+    {
+      "epoch": 0.6901098901098901,
+      "grad_norm": 1.9800787335964507,
+      "learning_rate": 4.628115742488119e-06,
+      "loss": 0.8376,
+      "step": 6594
+    },
+    {
+      "epoch": 0.6902145473574045,
+      "grad_norm": 2.2297898584366918,
+      "learning_rate": 4.625256950105188e-06,
+      "loss": 0.9368,
+      "step": 6595
+    },
+    {
+      "epoch": 0.6903192046049189,
+      "grad_norm": 1.9376950455128639,
+      "learning_rate": 4.622398775291129e-06,
+      "loss": 0.7967,
+      "step": 6596
+    },
+    {
+      "epoch": 0.6904238618524333,
+      "grad_norm": 2.1640506849550776,
+      "learning_rate": 4.6195412183743485e-06,
+      "loss": 0.965,
+      "step": 6597
+    },
+    {
+      "epoch": 0.6905285190999477,
+      "grad_norm": 2.1629346113510945,
+      "learning_rate": 4.616684279683193e-06,
+      "loss": 0.7332,
+      "step": 6598
+    },
+    {
+      "epoch": 0.690633176347462,
+      "grad_norm": 2.4687954434374264,
+      "learning_rate": 4.6138279595459255e-06,
+      "loss": 0.9789,
+      "step": 6599
+    },
+    {
+      "epoch": 0.6907378335949764,
+      "grad_norm": 1.9293003022663573,
+      "learning_rate": 4.610972258290745e-06,
+      "loss": 0.8768,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6908424908424908,
+      "grad_norm": 2.000231072203521,
+      "learning_rate": 4.608117176245773e-06,
+      "loss": 0.9758,
+      "step": 6601
+    },
+    {
+      "epoch": 0.6909471480900052,
+      "grad_norm": 2.223037956191113,
+      "learning_rate": 4.60526271373907e-06,
+      "loss": 0.9322,
+      "step": 6602
+    },
+    {
+      "epoch": 0.6910518053375196,
+      "grad_norm": 1.9628517839572384,
+      "learning_rate": 4.602408871098618e-06,
+      "loss": 0.8642,
+      "step": 6603
+    },
+    {
+      "epoch": 0.691156462585034,
+      "grad_norm": 1.985045530723375,
+      "learning_rate": 4.599555648652331e-06,
+      "loss": 0.9152,
+      "step": 6604
+    },
+    {
+      "epoch": 0.6912611198325485,
+      "grad_norm": 1.9758634460748625,
+      "learning_rate": 4.5967030467280475e-06,
+      "loss": 0.8055,
+      "step": 6605
+    },
+    {
+      "epoch": 0.6913657770800627,
+      "grad_norm": 3.103005744229717,
+      "learning_rate": 4.5938510656535325e-06,
+      "loss": 0.8094,
+      "step": 6606
+    },
+    {
+      "epoch": 0.6914704343275772,
+      "grad_norm": 2.1448807382230024,
+      "learning_rate": 4.590999705756495e-06,
+      "loss": 0.9655,
+      "step": 6607
+    },
+    {
+      "epoch": 0.6915750915750916,
+      "grad_norm": 1.8041089841225364,
+      "learning_rate": 4.588148967364555e-06,
+      "loss": 0.8029,
+      "step": 6608
+    },
+    {
+      "epoch": 0.691679748822606,
+      "grad_norm": 2.0056877275259106,
+      "learning_rate": 4.585298850805266e-06,
+      "loss": 0.8963,
+      "step": 6609
+    },
+    {
+      "epoch": 0.6917844060701204,
+      "grad_norm": 1.993991355774141,
+      "learning_rate": 4.58244935640612e-06,
+      "loss": 0.8601,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6918890633176348,
+      "grad_norm": 1.8982333609414117,
+      "learning_rate": 4.579600484494522e-06,
+      "loss": 0.9538,
+      "step": 6611
+    },
+    {
+      "epoch": 0.6919937205651492,
+      "grad_norm": 2.2109365168453623,
+      "learning_rate": 4.57675223539782e-06,
+      "loss": 0.95,
+      "step": 6612
+    },
+    {
+      "epoch": 0.6920983778126635,
+      "grad_norm": 2.077498771030345,
+      "learning_rate": 4.573904609443281e-06,
+      "loss": 0.9701,
+      "step": 6613
+    },
+    {
+      "epoch": 0.6922030350601779,
+      "grad_norm": 2.706301406949021,
+      "learning_rate": 4.571057606958102e-06,
+      "loss": 0.9979,
+      "step": 6614
+    },
+    {
+      "epoch": 0.6923076923076923,
+      "grad_norm": 1.9123792412988372,
+      "learning_rate": 4.568211228269408e-06,
+      "loss": 0.8161,
+      "step": 6615
+    },
+    {
+      "epoch": 0.6924123495552067,
+      "grad_norm": 2.0744022076341393,
+      "learning_rate": 4.565365473704253e-06,
+      "loss": 0.8749,
+      "step": 6616
+    },
+    {
+      "epoch": 0.6925170068027211,
+      "grad_norm": 1.9700768269508973,
+      "learning_rate": 4.5625203435896244e-06,
+      "loss": 0.7862,
+      "step": 6617
+    },
+    {
+      "epoch": 0.6926216640502355,
+      "grad_norm": 2.1381491046390257,
+      "learning_rate": 4.55967583825243e-06,
+      "loss": 0.8554,
+      "step": 6618
+    },
+    {
+      "epoch": 0.6927263212977499,
+      "grad_norm": 1.912057699434843,
+      "learning_rate": 4.556831958019512e-06,
+      "loss": 0.884,
+      "step": 6619
+    },
+    {
+      "epoch": 0.6928309785452642,
+      "grad_norm": 1.8093614568345422,
+      "learning_rate": 4.553988703217638e-06,
+      "loss": 0.9215,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6929356357927786,
+      "grad_norm": 1.9690519624929577,
+      "learning_rate": 4.551146074173497e-06,
+      "loss": 0.8911,
+      "step": 6621
+    },
+    {
+      "epoch": 0.693040293040293,
+      "grad_norm": 2.0811151718511525,
+      "learning_rate": 4.548304071213723e-06,
+      "loss": 1.0378,
+      "step": 6622
+    },
+    {
+      "epoch": 0.6931449502878074,
+      "grad_norm": 1.9485593151139378,
+      "learning_rate": 4.545462694664863e-06,
+      "loss": 0.8596,
+      "step": 6623
+    },
+    {
+      "epoch": 0.6932496075353218,
+      "grad_norm": 2.1609504606964784,
+      "learning_rate": 4.542621944853396e-06,
+      "loss": 0.8655,
+      "step": 6624
+    },
+    {
+      "epoch": 0.6933542647828362,
+      "grad_norm": 1.897884223321221,
+      "learning_rate": 4.539781822105728e-06,
+      "loss": 0.8386,
+      "step": 6625
+    },
+    {
+      "epoch": 0.6934589220303506,
+      "grad_norm": 1.72032226811971,
+      "learning_rate": 4.5369423267482004e-06,
+      "loss": 0.8291,
+      "step": 6626
+    },
+    {
+      "epoch": 0.6935635792778649,
+      "grad_norm": 2.2020630417755997,
+      "learning_rate": 4.534103459107071e-06,
+      "loss": 0.8895,
+      "step": 6627
+    },
+    {
+      "epoch": 0.6936682365253793,
+      "grad_norm": 2.0309417972154664,
+      "learning_rate": 4.5312652195085385e-06,
+      "loss": 0.9303,
+      "step": 6628
+    },
+    {
+      "epoch": 0.6937728937728938,
+      "grad_norm": 2.0639435237790953,
+      "learning_rate": 4.528427608278718e-06,
+      "loss": 0.9642,
+      "step": 6629
+    },
+    {
+      "epoch": 0.6938775510204082,
+      "grad_norm": 2.1146137247817527,
+      "learning_rate": 4.525590625743654e-06,
+      "loss": 0.8332,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6939822082679226,
+      "grad_norm": 2.218674423062428,
+      "learning_rate": 4.522754272229329e-06,
+      "loss": 0.99,
+      "step": 6631
+    },
+    {
+      "epoch": 0.694086865515437,
+      "grad_norm": 2.1718805339372045,
+      "learning_rate": 4.519918548061642e-06,
+      "loss": 0.974,
+      "step": 6632
+    },
+    {
+      "epoch": 0.6941915227629514,
+      "grad_norm": 1.925204404610261,
+      "learning_rate": 4.517083453566422e-06,
+      "loss": 0.8138,
+      "step": 6633
+    },
+    {
+      "epoch": 0.6942961800104658,
+      "grad_norm": 2.5317634962266538,
+      "learning_rate": 4.514248989069428e-06,
+      "loss": 1.0308,
+      "step": 6634
+    },
+    {
+      "epoch": 0.6944008372579801,
+      "grad_norm": 1.8774753895683007,
+      "learning_rate": 4.511415154896343e-06,
+      "loss": 0.8771,
+      "step": 6635
+    },
+    {
+      "epoch": 0.6945054945054945,
+      "grad_norm": 2.006100990453835,
+      "learning_rate": 4.508581951372783e-06,
+      "loss": 0.9325,
+      "step": 6636
+    },
+    {
+      "epoch": 0.6946101517530089,
+      "grad_norm": 1.9301716377433031,
+      "learning_rate": 4.505749378824294e-06,
+      "loss": 0.89,
+      "step": 6637
+    },
+    {
+      "epoch": 0.6947148090005233,
+      "grad_norm": 1.851932126585198,
+      "learning_rate": 4.502917437576338e-06,
+      "loss": 0.9756,
+      "step": 6638
+    },
+    {
+      "epoch": 0.6948194662480377,
+      "grad_norm": 1.8626235138561205,
+      "learning_rate": 4.500086127954313e-06,
+      "loss": 0.9258,
+      "step": 6639
+    },
+    {
+      "epoch": 0.6949241234955521,
+      "grad_norm": 1.8143489151160312,
+      "learning_rate": 4.497255450283537e-06,
+      "loss": 0.8582,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6950287807430665,
+      "grad_norm": 1.682072962368349,
+      "learning_rate": 4.4944254048892685e-06,
+      "loss": 0.8266,
+      "step": 6641
+    },
+    {
+      "epoch": 0.6951334379905808,
+      "grad_norm": 2.2156198148472828,
+      "learning_rate": 4.491595992096682e-06,
+      "loss": 0.9815,
+      "step": 6642
+    },
+    {
+      "epoch": 0.6952380952380952,
+      "grad_norm": 1.9277959774869025,
+      "learning_rate": 4.488767212230884e-06,
+      "loss": 0.9182,
+      "step": 6643
+    },
+    {
+      "epoch": 0.6953427524856096,
+      "grad_norm": 2.2062231983360925,
+      "learning_rate": 4.4859390656169e-06,
+      "loss": 0.9683,
+      "step": 6644
+    },
+    {
+      "epoch": 0.695447409733124,
+      "grad_norm": 1.9811122926788303,
+      "learning_rate": 4.4831115525796965e-06,
+      "loss": 0.8219,
+      "step": 6645
+    },
+    {
+      "epoch": 0.6955520669806384,
+      "grad_norm": 2.0834949431347014,
+      "learning_rate": 4.480284673444163e-06,
+      "loss": 1.0044,
+      "step": 6646
+    },
+    {
+      "epoch": 0.6956567242281528,
+      "grad_norm": 1.9488184187877493,
+      "learning_rate": 4.477458428535111e-06,
+      "loss": 0.9673,
+      "step": 6647
+    },
+    {
+      "epoch": 0.6957613814756672,
+      "grad_norm": 2.297182952318726,
+      "learning_rate": 4.47463281817728e-06,
+      "loss": 0.9403,
+      "step": 6648
+    },
+    {
+      "epoch": 0.6958660387231815,
+      "grad_norm": 1.7266007102300813,
+      "learning_rate": 4.471807842695339e-06,
+      "loss": 0.7784,
+      "step": 6649
+    },
+    {
+      "epoch": 0.6959706959706959,
+      "grad_norm": 2.228867198331409,
+      "learning_rate": 4.468983502413882e-06,
+      "loss": 0.9004,
+      "step": 6650
+    },
+    {
+      "epoch": 0.6960753532182103,
+      "grad_norm": 2.141327124826885,
+      "learning_rate": 4.466159797657434e-06,
+      "loss": 0.9162,
+      "step": 6651
+    },
+    {
+      "epoch": 0.6961800104657248,
+      "grad_norm": 2.0821483089510937,
+      "learning_rate": 4.463336728750446e-06,
+      "loss": 0.8988,
+      "step": 6652
+    },
+    {
+      "epoch": 0.6962846677132392,
+      "grad_norm": 2.0393438945172804,
+      "learning_rate": 4.460514296017286e-06,
+      "loss": 0.9121,
+      "step": 6653
+    },
+    {
+      "epoch": 0.6963893249607536,
+      "grad_norm": 2.407022437793897,
+      "learning_rate": 4.457692499782269e-06,
+      "loss": 0.9244,
+      "step": 6654
+    },
+    {
+      "epoch": 0.696493982208268,
+      "grad_norm": 2.222925129117027,
+      "learning_rate": 4.454871340369614e-06,
+      "loss": 0.9437,
+      "step": 6655
+    },
+    {
+      "epoch": 0.6965986394557823,
+      "grad_norm": 2.113703320766968,
+      "learning_rate": 4.452050818103487e-06,
+      "loss": 0.8637,
+      "step": 6656
+    },
+    {
+      "epoch": 0.6967032967032967,
+      "grad_norm": 1.79736876167149,
+      "learning_rate": 4.4492309333079685e-06,
+      "loss": 0.7871,
+      "step": 6657
+    },
+    {
+      "epoch": 0.6968079539508111,
+      "grad_norm": 2.035698628522382,
+      "learning_rate": 4.446411686307067e-06,
+      "loss": 1.034,
+      "step": 6658
+    },
+    {
+      "epoch": 0.6969126111983255,
+      "grad_norm": 1.9079323228404186,
+      "learning_rate": 4.443593077424718e-06,
+      "loss": 0.9502,
+      "step": 6659
+    },
+    {
+      "epoch": 0.6970172684458399,
+      "grad_norm": 2.018868747120195,
+      "learning_rate": 4.440775106984793e-06,
+      "loss": 0.8453,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6971219256933543,
+      "grad_norm": 1.9769821063615076,
+      "learning_rate": 4.437957775311073e-06,
+      "loss": 1.0526,
+      "step": 6661
+    },
+    {
+      "epoch": 0.6972265829408687,
+      "grad_norm": 1.8727187713600468,
+      "learning_rate": 4.435141082727285e-06,
+      "loss": 0.8593,
+      "step": 6662
+    },
+    {
+      "epoch": 0.697331240188383,
+      "grad_norm": 2.087270497186403,
+      "learning_rate": 4.432325029557067e-06,
+      "loss": 0.8752,
+      "step": 6663
+    },
+    {
+      "epoch": 0.6974358974358974,
+      "grad_norm": 2.1401689080867365,
+      "learning_rate": 4.429509616123987e-06,
+      "loss": 0.9799,
+      "step": 6664
+    },
+    {
+      "epoch": 0.6975405546834118,
+      "grad_norm": 1.7455887505421854,
+      "learning_rate": 4.426694842751548e-06,
+      "loss": 0.9056,
+      "step": 6665
+    },
+    {
+      "epoch": 0.6976452119309262,
+      "grad_norm": 2.0761648451950054,
+      "learning_rate": 4.4238807097631685e-06,
+      "loss": 0.6883,
+      "step": 6666
+    },
+    {
+      "epoch": 0.6977498691784406,
+      "grad_norm": 2.4630286085688673,
+      "learning_rate": 4.421067217482201e-06,
+      "loss": 0.8402,
+      "step": 6667
+    },
+    {
+      "epoch": 0.697854526425955,
+      "grad_norm": 2.2700430686821496,
+      "learning_rate": 4.418254366231919e-06,
+      "loss": 0.8151,
+      "step": 6668
+    },
+    {
+      "epoch": 0.6979591836734694,
+      "grad_norm": 2.106078902374301,
+      "learning_rate": 4.415442156335522e-06,
+      "loss": 0.974,
+      "step": 6669
+    },
+    {
+      "epoch": 0.6980638409209837,
+      "grad_norm": 2.1498422358932547,
+      "learning_rate": 4.412630588116144e-06,
+      "loss": 0.9164,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6981684981684981,
+      "grad_norm": 2.1170538037613413,
+      "learning_rate": 4.409819661896839e-06,
+      "loss": 0.9384,
+      "step": 6671
+    },
+    {
+      "epoch": 0.6982731554160125,
+      "grad_norm": 2.0390019015189864,
+      "learning_rate": 4.40700937800059e-06,
+      "loss": 0.9529,
+      "step": 6672
+    },
+    {
+      "epoch": 0.6983778126635269,
+      "grad_norm": 2.1816342624312215,
+      "learning_rate": 4.4041997367503e-06,
+      "loss": 0.7565,
+      "step": 6673
+    },
+    {
+      "epoch": 0.6984824699110413,
+      "grad_norm": 2.0528691851303353,
+      "learning_rate": 4.401390738468801e-06,
+      "loss": 0.9013,
+      "step": 6674
+    },
+    {
+      "epoch": 0.6985871271585558,
+      "grad_norm": 2.2025257423746725,
+      "learning_rate": 4.39858238347886e-06,
+      "loss": 0.8969,
+      "step": 6675
+    },
+    {
+      "epoch": 0.6986917844060702,
+      "grad_norm": 2.1117193529663916,
+      "learning_rate": 4.395774672103157e-06,
+      "loss": 0.9545,
+      "step": 6676
+    },
+    {
+      "epoch": 0.6987964416535846,
+      "grad_norm": 1.8309553111911094,
+      "learning_rate": 4.392967604664306e-06,
+      "loss": 0.8649,
+      "step": 6677
+    },
+    {
+      "epoch": 0.6989010989010989,
+      "grad_norm": 1.8989290566978339,
+      "learning_rate": 4.3901611814848396e-06,
+      "loss": 0.9028,
+      "step": 6678
+    },
+    {
+      "epoch": 0.6990057561486133,
+      "grad_norm": 2.326356063146579,
+      "learning_rate": 4.387355402887227e-06,
+      "loss": 0.9389,
+      "step": 6679
+    },
+    {
+      "epoch": 0.6991104133961277,
+      "grad_norm": 1.9814214365310432,
+      "learning_rate": 4.384550269193859e-06,
+      "loss": 0.947,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6992150706436421,
+      "grad_norm": 1.936796902489133,
+      "learning_rate": 4.381745780727049e-06,
+      "loss": 0.7648,
+      "step": 6681
+    },
+    {
+      "epoch": 0.6993197278911565,
+      "grad_norm": 2.3068454046130187,
+      "learning_rate": 4.378941937809039e-06,
+      "loss": 0.9382,
+      "step": 6682
+    },
+    {
+      "epoch": 0.6994243851386709,
+      "grad_norm": 2.0579861242161295,
+      "learning_rate": 4.3761387407619915e-06,
+      "loss": 0.8562,
+      "step": 6683
+    },
+    {
+      "epoch": 0.6995290423861853,
+      "grad_norm": 2.022770051321196,
+      "learning_rate": 4.373336189908007e-06,
+      "loss": 0.8727,
+      "step": 6684
+    },
+    {
+      "epoch": 0.6996336996336996,
+      "grad_norm": 1.8295973872025644,
+      "learning_rate": 4.3705342855691e-06,
+      "loss": 0.8223,
+      "step": 6685
+    },
+    {
+      "epoch": 0.699738356881214,
+      "grad_norm": 2.415516211385723,
+      "learning_rate": 4.367733028067217e-06,
+      "loss": 0.8492,
+      "step": 6686
+    },
+    {
+      "epoch": 0.6998430141287284,
+      "grad_norm": 2.5417469338554994,
+      "learning_rate": 4.364932417724222e-06,
+      "loss": 0.8159,
+      "step": 6687
+    },
+    {
+      "epoch": 0.6999476713762428,
+      "grad_norm": 1.7713596531603963,
+      "learning_rate": 4.362132454861916e-06,
+      "loss": 0.8118,
+      "step": 6688
+    },
+    {
+      "epoch": 0.7000523286237572,
+      "grad_norm": 1.9958944703716315,
+      "learning_rate": 4.359333139802024e-06,
+      "loss": 0.7404,
+      "step": 6689
+    },
+    {
+      "epoch": 0.7001569858712716,
+      "grad_norm": 2.0988662837685887,
+      "learning_rate": 4.356534472866189e-06,
+      "loss": 0.9781,
+      "step": 6690
+    },
+    {
+      "epoch": 0.700261643118786,
+      "grad_norm": 2.213783510735197,
+      "learning_rate": 4.353736454375983e-06,
+      "loss": 0.9575,
+      "step": 6691
+    },
+    {
+      "epoch": 0.7003663003663003,
+      "grad_norm": 2.17188348926095,
+      "learning_rate": 4.350939084652906e-06,
+      "loss": 0.9206,
+      "step": 6692
+    },
+    {
+      "epoch": 0.7004709576138147,
+      "grad_norm": 2.115726513977887,
+      "learning_rate": 4.348142364018375e-06,
+      "loss": 0.9514,
+      "step": 6693
+    },
+    {
+      "epoch": 0.7005756148613291,
+      "grad_norm": 1.9850295855218871,
+      "learning_rate": 4.345346292793748e-06,
+      "loss": 0.8236,
+      "step": 6694
+    },
+    {
+      "epoch": 0.7006802721088435,
+      "grad_norm": 1.94482693927181,
+      "learning_rate": 4.342550871300295e-06,
+      "loss": 0.8051,
+      "step": 6695
+    },
+    {
+      "epoch": 0.7007849293563579,
+      "grad_norm": 2.043191145044091,
+      "learning_rate": 4.339756099859213e-06,
+      "loss": 0.8906,
+      "step": 6696
+    },
+    {
+      "epoch": 0.7008895866038723,
+      "grad_norm": 2.128017700100571,
+      "learning_rate": 4.336961978791632e-06,
+      "loss": 0.9098,
+      "step": 6697
+    },
+    {
+      "epoch": 0.7009942438513868,
+      "grad_norm": 1.9203516360713857,
+      "learning_rate": 4.3341685084185965e-06,
+      "loss": 0.833,
+      "step": 6698
+    },
+    {
+      "epoch": 0.701098901098901,
+      "grad_norm": 1.9846571799966501,
+      "learning_rate": 4.331375689061089e-06,
+      "loss": 0.8093,
+      "step": 6699
+    },
+    {
+      "epoch": 0.7012035583464155,
+      "grad_norm": 2.6300245916251925,
+      "learning_rate": 4.328583521040006e-06,
+      "loss": 0.8283,
+      "step": 6700
+    },
+    {
+      "epoch": 0.7013082155939299,
+      "grad_norm": 2.227609828599236,
+      "learning_rate": 4.325792004676175e-06,
+      "loss": 0.9819,
+      "step": 6701
+    },
+    {
+      "epoch": 0.7014128728414443,
+      "grad_norm": 2.438764625145741,
+      "learning_rate": 4.32300114029034e-06,
+      "loss": 0.8809,
+      "step": 6702
+    },
+    {
+      "epoch": 0.7015175300889587,
+      "grad_norm": 2.2905189173309046,
+      "learning_rate": 4.320210928203187e-06,
+      "loss": 0.9054,
+      "step": 6703
+    },
+    {
+      "epoch": 0.7016221873364731,
+      "grad_norm": 2.089414522385608,
+      "learning_rate": 4.317421368735313e-06,
+      "loss": 0.9524,
+      "step": 6704
+    },
+    {
+      "epoch": 0.7017268445839875,
+      "grad_norm": 1.774274758188664,
+      "learning_rate": 4.314632462207239e-06,
+      "loss": 0.8178,
+      "step": 6705
+    },
+    {
+      "epoch": 0.7018315018315018,
+      "grad_norm": 2.147402104777586,
+      "learning_rate": 4.311844208939424e-06,
+      "loss": 0.7808,
+      "step": 6706
+    },
+    {
+      "epoch": 0.7019361590790162,
+      "grad_norm": 1.9194606735701536,
+      "learning_rate": 4.309056609252241e-06,
+      "loss": 0.9445,
+      "step": 6707
+    },
+    {
+      "epoch": 0.7020408163265306,
+      "grad_norm": 2.464691497275074,
+      "learning_rate": 4.306269663465986e-06,
+      "loss": 0.9314,
+      "step": 6708
+    },
+    {
+      "epoch": 0.702145473574045,
+      "grad_norm": 2.258686029857579,
+      "learning_rate": 4.3034833719008935e-06,
+      "loss": 0.9055,
+      "step": 6709
+    },
+    {
+      "epoch": 0.7022501308215594,
+      "grad_norm": 1.9254564663234373,
+      "learning_rate": 4.300697734877109e-06,
+      "loss": 0.9699,
+      "step": 6710
+    },
+    {
+      "epoch": 0.7023547880690738,
+      "grad_norm": 2.266755065720574,
+      "learning_rate": 4.297912752714709e-06,
+      "loss": 0.8583,
+      "step": 6711
+    },
+    {
+      "epoch": 0.7024594453165882,
+      "grad_norm": 1.9181752019643432,
+      "learning_rate": 4.29512842573369e-06,
+      "loss": 0.9748,
+      "step": 6712
+    },
+    {
+      "epoch": 0.7025641025641025,
+      "grad_norm": 1.8549903822927598,
+      "learning_rate": 4.2923447542539785e-06,
+      "loss": 0.7526,
+      "step": 6713
+    },
+    {
+      "epoch": 0.7026687598116169,
+      "grad_norm": 1.9363352003242609,
+      "learning_rate": 4.289561738595431e-06,
+      "loss": 0.9082,
+      "step": 6714
+    },
+    {
+      "epoch": 0.7027734170591313,
+      "grad_norm": 2.2343004147251095,
+      "learning_rate": 4.2867793790778136e-06,
+      "loss": 0.8977,
+      "step": 6715
+    },
+    {
+      "epoch": 0.7028780743066457,
+      "grad_norm": 2.0029331901560696,
+      "learning_rate": 4.283997676020829e-06,
+      "loss": 0.8673,
+      "step": 6716
+    },
+    {
+      "epoch": 0.7029827315541601,
+      "grad_norm": 1.93825276090313,
+      "learning_rate": 4.2812166297440946e-06,
+      "loss": 0.8511,
+      "step": 6717
+    },
+    {
+      "epoch": 0.7030873888016745,
+      "grad_norm": 2.1492481371612215,
+      "learning_rate": 4.278436240567166e-06,
+      "loss": 0.8906,
+      "step": 6718
+    },
+    {
+      "epoch": 0.703192046049189,
+      "grad_norm": 1.9356220967112918,
+      "learning_rate": 4.2756565088095125e-06,
+      "loss": 0.8833,
+      "step": 6719
+    },
+    {
+      "epoch": 0.7032967032967034,
+      "grad_norm": 1.9085420160045585,
+      "learning_rate": 4.272877434790531e-06,
+      "loss": 0.7844,
+      "step": 6720
+    },
+    {
+      "epoch": 0.7034013605442176,
+      "grad_norm": 2.1196564007597307,
+      "learning_rate": 4.270099018829539e-06,
+      "loss": 0.9646,
+      "step": 6721
+    },
+    {
+      "epoch": 0.703506017791732,
+      "grad_norm": 2.2766384603785346,
+      "learning_rate": 4.267321261245785e-06,
+      "loss": 0.8636,
+      "step": 6722
+    },
+    {
+      "epoch": 0.7036106750392465,
+      "grad_norm": 2.224471343348745,
+      "learning_rate": 4.264544162358443e-06,
+      "loss": 0.9361,
+      "step": 6723
+    },
+    {
+      "epoch": 0.7037153322867609,
+      "grad_norm": 1.8561557729274152,
+      "learning_rate": 4.2617677224866035e-06,
+      "loss": 0.7476,
+      "step": 6724
+    },
+    {
+      "epoch": 0.7038199895342753,
+      "grad_norm": 2.0486894102396254,
+      "learning_rate": 4.258991941949286e-06,
+      "loss": 0.7964,
+      "step": 6725
+    },
+    {
+      "epoch": 0.7039246467817897,
+      "grad_norm": 2.0441810946350936,
+      "learning_rate": 4.256216821065431e-06,
+      "loss": 0.9484,
+      "step": 6726
+    },
+    {
+      "epoch": 0.7040293040293041,
+      "grad_norm": 2.1128385722097507,
+      "learning_rate": 4.253442360153905e-06,
+      "loss": 0.9235,
+      "step": 6727
+    },
+    {
+      "epoch": 0.7041339612768184,
+      "grad_norm": 2.313931237729976,
+      "learning_rate": 4.250668559533504e-06,
+      "loss": 0.9055,
+      "step": 6728
+    },
+    {
+      "epoch": 0.7042386185243328,
+      "grad_norm": 1.8402369003536179,
+      "learning_rate": 4.247895419522941e-06,
+      "loss": 0.7616,
+      "step": 6729
+    },
+    {
+      "epoch": 0.7043432757718472,
+      "grad_norm": 1.8143536226822832,
+      "learning_rate": 4.245122940440855e-06,
+      "loss": 0.935,
+      "step": 6730
+    },
+    {
+      "epoch": 0.7044479330193616,
+      "grad_norm": 1.9312652171108742,
+      "learning_rate": 4.242351122605807e-06,
+      "loss": 0.8543,
+      "step": 6731
+    },
+    {
+      "epoch": 0.704552590266876,
+      "grad_norm": 1.9610927104433202,
+      "learning_rate": 4.239579966336286e-06,
+      "loss": 0.824,
+      "step": 6732
+    },
+    {
+      "epoch": 0.7046572475143904,
+      "grad_norm": 2.003342197333868,
+      "learning_rate": 4.236809471950708e-06,
+      "loss": 0.859,
+      "step": 6733
+    },
+    {
+      "epoch": 0.7047619047619048,
+      "grad_norm": 1.9433809349811346,
+      "learning_rate": 4.234039639767406e-06,
+      "loss": 0.902,
+      "step": 6734
+    },
+    {
+      "epoch": 0.7048665620094191,
+      "grad_norm": 2.178288945579384,
+      "learning_rate": 4.231270470104636e-06,
+      "loss": 0.8413,
+      "step": 6735
+    },
+    {
+      "epoch": 0.7049712192569335,
+      "grad_norm": 2.3256221301858866,
+      "learning_rate": 4.228501963280581e-06,
+      "loss": 0.9749,
+      "step": 6736
+    },
+    {
+      "epoch": 0.7050758765044479,
+      "grad_norm": 2.1566296507304763,
+      "learning_rate": 4.225734119613354e-06,
+      "loss": 0.9976,
+      "step": 6737
+    },
+    {
+      "epoch": 0.7051805337519623,
+      "grad_norm": 2.2955385696891453,
+      "learning_rate": 4.222966939420983e-06,
+      "loss": 0.9847,
+      "step": 6738
+    },
+    {
+      "epoch": 0.7052851909994767,
+      "grad_norm": 2.219932302560416,
+      "learning_rate": 4.220200423021421e-06,
+      "loss": 0.9591,
+      "step": 6739
+    },
+    {
+      "epoch": 0.7053898482469911,
+      "grad_norm": 2.171556004505835,
+      "learning_rate": 4.217434570732544e-06,
+      "loss": 0.9488,
+      "step": 6740
+    },
+    {
+      "epoch": 0.7054945054945055,
+      "grad_norm": 2.003742070991652,
+      "learning_rate": 4.214669382872157e-06,
+      "loss": 1.0053,
+      "step": 6741
+    },
+    {
+      "epoch": 0.7055991627420198,
+      "grad_norm": 2.2735271709277938,
+      "learning_rate": 4.2119048597579905e-06,
+      "loss": 0.7012,
+      "step": 6742
+    },
+    {
+      "epoch": 0.7057038199895342,
+      "grad_norm": 1.9509244256018325,
+      "learning_rate": 4.2091410017076884e-06,
+      "loss": 0.8871,
+      "step": 6743
+    },
+    {
+      "epoch": 0.7058084772370486,
+      "grad_norm": 1.803214751829208,
+      "learning_rate": 4.206377809038825e-06,
+      "loss": 0.8344,
+      "step": 6744
+    },
+    {
+      "epoch": 0.705913134484563,
+      "grad_norm": 2.1681624054889075,
+      "learning_rate": 4.203615282068896e-06,
+      "loss": 0.9701,
+      "step": 6745
+    },
+    {
+      "epoch": 0.7060177917320775,
+      "grad_norm": 2.003295420905956,
+      "learning_rate": 4.200853421115317e-06,
+      "loss": 0.9148,
+      "step": 6746
+    },
+    {
+      "epoch": 0.7061224489795919,
+      "grad_norm": 1.8542405762144707,
+      "learning_rate": 4.1980922264954415e-06,
+      "loss": 0.7981,
+      "step": 6747
+    },
+    {
+      "epoch": 0.7062271062271063,
+      "grad_norm": 2.2185412641129605,
+      "learning_rate": 4.195331698526526e-06,
+      "loss": 0.9382,
+      "step": 6748
+    },
+    {
+      "epoch": 0.7063317634746206,
+      "grad_norm": 1.9421670362200358,
+      "learning_rate": 4.19257183752577e-06,
+      "loss": 0.9211,
+      "step": 6749
+    },
+    {
+      "epoch": 0.706436420722135,
+      "grad_norm": 2.350163108156549,
+      "learning_rate": 4.189812643810282e-06,
+      "loss": 0.9539,
+      "step": 6750
+    },
+    {
+      "epoch": 0.7065410779696494,
+      "grad_norm": 2.1933671368986256,
+      "learning_rate": 4.187054117697097e-06,
+      "loss": 0.9205,
+      "step": 6751
+    },
+    {
+      "epoch": 0.7066457352171638,
+      "grad_norm": 1.8016529905030467,
+      "learning_rate": 4.184296259503181e-06,
+      "loss": 0.863,
+      "step": 6752
+    },
+    {
+      "epoch": 0.7067503924646782,
+      "grad_norm": 1.638442385400924,
+      "learning_rate": 4.181539069545414e-06,
+      "loss": 0.8414,
+      "step": 6753
+    },
+    {
+      "epoch": 0.7068550497121926,
+      "grad_norm": 2.2815059203943218,
+      "learning_rate": 4.1787825481406044e-06,
+      "loss": 0.9858,
+      "step": 6754
+    },
+    {
+      "epoch": 0.706959706959707,
+      "grad_norm": 2.0723833971887684,
+      "learning_rate": 4.176026695605476e-06,
+      "loss": 0.8515,
+      "step": 6755
+    },
+    {
+      "epoch": 0.7070643642072213,
+      "grad_norm": 2.0981663306838585,
+      "learning_rate": 4.17327151225669e-06,
+      "loss": 0.9072,
+      "step": 6756
+    },
+    {
+      "epoch": 0.7071690214547357,
+      "grad_norm": 1.885601962808024,
+      "learning_rate": 4.170516998410817e-06,
+      "loss": 0.9031,
+      "step": 6757
+    },
+    {
+      "epoch": 0.7072736787022501,
+      "grad_norm": 2.253291731810213,
+      "learning_rate": 4.1677631543843625e-06,
+      "loss": 0.9391,
+      "step": 6758
+    },
+    {
+      "epoch": 0.7073783359497645,
+      "grad_norm": 1.956313519509557,
+      "learning_rate": 4.165009980493742e-06,
+      "loss": 0.8985,
+      "step": 6759
+    },
+    {
+      "epoch": 0.7074829931972789,
+      "grad_norm": 2.2664708007932743,
+      "learning_rate": 4.1622574770553025e-06,
+      "loss": 0.8731,
+      "step": 6760
+    },
+    {
+      "epoch": 0.7075876504447933,
+      "grad_norm": 2.195150032088823,
+      "learning_rate": 4.159505644385316e-06,
+      "loss": 0.8524,
+      "step": 6761
+    },
+    {
+      "epoch": 0.7076923076923077,
+      "grad_norm": 1.9346641546170762,
+      "learning_rate": 4.1567544827999705e-06,
+      "loss": 0.8993,
+      "step": 6762
+    },
+    {
+      "epoch": 0.7077969649398221,
+      "grad_norm": 2.2504254337448453,
+      "learning_rate": 4.154003992615381e-06,
+      "loss": 0.9644,
+      "step": 6763
+    },
+    {
+      "epoch": 0.7079016221873364,
+      "grad_norm": 2.188401815842773,
+      "learning_rate": 4.151254174147584e-06,
+      "loss": 0.8787,
+      "step": 6764
+    },
+    {
+      "epoch": 0.7080062794348508,
+      "grad_norm": 2.037348430435397,
+      "learning_rate": 4.148505027712535e-06,
+      "loss": 0.8898,
+      "step": 6765
+    },
+    {
+      "epoch": 0.7081109366823652,
+      "grad_norm": 2.121364396425157,
+      "learning_rate": 4.1457565536261204e-06,
+      "loss": 0.9164,
+      "step": 6766
+    },
+    {
+      "epoch": 0.7082155939298796,
+      "grad_norm": 2.1227061314556104,
+      "learning_rate": 4.14300875220415e-06,
+      "loss": 0.8375,
+      "step": 6767
+    },
+    {
+      "epoch": 0.7083202511773941,
+      "grad_norm": 2.2345470752377112,
+      "learning_rate": 4.140261623762346e-06,
+      "loss": 1.0657,
+      "step": 6768
+    },
+    {
+      "epoch": 0.7084249084249085,
+      "grad_norm": 2.115378021426516,
+      "learning_rate": 4.13751516861636e-06,
+      "loss": 0.8428,
+      "step": 6769
+    },
+    {
+      "epoch": 0.7085295656724229,
+      "grad_norm": 1.8781409525877706,
+      "learning_rate": 4.134769387081764e-06,
+      "loss": 0.8158,
+      "step": 6770
+    },
+    {
+      "epoch": 0.7086342229199372,
+      "grad_norm": 2.870662361008902,
+      "learning_rate": 4.132024279474058e-06,
+      "loss": 0.8034,
+      "step": 6771
+    },
+    {
+      "epoch": 0.7087388801674516,
+      "grad_norm": 1.9910392284133023,
+      "learning_rate": 4.129279846108658e-06,
+      "loss": 0.8839,
+      "step": 6772
+    },
+    {
+      "epoch": 0.708843537414966,
+      "grad_norm": 2.7345053141599274,
+      "learning_rate": 4.1265360873009054e-06,
+      "loss": 0.8741,
+      "step": 6773
+    },
+    {
+      "epoch": 0.7089481946624804,
+      "grad_norm": 1.817693344021189,
+      "learning_rate": 4.123793003366059e-06,
+      "loss": 0.8432,
+      "step": 6774
+    },
+    {
+      "epoch": 0.7090528519099948,
+      "grad_norm": 1.9857696095073964,
+      "learning_rate": 4.121050594619308e-06,
+      "loss": 0.9486,
+      "step": 6775
+    },
+    {
+      "epoch": 0.7091575091575092,
+      "grad_norm": 2.138607157252604,
+      "learning_rate": 4.118308861375766e-06,
+      "loss": 0.9636,
+      "step": 6776
+    },
+    {
+      "epoch": 0.7092621664050236,
+      "grad_norm": 1.801323948809371,
+      "learning_rate": 4.11556780395046e-06,
+      "loss": 0.8476,
+      "step": 6777
+    },
+    {
+      "epoch": 0.7093668236525379,
+      "grad_norm": 2.120684116550481,
+      "learning_rate": 4.112827422658341e-06,
+      "loss": 0.9265,
+      "step": 6778
+    },
+    {
+      "epoch": 0.7094714809000523,
+      "grad_norm": 1.9405112301994334,
+      "learning_rate": 4.1100877178142825e-06,
+      "loss": 0.8766,
+      "step": 6779
+    },
+    {
+      "epoch": 0.7095761381475667,
+      "grad_norm": 2.33308294996142,
+      "learning_rate": 4.10734868973309e-06,
+      "loss": 0.8442,
+      "step": 6780
+    },
+    {
+      "epoch": 0.7096807953950811,
+      "grad_norm": 2.0092401960246016,
+      "learning_rate": 4.104610338729478e-06,
+      "loss": 0.9067,
+      "step": 6781
+    },
+    {
+      "epoch": 0.7097854526425955,
+      "grad_norm": 2.4295543764879763,
+      "learning_rate": 4.101872665118088e-06,
+      "loss": 0.9433,
+      "step": 6782
+    },
+    {
+      "epoch": 0.7098901098901099,
+      "grad_norm": 2.1008017438183924,
+      "learning_rate": 4.099135669213483e-06,
+      "loss": 0.8513,
+      "step": 6783
+    },
+    {
+      "epoch": 0.7099947671376243,
+      "grad_norm": 2.445983785008863,
+      "learning_rate": 4.096399351330156e-06,
+      "loss": 0.9137,
+      "step": 6784
+    },
+    {
+      "epoch": 0.7100994243851386,
+      "grad_norm": 2.1839137429933606,
+      "learning_rate": 4.093663711782507e-06,
+      "loss": 0.8465,
+      "step": 6785
+    },
+    {
+      "epoch": 0.710204081632653,
+      "grad_norm": 1.9874222387615066,
+      "learning_rate": 4.0909287508848745e-06,
+      "loss": 0.8232,
+      "step": 6786
+    },
+    {
+      "epoch": 0.7103087388801674,
+      "grad_norm": 1.8844372558093263,
+      "learning_rate": 4.0881944689515085e-06,
+      "loss": 0.8779,
+      "step": 6787
+    },
+    {
+      "epoch": 0.7104133961276818,
+      "grad_norm": 1.9799678528512243,
+      "learning_rate": 4.085460866296581e-06,
+      "loss": 0.9447,
+      "step": 6788
+    },
+    {
+      "epoch": 0.7105180533751962,
+      "grad_norm": 2.124231521122012,
+      "learning_rate": 4.082727943234187e-06,
+      "loss": 0.8971,
+      "step": 6789
+    },
+    {
+      "epoch": 0.7106227106227107,
+      "grad_norm": 1.8530431577227409,
+      "learning_rate": 4.079995700078352e-06,
+      "loss": 0.8753,
+      "step": 6790
+    },
+    {
+      "epoch": 0.7107273678702251,
+      "grad_norm": 2.0466031803847162,
+      "learning_rate": 4.077264137143012e-06,
+      "loss": 0.8169,
+      "step": 6791
+    },
+    {
+      "epoch": 0.7108320251177394,
+      "grad_norm": 2.183687382955889,
+      "learning_rate": 4.074533254742026e-06,
+      "loss": 0.937,
+      "step": 6792
+    },
+    {
+      "epoch": 0.7109366823652538,
+      "grad_norm": 1.8299109318292106,
+      "learning_rate": 4.071803053189184e-06,
+      "loss": 0.8489,
+      "step": 6793
+    },
+    {
+      "epoch": 0.7110413396127682,
+      "grad_norm": 2.48666879446886,
+      "learning_rate": 4.069073532798185e-06,
+      "loss": 0.9826,
+      "step": 6794
+    },
+    {
+      "epoch": 0.7111459968602826,
+      "grad_norm": 1.9966103200441245,
+      "learning_rate": 4.066344693882665e-06,
+      "loss": 0.9272,
+      "step": 6795
+    },
+    {
+      "epoch": 0.711250654107797,
+      "grad_norm": 2.0372668354924715,
+      "learning_rate": 4.063616536756168e-06,
+      "loss": 0.8891,
+      "step": 6796
+    },
+    {
+      "epoch": 0.7113553113553114,
+      "grad_norm": 2.0421407323614487,
+      "learning_rate": 4.060889061732165e-06,
+      "loss": 0.8493,
+      "step": 6797
+    },
+    {
+      "epoch": 0.7114599686028258,
+      "grad_norm": 2.051281230221165,
+      "learning_rate": 4.058162269124045e-06,
+      "loss": 0.8805,
+      "step": 6798
+    },
+    {
+      "epoch": 0.7115646258503401,
+      "grad_norm": 2.0417527238947617,
+      "learning_rate": 4.055436159245129e-06,
+      "loss": 0.8527,
+      "step": 6799
+    },
+    {
+      "epoch": 0.7116692830978545,
+      "grad_norm": 2.01692569770081,
+      "learning_rate": 4.0527107324086465e-06,
+      "loss": 0.8964,
+      "step": 6800
+    },
+    {
+      "epoch": 0.7117739403453689,
+      "grad_norm": 2.498291537658827,
+      "learning_rate": 4.04998598892776e-06,
+      "loss": 0.8871,
+      "step": 6801
+    },
+    {
+      "epoch": 0.7118785975928833,
+      "grad_norm": 1.7272578518417077,
+      "learning_rate": 4.047261929115546e-06,
+      "loss": 0.7592,
+      "step": 6802
+    },
+    {
+      "epoch": 0.7119832548403977,
+      "grad_norm": 2.067238416917133,
+      "learning_rate": 4.044538553285003e-06,
+      "loss": 0.8075,
+      "step": 6803
+    },
+    {
+      "epoch": 0.7120879120879121,
+      "grad_norm": 2.0306398676192043,
+      "learning_rate": 4.04181586174905e-06,
+      "loss": 0.9844,
+      "step": 6804
+    },
+    {
+      "epoch": 0.7121925693354265,
+      "grad_norm": 1.7327165325882197,
+      "learning_rate": 4.039093854820537e-06,
+      "loss": 0.8486,
+      "step": 6805
+    },
+    {
+      "epoch": 0.7122972265829409,
+      "grad_norm": 1.954070649531059,
+      "learning_rate": 4.036372532812224e-06,
+      "loss": 0.9154,
+      "step": 6806
+    },
+    {
+      "epoch": 0.7124018838304552,
+      "grad_norm": 2.113577347797847,
+      "learning_rate": 4.033651896036797e-06,
+      "loss": 0.8904,
+      "step": 6807
+    },
+    {
+      "epoch": 0.7125065410779696,
+      "grad_norm": 2.103592454954578,
+      "learning_rate": 4.030931944806857e-06,
+      "loss": 0.8943,
+      "step": 6808
+    },
+    {
+      "epoch": 0.712611198325484,
+      "grad_norm": 2.1336538127277405,
+      "learning_rate": 4.0282126794349385e-06,
+      "loss": 0.8553,
+      "step": 6809
+    },
+    {
+      "epoch": 0.7127158555729984,
+      "grad_norm": 1.919816279896638,
+      "learning_rate": 4.0254941002334914e-06,
+      "loss": 0.8675,
+      "step": 6810
+    },
+    {
+      "epoch": 0.7128205128205128,
+      "grad_norm": 1.8626947261175397,
+      "learning_rate": 4.022776207514885e-06,
+      "loss": 0.8163,
+      "step": 6811
+    },
+    {
+      "epoch": 0.7129251700680272,
+      "grad_norm": 2.2003625119304875,
+      "learning_rate": 4.020059001591409e-06,
+      "loss": 0.9314,
+      "step": 6812
+    },
+    {
+      "epoch": 0.7130298273155417,
+      "grad_norm": 2.1580488886290325,
+      "learning_rate": 4.0173424827752725e-06,
+      "loss": 0.9063,
+      "step": 6813
+    },
+    {
+      "epoch": 0.713134484563056,
+      "grad_norm": 1.813499024046145,
+      "learning_rate": 4.014626651378617e-06,
+      "loss": 0.8229,
+      "step": 6814
+    },
+    {
+      "epoch": 0.7132391418105704,
+      "grad_norm": 2.13054755191047,
+      "learning_rate": 4.011911507713493e-06,
+      "loss": 0.9653,
+      "step": 6815
+    },
+    {
+      "epoch": 0.7133437990580848,
+      "grad_norm": 2.3481830820459466,
+      "learning_rate": 4.0091970520918755e-06,
+      "loss": 0.9658,
+      "step": 6816
+    },
+    {
+      "epoch": 0.7134484563055992,
+      "grad_norm": 3.110141424492723,
+      "learning_rate": 4.006483284825658e-06,
+      "loss": 0.9323,
+      "step": 6817
+    },
+    {
+      "epoch": 0.7135531135531136,
+      "grad_norm": 1.9336469464972577,
+      "learning_rate": 4.00377020622666e-06,
+      "loss": 0.9056,
+      "step": 6818
+    },
+    {
+      "epoch": 0.713657770800628,
+      "grad_norm": 1.7890408343625537,
+      "learning_rate": 4.001057816606626e-06,
+      "loss": 0.8723,
+      "step": 6819
+    },
+    {
+      "epoch": 0.7137624280481424,
+      "grad_norm": 2.0458672431955427,
+      "learning_rate": 3.99834611627721e-06,
+      "loss": 0.9242,
+      "step": 6820
+    },
+    {
+      "epoch": 0.7138670852956567,
+      "grad_norm": 2.169252164372577,
+      "learning_rate": 3.995635105549991e-06,
+      "loss": 0.8732,
+      "step": 6821
+    },
+    {
+      "epoch": 0.7139717425431711,
+      "grad_norm": 2.1207025732586677,
+      "learning_rate": 3.99292478473647e-06,
+      "loss": 0.9312,
+      "step": 6822
+    },
+    {
+      "epoch": 0.7140763997906855,
+      "grad_norm": 1.8860336974481258,
+      "learning_rate": 3.990215154148067e-06,
+      "loss": 0.8711,
+      "step": 6823
+    },
+    {
+      "epoch": 0.7141810570381999,
+      "grad_norm": 1.7535400399995449,
+      "learning_rate": 3.987506214096128e-06,
+      "loss": 0.8432,
+      "step": 6824
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 1.9294483445338717,
+      "learning_rate": 3.984797964891914e-06,
+      "loss": 0.8547,
+      "step": 6825
+    },
+    {
+      "epoch": 0.7143903715332287,
+      "grad_norm": 2.1241763749956415,
+      "learning_rate": 3.982090406846605e-06,
+      "loss": 0.9769,
+      "step": 6826
+    },
+    {
+      "epoch": 0.7144950287807431,
+      "grad_norm": 1.617410025245596,
+      "learning_rate": 3.979383540271311e-06,
+      "loss": 0.8158,
+      "step": 6827
+    },
+    {
+      "epoch": 0.7145996860282574,
+      "grad_norm": 1.9851405262752755,
+      "learning_rate": 3.976677365477049e-06,
+      "loss": 0.8617,
+      "step": 6828
+    },
+    {
+      "epoch": 0.7147043432757718,
+      "grad_norm": 2.3520800347490125,
+      "learning_rate": 3.973971882774773e-06,
+      "loss": 0.8526,
+      "step": 6829
+    },
+    {
+      "epoch": 0.7148090005232862,
+      "grad_norm": 1.9514505903548016,
+      "learning_rate": 3.971267092475343e-06,
+      "loss": 0.9752,
+      "step": 6830
+    },
+    {
+      "epoch": 0.7149136577708006,
+      "grad_norm": 2.5826157191432997,
+      "learning_rate": 3.968562994889545e-06,
+      "loss": 0.9407,
+      "step": 6831
+    },
+    {
+      "epoch": 0.715018315018315,
+      "grad_norm": 2.300126908869992,
+      "learning_rate": 3.9658595903280836e-06,
+      "loss": 0.8193,
+      "step": 6832
+    },
+    {
+      "epoch": 0.7151229722658294,
+      "grad_norm": 1.8165911897167484,
+      "learning_rate": 3.963156879101591e-06,
+      "loss": 0.8707,
+      "step": 6833
+    },
+    {
+      "epoch": 0.7152276295133438,
+      "grad_norm": 2.059647993743263,
+      "learning_rate": 3.96045486152061e-06,
+      "loss": 0.9069,
+      "step": 6834
+    },
+    {
+      "epoch": 0.7153322867608581,
+      "grad_norm": 3.670243670467124,
+      "learning_rate": 3.957753537895607e-06,
+      "loss": 0.9077,
+      "step": 6835
+    },
+    {
+      "epoch": 0.7154369440083725,
+      "grad_norm": 2.4214679282114204,
+      "learning_rate": 3.9550529085369735e-06,
+      "loss": 0.9734,
+      "step": 6836
+    },
+    {
+      "epoch": 0.715541601255887,
+      "grad_norm": 1.908703225014119,
+      "learning_rate": 3.952352973755012e-06,
+      "loss": 0.9761,
+      "step": 6837
+    },
+    {
+      "epoch": 0.7156462585034014,
+      "grad_norm": 1.9494300614897873,
+      "learning_rate": 3.949653733859958e-06,
+      "loss": 0.9582,
+      "step": 6838
+    },
+    {
+      "epoch": 0.7157509157509158,
+      "grad_norm": 2.3349105601456652,
+      "learning_rate": 3.946955189161954e-06,
+      "loss": 0.9848,
+      "step": 6839
+    },
+    {
+      "epoch": 0.7158555729984302,
+      "grad_norm": 1.98806617370881,
+      "learning_rate": 3.944257339971071e-06,
+      "loss": 1.0294,
+      "step": 6840
+    },
+    {
+      "epoch": 0.7159602302459446,
+      "grad_norm": 2.032978994521527,
+      "learning_rate": 3.941560186597295e-06,
+      "loss": 0.7697,
+      "step": 6841
+    },
+    {
+      "epoch": 0.7160648874934589,
+      "grad_norm": 2.2859142638467196,
+      "learning_rate": 3.938863729350532e-06,
+      "loss": 0.9599,
+      "step": 6842
+    },
+    {
+      "epoch": 0.7161695447409733,
+      "grad_norm": 1.95473032037277,
+      "learning_rate": 3.936167968540617e-06,
+      "loss": 0.8378,
+      "step": 6843
+    },
+    {
+      "epoch": 0.7162742019884877,
+      "grad_norm": 2.2383483044471206,
+      "learning_rate": 3.933472904477291e-06,
+      "loss": 0.8798,
+      "step": 6844
+    },
+    {
+      "epoch": 0.7163788592360021,
+      "grad_norm": 1.6286590487546038,
+      "learning_rate": 3.930778537470231e-06,
+      "loss": 0.9243,
+      "step": 6845
+    },
+    {
+      "epoch": 0.7164835164835165,
+      "grad_norm": 1.8458670553744083,
+      "learning_rate": 3.928084867829021e-06,
+      "loss": 0.7944,
+      "step": 6846
+    },
+    {
+      "epoch": 0.7165881737310309,
+      "grad_norm": 2.270745347465646,
+      "learning_rate": 3.9253918958631635e-06,
+      "loss": 1.0004,
+      "step": 6847
+    },
+    {
+      "epoch": 0.7166928309785453,
+      "grad_norm": 1.8785468739801112,
+      "learning_rate": 3.922699621882097e-06,
+      "loss": 0.8825,
+      "step": 6848
+    },
+    {
+      "epoch": 0.7167974882260597,
+      "grad_norm": 2.05640939822826,
+      "learning_rate": 3.920008046195163e-06,
+      "loss": 0.8316,
+      "step": 6849
+    },
+    {
+      "epoch": 0.716902145473574,
+      "grad_norm": 2.2722674004261476,
+      "learning_rate": 3.917317169111629e-06,
+      "loss": 1.0579,
+      "step": 6850
+    },
+    {
+      "epoch": 0.7170068027210884,
+      "grad_norm": 2.4202451915007512,
+      "learning_rate": 3.9146269909406805e-06,
+      "loss": 0.9921,
+      "step": 6851
+    },
+    {
+      "epoch": 0.7171114599686028,
+      "grad_norm": 2.2412583484266295,
+      "learning_rate": 3.911937511991428e-06,
+      "loss": 0.8362,
+      "step": 6852
+    },
+    {
+      "epoch": 0.7172161172161172,
+      "grad_norm": 2.0935113321976733,
+      "learning_rate": 3.9092487325729e-06,
+      "loss": 0.7682,
+      "step": 6853
+    },
+    {
+      "epoch": 0.7173207744636316,
+      "grad_norm": 2.0222550806048267,
+      "learning_rate": 3.906560652994039e-06,
+      "loss": 0.8426,
+      "step": 6854
+    },
+    {
+      "epoch": 0.717425431711146,
+      "grad_norm": 1.9696243728220855,
+      "learning_rate": 3.903873273563713e-06,
+      "loss": 0.8273,
+      "step": 6855
+    },
+    {
+      "epoch": 0.7175300889586604,
+      "grad_norm": 1.8907957710602337,
+      "learning_rate": 3.901186594590702e-06,
+      "loss": 0.8415,
+      "step": 6856
+    },
+    {
+      "epoch": 0.7176347462061747,
+      "grad_norm": 1.8888147086385425,
+      "learning_rate": 3.898500616383718e-06,
+      "loss": 0.9082,
+      "step": 6857
+    },
+    {
+      "epoch": 0.7177394034536891,
+      "grad_norm": 1.9341244750710724,
+      "learning_rate": 3.895815339251382e-06,
+      "loss": 0.8821,
+      "step": 6858
+    },
+    {
+      "epoch": 0.7178440607012035,
+      "grad_norm": 2.2280568062526522,
+      "learning_rate": 3.893130763502239e-06,
+      "loss": 0.8867,
+      "step": 6859
+    },
+    {
+      "epoch": 0.717948717948718,
+      "grad_norm": 1.9985998533707623,
+      "learning_rate": 3.890446889444751e-06,
+      "loss": 0.8671,
+      "step": 6860
+    },
+    {
+      "epoch": 0.7180533751962324,
+      "grad_norm": 1.9789150948560013,
+      "learning_rate": 3.887763717387296e-06,
+      "loss": 0.9274,
+      "step": 6861
+    },
+    {
+      "epoch": 0.7181580324437468,
+      "grad_norm": 2.008681323462506,
+      "learning_rate": 3.8850812476381814e-06,
+      "loss": 0.918,
+      "step": 6862
+    },
+    {
+      "epoch": 0.7182626896912612,
+      "grad_norm": 1.6970462178438779,
+      "learning_rate": 3.882399480505632e-06,
+      "loss": 0.7673,
+      "step": 6863
+    },
+    {
+      "epoch": 0.7183673469387755,
+      "grad_norm": 1.9737100576957622,
+      "learning_rate": 3.879718416297783e-06,
+      "loss": 0.8705,
+      "step": 6864
+    },
+    {
+      "epoch": 0.7184720041862899,
+      "grad_norm": 2.116672494256888,
+      "learning_rate": 3.877038055322696e-06,
+      "loss": 0.9756,
+      "step": 6865
+    },
+    {
+      "epoch": 0.7185766614338043,
+      "grad_norm": 1.6446313933488226,
+      "learning_rate": 3.874358397888345e-06,
+      "loss": 0.7854,
+      "step": 6866
+    },
+    {
+      "epoch": 0.7186813186813187,
+      "grad_norm": 2.2698290566150323,
+      "learning_rate": 3.871679444302635e-06,
+      "loss": 0.8557,
+      "step": 6867
+    },
+    {
+      "epoch": 0.7187859759288331,
+      "grad_norm": 2.1256730677441578,
+      "learning_rate": 3.86900119487338e-06,
+      "loss": 0.8339,
+      "step": 6868
+    },
+    {
+      "epoch": 0.7188906331763475,
+      "grad_norm": 1.9881658784209284,
+      "learning_rate": 3.866323649908318e-06,
+      "loss": 0.9258,
+      "step": 6869
+    },
+    {
+      "epoch": 0.7189952904238619,
+      "grad_norm": 2.124198993385882,
+      "learning_rate": 3.8636468097150995e-06,
+      "loss": 0.883,
+      "step": 6870
+    },
+    {
+      "epoch": 0.7190999476713762,
+      "grad_norm": 1.9123797963557438,
+      "learning_rate": 3.860970674601301e-06,
+      "loss": 0.8941,
+      "step": 6871
+    },
+    {
+      "epoch": 0.7192046049188906,
+      "grad_norm": 2.0104065734716636,
+      "learning_rate": 3.85829524487442e-06,
+      "loss": 0.8806,
+      "step": 6872
+    },
+    {
+      "epoch": 0.719309262166405,
+      "grad_norm": 2.1166454011026707,
+      "learning_rate": 3.855620520841868e-06,
+      "loss": 0.923,
+      "step": 6873
+    },
+    {
+      "epoch": 0.7194139194139194,
+      "grad_norm": 1.9204632832965207,
+      "learning_rate": 3.8529465028109725e-06,
+      "loss": 0.8695,
+      "step": 6874
+    },
+    {
+      "epoch": 0.7195185766614338,
+      "grad_norm": 1.937070827448725,
+      "learning_rate": 3.8502731910889826e-06,
+      "loss": 0.835,
+      "step": 6875
+    },
+    {
+      "epoch": 0.7196232339089482,
+      "grad_norm": 2.423772623864867,
+      "learning_rate": 3.8476005859830725e-06,
+      "loss": 1.0035,
+      "step": 6876
+    },
+    {
+      "epoch": 0.7197278911564626,
+      "grad_norm": 2.2319802937639213,
+      "learning_rate": 3.844928687800328e-06,
+      "loss": 0.9172,
+      "step": 6877
+    },
+    {
+      "epoch": 0.7198325484039769,
+      "grad_norm": 1.8816969569554647,
+      "learning_rate": 3.842257496847751e-06,
+      "loss": 0.8183,
+      "step": 6878
+    },
+    {
+      "epoch": 0.7199372056514913,
+      "grad_norm": 2.328593060017702,
+      "learning_rate": 3.839587013432273e-06,
+      "loss": 0.9581,
+      "step": 6879
+    },
+    {
+      "epoch": 0.7200418628990057,
+      "grad_norm": 2.357599652370192,
+      "learning_rate": 3.836917237860738e-06,
+      "loss": 0.8882,
+      "step": 6880
+    },
+    {
+      "epoch": 0.7201465201465201,
+      "grad_norm": 2.112443593657366,
+      "learning_rate": 3.834248170439901e-06,
+      "loss": 0.8824,
+      "step": 6881
+    },
+    {
+      "epoch": 0.7202511773940345,
+      "grad_norm": 2.5342213753725265,
+      "learning_rate": 3.831579811476452e-06,
+      "loss": 0.8558,
+      "step": 6882
+    },
+    {
+      "epoch": 0.720355834641549,
+      "grad_norm": 2.187453453613242,
+      "learning_rate": 3.8289121612769885e-06,
+      "loss": 1.0146,
+      "step": 6883
+    },
+    {
+      "epoch": 0.7204604918890634,
+      "grad_norm": 2.0491956184185764,
+      "learning_rate": 3.8262452201480275e-06,
+      "loss": 0.9876,
+      "step": 6884
+    },
+    {
+      "epoch": 0.7205651491365777,
+      "grad_norm": 1.9619918810973644,
+      "learning_rate": 3.823578988396003e-06,
+      "loss": 0.9359,
+      "step": 6885
+    },
+    {
+      "epoch": 0.7206698063840921,
+      "grad_norm": 2.435486034108263,
+      "learning_rate": 3.820913466327276e-06,
+      "loss": 0.9142,
+      "step": 6886
+    },
+    {
+      "epoch": 0.7207744636316065,
+      "grad_norm": 2.128985121841604,
+      "learning_rate": 3.818248654248115e-06,
+      "loss": 0.8178,
+      "step": 6887
+    },
+    {
+      "epoch": 0.7208791208791209,
+      "grad_norm": 2.1846007264436063,
+      "learning_rate": 3.81558455246472e-06,
+      "loss": 0.8921,
+      "step": 6888
+    },
+    {
+      "epoch": 0.7209837781266353,
+      "grad_norm": 2.134017674581687,
+      "learning_rate": 3.8129211612831953e-06,
+      "loss": 0.9682,
+      "step": 6889
+    },
+    {
+      "epoch": 0.7210884353741497,
+      "grad_norm": 2.4046779312825386,
+      "learning_rate": 3.8102584810095687e-06,
+      "loss": 0.801,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7211930926216641,
+      "grad_norm": 2.2865554223892066,
+      "learning_rate": 3.8075965119497936e-06,
+      "loss": 0.9072,
+      "step": 6891
+    },
+    {
+      "epoch": 0.7212977498691785,
+      "grad_norm": 1.8509159996167859,
+      "learning_rate": 3.804935254409734e-06,
+      "loss": 0.88,
+      "step": 6892
+    },
+    {
+      "epoch": 0.7214024071166928,
+      "grad_norm": 2.2394314733138647,
+      "learning_rate": 3.8022747086951715e-06,
+      "loss": 0.904,
+      "step": 6893
+    },
+    {
+      "epoch": 0.7215070643642072,
+      "grad_norm": 1.9861651335567048,
+      "learning_rate": 3.7996148751118057e-06,
+      "loss": 0.8216,
+      "step": 6894
+    },
+    {
+      "epoch": 0.7216117216117216,
+      "grad_norm": 1.9218843095758162,
+      "learning_rate": 3.7969557539652636e-06,
+      "loss": 0.918,
+      "step": 6895
+    },
+    {
+      "epoch": 0.721716378859236,
+      "grad_norm": 1.9157472248957887,
+      "learning_rate": 3.7942973455610766e-06,
+      "loss": 0.8062,
+      "step": 6896
+    },
+    {
+      "epoch": 0.7218210361067504,
+      "grad_norm": 1.9189826257514782,
+      "learning_rate": 3.791639650204709e-06,
+      "loss": 0.8006,
+      "step": 6897
+    },
+    {
+      "epoch": 0.7219256933542648,
+      "grad_norm": 1.8660285862083692,
+      "learning_rate": 3.7889826682015306e-06,
+      "loss": 0.7262,
+      "step": 6898
+    },
+    {
+      "epoch": 0.7220303506017792,
+      "grad_norm": 2.0909598391703046,
+      "learning_rate": 3.7863263998568346e-06,
+      "loss": 0.9164,
+      "step": 6899
+    },
+    {
+      "epoch": 0.7221350078492935,
+      "grad_norm": 2.5637117016782507,
+      "learning_rate": 3.7836708454758287e-06,
+      "loss": 0.8562,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7222396650968079,
+      "grad_norm": 2.09685317781106,
+      "learning_rate": 3.781016005363648e-06,
+      "loss": 0.9648,
+      "step": 6901
+    },
+    {
+      "epoch": 0.7223443223443223,
+      "grad_norm": 1.832463976356295,
+      "learning_rate": 3.7783618798253354e-06,
+      "loss": 0.8759,
+      "step": 6902
+    },
+    {
+      "epoch": 0.7224489795918367,
+      "grad_norm": 1.9229291271734896,
+      "learning_rate": 3.7757084691658552e-06,
+      "loss": 0.8913,
+      "step": 6903
+    },
+    {
+      "epoch": 0.7225536368393511,
+      "grad_norm": 2.519016468836667,
+      "learning_rate": 3.7730557736900865e-06,
+      "loss": 0.9351,
+      "step": 6904
+    },
+    {
+      "epoch": 0.7226582940868655,
+      "grad_norm": 1.9402129919604572,
+      "learning_rate": 3.7704037937028324e-06,
+      "loss": 0.9447,
+      "step": 6905
+    },
+    {
+      "epoch": 0.72276295133438,
+      "grad_norm": 1.9677225873579622,
+      "learning_rate": 3.7677525295088146e-06,
+      "loss": 0.8379,
+      "step": 6906
+    },
+    {
+      "epoch": 0.7228676085818942,
+      "grad_norm": 2.3777845625897482,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.8652,
+      "step": 6907
+    },
+    {
+      "epoch": 0.7229722658294087,
+      "grad_norm": 1.9315748698397706,
+      "learning_rate": 3.762452149718938e-06,
+      "loss": 0.9468,
+      "step": 6908
+    },
+    {
+      "epoch": 0.7230769230769231,
+      "grad_norm": 2.205149403013181,
+      "learning_rate": 3.7598030347321e-06,
+      "loss": 0.8733,
+      "step": 6909
+    },
+    {
+      "epoch": 0.7231815803244375,
+      "grad_norm": 1.8693736948006856,
+      "learning_rate": 3.757154636756547e-06,
+      "loss": 0.8561,
+      "step": 6910
+    },
+    {
+      "epoch": 0.7232862375719519,
+      "grad_norm": 1.873969507521071,
+      "learning_rate": 3.7545069560965817e-06,
+      "loss": 0.8318,
+      "step": 6911
+    },
+    {
+      "epoch": 0.7233908948194663,
+      "grad_norm": 2.42283057627858,
+      "learning_rate": 3.751859993056428e-06,
+      "loss": 0.8737,
+      "step": 6912
+    },
+    {
+      "epoch": 0.7234955520669807,
+      "grad_norm": 1.865784635381646,
+      "learning_rate": 3.749213747940225e-06,
+      "loss": 1.0405,
+      "step": 6913
+    },
+    {
+      "epoch": 0.723600209314495,
+      "grad_norm": 2.6319699814934623,
+      "learning_rate": 3.7465682210520325e-06,
+      "loss": 0.934,
+      "step": 6914
+    },
+    {
+      "epoch": 0.7237048665620094,
+      "grad_norm": 1.9195838969682297,
+      "learning_rate": 3.7439234126958326e-06,
+      "loss": 0.8494,
+      "step": 6915
+    },
+    {
+      "epoch": 0.7238095238095238,
+      "grad_norm": 2.1904292611117726,
+      "learning_rate": 3.741279323175515e-06,
+      "loss": 0.9698,
+      "step": 6916
+    },
+    {
+      "epoch": 0.7239141810570382,
+      "grad_norm": 2.2002972862059833,
+      "learning_rate": 3.7386359527948914e-06,
+      "loss": 0.8893,
+      "step": 6917
+    },
+    {
+      "epoch": 0.7240188383045526,
+      "grad_norm": 2.3913339759973917,
+      "learning_rate": 3.7359933018576887e-06,
+      "loss": 0.8367,
+      "step": 6918
+    },
+    {
+      "epoch": 0.724123495552067,
+      "grad_norm": 2.1969314364288484,
+      "learning_rate": 3.733351370667552e-06,
+      "loss": 0.9061,
+      "step": 6919
+    },
+    {
+      "epoch": 0.7242281527995814,
+      "grad_norm": 2.024923979347576,
+      "learning_rate": 3.7307101595280495e-06,
+      "loss": 0.9443,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7243328100470957,
+      "grad_norm": 2.113986243612249,
+      "learning_rate": 3.7280696687426578e-06,
+      "loss": 0.9265,
+      "step": 6921
+    },
+    {
+      "epoch": 0.7244374672946101,
+      "grad_norm": 1.950614982154138,
+      "learning_rate": 3.725429898614773e-06,
+      "loss": 0.9245,
+      "step": 6922
+    },
+    {
+      "epoch": 0.7245421245421245,
+      "grad_norm": 2.1324732521905205,
+      "learning_rate": 3.7227908494477174e-06,
+      "loss": 0.9454,
+      "step": 6923
+    },
+    {
+      "epoch": 0.7246467817896389,
+      "grad_norm": 2.2386944352919906,
+      "learning_rate": 3.7201525215447133e-06,
+      "loss": 0.9405,
+      "step": 6924
+    },
+    {
+      "epoch": 0.7247514390371533,
+      "grad_norm": 2.0332904488764236,
+      "learning_rate": 3.7175149152089185e-06,
+      "loss": 0.982,
+      "step": 6925
+    },
+    {
+      "epoch": 0.7248560962846677,
+      "grad_norm": 1.8576178852621916,
+      "learning_rate": 3.7148780307433975e-06,
+      "loss": 0.7685,
+      "step": 6926
+    },
+    {
+      "epoch": 0.7249607535321821,
+      "grad_norm": 2.1484299818831945,
+      "learning_rate": 3.7122418684511306e-06,
+      "loss": 0.8236,
+      "step": 6927
+    },
+    {
+      "epoch": 0.7250654107796964,
+      "grad_norm": 2.516712152720112,
+      "learning_rate": 3.7096064286350164e-06,
+      "loss": 0.8525,
+      "step": 6928
+    },
+    {
+      "epoch": 0.7251700680272108,
+      "grad_norm": 2.1566899339097936,
+      "learning_rate": 3.706971711597879e-06,
+      "loss": 0.926,
+      "step": 6929
+    },
+    {
+      "epoch": 0.7252747252747253,
+      "grad_norm": 1.7869536827919366,
+      "learning_rate": 3.7043377176424467e-06,
+      "loss": 0.9757,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7253793825222397,
+      "grad_norm": 1.70469612797693,
+      "learning_rate": 3.701704447071376e-06,
+      "loss": 0.8288,
+      "step": 6931
+    },
+    {
+      "epoch": 0.7254840397697541,
+      "grad_norm": 1.8514012748832875,
+      "learning_rate": 3.699071900187232e-06,
+      "loss": 0.8338,
+      "step": 6932
+    },
+    {
+      "epoch": 0.7255886970172685,
+      "grad_norm": 2.3795752920291773,
+      "learning_rate": 3.696440077292497e-06,
+      "loss": 0.862,
+      "step": 6933
+    },
+    {
+      "epoch": 0.7256933542647829,
+      "grad_norm": 2.2429703352322004,
+      "learning_rate": 3.6938089786895783e-06,
+      "loss": 1.0258,
+      "step": 6934
+    },
+    {
+      "epoch": 0.7257980115122973,
+      "grad_norm": 1.986089459097724,
+      "learning_rate": 3.691178604680793e-06,
+      "loss": 0.8968,
+      "step": 6935
+    },
+    {
+      "epoch": 0.7259026687598116,
+      "grad_norm": 1.8760611340077848,
+      "learning_rate": 3.688548955568375e-06,
+      "loss": 0.8637,
+      "step": 6936
+    },
+    {
+      "epoch": 0.726007326007326,
+      "grad_norm": 1.8062260800693908,
+      "learning_rate": 3.685920031654476e-06,
+      "loss": 0.8974,
+      "step": 6937
+    },
+    {
+      "epoch": 0.7261119832548404,
+      "grad_norm": 2.0910032713345155,
+      "learning_rate": 3.683291833241163e-06,
+      "loss": 0.9255,
+      "step": 6938
+    },
+    {
+      "epoch": 0.7262166405023548,
+      "grad_norm": 1.952084219425161,
+      "learning_rate": 3.6806643606304226e-06,
+      "loss": 0.954,
+      "step": 6939
+    },
+    {
+      "epoch": 0.7263212977498692,
+      "grad_norm": 2.2475768165777756,
+      "learning_rate": 3.6780376141241624e-06,
+      "loss": 0.7996,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7264259549973836,
+      "grad_norm": 2.039613944161897,
+      "learning_rate": 3.6754115940241954e-06,
+      "loss": 0.8029,
+      "step": 6941
+    },
+    {
+      "epoch": 0.726530612244898,
+      "grad_norm": 2.4372491132455054,
+      "learning_rate": 3.6727863006322585e-06,
+      "loss": 0.8573,
+      "step": 6942
+    },
+    {
+      "epoch": 0.7266352694924123,
+      "grad_norm": 2.135064158968042,
+      "learning_rate": 3.6701617342499975e-06,
+      "loss": 0.9144,
+      "step": 6943
+    },
+    {
+      "epoch": 0.7267399267399267,
+      "grad_norm": 2.141195883147221,
+      "learning_rate": 3.66753789517899e-06,
+      "loss": 0.9424,
+      "step": 6944
+    },
+    {
+      "epoch": 0.7268445839874411,
+      "grad_norm": 2.3453010874226496,
+      "learning_rate": 3.6649147837207143e-06,
+      "loss": 0.8654,
+      "step": 6945
+    },
+    {
+      "epoch": 0.7269492412349555,
+      "grad_norm": 1.7642203173816415,
+      "learning_rate": 3.6622924001765725e-06,
+      "loss": 0.8269,
+      "step": 6946
+    },
+    {
+      "epoch": 0.7270538984824699,
+      "grad_norm": 2.2323698391066835,
+      "learning_rate": 3.6596707448478796e-06,
+      "loss": 0.9424,
+      "step": 6947
+    },
+    {
+      "epoch": 0.7271585557299843,
+      "grad_norm": 2.631836450156995,
+      "learning_rate": 3.6570498180358705e-06,
+      "loss": 0.9094,
+      "step": 6948
+    },
+    {
+      "epoch": 0.7272632129774987,
+      "grad_norm": 1.9050273254003347,
+      "learning_rate": 3.6544296200416994e-06,
+      "loss": 0.9081,
+      "step": 6949
+    },
+    {
+      "epoch": 0.727367870225013,
+      "grad_norm": 2.1745905818103606,
+      "learning_rate": 3.6518101511664284e-06,
+      "loss": 0.9148,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7274725274725274,
+      "grad_norm": 2.1487595088513487,
+      "learning_rate": 3.6491914117110405e-06,
+      "loss": 0.9708,
+      "step": 6951
+    },
+    {
+      "epoch": 0.7275771847200418,
+      "grad_norm": 2.171265284624992,
+      "learning_rate": 3.646573401976431e-06,
+      "loss": 0.9458,
+      "step": 6952
+    },
+    {
+      "epoch": 0.7276818419675563,
+      "grad_norm": 2.762939860481466,
+      "learning_rate": 3.64395612226342e-06,
+      "loss": 0.9506,
+      "step": 6953
+    },
+    {
+      "epoch": 0.7277864992150707,
+      "grad_norm": 2.076483034212252,
+      "learning_rate": 3.641339572872735e-06,
+      "loss": 0.8965,
+      "step": 6954
+    },
+    {
+      "epoch": 0.7278911564625851,
+      "grad_norm": 1.8982571709599116,
+      "learning_rate": 3.638723754105025e-06,
+      "loss": 0.8545,
+      "step": 6955
+    },
+    {
+      "epoch": 0.7279958137100995,
+      "grad_norm": 1.6455039950849177,
+      "learning_rate": 3.636108666260847e-06,
+      "loss": 0.7816,
+      "step": 6956
+    },
+    {
+      "epoch": 0.7281004709576138,
+      "grad_norm": 2.1303615491352286,
+      "learning_rate": 3.6334943096406873e-06,
+      "loss": 0.967,
+      "step": 6957
+    },
+    {
+      "epoch": 0.7282051282051282,
+      "grad_norm": 2.3205676424264197,
+      "learning_rate": 3.630880684544934e-06,
+      "loss": 0.9537,
+      "step": 6958
+    },
+    {
+      "epoch": 0.7283097854526426,
+      "grad_norm": 1.9729345731294923,
+      "learning_rate": 3.628267791273906e-06,
+      "loss": 0.8789,
+      "step": 6959
+    },
+    {
+      "epoch": 0.728414442700157,
+      "grad_norm": 2.0365158246102704,
+      "learning_rate": 3.625655630127826e-06,
+      "loss": 0.8889,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7285190999476714,
+      "grad_norm": 2.0572596307136086,
+      "learning_rate": 3.6230442014068346e-06,
+      "loss": 0.8673,
+      "step": 6961
+    },
+    {
+      "epoch": 0.7286237571951858,
+      "grad_norm": 2.265474450323085,
+      "learning_rate": 3.6204335054109897e-06,
+      "loss": 0.8654,
+      "step": 6962
+    },
+    {
+      "epoch": 0.7287284144427002,
+      "grad_norm": 2.036780676342137,
+      "learning_rate": 3.6178235424402707e-06,
+      "loss": 0.9066,
+      "step": 6963
+    },
+    {
+      "epoch": 0.7288330716902145,
+      "grad_norm": 2.3072772153174985,
+      "learning_rate": 3.6152143127945647e-06,
+      "loss": 0.8402,
+      "step": 6964
+    },
+    {
+      "epoch": 0.7289377289377289,
+      "grad_norm": 1.5267072530528316,
+      "learning_rate": 3.6126058167736742e-06,
+      "loss": 0.7113,
+      "step": 6965
+    },
+    {
+      "epoch": 0.7290423861852433,
+      "grad_norm": 2.312199363621423,
+      "learning_rate": 3.609998054677327e-06,
+      "loss": 0.894,
+      "step": 6966
+    },
+    {
+      "epoch": 0.7291470434327577,
+      "grad_norm": 2.1365598908923964,
+      "learning_rate": 3.607391026805155e-06,
+      "loss": 0.8484,
+      "step": 6967
+    },
+    {
+      "epoch": 0.7292517006802721,
+      "grad_norm": 1.814578926920421,
+      "learning_rate": 3.6047847334567153e-06,
+      "loss": 0.8571,
+      "step": 6968
+    },
+    {
+      "epoch": 0.7293563579277865,
+      "grad_norm": 2.5103976495037257,
+      "learning_rate": 3.602179174931475e-06,
+      "loss": 0.874,
+      "step": 6969
+    },
+    {
+      "epoch": 0.7294610151753009,
+      "grad_norm": 1.9816767917289961,
+      "learning_rate": 3.5995743515288163e-06,
+      "loss": 0.9487,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7295656724228152,
+      "grad_norm": 1.6802672691339464,
+      "learning_rate": 3.5969702635480374e-06,
+      "loss": 0.9187,
+      "step": 6971
+    },
+    {
+      "epoch": 0.7296703296703296,
+      "grad_norm": 1.9087881371348294,
+      "learning_rate": 3.594366911288358e-06,
+      "loss": 0.8732,
+      "step": 6972
+    },
+    {
+      "epoch": 0.729774986917844,
+      "grad_norm": 2.097973491871284,
+      "learning_rate": 3.5917642950489062e-06,
+      "loss": 1.0095,
+      "step": 6973
+    },
+    {
+      "epoch": 0.7298796441653584,
+      "grad_norm": 2.0996759377189043,
+      "learning_rate": 3.5891624151287253e-06,
+      "loss": 0.9249,
+      "step": 6974
+    },
+    {
+      "epoch": 0.7299843014128728,
+      "grad_norm": 2.0231942048602405,
+      "learning_rate": 3.5865612718267807e-06,
+      "loss": 0.8537,
+      "step": 6975
+    },
+    {
+      "epoch": 0.7300889586603873,
+      "grad_norm": 2.2582186780087894,
+      "learning_rate": 3.583960865441949e-06,
+      "loss": 0.8968,
+      "step": 6976
+    },
+    {
+      "epoch": 0.7301936159079017,
+      "grad_norm": 2.1127814779861462,
+      "learning_rate": 3.5813611962730165e-06,
+      "loss": 0.8105,
+      "step": 6977
+    },
+    {
+      "epoch": 0.7302982731554161,
+      "grad_norm": 2.1735847999153677,
+      "learning_rate": 3.578762264618697e-06,
+      "loss": 0.9232,
+      "step": 6978
+    },
+    {
+      "epoch": 0.7304029304029304,
+      "grad_norm": 2.086077873057493,
+      "learning_rate": 3.5761640707776115e-06,
+      "loss": 0.9935,
+      "step": 6979
+    },
+    {
+      "epoch": 0.7305075876504448,
+      "grad_norm": 1.926989359894305,
+      "learning_rate": 3.573566615048297e-06,
+      "loss": 0.9609,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7306122448979592,
+      "grad_norm": 1.7955588355729455,
+      "learning_rate": 3.570969897729202e-06,
+      "loss": 0.9249,
+      "step": 6981
+    },
+    {
+      "epoch": 0.7307169021454736,
+      "grad_norm": 2.0822537293283134,
+      "learning_rate": 3.5683739191187027e-06,
+      "loss": 0.92,
+      "step": 6982
+    },
+    {
+      "epoch": 0.730821559392988,
+      "grad_norm": 1.988495753359855,
+      "learning_rate": 3.565778679515075e-06,
+      "loss": 0.7726,
+      "step": 6983
+    },
+    {
+      "epoch": 0.7309262166405024,
+      "grad_norm": 2.1187421281363172,
+      "learning_rate": 3.5631841792165236e-06,
+      "loss": 0.8689,
+      "step": 6984
+    },
+    {
+      "epoch": 0.7310308738880168,
+      "grad_norm": 1.8611241031497499,
+      "learning_rate": 3.5605904185211582e-06,
+      "loss": 0.931,
+      "step": 6985
+    },
+    {
+      "epoch": 0.7311355311355311,
+      "grad_norm": 1.8366170447158119,
+      "learning_rate": 3.5579973977270057e-06,
+      "loss": 0.864,
+      "step": 6986
+    },
+    {
+      "epoch": 0.7312401883830455,
+      "grad_norm": 2.204065983976754,
+      "learning_rate": 3.5554051171320136e-06,
+      "loss": 0.9756,
+      "step": 6987
+    },
+    {
+      "epoch": 0.7313448456305599,
+      "grad_norm": 2.133144515554238,
+      "learning_rate": 3.552813577034039e-06,
+      "loss": 0.9533,
+      "step": 6988
+    },
+    {
+      "epoch": 0.7314495028780743,
+      "grad_norm": 1.9662548833887095,
+      "learning_rate": 3.550222777730854e-06,
+      "loss": 0.8933,
+      "step": 6989
+    },
+    {
+      "epoch": 0.7315541601255887,
+      "grad_norm": 1.8080125380457004,
+      "learning_rate": 3.5476327195201463e-06,
+      "loss": 0.8176,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7316588173731031,
+      "grad_norm": 2.097754039027627,
+      "learning_rate": 3.545043402699514e-06,
+      "loss": 0.9289,
+      "step": 6991
+    },
+    {
+      "epoch": 0.7317634746206175,
+      "grad_norm": 1.9647814759624955,
+      "learning_rate": 3.542454827566486e-06,
+      "loss": 0.9348,
+      "step": 6992
+    },
+    {
+      "epoch": 0.7318681318681318,
+      "grad_norm": 2.183568175809098,
+      "learning_rate": 3.5398669944184894e-06,
+      "loss": 0.8125,
+      "step": 6993
+    },
+    {
+      "epoch": 0.7319727891156462,
+      "grad_norm": 2.2546413328737906,
+      "learning_rate": 3.5372799035528716e-06,
+      "loss": 0.8066,
+      "step": 6994
+    },
+    {
+      "epoch": 0.7320774463631606,
+      "grad_norm": 2.018409429501914,
+      "learning_rate": 3.5346935552668936e-06,
+      "loss": 0.876,
+      "step": 6995
+    },
+    {
+      "epoch": 0.732182103610675,
+      "grad_norm": 1.8056317532440103,
+      "learning_rate": 3.5321079498577292e-06,
+      "loss": 0.7721,
+      "step": 6996
+    },
+    {
+      "epoch": 0.7322867608581894,
+      "grad_norm": 2.3671016869226147,
+      "learning_rate": 3.5295230876224763e-06,
+      "loss": 0.9053,
+      "step": 6997
+    },
+    {
+      "epoch": 0.7323914181057039,
+      "grad_norm": 2.26537165470211,
+      "learning_rate": 3.526938968858139e-06,
+      "loss": 0.8924,
+      "step": 6998
+    },
+    {
+      "epoch": 0.7324960753532183,
+      "grad_norm": 2.261177073983615,
+      "learning_rate": 3.5243555938616348e-06,
+      "loss": 0.7816,
+      "step": 6999
+    },
+    {
+      "epoch": 0.7326007326007326,
+      "grad_norm": 2.0780543997404592,
+      "learning_rate": 3.521772962929798e-06,
+      "loss": 0.9149,
+      "step": 7000
+    },
+    {
+      "epoch": 0.732705389848247,
+      "grad_norm": 2.0900590029616897,
+      "learning_rate": 3.5191910763593808e-06,
+      "loss": 0.8636,
+      "step": 7001
+    },
+    {
+      "epoch": 0.7328100470957614,
+      "grad_norm": 2.0840433266339184,
+      "learning_rate": 3.5166099344470493e-06,
+      "loss": 0.8882,
+      "step": 7002
+    },
+    {
+      "epoch": 0.7329147043432758,
+      "grad_norm": 2.1351284570724713,
+      "learning_rate": 3.5140295374893797e-06,
+      "loss": 0.8069,
+      "step": 7003
+    },
+    {
+      "epoch": 0.7330193615907902,
+      "grad_norm": 2.104817813524145,
+      "learning_rate": 3.511449885782865e-06,
+      "loss": 0.9998,
+      "step": 7004
+    },
+    {
+      "epoch": 0.7331240188383046,
+      "grad_norm": 1.977905622536921,
+      "learning_rate": 3.5088709796239086e-06,
+      "loss": 0.8087,
+      "step": 7005
+    },
+    {
+      "epoch": 0.733228676085819,
+      "grad_norm": 1.803351442799279,
+      "learning_rate": 3.5062928193088385e-06,
+      "loss": 0.8974,
+      "step": 7006
+    },
+    {
+      "epoch": 0.7333333333333333,
+      "grad_norm": 2.1137702304154575,
+      "learning_rate": 3.5037154051338883e-06,
+      "loss": 0.9726,
+      "step": 7007
+    },
+    {
+      "epoch": 0.7334379905808477,
+      "grad_norm": 2.2621279545247024,
+      "learning_rate": 3.501138737395208e-06,
+      "loss": 0.9225,
+      "step": 7008
+    },
+    {
+      "epoch": 0.7335426478283621,
+      "grad_norm": 2.042349787240519,
+      "learning_rate": 3.4985628163888564e-06,
+      "loss": 1.043,
+      "step": 7009
+    },
+    {
+      "epoch": 0.7336473050758765,
+      "grad_norm": 2.1738234952277793,
+      "learning_rate": 3.4959876424108173e-06,
+      "loss": 0.9549,
+      "step": 7010
+    },
+    {
+      "epoch": 0.7337519623233909,
+      "grad_norm": 1.8561044157153646,
+      "learning_rate": 3.4934132157569866e-06,
+      "loss": 0.9149,
+      "step": 7011
+    },
+    {
+      "epoch": 0.7338566195709053,
+      "grad_norm": 1.8449197732704996,
+      "learning_rate": 3.4908395367231683e-06,
+      "loss": 0.7587,
+      "step": 7012
+    },
+    {
+      "epoch": 0.7339612768184197,
+      "grad_norm": 2.0696609987749945,
+      "learning_rate": 3.488266605605082e-06,
+      "loss": 0.9359,
+      "step": 7013
+    },
+    {
+      "epoch": 0.734065934065934,
+      "grad_norm": 2.138497203840195,
+      "learning_rate": 3.4856944226983637e-06,
+      "loss": 0.8327,
+      "step": 7014
+    },
+    {
+      "epoch": 0.7341705913134484,
+      "grad_norm": 1.6468331397988603,
+      "learning_rate": 3.4831229882985585e-06,
+      "loss": 0.8576,
+      "step": 7015
+    },
+    {
+      "epoch": 0.7342752485609628,
+      "grad_norm": 1.977918872042594,
+      "learning_rate": 3.4805523027011368e-06,
+      "loss": 0.9462,
+      "step": 7016
+    },
+    {
+      "epoch": 0.7343799058084772,
+      "grad_norm": 1.8654816094421645,
+      "learning_rate": 3.477982366201468e-06,
+      "loss": 0.8792,
+      "step": 7017
+    },
+    {
+      "epoch": 0.7344845630559916,
+      "grad_norm": 2.2589654681645936,
+      "learning_rate": 3.475413179094851e-06,
+      "loss": 0.9523,
+      "step": 7018
+    },
+    {
+      "epoch": 0.734589220303506,
+      "grad_norm": 1.9481397990624987,
+      "learning_rate": 3.4728447416764854e-06,
+      "loss": 0.9307,
+      "step": 7019
+    },
+    {
+      "epoch": 0.7346938775510204,
+      "grad_norm": 1.8809532054531273,
+      "learning_rate": 3.470277054241488e-06,
+      "loss": 0.8929,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7347985347985349,
+      "grad_norm": 1.7678658679826986,
+      "learning_rate": 3.467710117084897e-06,
+      "loss": 0.8018,
+      "step": 7021
+    },
+    {
+      "epoch": 0.7349031920460491,
+      "grad_norm": 2.14212227669439,
+      "learning_rate": 3.4651439305016565e-06,
+      "loss": 0.8358,
+      "step": 7022
+    },
+    {
+      "epoch": 0.7350078492935636,
+      "grad_norm": 2.0802743972041218,
+      "learning_rate": 3.4625784947866258e-06,
+      "loss": 0.8696,
+      "step": 7023
+    },
+    {
+      "epoch": 0.735112506541078,
+      "grad_norm": 1.945003157814486,
+      "learning_rate": 3.460013810234576e-06,
+      "loss": 0.9213,
+      "step": 7024
+    },
+    {
+      "epoch": 0.7352171637885924,
+      "grad_norm": 2.0651692861153945,
+      "learning_rate": 3.4574498771401997e-06,
+      "loss": 0.8322,
+      "step": 7025
+    },
+    {
+      "epoch": 0.7353218210361068,
+      "grad_norm": 1.8951231983789885,
+      "learning_rate": 3.454886695798093e-06,
+      "loss": 0.9921,
+      "step": 7026
+    },
+    {
+      "epoch": 0.7354264782836212,
+      "grad_norm": 2.034145921283138,
+      "learning_rate": 3.452324266502777e-06,
+      "loss": 0.9316,
+      "step": 7027
+    },
+    {
+      "epoch": 0.7355311355311356,
+      "grad_norm": 2.632863001346807,
+      "learning_rate": 3.4497625895486755e-06,
+      "loss": 0.9863,
+      "step": 7028
+    },
+    {
+      "epoch": 0.7356357927786499,
+      "grad_norm": 1.8786479307461825,
+      "learning_rate": 3.4472016652301276e-06,
+      "loss": 0.8824,
+      "step": 7029
+    },
+    {
+      "epoch": 0.7357404500261643,
+      "grad_norm": 2.0135193900407584,
+      "learning_rate": 3.4446414938413964e-06,
+      "loss": 0.9466,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7358451072736787,
+      "grad_norm": 2.2920626537087316,
+      "learning_rate": 3.442082075676646e-06,
+      "loss": 0.864,
+      "step": 7031
+    },
+    {
+      "epoch": 0.7359497645211931,
+      "grad_norm": 1.9835068207886113,
+      "learning_rate": 3.4395234110299593e-06,
+      "loss": 1.0386,
+      "step": 7032
+    },
+    {
+      "epoch": 0.7360544217687075,
+      "grad_norm": 1.8964138450221253,
+      "learning_rate": 3.4369655001953327e-06,
+      "loss": 0.8778,
+      "step": 7033
+    },
+    {
+      "epoch": 0.7361590790162219,
+      "grad_norm": 1.9000636515942269,
+      "learning_rate": 3.4344083434666707e-06,
+      "loss": 0.9043,
+      "step": 7034
+    },
+    {
+      "epoch": 0.7362637362637363,
+      "grad_norm": 1.885698255986817,
+      "learning_rate": 3.4318519411378006e-06,
+      "loss": 0.8209,
+      "step": 7035
+    },
+    {
+      "epoch": 0.7363683935112506,
+      "grad_norm": 1.9675198172784103,
+      "learning_rate": 3.429296293502461e-06,
+      "loss": 0.9245,
+      "step": 7036
+    },
+    {
+      "epoch": 0.736473050758765,
+      "grad_norm": 2.3490174091108518,
+      "learning_rate": 3.4267414008542967e-06,
+      "loss": 0.8804,
+      "step": 7037
+    },
+    {
+      "epoch": 0.7365777080062794,
+      "grad_norm": 1.981563826198973,
+      "learning_rate": 3.424187263486871e-06,
+      "loss": 0.9156,
+      "step": 7038
+    },
+    {
+      "epoch": 0.7366823652537938,
+      "grad_norm": 2.0151298287952373,
+      "learning_rate": 3.421633881693657e-06,
+      "loss": 0.8659,
+      "step": 7039
+    },
+    {
+      "epoch": 0.7367870225013082,
+      "grad_norm": 1.7892471082853696,
+      "learning_rate": 3.4190812557680487e-06,
+      "loss": 0.8883,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7368916797488226,
+      "grad_norm": 2.6350578287776822,
+      "learning_rate": 3.4165293860033444e-06,
+      "loss": 0.9169,
+      "step": 7041
+    },
+    {
+      "epoch": 0.736996336996337,
+      "grad_norm": 1.9233102606504318,
+      "learning_rate": 3.4139782726927597e-06,
+      "loss": 0.8025,
+      "step": 7042
+    },
+    {
+      "epoch": 0.7371009942438513,
+      "grad_norm": 1.8506753750571818,
+      "learning_rate": 3.4114279161294195e-06,
+      "loss": 0.9393,
+      "step": 7043
+    },
+    {
+      "epoch": 0.7372056514913657,
+      "grad_norm": 1.850056642327913,
+      "learning_rate": 3.4088783166063686e-06,
+      "loss": 0.8943,
+      "step": 7044
+    },
+    {
+      "epoch": 0.7373103087388801,
+      "grad_norm": 2.153444187423901,
+      "learning_rate": 3.406329474416563e-06,
+      "loss": 1.0154,
+      "step": 7045
+    },
+    {
+      "epoch": 0.7374149659863946,
+      "grad_norm": 2.0167001869694383,
+      "learning_rate": 3.4037813898528682e-06,
+      "loss": 0.8576,
+      "step": 7046
+    },
+    {
+      "epoch": 0.737519623233909,
+      "grad_norm": 2.065120817248703,
+      "learning_rate": 3.401234063208064e-06,
+      "loss": 0.839,
+      "step": 7047
+    },
+    {
+      "epoch": 0.7376242804814234,
+      "grad_norm": 2.066946242767024,
+      "learning_rate": 3.3986874947748428e-06,
+      "loss": 0.9666,
+      "step": 7048
+    },
+    {
+      "epoch": 0.7377289377289378,
+      "grad_norm": 1.8889031726518197,
+      "learning_rate": 3.3961416848458073e-06,
+      "loss": 0.8222,
+      "step": 7049
+    },
+    {
+      "epoch": 0.7378335949764521,
+      "grad_norm": 1.9632185729176141,
+      "learning_rate": 3.393596633713483e-06,
+      "loss": 0.8447,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7379382522239665,
+      "grad_norm": 2.0510910347919413,
+      "learning_rate": 3.3910523416702977e-06,
+      "loss": 1.0462,
+      "step": 7051
+    },
+    {
+      "epoch": 0.7380429094714809,
+      "grad_norm": 2.12388351552657,
+      "learning_rate": 3.388508809008594e-06,
+      "loss": 0.8637,
+      "step": 7052
+    },
+    {
+      "epoch": 0.7381475667189953,
+      "grad_norm": 1.8703899844326586,
+      "learning_rate": 3.3859660360206336e-06,
+      "loss": 0.8867,
+      "step": 7053
+    },
+    {
+      "epoch": 0.7382522239665097,
+      "grad_norm": 2.022816664688723,
+      "learning_rate": 3.383424022998579e-06,
+      "loss": 0.8605,
+      "step": 7054
+    },
+    {
+      "epoch": 0.7383568812140241,
+      "grad_norm": 2.1134498670646216,
+      "learning_rate": 3.3808827702345227e-06,
+      "loss": 0.8655,
+      "step": 7055
+    },
+    {
+      "epoch": 0.7384615384615385,
+      "grad_norm": 1.9780213401794018,
+      "learning_rate": 3.3783422780204535e-06,
+      "loss": 0.9235,
+      "step": 7056
+    },
+    {
+      "epoch": 0.7385661957090528,
+      "grad_norm": 1.9464458644225906,
+      "learning_rate": 3.37580254664828e-06,
+      "loss": 0.9951,
+      "step": 7057
+    },
+    {
+      "epoch": 0.7386708529565672,
+      "grad_norm": 1.8745039327238657,
+      "learning_rate": 3.37326357640982e-06,
+      "loss": 0.7171,
+      "step": 7058
+    },
+    {
+      "epoch": 0.7387755102040816,
+      "grad_norm": 2.037899312393001,
+      "learning_rate": 3.370725367596811e-06,
+      "loss": 0.9593,
+      "step": 7059
+    },
+    {
+      "epoch": 0.738880167451596,
+      "grad_norm": 2.1099882075886844,
+      "learning_rate": 3.368187920500897e-06,
+      "loss": 0.9113,
+      "step": 7060
+    },
+    {
+      "epoch": 0.7389848246991104,
+      "grad_norm": 2.6349773976618667,
+      "learning_rate": 3.3656512354136317e-06,
+      "loss": 0.7943,
+      "step": 7061
+    },
+    {
+      "epoch": 0.7390894819466248,
+      "grad_norm": 2.262938094716051,
+      "learning_rate": 3.3631153126264915e-06,
+      "loss": 0.8173,
+      "step": 7062
+    },
+    {
+      "epoch": 0.7391941391941392,
+      "grad_norm": 1.9162832453637948,
+      "learning_rate": 3.3605801524308535e-06,
+      "loss": 0.7447,
+      "step": 7063
+    },
+    {
+      "epoch": 0.7392987964416536,
+      "grad_norm": 2.4566628443132372,
+      "learning_rate": 3.358045755118019e-06,
+      "loss": 0.7898,
+      "step": 7064
+    },
+    {
+      "epoch": 0.7394034536891679,
+      "grad_norm": 2.1132700561524316,
+      "learning_rate": 3.3555121209791906e-06,
+      "loss": 0.8397,
+      "step": 7065
+    },
+    {
+      "epoch": 0.7395081109366823,
+      "grad_norm": 2.6164728972304645,
+      "learning_rate": 3.3529792503054903e-06,
+      "loss": 0.9608,
+      "step": 7066
+    },
+    {
+      "epoch": 0.7396127681841967,
+      "grad_norm": 2.3074332053564137,
+      "learning_rate": 3.3504471433879493e-06,
+      "loss": 0.9128,
+      "step": 7067
+    },
+    {
+      "epoch": 0.7397174254317112,
+      "grad_norm": 2.0330030779609234,
+      "learning_rate": 3.3479158005175082e-06,
+      "loss": 0.9589,
+      "step": 7068
+    },
+    {
+      "epoch": 0.7398220826792256,
+      "grad_norm": 2.120921500126816,
+      "learning_rate": 3.345385221985026e-06,
+      "loss": 0.9727,
+      "step": 7069
+    },
+    {
+      "epoch": 0.73992673992674,
+      "grad_norm": 2.6676041899583107,
+      "learning_rate": 3.342855408081276e-06,
+      "loss": 0.912,
+      "step": 7070
+    },
+    {
+      "epoch": 0.7400313971742544,
+      "grad_norm": 1.9000272361188917,
+      "learning_rate": 3.340326359096935e-06,
+      "loss": 0.7653,
+      "step": 7071
+    },
+    {
+      "epoch": 0.7401360544217687,
+      "grad_norm": 2.1277226935445817,
+      "learning_rate": 3.337798075322596e-06,
+      "loss": 0.835,
+      "step": 7072
+    },
+    {
+      "epoch": 0.7402407116692831,
+      "grad_norm": 2.1217486595239565,
+      "learning_rate": 3.3352705570487598e-06,
+      "loss": 0.9353,
+      "step": 7073
+    },
+    {
+      "epoch": 0.7403453689167975,
+      "grad_norm": 2.099696714940113,
+      "learning_rate": 3.332743804565851e-06,
+      "loss": 0.9437,
+      "step": 7074
+    },
+    {
+      "epoch": 0.7404500261643119,
+      "grad_norm": 2.310679915225124,
+      "learning_rate": 3.330217818164195e-06,
+      "loss": 0.8943,
+      "step": 7075
+    },
+    {
+      "epoch": 0.7405546834118263,
+      "grad_norm": 1.9472970296459848,
+      "learning_rate": 3.3276925981340336e-06,
+      "loss": 0.8915,
+      "step": 7076
+    },
+    {
+      "epoch": 0.7406593406593407,
+      "grad_norm": 2.3805641441348975,
+      "learning_rate": 3.325168144765515e-06,
+      "loss": 0.9253,
+      "step": 7077
+    },
+    {
+      "epoch": 0.7407639979068551,
+      "grad_norm": 1.9193799868580086,
+      "learning_rate": 3.3226444583487085e-06,
+      "loss": 1.0081,
+      "step": 7078
+    },
+    {
+      "epoch": 0.7408686551543694,
+      "grad_norm": 2.2795516135400287,
+      "learning_rate": 3.320121539173592e-06,
+      "loss": 0.8681,
+      "step": 7079
+    },
+    {
+      "epoch": 0.7409733124018838,
+      "grad_norm": 1.9038999366774318,
+      "learning_rate": 3.3175993875300527e-06,
+      "loss": 0.8152,
+      "step": 7080
+    },
+    {
+      "epoch": 0.7410779696493982,
+      "grad_norm": 2.209622572681599,
+      "learning_rate": 3.315078003707891e-06,
+      "loss": 1.0209,
+      "step": 7081
+    },
+    {
+      "epoch": 0.7411826268969126,
+      "grad_norm": 2.3636374036749004,
+      "learning_rate": 3.3125573879968154e-06,
+      "loss": 0.8659,
+      "step": 7082
+    },
+    {
+      "epoch": 0.741287284144427,
+      "grad_norm": 2.125287520788546,
+      "learning_rate": 3.310037540686455e-06,
+      "loss": 1.0528,
+      "step": 7083
+    },
+    {
+      "epoch": 0.7413919413919414,
+      "grad_norm": 2.015778138651614,
+      "learning_rate": 3.307518462066344e-06,
+      "loss": 0.8942,
+      "step": 7084
+    },
+    {
+      "epoch": 0.7414965986394558,
+      "grad_norm": 2.4686495141026987,
+      "learning_rate": 3.305000152425928e-06,
+      "loss": 0.8532,
+      "step": 7085
+    },
+    {
+      "epoch": 0.7416012558869701,
+      "grad_norm": 1.9474568598951196,
+      "learning_rate": 3.3024826120545673e-06,
+      "loss": 0.9053,
+      "step": 7086
+    },
+    {
+      "epoch": 0.7417059131344845,
+      "grad_norm": 1.6828735085636821,
+      "learning_rate": 3.299965841241525e-06,
+      "loss": 0.8393,
+      "step": 7087
+    },
+    {
+      "epoch": 0.7418105703819989,
+      "grad_norm": 2.081597291218235,
+      "learning_rate": 3.297449840275996e-06,
+      "loss": 0.9671,
+      "step": 7088
+    },
+    {
+      "epoch": 0.7419152276295133,
+      "grad_norm": 1.6821107578784833,
+      "learning_rate": 3.294934609447068e-06,
+      "loss": 0.7459,
+      "step": 7089
+    },
+    {
+      "epoch": 0.7420198848770277,
+      "grad_norm": 2.2858216269881684,
+      "learning_rate": 3.292420149043747e-06,
+      "loss": 0.9515,
+      "step": 7090
+    },
+    {
+      "epoch": 0.7421245421245422,
+      "grad_norm": 2.1396541624839793,
+      "learning_rate": 3.2899064593549477e-06,
+      "loss": 0.9349,
+      "step": 7091
+    },
+    {
+      "epoch": 0.7422291993720566,
+      "grad_norm": 2.172610516131483,
+      "learning_rate": 3.2873935406694956e-06,
+      "loss": 0.9582,
+      "step": 7092
+    },
+    {
+      "epoch": 0.7423338566195709,
+      "grad_norm": 2.1072743883866374,
+      "learning_rate": 3.284881393276137e-06,
+      "loss": 0.9638,
+      "step": 7093
+    },
+    {
+      "epoch": 0.7424385138670853,
+      "grad_norm": 2.30997826595737,
+      "learning_rate": 3.2823700174635185e-06,
+      "loss": 0.9079,
+      "step": 7094
+    },
+    {
+      "epoch": 0.7425431711145997,
+      "grad_norm": 2.079680297310184,
+      "learning_rate": 3.2798594135202012e-06,
+      "loss": 0.9564,
+      "step": 7095
+    },
+    {
+      "epoch": 0.7426478283621141,
+      "grad_norm": 2.2079477477185763,
+      "learning_rate": 3.2773495817346636e-06,
+      "loss": 0.8085,
+      "step": 7096
+    },
+    {
+      "epoch": 0.7427524856096285,
+      "grad_norm": 1.9783748322647048,
+      "learning_rate": 3.274840522395283e-06,
+      "loss": 0.9652,
+      "step": 7097
+    },
+    {
+      "epoch": 0.7428571428571429,
+      "grad_norm": 2.013884564094195,
+      "learning_rate": 3.272332235790363e-06,
+      "loss": 0.8247,
+      "step": 7098
+    },
+    {
+      "epoch": 0.7429618001046573,
+      "grad_norm": 1.8380568856422557,
+      "learning_rate": 3.269824722208108e-06,
+      "loss": 0.7911,
+      "step": 7099
+    },
+    {
+      "epoch": 0.7430664573521716,
+      "grad_norm": 1.9699143563198767,
+      "learning_rate": 3.2673179819366363e-06,
+      "loss": 0.917,
+      "step": 7100
+    },
+    {
+      "epoch": 0.743171114599686,
+      "grad_norm": 1.8324074187595984,
+      "learning_rate": 3.264812015263973e-06,
+      "loss": 0.8143,
+      "step": 7101
+    },
+    {
+      "epoch": 0.7432757718472004,
+      "grad_norm": 2.0807751148157196,
+      "learning_rate": 3.2623068224780663e-06,
+      "loss": 0.9625,
+      "step": 7102
+    },
+    {
+      "epoch": 0.7433804290947148,
+      "grad_norm": 2.0925637950807556,
+      "learning_rate": 3.2598024038667655e-06,
+      "loss": 0.8431,
+      "step": 7103
+    },
+    {
+      "epoch": 0.7434850863422292,
+      "grad_norm": 2.3755126784483402,
+      "learning_rate": 3.2572987597178274e-06,
+      "loss": 0.8246,
+      "step": 7104
+    },
+    {
+      "epoch": 0.7435897435897436,
+      "grad_norm": 1.9385820118174737,
+      "learning_rate": 3.254795890318935e-06,
+      "loss": 0.8357,
+      "step": 7105
+    },
+    {
+      "epoch": 0.743694400837258,
+      "grad_norm": 1.9631332561496653,
+      "learning_rate": 3.2522937959576684e-06,
+      "loss": 0.9545,
+      "step": 7106
+    },
+    {
+      "epoch": 0.7437990580847724,
+      "grad_norm": 1.9072432029408546,
+      "learning_rate": 3.2497924769215206e-06,
+      "loss": 0.9965,
+      "step": 7107
+    },
+    {
+      "epoch": 0.7439037153322867,
+      "grad_norm": 2.3417750032717493,
+      "learning_rate": 3.2472919334979034e-06,
+      "loss": 1.0051,
+      "step": 7108
+    },
+    {
+      "epoch": 0.7440083725798011,
+      "grad_norm": 2.00729117210432,
+      "learning_rate": 3.2447921659741333e-06,
+      "loss": 0.9051,
+      "step": 7109
+    },
+    {
+      "epoch": 0.7441130298273155,
+      "grad_norm": 1.8562956071816943,
+      "learning_rate": 3.2422931746374375e-06,
+      "loss": 0.8059,
+      "step": 7110
+    },
+    {
+      "epoch": 0.7442176870748299,
+      "grad_norm": 2.5424170068230647,
+      "learning_rate": 3.2397949597749525e-06,
+      "loss": 0.767,
+      "step": 7111
+    },
+    {
+      "epoch": 0.7443223443223443,
+      "grad_norm": 2.2671400535027204,
+      "learning_rate": 3.2372975216737335e-06,
+      "loss": 0.9178,
+      "step": 7112
+    },
+    {
+      "epoch": 0.7444270015698587,
+      "grad_norm": 2.3361550774786988,
+      "learning_rate": 3.234800860620736e-06,
+      "loss": 0.7264,
+      "step": 7113
+    },
+    {
+      "epoch": 0.7445316588173732,
+      "grad_norm": 1.9859089354809334,
+      "learning_rate": 3.232304976902837e-06,
+      "loss": 0.9158,
+      "step": 7114
+    },
+    {
+      "epoch": 0.7446363160648874,
+      "grad_norm": 1.7913360410024908,
+      "learning_rate": 3.229809870806815e-06,
+      "loss": 0.8756,
+      "step": 7115
+    },
+    {
+      "epoch": 0.7447409733124019,
+      "grad_norm": 2.122253887975913,
+      "learning_rate": 3.2273155426193613e-06,
+      "loss": 0.6974,
+      "step": 7116
+    },
+    {
+      "epoch": 0.7448456305599163,
+      "grad_norm": 1.790135145026046,
+      "learning_rate": 3.224821992627084e-06,
+      "loss": 0.8797,
+      "step": 7117
+    },
+    {
+      "epoch": 0.7449502878074307,
+      "grad_norm": 2.0103996789911998,
+      "learning_rate": 3.2223292211164946e-06,
+      "loss": 0.7686,
+      "step": 7118
+    },
+    {
+      "epoch": 0.7450549450549451,
+      "grad_norm": 2.3239725196592413,
+      "learning_rate": 3.2198372283740176e-06,
+      "loss": 0.8821,
+      "step": 7119
+    },
+    {
+      "epoch": 0.7451596023024595,
+      "grad_norm": 2.4439097088179667,
+      "learning_rate": 3.217346014685985e-06,
+      "loss": 0.8314,
+      "step": 7120
+    },
+    {
+      "epoch": 0.7452642595499739,
+      "grad_norm": 1.8200085048128307,
+      "learning_rate": 3.214855580338644e-06,
+      "loss": 0.8329,
+      "step": 7121
+    },
+    {
+      "epoch": 0.7453689167974882,
+      "grad_norm": 1.8651829102105335,
+      "learning_rate": 3.212365925618156e-06,
+      "loss": 0.8808,
+      "step": 7122
+    },
+    {
+      "epoch": 0.7454735740450026,
+      "grad_norm": 1.954259108086037,
+      "learning_rate": 3.2098770508105825e-06,
+      "loss": 0.9463,
+      "step": 7123
+    },
+    {
+      "epoch": 0.745578231292517,
+      "grad_norm": 2.026272455843576,
+      "learning_rate": 3.207388956201901e-06,
+      "loss": 0.8874,
+      "step": 7124
+    },
+    {
+      "epoch": 0.7456828885400314,
+      "grad_norm": 1.9861545009241273,
+      "learning_rate": 3.204901642077999e-06,
+      "loss": 0.836,
+      "step": 7125
+    },
+    {
+      "epoch": 0.7457875457875458,
+      "grad_norm": 2.1476725693471126,
+      "learning_rate": 3.2024151087246704e-06,
+      "loss": 0.9005,
+      "step": 7126
+    },
+    {
+      "epoch": 0.7458922030350602,
+      "grad_norm": 2.3373997568981313,
+      "learning_rate": 3.199929356427628e-06,
+      "loss": 0.8782,
+      "step": 7127
+    },
+    {
+      "epoch": 0.7459968602825746,
+      "grad_norm": 2.0228477518559522,
+      "learning_rate": 3.197444385472489e-06,
+      "loss": 0.8036,
+      "step": 7128
+    },
+    {
+      "epoch": 0.7461015175300889,
+      "grad_norm": 1.728215649932167,
+      "learning_rate": 3.1949601961447795e-06,
+      "loss": 0.7443,
+      "step": 7129
+    },
+    {
+      "epoch": 0.7462061747776033,
+      "grad_norm": 2.1796559194931606,
+      "learning_rate": 3.1924767887299357e-06,
+      "loss": 0.8856,
+      "step": 7130
+    },
+    {
+      "epoch": 0.7463108320251177,
+      "grad_norm": 2.171160116743099,
+      "learning_rate": 3.1899941635133092e-06,
+      "loss": 0.8125,
+      "step": 7131
+    },
+    {
+      "epoch": 0.7464154892726321,
+      "grad_norm": 1.688400395699494,
+      "learning_rate": 3.1875123207801616e-06,
+      "loss": 0.7771,
+      "step": 7132
+    },
+    {
+      "epoch": 0.7465201465201465,
+      "grad_norm": 2.6231117402138593,
+      "learning_rate": 3.1850312608156596e-06,
+      "loss": 0.9425,
+      "step": 7133
+    },
+    {
+      "epoch": 0.7466248037676609,
+      "grad_norm": 1.8438824122826507,
+      "learning_rate": 3.1825509839048806e-06,
+      "loss": 0.788,
+      "step": 7134
+    },
+    {
+      "epoch": 0.7467294610151753,
+      "grad_norm": 2.2251217003891828,
+      "learning_rate": 3.1800714903328102e-06,
+      "loss": 0.9457,
+      "step": 7135
+    },
+    {
+      "epoch": 0.7468341182626896,
+      "grad_norm": 2.1306082735607808,
+      "learning_rate": 3.1775927803843543e-06,
+      "loss": 0.9576,
+      "step": 7136
+    },
+    {
+      "epoch": 0.746938775510204,
+      "grad_norm": 2.1790105073105757,
+      "learning_rate": 3.1751148543443187e-06,
+      "loss": 0.8739,
+      "step": 7137
+    },
+    {
+      "epoch": 0.7470434327577185,
+      "grad_norm": 2.239558304766855,
+      "learning_rate": 3.1726377124974217e-06,
+      "loss": 0.8207,
+      "step": 7138
+    },
+    {
+      "epoch": 0.7471480900052329,
+      "grad_norm": 2.008095181346031,
+      "learning_rate": 3.1701613551282893e-06,
+      "loss": 0.867,
+      "step": 7139
+    },
+    {
+      "epoch": 0.7472527472527473,
+      "grad_norm": 2.3260833446259777,
+      "learning_rate": 3.1676857825214623e-06,
+      "loss": 0.8897,
+      "step": 7140
+    },
+    {
+      "epoch": 0.7473574045002617,
+      "grad_norm": 2.070498373715419,
+      "learning_rate": 3.165210994961393e-06,
+      "loss": 0.8824,
+      "step": 7141
+    },
+    {
+      "epoch": 0.7474620617477761,
+      "grad_norm": 2.0573703279887807,
+      "learning_rate": 3.162736992732436e-06,
+      "loss": 0.8889,
+      "step": 7142
+    },
+    {
+      "epoch": 0.7475667189952904,
+      "grad_norm": 2.3464698580212304,
+      "learning_rate": 3.160263776118858e-06,
+      "loss": 0.9127,
+      "step": 7143
+    },
+    {
+      "epoch": 0.7476713762428048,
+      "grad_norm": 2.314229707278681,
+      "learning_rate": 3.1577913454048393e-06,
+      "loss": 0.9432,
+      "step": 7144
+    },
+    {
+      "epoch": 0.7477760334903192,
+      "grad_norm": 2.2903886734227856,
+      "learning_rate": 3.1553197008744607e-06,
+      "loss": 0.9033,
+      "step": 7145
+    },
+    {
+      "epoch": 0.7478806907378336,
+      "grad_norm": 2.26707032536701,
+      "learning_rate": 3.1528488428117287e-06,
+      "loss": 0.8826,
+      "step": 7146
+    },
+    {
+      "epoch": 0.747985347985348,
+      "grad_norm": 2.201793259018066,
+      "learning_rate": 3.150378771500542e-06,
+      "loss": 0.8848,
+      "step": 7147
+    },
+    {
+      "epoch": 0.7480900052328624,
+      "grad_norm": 2.153815935771742,
+      "learning_rate": 3.147909487224723e-06,
+      "loss": 1.0046,
+      "step": 7148
+    },
+    {
+      "epoch": 0.7481946624803768,
+      "grad_norm": 2.23656638946847,
+      "learning_rate": 3.145440990267994e-06,
+      "loss": 0.8385,
+      "step": 7149
+    },
+    {
+      "epoch": 0.7482993197278912,
+      "grad_norm": 1.9518763135706487,
+      "learning_rate": 3.142973280913988e-06,
+      "loss": 0.8302,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7484039769754055,
+      "grad_norm": 1.9560563182373518,
+      "learning_rate": 3.140506359446256e-06,
+      "loss": 0.8629,
+      "step": 7151
+    },
+    {
+      "epoch": 0.7485086342229199,
+      "grad_norm": 1.827294647023681,
+      "learning_rate": 3.138040226148249e-06,
+      "loss": 0.8265,
+      "step": 7152
+    },
+    {
+      "epoch": 0.7486132914704343,
+      "grad_norm": 2.1222980400949245,
+      "learning_rate": 3.1355748813033305e-06,
+      "loss": 0.8273,
+      "step": 7153
+    },
+    {
+      "epoch": 0.7487179487179487,
+      "grad_norm": 2.1538492848488855,
+      "learning_rate": 3.1331103251947703e-06,
+      "loss": 0.8809,
+      "step": 7154
+    },
+    {
+      "epoch": 0.7488226059654631,
+      "grad_norm": 1.9153204561792283,
+      "learning_rate": 3.130646558105758e-06,
+      "loss": 0.8623,
+      "step": 7155
+    },
+    {
+      "epoch": 0.7489272632129775,
+      "grad_norm": 2.146577046698544,
+      "learning_rate": 3.128183580319378e-06,
+      "loss": 0.9102,
+      "step": 7156
+    },
+    {
+      "epoch": 0.7490319204604919,
+      "grad_norm": 2.2253981483892598,
+      "learning_rate": 3.125721392118639e-06,
+      "loss": 0.8395,
+      "step": 7157
+    },
+    {
+      "epoch": 0.7491365777080062,
+      "grad_norm": 2.252424457310398,
+      "learning_rate": 3.1232599937864483e-06,
+      "loss": 1.0134,
+      "step": 7158
+    },
+    {
+      "epoch": 0.7492412349555206,
+      "grad_norm": 2.160152684893339,
+      "learning_rate": 3.1207993856056205e-06,
+      "loss": 0.9069,
+      "step": 7159
+    },
+    {
+      "epoch": 0.749345892203035,
+      "grad_norm": 2.163254492760139,
+      "learning_rate": 3.118339567858892e-06,
+      "loss": 0.8737,
+      "step": 7160
+    },
+    {
+      "epoch": 0.7494505494505495,
+      "grad_norm": 1.6978836150995644,
+      "learning_rate": 3.1158805408288995e-06,
+      "loss": 0.8209,
+      "step": 7161
+    },
+    {
+      "epoch": 0.7495552066980639,
+      "grad_norm": 1.9948469938453408,
+      "learning_rate": 3.1134223047981872e-06,
+      "loss": 0.947,
+      "step": 7162
+    },
+    {
+      "epoch": 0.7496598639455783,
+      "grad_norm": 2.261922765416398,
+      "learning_rate": 3.1109648600492126e-06,
+      "loss": 0.8666,
+      "step": 7163
+    },
+    {
+      "epoch": 0.7497645211930927,
+      "grad_norm": 2.004553104391324,
+      "learning_rate": 3.108508206864338e-06,
+      "loss": 0.8072,
+      "step": 7164
+    },
+    {
+      "epoch": 0.749869178440607,
+      "grad_norm": 1.9717258623857916,
+      "learning_rate": 3.1060523455258406e-06,
+      "loss": 0.8721,
+      "step": 7165
+    },
+    {
+      "epoch": 0.7499738356881214,
+      "grad_norm": 2.1108494565639857,
+      "learning_rate": 3.1035972763159074e-06,
+      "loss": 0.8608,
+      "step": 7166
+    },
+    {
+      "epoch": 0.7500784929356358,
+      "grad_norm": 2.0578079658481423,
+      "learning_rate": 3.1011429995166288e-06,
+      "loss": 0.8963,
+      "step": 7167
+    },
+    {
+      "epoch": 0.7501831501831502,
+      "grad_norm": 1.983828357087619,
+      "learning_rate": 3.098689515410004e-06,
+      "loss": 0.899,
+      "step": 7168
+    },
+    {
+      "epoch": 0.7502878074306646,
+      "grad_norm": 2.2998223358846333,
+      "learning_rate": 3.0962368242779406e-06,
+      "loss": 0.8537,
+      "step": 7169
+    },
+    {
+      "epoch": 0.750392464678179,
+      "grad_norm": 1.8820805197314077,
+      "learning_rate": 3.0937849264022658e-06,
+      "loss": 0.8407,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7504971219256934,
+      "grad_norm": 2.177356874448031,
+      "learning_rate": 3.0913338220647028e-06,
+      "loss": 0.9178,
+      "step": 7171
+    },
+    {
+      "epoch": 0.7506017791732077,
+      "grad_norm": 2.198171422534131,
+      "learning_rate": 3.0888835115468883e-06,
+      "loss": 0.9125,
+      "step": 7172
+    },
+    {
+      "epoch": 0.7507064364207221,
+      "grad_norm": 2.031436579163277,
+      "learning_rate": 3.086433995130367e-06,
+      "loss": 0.8067,
+      "step": 7173
+    },
+    {
+      "epoch": 0.7508110936682365,
+      "grad_norm": 1.8475737903440372,
+      "learning_rate": 3.0839852730965934e-06,
+      "loss": 0.8454,
+      "step": 7174
+    },
+    {
+      "epoch": 0.7509157509157509,
+      "grad_norm": 2.25678437649617,
+      "learning_rate": 3.081537345726936e-06,
+      "loss": 0.839,
+      "step": 7175
+    },
+    {
+      "epoch": 0.7510204081632653,
+      "grad_norm": 1.9401602491295404,
+      "learning_rate": 3.0790902133026625e-06,
+      "loss": 0.8353,
+      "step": 7176
+    },
+    {
+      "epoch": 0.7511250654107797,
+      "grad_norm": 2.0825066629501694,
+      "learning_rate": 3.0766438761049544e-06,
+      "loss": 0.8053,
+      "step": 7177
+    },
+    {
+      "epoch": 0.7512297226582941,
+      "grad_norm": 2.451954510004089,
+      "learning_rate": 3.074198334414896e-06,
+      "loss": 0.851,
+      "step": 7178
+    },
+    {
+      "epoch": 0.7513343799058084,
+      "grad_norm": 2.048655055859803,
+      "learning_rate": 3.071753588513493e-06,
+      "loss": 0.8167,
+      "step": 7179
+    },
+    {
+      "epoch": 0.7514390371533228,
+      "grad_norm": 2.177508435886103,
+      "learning_rate": 3.069309638681647e-06,
+      "loss": 0.9355,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7515436944008372,
+      "grad_norm": 2.144395053127334,
+      "learning_rate": 3.066866485200174e-06,
+      "loss": 1.0297,
+      "step": 7181
+    },
+    {
+      "epoch": 0.7516483516483516,
+      "grad_norm": 1.8941507496621728,
+      "learning_rate": 3.0644241283497934e-06,
+      "loss": 0.7487,
+      "step": 7182
+    },
+    {
+      "epoch": 0.751753008895866,
+      "grad_norm": 2.03238588774476,
+      "learning_rate": 3.0619825684111425e-06,
+      "loss": 0.9322,
+      "step": 7183
+    },
+    {
+      "epoch": 0.7518576661433805,
+      "grad_norm": 2.232382445814665,
+      "learning_rate": 3.0595418056647574e-06,
+      "loss": 1.0348,
+      "step": 7184
+    },
+    {
+      "epoch": 0.7519623233908949,
+      "grad_norm": 2.027939873344001,
+      "learning_rate": 3.0571018403910914e-06,
+      "loss": 0.8486,
+      "step": 7185
+    },
+    {
+      "epoch": 0.7520669806384092,
+      "grad_norm": 1.7087518006551394,
+      "learning_rate": 3.0546626728704986e-06,
+      "loss": 0.7637,
+      "step": 7186
+    },
+    {
+      "epoch": 0.7521716378859236,
+      "grad_norm": 2.090097485492809,
+      "learning_rate": 3.0522243033832455e-06,
+      "loss": 0.8497,
+      "step": 7187
+    },
+    {
+      "epoch": 0.752276295133438,
+      "grad_norm": 1.9895769202770441,
+      "learning_rate": 3.0497867322094998e-06,
+      "loss": 0.8822,
+      "step": 7188
+    },
+    {
+      "epoch": 0.7523809523809524,
+      "grad_norm": 2.40584150601798,
+      "learning_rate": 3.047349959629352e-06,
+      "loss": 0.9995,
+      "step": 7189
+    },
+    {
+      "epoch": 0.7524856096284668,
+      "grad_norm": 2.0368993311848977,
+      "learning_rate": 3.0449139859227883e-06,
+      "loss": 0.8247,
+      "step": 7190
+    },
+    {
+      "epoch": 0.7525902668759812,
+      "grad_norm": 2.056257702126359,
+      "learning_rate": 3.0424788113697036e-06,
+      "loss": 0.9778,
+      "step": 7191
+    },
+    {
+      "epoch": 0.7526949241234956,
+      "grad_norm": 1.9693791719687486,
+      "learning_rate": 3.0400444362499115e-06,
+      "loss": 0.9017,
+      "step": 7192
+    },
+    {
+      "epoch": 0.75279958137101,
+      "grad_norm": 2.0591702890964654,
+      "learning_rate": 3.0376108608431188e-06,
+      "loss": 0.9149,
+      "step": 7193
+    },
+    {
+      "epoch": 0.7529042386185243,
+      "grad_norm": 2.0700766255920944,
+      "learning_rate": 3.035178085428957e-06,
+      "loss": 0.8404,
+      "step": 7194
+    },
+    {
+      "epoch": 0.7530088958660387,
+      "grad_norm": 1.8951621768222429,
+      "learning_rate": 3.0327461102869514e-06,
+      "loss": 0.8727,
+      "step": 7195
+    },
+    {
+      "epoch": 0.7531135531135531,
+      "grad_norm": 1.760732012119775,
+      "learning_rate": 3.0303149356965424e-06,
+      "loss": 0.7707,
+      "step": 7196
+    },
+    {
+      "epoch": 0.7532182103610675,
+      "grad_norm": 2.0503813015447903,
+      "learning_rate": 3.027884561937072e-06,
+      "loss": 0.9688,
+      "step": 7197
+    },
+    {
+      "epoch": 0.7533228676085819,
+      "grad_norm": 2.0021170411447766,
+      "learning_rate": 3.0254549892878038e-06,
+      "loss": 0.922,
+      "step": 7198
+    },
+    {
+      "epoch": 0.7534275248560963,
+      "grad_norm": 1.8860849552256713,
+      "learning_rate": 3.0230262180278925e-06,
+      "loss": 0.9059,
+      "step": 7199
+    },
+    {
+      "epoch": 0.7535321821036107,
+      "grad_norm": 2.1818774407352404,
+      "learning_rate": 3.020598248436415e-06,
+      "loss": 0.9216,
+      "step": 7200
+    },
+    {
+      "epoch": 0.753636839351125,
+      "grad_norm": 1.8584792324138004,
+      "learning_rate": 3.0181710807923492e-06,
+      "loss": 0.7917,
+      "step": 7201
+    },
+    {
+      "epoch": 0.7537414965986394,
+      "grad_norm": 1.819519299711091,
+      "learning_rate": 3.015744715374579e-06,
+      "loss": 0.8911,
+      "step": 7202
+    },
+    {
+      "epoch": 0.7538461538461538,
+      "grad_norm": 2.164007510379012,
+      "learning_rate": 3.0133191524618956e-06,
+      "loss": 0.9111,
+      "step": 7203
+    },
+    {
+      "epoch": 0.7539508110936682,
+      "grad_norm": 1.8081808504852512,
+      "learning_rate": 3.010894392333009e-06,
+      "loss": 1.0013,
+      "step": 7204
+    },
+    {
+      "epoch": 0.7540554683411826,
+      "grad_norm": 1.9661369597242528,
+      "learning_rate": 3.008470435266525e-06,
+      "loss": 0.9094,
+      "step": 7205
+    },
+    {
+      "epoch": 0.754160125588697,
+      "grad_norm": 2.1709315190330494,
+      "learning_rate": 3.0060472815409614e-06,
+      "loss": 0.856,
+      "step": 7206
+    },
+    {
+      "epoch": 0.7542647828362115,
+      "grad_norm": 2.077801523128462,
+      "learning_rate": 3.00362493143474e-06,
+      "loss": 0.8178,
+      "step": 7207
+    },
+    {
+      "epoch": 0.7543694400837258,
+      "grad_norm": 2.0343924494049737,
+      "learning_rate": 3.001203385226198e-06,
+      "loss": 0.9672,
+      "step": 7208
+    },
+    {
+      "epoch": 0.7544740973312402,
+      "grad_norm": 2.612125199657769,
+      "learning_rate": 2.9987826431935773e-06,
+      "loss": 0.7652,
+      "step": 7209
+    },
+    {
+      "epoch": 0.7545787545787546,
+      "grad_norm": 1.9881607065905555,
+      "learning_rate": 2.9963627056150234e-06,
+      "loss": 0.8869,
+      "step": 7210
+    },
+    {
+      "epoch": 0.754683411826269,
+      "grad_norm": 2.2653350892421775,
+      "learning_rate": 2.993943572768594e-06,
+      "loss": 1.0044,
+      "step": 7211
+    },
+    {
+      "epoch": 0.7547880690737834,
+      "grad_norm": 1.9442254836980624,
+      "learning_rate": 2.9915252449322463e-06,
+      "loss": 0.8595,
+      "step": 7212
+    },
+    {
+      "epoch": 0.7548927263212978,
+      "grad_norm": 1.7828004495021605,
+      "learning_rate": 2.9891077223838594e-06,
+      "loss": 0.7383,
+      "step": 7213
+    },
+    {
+      "epoch": 0.7549973835688122,
+      "grad_norm": 2.1375811309701143,
+      "learning_rate": 2.9866910054012078e-06,
+      "loss": 0.8917,
+      "step": 7214
+    },
+    {
+      "epoch": 0.7551020408163265,
+      "grad_norm": 1.8434137581998618,
+      "learning_rate": 2.9842750942619767e-06,
+      "loss": 0.8375,
+      "step": 7215
+    },
+    {
+      "epoch": 0.7552066980638409,
+      "grad_norm": 1.8544288019845543,
+      "learning_rate": 2.9818599892437572e-06,
+      "loss": 0.8359,
+      "step": 7216
+    },
+    {
+      "epoch": 0.7553113553113553,
+      "grad_norm": 2.190544453353631,
+      "learning_rate": 2.979445690624051e-06,
+      "loss": 0.9268,
+      "step": 7217
+    },
+    {
+      "epoch": 0.7554160125588697,
+      "grad_norm": 1.6724087178652072,
+      "learning_rate": 2.977032198680272e-06,
+      "loss": 0.788,
+      "step": 7218
+    },
+    {
+      "epoch": 0.7555206698063841,
+      "grad_norm": 2.444110327040675,
+      "learning_rate": 2.9746195136897294e-06,
+      "loss": 1.0125,
+      "step": 7219
+    },
+    {
+      "epoch": 0.7556253270538985,
+      "grad_norm": 2.395682775915913,
+      "learning_rate": 2.9722076359296457e-06,
+      "loss": 0.9869,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7557299843014129,
+      "grad_norm": 2.1788752080998837,
+      "learning_rate": 2.9697965656771534e-06,
+      "loss": 0.8759,
+      "step": 7221
+    },
+    {
+      "epoch": 0.7558346415489272,
+      "grad_norm": 2.3785762750866484,
+      "learning_rate": 2.967386303209283e-06,
+      "loss": 0.9297,
+      "step": 7222
+    },
+    {
+      "epoch": 0.7559392987964416,
+      "grad_norm": 1.9232465597929709,
+      "learning_rate": 2.9649768488029862e-06,
+      "loss": 0.867,
+      "step": 7223
+    },
+    {
+      "epoch": 0.756043956043956,
+      "grad_norm": 1.9326752300709829,
+      "learning_rate": 2.962568202735111e-06,
+      "loss": 0.8029,
+      "step": 7224
+    },
+    {
+      "epoch": 0.7561486132914704,
+      "grad_norm": 2.127211853559956,
+      "learning_rate": 2.960160365282413e-06,
+      "loss": 0.855,
+      "step": 7225
+    },
+    {
+      "epoch": 0.7562532705389848,
+      "grad_norm": 2.268097028045051,
+      "learning_rate": 2.957753336721563e-06,
+      "loss": 0.9204,
+      "step": 7226
+    },
+    {
+      "epoch": 0.7563579277864992,
+      "grad_norm": 2.0703952112572157,
+      "learning_rate": 2.955347117329127e-06,
+      "loss": 0.9523,
+      "step": 7227
+    },
+    {
+      "epoch": 0.7564625850340136,
+      "grad_norm": 2.6447867664491347,
+      "learning_rate": 2.9529417073815925e-06,
+      "loss": 0.8857,
+      "step": 7228
+    },
+    {
+      "epoch": 0.7565672422815279,
+      "grad_norm": 2.5816773745873927,
+      "learning_rate": 2.950537107155341e-06,
+      "loss": 0.9542,
+      "step": 7229
+    },
+    {
+      "epoch": 0.7566718995290423,
+      "grad_norm": 2.16016794368315,
+      "learning_rate": 2.9481333169266667e-06,
+      "loss": 0.9248,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7567765567765568,
+      "grad_norm": 2.0925825053657405,
+      "learning_rate": 2.945730336971767e-06,
+      "loss": 0.8558,
+      "step": 7231
+    },
+    {
+      "epoch": 0.7568812140240712,
+      "grad_norm": 2.0853760218196413,
+      "learning_rate": 2.9433281675667545e-06,
+      "loss": 0.9312,
+      "step": 7232
+    },
+    {
+      "epoch": 0.7569858712715856,
+      "grad_norm": 2.2418754596740755,
+      "learning_rate": 2.940926808987642e-06,
+      "loss": 0.7494,
+      "step": 7233
+    },
+    {
+      "epoch": 0.7570905285191,
+      "grad_norm": 1.958769862807278,
+      "learning_rate": 2.938526261510346e-06,
+      "loss": 0.9602,
+      "step": 7234
+    },
+    {
+      "epoch": 0.7571951857666144,
+      "grad_norm": 1.9315836818382839,
+      "learning_rate": 2.9361265254106997e-06,
+      "loss": 0.9757,
+      "step": 7235
+    },
+    {
+      "epoch": 0.7572998430141288,
+      "grad_norm": 2.2173927927463297,
+      "learning_rate": 2.933727600964433e-06,
+      "loss": 0.8727,
+      "step": 7236
+    },
+    {
+      "epoch": 0.7574045002616431,
+      "grad_norm": 1.7683552818088055,
+      "learning_rate": 2.9313294884471945e-06,
+      "loss": 0.7764,
+      "step": 7237
+    },
+    {
+      "epoch": 0.7575091575091575,
+      "grad_norm": 2.042604009056469,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.8552,
+      "step": 7238
+    },
+    {
+      "epoch": 0.7576138147566719,
+      "grad_norm": 1.963283487176052,
+      "learning_rate": 2.9265357003018836e-06,
+      "loss": 0.8042,
+      "step": 7239
+    },
+    {
+      "epoch": 0.7577184720041863,
+      "grad_norm": 2.181411025825987,
+      "learning_rate": 2.924140025224629e-06,
+      "loss": 0.9476,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7578231292517007,
+      "grad_norm": 1.7246146896003263,
+      "learning_rate": 2.921745163178026e-06,
+      "loss": 0.7424,
+      "step": 7241
+    },
+    {
+      "epoch": 0.7579277864992151,
+      "grad_norm": 1.679941753555536,
+      "learning_rate": 2.9193511144372565e-06,
+      "loss": 0.8579,
+      "step": 7242
+    },
+    {
+      "epoch": 0.7580324437467295,
+      "grad_norm": 2.166472002968409,
+      "learning_rate": 2.9169578792773944e-06,
+      "loss": 0.9046,
+      "step": 7243
+    },
+    {
+      "epoch": 0.7581371009942438,
+      "grad_norm": 1.6957884419729579,
+      "learning_rate": 2.9145654579734352e-06,
+      "loss": 0.7482,
+      "step": 7244
+    },
+    {
+      "epoch": 0.7582417582417582,
+      "grad_norm": 2.2702186393221044,
+      "learning_rate": 2.9121738508002675e-06,
+      "loss": 0.8589,
+      "step": 7245
+    },
+    {
+      "epoch": 0.7583464154892726,
+      "grad_norm": 2.1812945897899727,
+      "learning_rate": 2.90978305803269e-06,
+      "loss": 0.7682,
+      "step": 7246
+    },
+    {
+      "epoch": 0.758451072736787,
+      "grad_norm": 1.7927929123690813,
+      "learning_rate": 2.9073930799454153e-06,
+      "loss": 0.7735,
+      "step": 7247
+    },
+    {
+      "epoch": 0.7585557299843014,
+      "grad_norm": 2.353877525391729,
+      "learning_rate": 2.9050039168130537e-06,
+      "loss": 0.9638,
+      "step": 7248
+    },
+    {
+      "epoch": 0.7586603872318158,
+      "grad_norm": 2.653249641565394,
+      "learning_rate": 2.9026155689101256e-06,
+      "loss": 0.8886,
+      "step": 7249
+    },
+    {
+      "epoch": 0.7587650444793302,
+      "grad_norm": 2.343195015332429,
+      "learning_rate": 2.9002280365110534e-06,
+      "loss": 0.8646,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7588697017268445,
+      "grad_norm": 7.944090802217359,
+      "learning_rate": 2.8978413198901754e-06,
+      "loss": 0.7561,
+      "step": 7251
+    },
+    {
+      "epoch": 0.7589743589743589,
+      "grad_norm": 1.7220024064319135,
+      "learning_rate": 2.8954554193217254e-06,
+      "loss": 0.8006,
+      "step": 7252
+    },
+    {
+      "epoch": 0.7590790162218733,
+      "grad_norm": 2.230587821004055,
+      "learning_rate": 2.893070335079852e-06,
+      "loss": 0.8161,
+      "step": 7253
+    },
+    {
+      "epoch": 0.7591836734693878,
+      "grad_norm": 2.278411767642871,
+      "learning_rate": 2.890686067438605e-06,
+      "loss": 0.8948,
+      "step": 7254
+    },
+    {
+      "epoch": 0.7592883307169022,
+      "grad_norm": 2.3002079292626387,
+      "learning_rate": 2.8883026166719374e-06,
+      "loss": 1.0273,
+      "step": 7255
+    },
+    {
+      "epoch": 0.7593929879644166,
+      "grad_norm": 2.3979608311717864,
+      "learning_rate": 2.8859199830537188e-06,
+      "loss": 0.8825,
+      "step": 7256
+    },
+    {
+      "epoch": 0.759497645211931,
+      "grad_norm": 1.9704760095316236,
+      "learning_rate": 2.8835381668577167e-06,
+      "loss": 0.9408,
+      "step": 7257
+    },
+    {
+      "epoch": 0.7596023024594453,
+      "grad_norm": 1.9634553592169977,
+      "learning_rate": 2.8811571683576047e-06,
+      "loss": 0.8909,
+      "step": 7258
+    },
+    {
+      "epoch": 0.7597069597069597,
+      "grad_norm": 2.0676146544534997,
+      "learning_rate": 2.8787769878269667e-06,
+      "loss": 0.8999,
+      "step": 7259
+    },
+    {
+      "epoch": 0.7598116169544741,
+      "grad_norm": 1.902506916202938,
+      "learning_rate": 2.8763976255392854e-06,
+      "loss": 0.7752,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7599162742019885,
+      "grad_norm": 1.784936461204701,
+      "learning_rate": 2.874019081767958e-06,
+      "loss": 0.9145,
+      "step": 7261
+    },
+    {
+      "epoch": 0.7600209314495029,
+      "grad_norm": 1.9199931698650219,
+      "learning_rate": 2.8716413567862865e-06,
+      "loss": 0.8145,
+      "step": 7262
+    },
+    {
+      "epoch": 0.7601255886970173,
+      "grad_norm": 2.053832292499548,
+      "learning_rate": 2.869264450867475e-06,
+      "loss": 0.8412,
+      "step": 7263
+    },
+    {
+      "epoch": 0.7602302459445317,
+      "grad_norm": 1.8965072652709092,
+      "learning_rate": 2.8668883642846325e-06,
+      "loss": 0.8407,
+      "step": 7264
+    },
+    {
+      "epoch": 0.760334903192046,
+      "grad_norm": 2.121023584744659,
+      "learning_rate": 2.864513097310775e-06,
+      "loss": 0.9228,
+      "step": 7265
+    },
+    {
+      "epoch": 0.7604395604395604,
+      "grad_norm": 2.2573579326522,
+      "learning_rate": 2.8621386502188296e-06,
+      "loss": 0.8228,
+      "step": 7266
+    },
+    {
+      "epoch": 0.7605442176870748,
+      "grad_norm": 2.0668406704888103,
+      "learning_rate": 2.8597650232816245e-06,
+      "loss": 0.8929,
+      "step": 7267
+    },
+    {
+      "epoch": 0.7606488749345892,
+      "grad_norm": 2.249728883019577,
+      "learning_rate": 2.8573922167718927e-06,
+      "loss": 0.8714,
+      "step": 7268
+    },
+    {
+      "epoch": 0.7607535321821036,
+      "grad_norm": 2.140716219541326,
+      "learning_rate": 2.855020230962271e-06,
+      "loss": 0.9412,
+      "step": 7269
+    },
+    {
+      "epoch": 0.760858189429618,
+      "grad_norm": 2.313215833515219,
+      "learning_rate": 2.8526490661253104e-06,
+      "loss": 0.9789,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7609628466771324,
+      "grad_norm": 2.1190694336129585,
+      "learning_rate": 2.8502787225334636e-06,
+      "loss": 0.8946,
+      "step": 7271
+    },
+    {
+      "epoch": 0.7610675039246467,
+      "grad_norm": 2.4208134953494795,
+      "learning_rate": 2.8479092004590857e-06,
+      "loss": 0.7714,
+      "step": 7272
+    },
+    {
+      "epoch": 0.7611721611721611,
+      "grad_norm": 2.3515481614362646,
+      "learning_rate": 2.8455405001744397e-06,
+      "loss": 0.7356,
+      "step": 7273
+    },
+    {
+      "epoch": 0.7612768184196755,
+      "grad_norm": 1.8723314249073932,
+      "learning_rate": 2.8431726219516896e-06,
+      "loss": 0.7776,
+      "step": 7274
+    },
+    {
+      "epoch": 0.7613814756671899,
+      "grad_norm": 1.7706806040241194,
+      "learning_rate": 2.8408055660629185e-06,
+      "loss": 0.8914,
+      "step": 7275
+    },
+    {
+      "epoch": 0.7614861329147044,
+      "grad_norm": 1.923104823220645,
+      "learning_rate": 2.8384393327800997e-06,
+      "loss": 0.8769,
+      "step": 7276
+    },
+    {
+      "epoch": 0.7615907901622188,
+      "grad_norm": 2.094266643745062,
+      "learning_rate": 2.83607392237512e-06,
+      "loss": 0.9438,
+      "step": 7277
+    },
+    {
+      "epoch": 0.7616954474097332,
+      "grad_norm": 1.7758737658741865,
+      "learning_rate": 2.8337093351197664e-06,
+      "loss": 0.8055,
+      "step": 7278
+    },
+    {
+      "epoch": 0.7618001046572476,
+      "grad_norm": 2.2135817220161376,
+      "learning_rate": 2.8313455712857408e-06,
+      "loss": 0.8821,
+      "step": 7279
+    },
+    {
+      "epoch": 0.7619047619047619,
+      "grad_norm": 1.933667595189823,
+      "learning_rate": 2.828982631144639e-06,
+      "loss": 0.8911,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7620094191522763,
+      "grad_norm": 2.5068411421201553,
+      "learning_rate": 2.8266205149679717e-06,
+      "loss": 1.022,
+      "step": 7281
+    },
+    {
+      "epoch": 0.7621140763997907,
+      "grad_norm": 2.109661522511011,
+      "learning_rate": 2.8242592230271506e-06,
+      "loss": 0.9163,
+      "step": 7282
+    },
+    {
+      "epoch": 0.7622187336473051,
+      "grad_norm": 2.2212961677842493,
+      "learning_rate": 2.821898755593491e-06,
+      "loss": 0.9055,
+      "step": 7283
+    },
+    {
+      "epoch": 0.7623233908948195,
+      "grad_norm": 2.4633904814759053,
+      "learning_rate": 2.819539112938212e-06,
+      "loss": 0.7539,
+      "step": 7284
+    },
+    {
+      "epoch": 0.7624280481423339,
+      "grad_norm": 1.8332947615386888,
+      "learning_rate": 2.817180295332449e-06,
+      "loss": 0.8608,
+      "step": 7285
+    },
+    {
+      "epoch": 0.7625327053898483,
+      "grad_norm": 2.252313473361386,
+      "learning_rate": 2.8148223030472287e-06,
+      "loss": 0.8452,
+      "step": 7286
+    },
+    {
+      "epoch": 0.7626373626373626,
+      "grad_norm": 2.430798136578984,
+      "learning_rate": 2.812465136353494e-06,
+      "loss": 0.9796,
+      "step": 7287
+    },
+    {
+      "epoch": 0.762742019884877,
+      "grad_norm": 2.145484648346231,
+      "learning_rate": 2.8101087955220864e-06,
+      "loss": 0.9313,
+      "step": 7288
+    },
+    {
+      "epoch": 0.7628466771323914,
+      "grad_norm": 2.30890945546967,
+      "learning_rate": 2.8077532808237497e-06,
+      "loss": 0.9836,
+      "step": 7289
+    },
+    {
+      "epoch": 0.7629513343799058,
+      "grad_norm": 2.4395138433728456,
+      "learning_rate": 2.805398592529145e-06,
+      "loss": 0.7632,
+      "step": 7290
+    },
+    {
+      "epoch": 0.7630559916274202,
+      "grad_norm": 2.1560151415099833,
+      "learning_rate": 2.803044730908826e-06,
+      "loss": 0.938,
+      "step": 7291
+    },
+    {
+      "epoch": 0.7631606488749346,
+      "grad_norm": 1.799807577137677,
+      "learning_rate": 2.8006916962332586e-06,
+      "loss": 0.749,
+      "step": 7292
+    },
+    {
+      "epoch": 0.763265306122449,
+      "grad_norm": 2.6437825820162133,
+      "learning_rate": 2.7983394887728054e-06,
+      "loss": 0.9678,
+      "step": 7293
+    },
+    {
+      "epoch": 0.7633699633699633,
+      "grad_norm": 2.0001658220701786,
+      "learning_rate": 2.795988108797748e-06,
+      "loss": 0.879,
+      "step": 7294
+    },
+    {
+      "epoch": 0.7634746206174777,
+      "grad_norm": 2.4220922582417654,
+      "learning_rate": 2.793637556578258e-06,
+      "loss": 0.8539,
+      "step": 7295
+    },
+    {
+      "epoch": 0.7635792778649921,
+      "grad_norm": 2.326513581121549,
+      "learning_rate": 2.791287832384424e-06,
+      "loss": 0.9231,
+      "step": 7296
+    },
+    {
+      "epoch": 0.7636839351125065,
+      "grad_norm": 2.0582333962241037,
+      "learning_rate": 2.788938936486232e-06,
+      "loss": 0.817,
+      "step": 7297
+    },
+    {
+      "epoch": 0.763788592360021,
+      "grad_norm": 1.964193500525283,
+      "learning_rate": 2.7865908691535746e-06,
+      "loss": 0.7875,
+      "step": 7298
+    },
+    {
+      "epoch": 0.7638932496075354,
+      "grad_norm": 1.9120684932559597,
+      "learning_rate": 2.784243630656247e-06,
+      "loss": 0.9067,
+      "step": 7299
+    },
+    {
+      "epoch": 0.7639979068550498,
+      "grad_norm": 2.203369880119307,
+      "learning_rate": 2.781897221263956e-06,
+      "loss": 0.9302,
+      "step": 7300
+    },
+    {
+      "epoch": 0.764102564102564,
+      "grad_norm": 1.8664131554131014,
+      "learning_rate": 2.7795516412463077e-06,
+      "loss": 0.8078,
+      "step": 7301
+    },
+    {
+      "epoch": 0.7642072213500785,
+      "grad_norm": 2.152251063216064,
+      "learning_rate": 2.7772068908728133e-06,
+      "loss": 0.8731,
+      "step": 7302
+    },
+    {
+      "epoch": 0.7643118785975929,
+      "grad_norm": 1.9704503123853456,
+      "learning_rate": 2.774862970412886e-06,
+      "loss": 0.9309,
+      "step": 7303
+    },
+    {
+      "epoch": 0.7644165358451073,
+      "grad_norm": 2.217027087214899,
+      "learning_rate": 2.77251988013585e-06,
+      "loss": 0.9636,
+      "step": 7304
+    },
+    {
+      "epoch": 0.7645211930926217,
+      "grad_norm": 2.4945137079215836,
+      "learning_rate": 2.7701776203109342e-06,
+      "loss": 0.9788,
+      "step": 7305
+    },
+    {
+      "epoch": 0.7646258503401361,
+      "grad_norm": 1.979000751747689,
+      "learning_rate": 2.767836191207267e-06,
+      "loss": 0.9692,
+      "step": 7306
+    },
+    {
+      "epoch": 0.7647305075876505,
+      "grad_norm": 1.9320377725750433,
+      "learning_rate": 2.765495593093882e-06,
+      "loss": 0.8792,
+      "step": 7307
+    },
+    {
+      "epoch": 0.7648351648351648,
+      "grad_norm": 1.661537117127484,
+      "learning_rate": 2.7631558262397164e-06,
+      "loss": 0.7901,
+      "step": 7308
+    },
+    {
+      "epoch": 0.7649398220826792,
+      "grad_norm": 2.066054899073813,
+      "learning_rate": 2.7608168909136203e-06,
+      "loss": 0.789,
+      "step": 7309
+    },
+    {
+      "epoch": 0.7650444793301936,
+      "grad_norm": 2.309887605904426,
+      "learning_rate": 2.7584787873843376e-06,
+      "loss": 0.9919,
+      "step": 7310
+    },
+    {
+      "epoch": 0.765149136577708,
+      "grad_norm": 2.219400510624547,
+      "learning_rate": 2.756141515920524e-06,
+      "loss": 0.8606,
+      "step": 7311
+    },
+    {
+      "epoch": 0.7652537938252224,
+      "grad_norm": 2.1735797860738977,
+      "learning_rate": 2.753805076790731e-06,
+      "loss": 0.854,
+      "step": 7312
+    },
+    {
+      "epoch": 0.7653584510727368,
+      "grad_norm": 2.132548975039601,
+      "learning_rate": 2.7514694702634237e-06,
+      "loss": 0.8229,
+      "step": 7313
+    },
+    {
+      "epoch": 0.7654631083202512,
+      "grad_norm": 2.2324206156113777,
+      "learning_rate": 2.7491346966069734e-06,
+      "loss": 0.8859,
+      "step": 7314
+    },
+    {
+      "epoch": 0.7655677655677655,
+      "grad_norm": 2.0237282392424114,
+      "learning_rate": 2.7468007560896435e-06,
+      "loss": 0.817,
+      "step": 7315
+    },
+    {
+      "epoch": 0.7656724228152799,
+      "grad_norm": 1.8891116758472233,
+      "learning_rate": 2.744467648979612e-06,
+      "loss": 0.8634,
+      "step": 7316
+    },
+    {
+      "epoch": 0.7657770800627943,
+      "grad_norm": 1.9422661279068079,
+      "learning_rate": 2.7421353755449552e-06,
+      "loss": 0.9092,
+      "step": 7317
+    },
+    {
+      "epoch": 0.7658817373103087,
+      "grad_norm": 2.2850690583866644,
+      "learning_rate": 2.7398039360536543e-06,
+      "loss": 0.8976,
+      "step": 7318
+    },
+    {
+      "epoch": 0.7659863945578231,
+      "grad_norm": 2.0211359219333467,
+      "learning_rate": 2.7374733307736025e-06,
+      "loss": 0.8664,
+      "step": 7319
+    },
+    {
+      "epoch": 0.7660910518053375,
+      "grad_norm": 1.8583554118941399,
+      "learning_rate": 2.7351435599725874e-06,
+      "loss": 0.8905,
+      "step": 7320
+    },
+    {
+      "epoch": 0.766195709052852,
+      "grad_norm": 1.9734674152504228,
+      "learning_rate": 2.7328146239183007e-06,
+      "loss": 0.8991,
+      "step": 7321
+    },
+    {
+      "epoch": 0.7663003663003664,
+      "grad_norm": 1.7218591246248722,
+      "learning_rate": 2.7304865228783507e-06,
+      "loss": 0.8628,
+      "step": 7322
+    },
+    {
+      "epoch": 0.7664050235478806,
+      "grad_norm": 1.7601071260258068,
+      "learning_rate": 2.7281592571202307e-06,
+      "loss": 0.8435,
+      "step": 7323
+    },
+    {
+      "epoch": 0.766509680795395,
+      "grad_norm": 2.5384380504587583,
+      "learning_rate": 2.725832826911359e-06,
+      "loss": 0.9552,
+      "step": 7324
+    },
+    {
+      "epoch": 0.7666143380429095,
+      "grad_norm": 1.9897641468927232,
+      "learning_rate": 2.7235072325190404e-06,
+      "loss": 0.8246,
+      "step": 7325
+    },
+    {
+      "epoch": 0.7667189952904239,
+      "grad_norm": 2.18530144040617,
+      "learning_rate": 2.721182474210492e-06,
+      "loss": 1.0168,
+      "step": 7326
+    },
+    {
+      "epoch": 0.7668236525379383,
+      "grad_norm": 1.9476014865880453,
+      "learning_rate": 2.71885855225283e-06,
+      "loss": 0.8297,
+      "step": 7327
+    },
+    {
+      "epoch": 0.7669283097854527,
+      "grad_norm": 2.0184306914039083,
+      "learning_rate": 2.716535466913084e-06,
+      "loss": 0.8521,
+      "step": 7328
+    },
+    {
+      "epoch": 0.7670329670329671,
+      "grad_norm": 2.1006322169829144,
+      "learning_rate": 2.714213218458178e-06,
+      "loss": 0.7977,
+      "step": 7329
+    },
+    {
+      "epoch": 0.7671376242804814,
+      "grad_norm": 2.0262143165867337,
+      "learning_rate": 2.7118918071549395e-06,
+      "loss": 0.8349,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7672422815279958,
+      "grad_norm": 2.269865382008233,
+      "learning_rate": 2.7095712332701108e-06,
+      "loss": 0.9142,
+      "step": 7331
+    },
+    {
+      "epoch": 0.7673469387755102,
+      "grad_norm": 2.3038111471769858,
+      "learning_rate": 2.7072514970703224e-06,
+      "loss": 0.9053,
+      "step": 7332
+    },
+    {
+      "epoch": 0.7674515960230246,
+      "grad_norm": 1.936577382811956,
+      "learning_rate": 2.7049325988221255e-06,
+      "loss": 0.8464,
+      "step": 7333
+    },
+    {
+      "epoch": 0.767556253270539,
+      "grad_norm": 2.5043378818478494,
+      "learning_rate": 2.70261453879196e-06,
+      "loss": 0.7622,
+      "step": 7334
+    },
+    {
+      "epoch": 0.7676609105180534,
+      "grad_norm": 2.0270141293438635,
+      "learning_rate": 2.7002973172461775e-06,
+      "loss": 0.8564,
+      "step": 7335
+    },
+    {
+      "epoch": 0.7677655677655678,
+      "grad_norm": 2.238914386289856,
+      "learning_rate": 2.6979809344510323e-06,
+      "loss": 0.949,
+      "step": 7336
+    },
+    {
+      "epoch": 0.7678702250130821,
+      "grad_norm": 2.112977680901526,
+      "learning_rate": 2.695665390672677e-06,
+      "loss": 0.8794,
+      "step": 7337
+    },
+    {
+      "epoch": 0.7679748822605965,
+      "grad_norm": 1.7578779088455374,
+      "learning_rate": 2.6933506861771764e-06,
+      "loss": 0.9656,
+      "step": 7338
+    },
+    {
+      "epoch": 0.7680795395081109,
+      "grad_norm": 2.025671558545498,
+      "learning_rate": 2.691036821230496e-06,
+      "loss": 0.8891,
+      "step": 7339
+    },
+    {
+      "epoch": 0.7681841967556253,
+      "grad_norm": 2.2330212723105913,
+      "learning_rate": 2.6887237960985024e-06,
+      "loss": 0.9527,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7682888540031397,
+      "grad_norm": 2.312102639631419,
+      "learning_rate": 2.6864116110469664e-06,
+      "loss": 0.9365,
+      "step": 7341
+    },
+    {
+      "epoch": 0.7683935112506541,
+      "grad_norm": 2.2905403470328847,
+      "learning_rate": 2.684100266341558e-06,
+      "loss": 0.9652,
+      "step": 7342
+    },
+    {
+      "epoch": 0.7684981684981685,
+      "grad_norm": 1.7444191920341985,
+      "learning_rate": 2.681789762247864e-06,
+      "loss": 0.7956,
+      "step": 7343
+    },
+    {
+      "epoch": 0.7686028257456828,
+      "grad_norm": 2.125810203110687,
+      "learning_rate": 2.6794800990313617e-06,
+      "loss": 0.8981,
+      "step": 7344
+    },
+    {
+      "epoch": 0.7687074829931972,
+      "grad_norm": 2.0640436363199584,
+      "learning_rate": 2.6771712769574364e-06,
+      "loss": 0.9515,
+      "step": 7345
+    },
+    {
+      "epoch": 0.7688121402407117,
+      "grad_norm": 2.2540317780667283,
+      "learning_rate": 2.6748632962913732e-06,
+      "loss": 0.9397,
+      "step": 7346
+    },
+    {
+      "epoch": 0.7689167974882261,
+      "grad_norm": 2.2939707078082434,
+      "learning_rate": 2.672556157298367e-06,
+      "loss": 0.873,
+      "step": 7347
+    },
+    {
+      "epoch": 0.7690214547357405,
+      "grad_norm": 2.0271839169120343,
+      "learning_rate": 2.6702498602435155e-06,
+      "loss": 0.9098,
+      "step": 7348
+    },
+    {
+      "epoch": 0.7691261119832549,
+      "grad_norm": 2.194903694850483,
+      "learning_rate": 2.6679444053918137e-06,
+      "loss": 0.7982,
+      "step": 7349
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 1.8945213031067587,
+      "learning_rate": 2.6656397930081635e-06,
+      "loss": 0.9307,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7693354264782836,
+      "grad_norm": 1.9063057362434497,
+      "learning_rate": 2.6633360233573656e-06,
+      "loss": 0.9288,
+      "step": 7351
+    },
+    {
+      "epoch": 0.769440083725798,
+      "grad_norm": 2.318201908671572,
+      "learning_rate": 2.6610330967041366e-06,
+      "loss": 0.9832,
+      "step": 7352
+    },
+    {
+      "epoch": 0.7695447409733124,
+      "grad_norm": 2.087584884442611,
+      "learning_rate": 2.6587310133130805e-06,
+      "loss": 0.8781,
+      "step": 7353
+    },
+    {
+      "epoch": 0.7696493982208268,
+      "grad_norm": 1.8841301045621537,
+      "learning_rate": 2.6564297734487144e-06,
+      "loss": 0.7486,
+      "step": 7354
+    },
+    {
+      "epoch": 0.7697540554683412,
+      "grad_norm": 2.048554104308666,
+      "learning_rate": 2.654129377375454e-06,
+      "loss": 0.7509,
+      "step": 7355
+    },
+    {
+      "epoch": 0.7698587127158556,
+      "grad_norm": 1.9448133038229751,
+      "learning_rate": 2.6518298253576167e-06,
+      "loss": 0.8034,
+      "step": 7356
+    },
+    {
+      "epoch": 0.76996336996337,
+      "grad_norm": 2.3334382875645465,
+      "learning_rate": 2.6495311176594286e-06,
+      "loss": 1.0057,
+      "step": 7357
+    },
+    {
+      "epoch": 0.7700680272108843,
+      "grad_norm": 2.0609258709754203,
+      "learning_rate": 2.6472332545450195e-06,
+      "loss": 0.9352,
+      "step": 7358
+    },
+    {
+      "epoch": 0.7701726844583987,
+      "grad_norm": 2.043836151919123,
+      "learning_rate": 2.6449362362784147e-06,
+      "loss": 0.8086,
+      "step": 7359
+    },
+    {
+      "epoch": 0.7702773417059131,
+      "grad_norm": 1.9222038165954007,
+      "learning_rate": 2.642640063123546e-06,
+      "loss": 0.8762,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7703819989534275,
+      "grad_norm": 2.272474659914367,
+      "learning_rate": 2.640344735344247e-06,
+      "loss": 0.887,
+      "step": 7361
+    },
+    {
+      "epoch": 0.7704866562009419,
+      "grad_norm": 2.0559426832557066,
+      "learning_rate": 2.6380502532042608e-06,
+      "loss": 0.8748,
+      "step": 7362
+    },
+    {
+      "epoch": 0.7705913134484563,
+      "grad_norm": 1.9381978766078287,
+      "learning_rate": 2.635756616967223e-06,
+      "loss": 0.8543,
+      "step": 7363
+    },
+    {
+      "epoch": 0.7706959706959707,
+      "grad_norm": 1.885340263704655,
+      "learning_rate": 2.6334638268966773e-06,
+      "loss": 0.8073,
+      "step": 7364
+    },
+    {
+      "epoch": 0.7708006279434851,
+      "grad_norm": 1.939561768160846,
+      "learning_rate": 2.631171883256074e-06,
+      "loss": 0.8754,
+      "step": 7365
+    },
+    {
+      "epoch": 0.7709052851909994,
+      "grad_norm": 1.7916271930244994,
+      "learning_rate": 2.628880786308756e-06,
+      "loss": 0.9365,
+      "step": 7366
+    },
+    {
+      "epoch": 0.7710099424385138,
+      "grad_norm": 1.9214226616582117,
+      "learning_rate": 2.626590536317982e-06,
+      "loss": 0.9126,
+      "step": 7367
+    },
+    {
+      "epoch": 0.7711145996860282,
+      "grad_norm": 1.6201683681441432,
+      "learning_rate": 2.6243011335469027e-06,
+      "loss": 0.7946,
+      "step": 7368
+    },
+    {
+      "epoch": 0.7712192569335427,
+      "grad_norm": 2.0135703962696954,
+      "learning_rate": 2.622012578258576e-06,
+      "loss": 0.9228,
+      "step": 7369
+    },
+    {
+      "epoch": 0.7713239141810571,
+      "grad_norm": 2.1109014783062747,
+      "learning_rate": 2.6197248707159575e-06,
+      "loss": 0.8835,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7714285714285715,
+      "grad_norm": 2.004035053535731,
+      "learning_rate": 2.6174380111819144e-06,
+      "loss": 0.8671,
+      "step": 7371
+    },
+    {
+      "epoch": 0.7715332286760859,
+      "grad_norm": 2.027444223995198,
+      "learning_rate": 2.615151999919211e-06,
+      "loss": 0.9357,
+      "step": 7372
+    },
+    {
+      "epoch": 0.7716378859236002,
+      "grad_norm": 1.9959569059145792,
+      "learning_rate": 2.61286683719051e-06,
+      "loss": 0.9009,
+      "step": 7373
+    },
+    {
+      "epoch": 0.7717425431711146,
+      "grad_norm": 1.9792281441894708,
+      "learning_rate": 2.610582523258388e-06,
+      "loss": 0.8614,
+      "step": 7374
+    },
+    {
+      "epoch": 0.771847200418629,
+      "grad_norm": 1.983946749775523,
+      "learning_rate": 2.608299058385314e-06,
+      "loss": 0.8806,
+      "step": 7375
+    },
+    {
+      "epoch": 0.7719518576661434,
+      "grad_norm": 1.9428206558603256,
+      "learning_rate": 2.60601644283366e-06,
+      "loss": 0.8755,
+      "step": 7376
+    },
+    {
+      "epoch": 0.7720565149136578,
+      "grad_norm": 2.0749303955604703,
+      "learning_rate": 2.603734676865708e-06,
+      "loss": 0.8801,
+      "step": 7377
+    },
+    {
+      "epoch": 0.7721611721611722,
+      "grad_norm": 1.9352525690019882,
+      "learning_rate": 2.6014537607436365e-06,
+      "loss": 0.9563,
+      "step": 7378
+    },
+    {
+      "epoch": 0.7722658294086866,
+      "grad_norm": 2.0793047689529844,
+      "learning_rate": 2.5991736947295254e-06,
+      "loss": 0.8347,
+      "step": 7379
+    },
+    {
+      "epoch": 0.7723704866562009,
+      "grad_norm": 2.293524394942549,
+      "learning_rate": 2.596894479085357e-06,
+      "loss": 0.8757,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7724751439037153,
+      "grad_norm": 1.751731557615987,
+      "learning_rate": 2.5946161140730243e-06,
+      "loss": 0.8043,
+      "step": 7381
+    },
+    {
+      "epoch": 0.7725798011512297,
+      "grad_norm": 2.25419896408031,
+      "learning_rate": 2.592338599954308e-06,
+      "loss": 0.7497,
+      "step": 7382
+    },
+    {
+      "epoch": 0.7726844583987441,
+      "grad_norm": 2.0024558583214342,
+      "learning_rate": 2.5900619369909074e-06,
+      "loss": 0.8527,
+      "step": 7383
+    },
+    {
+      "epoch": 0.7727891156462585,
+      "grad_norm": 2.0969280155721552,
+      "learning_rate": 2.5877861254444115e-06,
+      "loss": 0.7273,
+      "step": 7384
+    },
+    {
+      "epoch": 0.7728937728937729,
+      "grad_norm": 2.1227745715824717,
+      "learning_rate": 2.5855111655763134e-06,
+      "loss": 0.86,
+      "step": 7385
+    },
+    {
+      "epoch": 0.7729984301412873,
+      "grad_norm": 1.8610864043360293,
+      "learning_rate": 2.583237057648016e-06,
+      "loss": 0.8691,
+      "step": 7386
+    },
+    {
+      "epoch": 0.7731030873888016,
+      "grad_norm": 2.1425965300073884,
+      "learning_rate": 2.580963801920816e-06,
+      "loss": 0.8999,
+      "step": 7387
+    },
+    {
+      "epoch": 0.773207744636316,
+      "grad_norm": 2.042623189976076,
+      "learning_rate": 2.5786913986559147e-06,
+      "loss": 0.8993,
+      "step": 7388
+    },
+    {
+      "epoch": 0.7733124018838304,
+      "grad_norm": 1.9775430060536299,
+      "learning_rate": 2.5764198481144164e-06,
+      "loss": 0.9127,
+      "step": 7389
+    },
+    {
+      "epoch": 0.7734170591313448,
+      "grad_norm": 2.000333442979551,
+      "learning_rate": 2.5741491505573213e-06,
+      "loss": 0.9102,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7735217163788592,
+      "grad_norm": 1.9614405240639192,
+      "learning_rate": 2.5718793062455472e-06,
+      "loss": 0.8835,
+      "step": 7391
+    },
+    {
+      "epoch": 0.7736263736263737,
+      "grad_norm": 2.4448514916772806,
+      "learning_rate": 2.5696103154399e-06,
+      "loss": 0.9974,
+      "step": 7392
+    },
+    {
+      "epoch": 0.7737310308738881,
+      "grad_norm": 1.9778536256057575,
+      "learning_rate": 2.567342178401091e-06,
+      "loss": 0.8884,
+      "step": 7393
+    },
+    {
+      "epoch": 0.7738356881214024,
+      "grad_norm": 1.965676707722803,
+      "learning_rate": 2.565074895389733e-06,
+      "loss": 0.8692,
+      "step": 7394
+    },
+    {
+      "epoch": 0.7739403453689168,
+      "grad_norm": 2.1903706642179195,
+      "learning_rate": 2.562808466666338e-06,
+      "loss": 0.8845,
+      "step": 7395
+    },
+    {
+      "epoch": 0.7740450026164312,
+      "grad_norm": 1.9841184192672507,
+      "learning_rate": 2.5605428924913312e-06,
+      "loss": 0.8959,
+      "step": 7396
+    },
+    {
+      "epoch": 0.7741496598639456,
+      "grad_norm": 2.169747156193136,
+      "learning_rate": 2.558278173125026e-06,
+      "loss": 0.8318,
+      "step": 7397
+    },
+    {
+      "epoch": 0.77425431711146,
+      "grad_norm": 2.742205301494219,
+      "learning_rate": 2.5560143088276456e-06,
+      "loss": 0.9708,
+      "step": 7398
+    },
+    {
+      "epoch": 0.7743589743589744,
+      "grad_norm": 2.1736256054455145,
+      "learning_rate": 2.553751299859308e-06,
+      "loss": 0.8634,
+      "step": 7399
+    },
+    {
+      "epoch": 0.7744636316064888,
+      "grad_norm": 2.0298243788952792,
+      "learning_rate": 2.551489146480042e-06,
+      "loss": 0.98,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7745682888540032,
+      "grad_norm": 1.8171963930524047,
+      "learning_rate": 2.5492278489497745e-06,
+      "loss": 0.8868,
+      "step": 7401
+    },
+    {
+      "epoch": 0.7746729461015175,
+      "grad_norm": 1.8750562250635816,
+      "learning_rate": 2.5469674075283325e-06,
+      "loss": 0.8463,
+      "step": 7402
+    },
+    {
+      "epoch": 0.7747776033490319,
+      "grad_norm": 2.0170705567329934,
+      "learning_rate": 2.544707822475444e-06,
+      "loss": 0.8602,
+      "step": 7403
+    },
+    {
+      "epoch": 0.7748822605965463,
+      "grad_norm": 2.0889797065511657,
+      "learning_rate": 2.5424490940507373e-06,
+      "loss": 0.8067,
+      "step": 7404
+    },
+    {
+      "epoch": 0.7749869178440607,
+      "grad_norm": 2.084921965591833,
+      "learning_rate": 2.5401912225137504e-06,
+      "loss": 0.8538,
+      "step": 7405
+    },
+    {
+      "epoch": 0.7750915750915751,
+      "grad_norm": 1.999119792459058,
+      "learning_rate": 2.5379342081239157e-06,
+      "loss": 0.954,
+      "step": 7406
+    },
+    {
+      "epoch": 0.7751962323390895,
+      "grad_norm": 1.9723776486252687,
+      "learning_rate": 2.5356780511405666e-06,
+      "loss": 0.8641,
+      "step": 7407
+    },
+    {
+      "epoch": 0.7753008895866039,
+      "grad_norm": 2.07024245752328,
+      "learning_rate": 2.533422751822938e-06,
+      "loss": 0.8942,
+      "step": 7408
+    },
+    {
+      "epoch": 0.7754055468341182,
+      "grad_norm": 1.8764528707975676,
+      "learning_rate": 2.531168310430172e-06,
+      "loss": 0.9478,
+      "step": 7409
+    },
+    {
+      "epoch": 0.7755102040816326,
+      "grad_norm": 1.8645837015695488,
+      "learning_rate": 2.528914727221311e-06,
+      "loss": 0.8407,
+      "step": 7410
+    },
+    {
+      "epoch": 0.775614861329147,
+      "grad_norm": 2.172900648266963,
+      "learning_rate": 2.5266620024552937e-06,
+      "loss": 0.856,
+      "step": 7411
+    },
+    {
+      "epoch": 0.7757195185766614,
+      "grad_norm": 2.0753861210504425,
+      "learning_rate": 2.524410136390961e-06,
+      "loss": 0.9372,
+      "step": 7412
+    },
+    {
+      "epoch": 0.7758241758241758,
+      "grad_norm": 2.0279346463791588,
+      "learning_rate": 2.5221591292870595e-06,
+      "loss": 0.8748,
+      "step": 7413
+    },
+    {
+      "epoch": 0.7759288330716902,
+      "grad_norm": 1.9053518322080332,
+      "learning_rate": 2.5199089814022293e-06,
+      "loss": 0.846,
+      "step": 7414
+    },
+    {
+      "epoch": 0.7760334903192047,
+      "grad_norm": 1.951751195516781,
+      "learning_rate": 2.517659692995025e-06,
+      "loss": 0.7616,
+      "step": 7415
+    },
+    {
+      "epoch": 0.776138147566719,
+      "grad_norm": 2.258508279860391,
+      "learning_rate": 2.515411264323887e-06,
+      "loss": 0.9151,
+      "step": 7416
+    },
+    {
+      "epoch": 0.7762428048142334,
+      "grad_norm": 2.318867938802657,
+      "learning_rate": 2.5131636956471696e-06,
+      "loss": 0.8769,
+      "step": 7417
+    },
+    {
+      "epoch": 0.7763474620617478,
+      "grad_norm": 1.867705727278113,
+      "learning_rate": 2.510916987223122e-06,
+      "loss": 0.9135,
+      "step": 7418
+    },
+    {
+      "epoch": 0.7764521193092622,
+      "grad_norm": 2.100130971915618,
+      "learning_rate": 2.5086711393098906e-06,
+      "loss": 0.85,
+      "step": 7419
+    },
+    {
+      "epoch": 0.7765567765567766,
+      "grad_norm": 2.0579255004196852,
+      "learning_rate": 2.5064261521655355e-06,
+      "loss": 0.8838,
+      "step": 7420
+    },
+    {
+      "epoch": 0.776661433804291,
+      "grad_norm": 2.0359736419827033,
+      "learning_rate": 2.5041820260480063e-06,
+      "loss": 0.891,
+      "step": 7421
+    },
+    {
+      "epoch": 0.7767660910518054,
+      "grad_norm": 2.007394652518043,
+      "learning_rate": 2.501938761215158e-06,
+      "loss": 0.8018,
+      "step": 7422
+    },
+    {
+      "epoch": 0.7768707482993197,
+      "grad_norm": 2.309114884247499,
+      "learning_rate": 2.4996963579247433e-06,
+      "loss": 0.8789,
+      "step": 7423
+    },
+    {
+      "epoch": 0.7769754055468341,
+      "grad_norm": 2.1276331305118017,
+      "learning_rate": 2.497454816434425e-06,
+      "loss": 0.9842,
+      "step": 7424
+    },
+    {
+      "epoch": 0.7770800627943485,
+      "grad_norm": 2.0491657489397985,
+      "learning_rate": 2.4952141370017536e-06,
+      "loss": 0.8562,
+      "step": 7425
+    },
+    {
+      "epoch": 0.7771847200418629,
+      "grad_norm": 1.780804392760166,
+      "learning_rate": 2.492974319884196e-06,
+      "loss": 0.8353,
+      "step": 7426
+    },
+    {
+      "epoch": 0.7772893772893773,
+      "grad_norm": 2.829423926147118,
+      "learning_rate": 2.4907353653391062e-06,
+      "loss": 0.8086,
+      "step": 7427
+    },
+    {
+      "epoch": 0.7773940345368917,
+      "grad_norm": 2.1527688872034516,
+      "learning_rate": 2.4884972736237433e-06,
+      "loss": 0.8685,
+      "step": 7428
+    },
+    {
+      "epoch": 0.7774986917844061,
+      "grad_norm": 1.7756364808096,
+      "learning_rate": 2.4862600449952744e-06,
+      "loss": 0.8753,
+      "step": 7429
+    },
+    {
+      "epoch": 0.7776033490319204,
+      "grad_norm": 2.193742164532645,
+      "learning_rate": 2.4840236797107575e-06,
+      "loss": 0.7831,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7777080062794348,
+      "grad_norm": 2.415166469626955,
+      "learning_rate": 2.481788178027157e-06,
+      "loss": 0.8495,
+      "step": 7431
+    },
+    {
+      "epoch": 0.7778126635269492,
+      "grad_norm": 2.2290675901663652,
+      "learning_rate": 2.479553540201335e-06,
+      "loss": 0.9439,
+      "step": 7432
+    },
+    {
+      "epoch": 0.7779173207744636,
+      "grad_norm": 2.2829886391415255,
+      "learning_rate": 2.4773197664900538e-06,
+      "loss": 1.0281,
+      "step": 7433
+    },
+    {
+      "epoch": 0.778021978021978,
+      "grad_norm": 2.142967693368868,
+      "learning_rate": 2.475086857149982e-06,
+      "loss": 0.9277,
+      "step": 7434
+    },
+    {
+      "epoch": 0.7781266352694924,
+      "grad_norm": 2.1352013664869403,
+      "learning_rate": 2.4728548124376882e-06,
+      "loss": 1.0059,
+      "step": 7435
+    },
+    {
+      "epoch": 0.7782312925170068,
+      "grad_norm": 2.1208486875780537,
+      "learning_rate": 2.4706236326096357e-06,
+      "loss": 0.9999,
+      "step": 7436
+    },
+    {
+      "epoch": 0.7783359497645211,
+      "grad_norm": 2.351231832444182,
+      "learning_rate": 2.4683933179221908e-06,
+      "loss": 0.9088,
+      "step": 7437
+    },
+    {
+      "epoch": 0.7784406070120355,
+      "grad_norm": 2.612668770820266,
+      "learning_rate": 2.4661638686316193e-06,
+      "loss": 0.8534,
+      "step": 7438
+    },
+    {
+      "epoch": 0.77854526425955,
+      "grad_norm": 1.9406158005630711,
+      "learning_rate": 2.4639352849940947e-06,
+      "loss": 0.8145,
+      "step": 7439
+    },
+    {
+      "epoch": 0.7786499215070644,
+      "grad_norm": 2.000613163110507,
+      "learning_rate": 2.461707567265682e-06,
+      "loss": 0.8868,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7787545787545788,
+      "grad_norm": 2.316740372445306,
+      "learning_rate": 2.4594807157023525e-06,
+      "loss": 0.8466,
+      "step": 7441
+    },
+    {
+      "epoch": 0.7788592360020932,
+      "grad_norm": 1.9850899017077899,
+      "learning_rate": 2.4572547305599727e-06,
+      "loss": 0.9479,
+      "step": 7442
+    },
+    {
+      "epoch": 0.7789638932496076,
+      "grad_norm": 2.2991940584000825,
+      "learning_rate": 2.455029612094314e-06,
+      "loss": 0.8785,
+      "step": 7443
+    },
+    {
+      "epoch": 0.779068550497122,
+      "grad_norm": 2.0602886300262164,
+      "learning_rate": 2.4528053605610513e-06,
+      "loss": 0.9742,
+      "step": 7444
+    },
+    {
+      "epoch": 0.7791732077446363,
+      "grad_norm": 1.7206711855095842,
+      "learning_rate": 2.4505819762157534e-06,
+      "loss": 0.7868,
+      "step": 7445
+    },
+    {
+      "epoch": 0.7792778649921507,
+      "grad_norm": 2.1321280190827805,
+      "learning_rate": 2.4483594593138894e-06,
+      "loss": 0.9418,
+      "step": 7446
+    },
+    {
+      "epoch": 0.7793825222396651,
+      "grad_norm": 2.2378800662787888,
+      "learning_rate": 2.4461378101108333e-06,
+      "loss": 1.0102,
+      "step": 7447
+    },
+    {
+      "epoch": 0.7794871794871795,
+      "grad_norm": 1.891205894491629,
+      "learning_rate": 2.443917028861853e-06,
+      "loss": 0.8933,
+      "step": 7448
+    },
+    {
+      "epoch": 0.7795918367346939,
+      "grad_norm": 2.21558145359656,
+      "learning_rate": 2.441697115822126e-06,
+      "loss": 0.878,
+      "step": 7449
+    },
+    {
+      "epoch": 0.7796964939822083,
+      "grad_norm": 2.072454113701976,
+      "learning_rate": 2.4394780712467236e-06,
+      "loss": 0.7732,
+      "step": 7450
+    },
+    {
+      "epoch": 0.7798011512297227,
+      "grad_norm": 2.10593743081749,
+      "learning_rate": 2.4372598953906133e-06,
+      "loss": 0.768,
+      "step": 7451
+    },
+    {
+      "epoch": 0.779905808477237,
+      "grad_norm": 1.9053195631944175,
+      "learning_rate": 2.435042588508677e-06,
+      "loss": 0.8531,
+      "step": 7452
+    },
+    {
+      "epoch": 0.7800104657247514,
+      "grad_norm": 1.7533274864047492,
+      "learning_rate": 2.4328261508556795e-06,
+      "loss": 0.9297,
+      "step": 7453
+    },
+    {
+      "epoch": 0.7801151229722658,
+      "grad_norm": 2.074809239844302,
+      "learning_rate": 2.4306105826862993e-06,
+      "loss": 0.8273,
+      "step": 7454
+    },
+    {
+      "epoch": 0.7802197802197802,
+      "grad_norm": 2.125111767929294,
+      "learning_rate": 2.428395884255109e-06,
+      "loss": 0.7983,
+      "step": 7455
+    },
+    {
+      "epoch": 0.7803244374672946,
+      "grad_norm": 2.1801053470480323,
+      "learning_rate": 2.4261820558165804e-06,
+      "loss": 0.858,
+      "step": 7456
+    },
+    {
+      "epoch": 0.780429094714809,
+      "grad_norm": 1.8010107947795544,
+      "learning_rate": 2.4239690976250854e-06,
+      "loss": 0.7268,
+      "step": 7457
+    },
+    {
+      "epoch": 0.7805337519623234,
+      "grad_norm": 2.26916467402209,
+      "learning_rate": 2.421757009934901e-06,
+      "loss": 0.8247,
+      "step": 7458
+    },
+    {
+      "epoch": 0.7806384092098377,
+      "grad_norm": 1.9033771203519017,
+      "learning_rate": 2.4195457930001998e-06,
+      "loss": 0.7597,
+      "step": 7459
+    },
+    {
+      "epoch": 0.7807430664573521,
+      "grad_norm": 1.9507378053947775,
+      "learning_rate": 2.417335447075051e-06,
+      "loss": 0.8733,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7808477237048665,
+      "grad_norm": 2.099170592286283,
+      "learning_rate": 2.4151259724134336e-06,
+      "loss": 0.824,
+      "step": 7461
+    },
+    {
+      "epoch": 0.780952380952381,
+      "grad_norm": 2.227444898261774,
+      "learning_rate": 2.4129173692692155e-06,
+      "loss": 0.8599,
+      "step": 7462
+    },
+    {
+      "epoch": 0.7810570381998954,
+      "grad_norm": 1.9903419632586217,
+      "learning_rate": 2.4107096378961746e-06,
+      "loss": 0.9733,
+      "step": 7463
+    },
+    {
+      "epoch": 0.7811616954474098,
+      "grad_norm": 2.037233493155889,
+      "learning_rate": 2.408502778547982e-06,
+      "loss": 0.7757,
+      "step": 7464
+    },
+    {
+      "epoch": 0.7812663526949242,
+      "grad_norm": 2.0708333939156134,
+      "learning_rate": 2.4062967914782086e-06,
+      "loss": 0.8146,
+      "step": 7465
+    },
+    {
+      "epoch": 0.7813710099424385,
+      "grad_norm": 2.0344131802341976,
+      "learning_rate": 2.404091676940329e-06,
+      "loss": 0.9128,
+      "step": 7466
+    },
+    {
+      "epoch": 0.7814756671899529,
+      "grad_norm": 2.032647250442873,
+      "learning_rate": 2.401887435187712e-06,
+      "loss": 0.9985,
+      "step": 7467
+    },
+    {
+      "epoch": 0.7815803244374673,
+      "grad_norm": 2.431120027142936,
+      "learning_rate": 2.39968406647363e-06,
+      "loss": 0.8236,
+      "step": 7468
+    },
+    {
+      "epoch": 0.7816849816849817,
+      "grad_norm": 2.1068090556259738,
+      "learning_rate": 2.39748157105126e-06,
+      "loss": 0.9063,
+      "step": 7469
+    },
+    {
+      "epoch": 0.7817896389324961,
+      "grad_norm": 2.234104108726407,
+      "learning_rate": 2.39527994917367e-06,
+      "loss": 0.9853,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7818942961800105,
+      "grad_norm": 2.115693775219522,
+      "learning_rate": 2.3930792010938286e-06,
+      "loss": 0.9584,
+      "step": 7471
+    },
+    {
+      "epoch": 0.7819989534275249,
+      "grad_norm": 2.173122868655898,
+      "learning_rate": 2.3908793270646057e-06,
+      "loss": 0.8623,
+      "step": 7472
+    },
+    {
+      "epoch": 0.7821036106750392,
+      "grad_norm": 2.158065559668168,
+      "learning_rate": 2.388680327338777e-06,
+      "loss": 0.9185,
+      "step": 7473
+    },
+    {
+      "epoch": 0.7822082679225536,
+      "grad_norm": 2.4956821194494454,
+      "learning_rate": 2.386482202169008e-06,
+      "loss": 1.0272,
+      "step": 7474
+    },
+    {
+      "epoch": 0.782312925170068,
+      "grad_norm": 2.0104495991381888,
+      "learning_rate": 2.384284951807868e-06,
+      "loss": 0.8998,
+      "step": 7475
+    },
+    {
+      "epoch": 0.7824175824175824,
+      "grad_norm": 1.8416688486042285,
+      "learning_rate": 2.3820885765078226e-06,
+      "loss": 0.8212,
+      "step": 7476
+    },
+    {
+      "epoch": 0.7825222396650968,
+      "grad_norm": 2.302585517865405,
+      "learning_rate": 2.3798930765212435e-06,
+      "loss": 0.9272,
+      "step": 7477
+    },
+    {
+      "epoch": 0.7826268969126112,
+      "grad_norm": 2.13934576419671,
+      "learning_rate": 2.3776984521003997e-06,
+      "loss": 0.7576,
+      "step": 7478
+    },
+    {
+      "epoch": 0.7827315541601256,
+      "grad_norm": 2.136647066906963,
+      "learning_rate": 2.3755047034974567e-06,
+      "loss": 0.9061,
+      "step": 7479
+    },
+    {
+      "epoch": 0.7828362114076399,
+      "grad_norm": 1.8057322334613368,
+      "learning_rate": 2.373311830964479e-06,
+      "loss": 0.8824,
+      "step": 7480
+    },
+    {
+      "epoch": 0.7829408686551543,
+      "grad_norm": 1.9635915461888118,
+      "learning_rate": 2.3711198347534305e-06,
+      "loss": 0.8677,
+      "step": 7481
+    },
+    {
+      "epoch": 0.7830455259026687,
+      "grad_norm": 2.1726928757916673,
+      "learning_rate": 2.36892871511618e-06,
+      "loss": 0.8336,
+      "step": 7482
+    },
+    {
+      "epoch": 0.7831501831501831,
+      "grad_norm": 2.002270471261225,
+      "learning_rate": 2.3667384723044918e-06,
+      "loss": 0.923,
+      "step": 7483
+    },
+    {
+      "epoch": 0.7832548403976975,
+      "grad_norm": 2.0664098657994883,
+      "learning_rate": 2.3645491065700257e-06,
+      "loss": 0.8348,
+      "step": 7484
+    },
+    {
+      "epoch": 0.783359497645212,
+      "grad_norm": 2.2593815059893734,
+      "learning_rate": 2.3623606181643465e-06,
+      "loss": 0.9542,
+      "step": 7485
+    },
+    {
+      "epoch": 0.7834641548927264,
+      "grad_norm": 2.025749143560053,
+      "learning_rate": 2.3601730073389096e-06,
+      "loss": 0.9224,
+      "step": 7486
+    },
+    {
+      "epoch": 0.7835688121402408,
+      "grad_norm": 2.0959469108319047,
+      "learning_rate": 2.3579862743450877e-06,
+      "loss": 0.8847,
+      "step": 7487
+    },
+    {
+      "epoch": 0.7836734693877551,
+      "grad_norm": 2.2588339335795173,
+      "learning_rate": 2.3558004194341343e-06,
+      "loss": 0.8129,
+      "step": 7488
+    },
+    {
+      "epoch": 0.7837781266352695,
+      "grad_norm": 1.9031430821916164,
+      "learning_rate": 2.35361544285721e-06,
+      "loss": 0.8896,
+      "step": 7489
+    },
+    {
+      "epoch": 0.7838827838827839,
+      "grad_norm": 2.2458748833211506,
+      "learning_rate": 2.3514313448653715e-06,
+      "loss": 0.965,
+      "step": 7490
+    },
+    {
+      "epoch": 0.7839874411302983,
+      "grad_norm": 1.9843366337068091,
+      "learning_rate": 2.3492481257095746e-06,
+      "loss": 0.9045,
+      "step": 7491
+    },
+    {
+      "epoch": 0.7840920983778127,
+      "grad_norm": 2.3071137715563186,
+      "learning_rate": 2.34706578564068e-06,
+      "loss": 0.8596,
+      "step": 7492
+    },
+    {
+      "epoch": 0.7841967556253271,
+      "grad_norm": 2.1993251510987437,
+      "learning_rate": 2.34488432490944e-06,
+      "loss": 1.0281,
+      "step": 7493
+    },
+    {
+      "epoch": 0.7843014128728415,
+      "grad_norm": 1.767074435073521,
+      "learning_rate": 2.342703743766508e-06,
+      "loss": 0.9149,
+      "step": 7494
+    },
+    {
+      "epoch": 0.7844060701203558,
+      "grad_norm": 1.9658161545730846,
+      "learning_rate": 2.340524042462441e-06,
+      "loss": 0.8962,
+      "step": 7495
+    },
+    {
+      "epoch": 0.7845107273678702,
+      "grad_norm": 2.1705849889367888,
+      "learning_rate": 2.3383452212476866e-06,
+      "loss": 0.9249,
+      "step": 7496
+    },
+    {
+      "epoch": 0.7846153846153846,
+      "grad_norm": 2.0783706544712217,
+      "learning_rate": 2.3361672803725997e-06,
+      "loss": 0.972,
+      "step": 7497
+    },
+    {
+      "epoch": 0.784720041862899,
+      "grad_norm": 1.8899752350082029,
+      "learning_rate": 2.33399022008743e-06,
+      "loss": 0.8817,
+      "step": 7498
+    },
+    {
+      "epoch": 0.7848246991104134,
+      "grad_norm": 2.5517986810451783,
+      "learning_rate": 2.331814040642324e-06,
+      "loss": 0.8759,
+      "step": 7499
+    },
+    {
+      "epoch": 0.7849293563579278,
+      "grad_norm": 1.8563647336761964,
+      "learning_rate": 2.329638742287327e-06,
+      "loss": 0.8498,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7850340136054422,
+      "grad_norm": 1.7691375881118099,
+      "learning_rate": 2.3274643252723904e-06,
+      "loss": 0.7296,
+      "step": 7501
+    },
+    {
+      "epoch": 0.7851386708529565,
+      "grad_norm": 2.1783402785198667,
+      "learning_rate": 2.3252907898473564e-06,
+      "loss": 0.8703,
+      "step": 7502
+    },
+    {
+      "epoch": 0.7852433281004709,
+      "grad_norm": 2.205810214469163,
+      "learning_rate": 2.323118136261967e-06,
+      "loss": 0.9018,
+      "step": 7503
+    },
+    {
+      "epoch": 0.7853479853479853,
+      "grad_norm": 1.8545931677630307,
+      "learning_rate": 2.320946364765869e-06,
+      "loss": 0.8532,
+      "step": 7504
+    },
+    {
+      "epoch": 0.7854526425954997,
+      "grad_norm": 2.0691817245860333,
+      "learning_rate": 2.3187754756086e-06,
+      "loss": 0.9241,
+      "step": 7505
+    },
+    {
+      "epoch": 0.7855572998430141,
+      "grad_norm": 1.830258047422379,
+      "learning_rate": 2.316605469039599e-06,
+      "loss": 0.777,
+      "step": 7506
+    },
+    {
+      "epoch": 0.7856619570905286,
+      "grad_norm": 2.1983345444045077,
+      "learning_rate": 2.3144363453082076e-06,
+      "loss": 0.9011,
+      "step": 7507
+    },
+    {
+      "epoch": 0.785766614338043,
+      "grad_norm": 2.1342406030341867,
+      "learning_rate": 2.31226810466366e-06,
+      "loss": 0.9166,
+      "step": 7508
+    },
+    {
+      "epoch": 0.7858712715855573,
+      "grad_norm": 2.1953438695436276,
+      "learning_rate": 2.310100747355092e-06,
+      "loss": 0.9392,
+      "step": 7509
+    },
+    {
+      "epoch": 0.7859759288330717,
+      "grad_norm": 2.0745384468715766,
+      "learning_rate": 2.3079342736315335e-06,
+      "loss": 0.85,
+      "step": 7510
+    },
+    {
+      "epoch": 0.7860805860805861,
+      "grad_norm": 2.0531247089902953,
+      "learning_rate": 2.3057686837419246e-06,
+      "loss": 0.9608,
+      "step": 7511
+    },
+    {
+      "epoch": 0.7861852433281005,
+      "grad_norm": 2.0646468979200194,
+      "learning_rate": 2.303603977935088e-06,
+      "loss": 0.8047,
+      "step": 7512
+    },
+    {
+      "epoch": 0.7862899005756149,
+      "grad_norm": 2.232605413909429,
+      "learning_rate": 2.30144015645976e-06,
+      "loss": 0.8882,
+      "step": 7513
+    },
+    {
+      "epoch": 0.7863945578231293,
+      "grad_norm": 1.9565581752784238,
+      "learning_rate": 2.299277219564565e-06,
+      "loss": 0.8375,
+      "step": 7514
+    },
+    {
+      "epoch": 0.7864992150706437,
+      "grad_norm": 2.0031871819667497,
+      "learning_rate": 2.2971151674980254e-06,
+      "loss": 0.8445,
+      "step": 7515
+    },
+    {
+      "epoch": 0.786603872318158,
+      "grad_norm": 2.222141146594989,
+      "learning_rate": 2.29495400050857e-06,
+      "loss": 0.9172,
+      "step": 7516
+    },
+    {
+      "epoch": 0.7867085295656724,
+      "grad_norm": 1.9729975164234717,
+      "learning_rate": 2.2927937188445217e-06,
+      "loss": 0.9044,
+      "step": 7517
+    },
+    {
+      "epoch": 0.7868131868131868,
+      "grad_norm": 1.8762093594300924,
+      "learning_rate": 2.2906343227540973e-06,
+      "loss": 0.8125,
+      "step": 7518
+    },
+    {
+      "epoch": 0.7869178440607012,
+      "grad_norm": 2.134221883980738,
+      "learning_rate": 2.2884758124854156e-06,
+      "loss": 0.8042,
+      "step": 7519
+    },
+    {
+      "epoch": 0.7870225013082156,
+      "grad_norm": 1.9860025023180254,
+      "learning_rate": 2.2863181882864995e-06,
+      "loss": 0.8895,
+      "step": 7520
+    },
+    {
+      "epoch": 0.78712715855573,
+      "grad_norm": 1.879511345324259,
+      "learning_rate": 2.2841614504052576e-06,
+      "loss": 0.8006,
+      "step": 7521
+    },
+    {
+      "epoch": 0.7872318158032444,
+      "grad_norm": 1.9777761746411802,
+      "learning_rate": 2.28200559908951e-06,
+      "loss": 0.9588,
+      "step": 7522
+    },
+    {
+      "epoch": 0.7873364730507587,
+      "grad_norm": 2.0015943184841034,
+      "learning_rate": 2.279850634586963e-06,
+      "loss": 0.8274,
+      "step": 7523
+    },
+    {
+      "epoch": 0.7874411302982731,
+      "grad_norm": 2.126665132820641,
+      "learning_rate": 2.2776965571452305e-06,
+      "loss": 0.8101,
+      "step": 7524
+    },
+    {
+      "epoch": 0.7875457875457875,
+      "grad_norm": 1.9386922457161486,
+      "learning_rate": 2.2755433670118156e-06,
+      "loss": 0.9006,
+      "step": 7525
+    },
+    {
+      "epoch": 0.7876504447933019,
+      "grad_norm": 2.1069264857135974,
+      "learning_rate": 2.2733910644341284e-06,
+      "loss": 0.8787,
+      "step": 7526
+    },
+    {
+      "epoch": 0.7877551020408163,
+      "grad_norm": 2.312845843018434,
+      "learning_rate": 2.2712396496594724e-06,
+      "loss": 0.8824,
+      "step": 7527
+    },
+    {
+      "epoch": 0.7878597592883307,
+      "grad_norm": 2.410379692069536,
+      "learning_rate": 2.2690891229350474e-06,
+      "loss": 0.8534,
+      "step": 7528
+    },
+    {
+      "epoch": 0.7879644165358451,
+      "grad_norm": 1.6004259225754003,
+      "learning_rate": 2.266939484507953e-06,
+      "loss": 0.7655,
+      "step": 7529
+    },
+    {
+      "epoch": 0.7880690737833596,
+      "grad_norm": 1.9401251846318475,
+      "learning_rate": 2.2647907346251874e-06,
+      "loss": 0.9055,
+      "step": 7530
+    },
+    {
+      "epoch": 0.7881737310308738,
+      "grad_norm": 1.8259854422426756,
+      "learning_rate": 2.2626428735336504e-06,
+      "loss": 0.8849,
+      "step": 7531
+    },
+    {
+      "epoch": 0.7882783882783883,
+      "grad_norm": 2.086066746215832,
+      "learning_rate": 2.260495901480132e-06,
+      "loss": 0.9008,
+      "step": 7532
+    },
+    {
+      "epoch": 0.7883830455259027,
+      "grad_norm": 1.920464809197468,
+      "learning_rate": 2.2583498187113253e-06,
+      "loss": 0.9415,
+      "step": 7533
+    },
+    {
+      "epoch": 0.7884877027734171,
+      "grad_norm": 2.266413825635565,
+      "learning_rate": 2.2562046254738144e-06,
+      "loss": 0.8273,
+      "step": 7534
+    },
+    {
+      "epoch": 0.7885923600209315,
+      "grad_norm": 1.737401277999669,
+      "learning_rate": 2.2540603220140934e-06,
+      "loss": 0.9201,
+      "step": 7535
+    },
+    {
+      "epoch": 0.7886970172684459,
+      "grad_norm": 2.1096363498094854,
+      "learning_rate": 2.251916908578544e-06,
+      "loss": 0.9345,
+      "step": 7536
+    },
+    {
+      "epoch": 0.7888016745159603,
+      "grad_norm": 2.150831886317924,
+      "learning_rate": 2.2497743854134493e-06,
+      "loss": 0.844,
+      "step": 7537
+    },
+    {
+      "epoch": 0.7889063317634746,
+      "grad_norm": 2.0401061997847547,
+      "learning_rate": 2.2476327527649865e-06,
+      "loss": 0.7954,
+      "step": 7538
+    },
+    {
+      "epoch": 0.789010989010989,
+      "grad_norm": 2.337995915788079,
+      "learning_rate": 2.2454920108792354e-06,
+      "loss": 0.9412,
+      "step": 7539
+    },
+    {
+      "epoch": 0.7891156462585034,
+      "grad_norm": 1.9830974874569107,
+      "learning_rate": 2.2433521600021744e-06,
+      "loss": 0.8127,
+      "step": 7540
+    },
+    {
+      "epoch": 0.7892203035060178,
+      "grad_norm": 2.167769265794437,
+      "learning_rate": 2.241213200379676e-06,
+      "loss": 0.7914,
+      "step": 7541
+    },
+    {
+      "epoch": 0.7893249607535322,
+      "grad_norm": 2.33273978878609,
+      "learning_rate": 2.239075132257509e-06,
+      "loss": 0.8763,
+      "step": 7542
+    },
+    {
+      "epoch": 0.7894296180010466,
+      "grad_norm": 2.030942504421985,
+      "learning_rate": 2.236937955881343e-06,
+      "loss": 0.9236,
+      "step": 7543
+    },
+    {
+      "epoch": 0.789534275248561,
+      "grad_norm": 1.7829580923400006,
+      "learning_rate": 2.234801671496739e-06,
+      "loss": 0.7905,
+      "step": 7544
+    },
+    {
+      "epoch": 0.7896389324960753,
+      "grad_norm": 1.9662829211533972,
+      "learning_rate": 2.232666279349168e-06,
+      "loss": 0.8887,
+      "step": 7545
+    },
+    {
+      "epoch": 0.7897435897435897,
+      "grad_norm": 2.0352979592881444,
+      "learning_rate": 2.2305317796839887e-06,
+      "loss": 0.7428,
+      "step": 7546
+    },
+    {
+      "epoch": 0.7898482469911041,
+      "grad_norm": 1.6539156555371572,
+      "learning_rate": 2.2283981727464544e-06,
+      "loss": 0.7944,
+      "step": 7547
+    },
+    {
+      "epoch": 0.7899529042386185,
+      "grad_norm": 2.1614558820180783,
+      "learning_rate": 2.226265458781728e-06,
+      "loss": 0.8095,
+      "step": 7548
+    },
+    {
+      "epoch": 0.7900575614861329,
+      "grad_norm": 1.908631638174456,
+      "learning_rate": 2.2241336380348557e-06,
+      "loss": 0.8981,
+      "step": 7549
+    },
+    {
+      "epoch": 0.7901622187336473,
+      "grad_norm": 2.073463123923285,
+      "learning_rate": 2.222002710750796e-06,
+      "loss": 0.8807,
+      "step": 7550
+    },
+    {
+      "epoch": 0.7902668759811617,
+      "grad_norm": 2.0673778416845185,
+      "learning_rate": 2.2198726771743905e-06,
+      "loss": 0.8022,
+      "step": 7551
+    },
+    {
+      "epoch": 0.790371533228676,
+      "grad_norm": 1.6946057542306046,
+      "learning_rate": 2.2177435375503874e-06,
+      "loss": 0.7675,
+      "step": 7552
+    },
+    {
+      "epoch": 0.7904761904761904,
+      "grad_norm": 2.04567393679294,
+      "learning_rate": 2.2156152921234254e-06,
+      "loss": 0.9319,
+      "step": 7553
+    },
+    {
+      "epoch": 0.7905808477237048,
+      "grad_norm": 2.3078311593974514,
+      "learning_rate": 2.2134879411380505e-06,
+      "loss": 0.9253,
+      "step": 7554
+    },
+    {
+      "epoch": 0.7906855049712193,
+      "grad_norm": 2.115100849729748,
+      "learning_rate": 2.2113614848386924e-06,
+      "loss": 0.9134,
+      "step": 7555
+    },
+    {
+      "epoch": 0.7907901622187337,
+      "grad_norm": 2.20323705537602,
+      "learning_rate": 2.2092359234696916e-06,
+      "loss": 0.977,
+      "step": 7556
+    },
+    {
+      "epoch": 0.7908948194662481,
+      "grad_norm": 1.9616275621121293,
+      "learning_rate": 2.207111257275276e-06,
+      "loss": 0.9302,
+      "step": 7557
+    },
+    {
+      "epoch": 0.7909994767137625,
+      "grad_norm": 2.059271815772192,
+      "learning_rate": 2.2049874864995735e-06,
+      "loss": 0.9602,
+      "step": 7558
+    },
+    {
+      "epoch": 0.7911041339612768,
+      "grad_norm": 1.9637722265176025,
+      "learning_rate": 2.2028646113866127e-06,
+      "loss": 0.8741,
+      "step": 7559
+    },
+    {
+      "epoch": 0.7912087912087912,
+      "grad_norm": 2.2232201652151837,
+      "learning_rate": 2.200742632180315e-06,
+      "loss": 0.8846,
+      "step": 7560
+    },
+    {
+      "epoch": 0.7913134484563056,
+      "grad_norm": 2.30436743903849,
+      "learning_rate": 2.1986215491245e-06,
+      "loss": 0.9266,
+      "step": 7561
+    },
+    {
+      "epoch": 0.79141810570382,
+      "grad_norm": 2.364762077617069,
+      "learning_rate": 2.196501362462883e-06,
+      "loss": 0.9375,
+      "step": 7562
+    },
+    {
+      "epoch": 0.7915227629513344,
+      "grad_norm": 2.353159650570667,
+      "learning_rate": 2.1943820724390763e-06,
+      "loss": 0.8549,
+      "step": 7563
+    },
+    {
+      "epoch": 0.7916274201988488,
+      "grad_norm": 2.032814738216188,
+      "learning_rate": 2.1922636792965935e-06,
+      "loss": 0.7728,
+      "step": 7564
+    },
+    {
+      "epoch": 0.7917320774463632,
+      "grad_norm": 1.8205772867039858,
+      "learning_rate": 2.1901461832788453e-06,
+      "loss": 0.9063,
+      "step": 7565
+    },
+    {
+      "epoch": 0.7918367346938775,
+      "grad_norm": 2.1878001602801254,
+      "learning_rate": 2.188029584629131e-06,
+      "loss": 0.8742,
+      "step": 7566
+    },
+    {
+      "epoch": 0.7919413919413919,
+      "grad_norm": 2.2177495247640664,
+      "learning_rate": 2.1859138835906557e-06,
+      "loss": 0.8931,
+      "step": 7567
+    },
+    {
+      "epoch": 0.7920460491889063,
+      "grad_norm": 2.5843532600798915,
+      "learning_rate": 2.1837990804065125e-06,
+      "loss": 0.93,
+      "step": 7568
+    },
+    {
+      "epoch": 0.7921507064364207,
+      "grad_norm": 1.7462838713565032,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.7721,
+      "step": 7569
+    },
+    {
+      "epoch": 0.7922553636839351,
+      "grad_norm": 2.7345603215828964,
+      "learning_rate": 2.1795721685731153e-06,
+      "loss": 0.7922,
+      "step": 7570
+    },
+    {
+      "epoch": 0.7923600209314495,
+      "grad_norm": 2.458606643347218,
+      "learning_rate": 2.1774600604095407e-06,
+      "loss": 0.893,
+      "step": 7571
+    },
+    {
+      "epoch": 0.7924646781789639,
+      "grad_norm": 2.0422870744164956,
+      "learning_rate": 2.175348851071659e-06,
+      "loss": 0.8463,
+      "step": 7572
+    },
+    {
+      "epoch": 0.7925693354264783,
+      "grad_norm": 2.220719234475042,
+      "learning_rate": 2.1732385408020575e-06,
+      "loss": 0.941,
+      "step": 7573
+    },
+    {
+      "epoch": 0.7926739926739926,
+      "grad_norm": 1.975429690998908,
+      "learning_rate": 2.1711291298432157e-06,
+      "loss": 0.9173,
+      "step": 7574
+    },
+    {
+      "epoch": 0.792778649921507,
+      "grad_norm": 2.036016848312796,
+      "learning_rate": 2.1690206184375086e-06,
+      "loss": 0.8766,
+      "step": 7575
+    },
+    {
+      "epoch": 0.7928833071690214,
+      "grad_norm": 1.8743908375351486,
+      "learning_rate": 2.166913006827207e-06,
+      "loss": 0.8244,
+      "step": 7576
+    },
+    {
+      "epoch": 0.7929879644165359,
+      "grad_norm": 1.933946634791488,
+      "learning_rate": 2.164806295254478e-06,
+      "loss": 0.8598,
+      "step": 7577
+    },
+    {
+      "epoch": 0.7930926216640503,
+      "grad_norm": 1.7813235104033394,
+      "learning_rate": 2.162700483961392e-06,
+      "loss": 0.8794,
+      "step": 7578
+    },
+    {
+      "epoch": 0.7931972789115647,
+      "grad_norm": 2.2090460695960026,
+      "learning_rate": 2.1605955731899086e-06,
+      "loss": 0.9021,
+      "step": 7579
+    },
+    {
+      "epoch": 0.7933019361590791,
+      "grad_norm": 2.0974899535355616,
+      "learning_rate": 2.158491563181885e-06,
+      "loss": 0.8928,
+      "step": 7580
+    },
+    {
+      "epoch": 0.7934065934065934,
+      "grad_norm": 2.0169609622705154,
+      "learning_rate": 2.1563884541790747e-06,
+      "loss": 0.7913,
+      "step": 7581
+    },
+    {
+      "epoch": 0.7935112506541078,
+      "grad_norm": 2.199278005040713,
+      "learning_rate": 2.154286246423135e-06,
+      "loss": 0.9365,
+      "step": 7582
+    },
+    {
+      "epoch": 0.7936159079016222,
+      "grad_norm": 1.9357028205928937,
+      "learning_rate": 2.152184940155607e-06,
+      "loss": 0.8404,
+      "step": 7583
+    },
+    {
+      "epoch": 0.7937205651491366,
+      "grad_norm": 2.299234559003062,
+      "learning_rate": 2.150084535617942e-06,
+      "loss": 0.8855,
+      "step": 7584
+    },
+    {
+      "epoch": 0.793825222396651,
+      "grad_norm": 2.1127780268188454,
+      "learning_rate": 2.147985033051476e-06,
+      "loss": 0.8773,
+      "step": 7585
+    },
+    {
+      "epoch": 0.7939298796441654,
+      "grad_norm": 1.9778297585940738,
+      "learning_rate": 2.1458864326974495e-06,
+      "loss": 0.7991,
+      "step": 7586
+    },
+    {
+      "epoch": 0.7940345368916798,
+      "grad_norm": 2.815180081463802,
+      "learning_rate": 2.14378873479699e-06,
+      "loss": 1.1133,
+      "step": 7587
+    },
+    {
+      "epoch": 0.7941391941391941,
+      "grad_norm": 2.2536327406341474,
+      "learning_rate": 2.1416919395911327e-06,
+      "loss": 0.9107,
+      "step": 7588
+    },
+    {
+      "epoch": 0.7942438513867085,
+      "grad_norm": 2.0517416713013,
+      "learning_rate": 2.139596047320803e-06,
+      "loss": 0.9105,
+      "step": 7589
+    },
+    {
+      "epoch": 0.7943485086342229,
+      "grad_norm": 1.9716793290444405,
+      "learning_rate": 2.1375010582268194e-06,
+      "loss": 0.9514,
+      "step": 7590
+    },
+    {
+      "epoch": 0.7944531658817373,
+      "grad_norm": 1.7217851253182228,
+      "learning_rate": 2.135406972549906e-06,
+      "loss": 0.776,
+      "step": 7591
+    },
+    {
+      "epoch": 0.7945578231292517,
+      "grad_norm": 1.9071443124007157,
+      "learning_rate": 2.1333137905306723e-06,
+      "loss": 0.953,
+      "step": 7592
+    },
+    {
+      "epoch": 0.7946624803767661,
+      "grad_norm": 1.789970511220544,
+      "learning_rate": 2.1312215124096337e-06,
+      "loss": 0.7764,
+      "step": 7593
+    },
+    {
+      "epoch": 0.7947671376242805,
+      "grad_norm": 2.728962359247706,
+      "learning_rate": 2.1291301384271955e-06,
+      "loss": 0.8014,
+      "step": 7594
+    },
+    {
+      "epoch": 0.7948717948717948,
+      "grad_norm": 1.9322455444724893,
+      "learning_rate": 2.1270396688236595e-06,
+      "loss": 0.7246,
+      "step": 7595
+    },
+    {
+      "epoch": 0.7949764521193092,
+      "grad_norm": 2.0774223052480107,
+      "learning_rate": 2.124950103839224e-06,
+      "loss": 1.0114,
+      "step": 7596
+    },
+    {
+      "epoch": 0.7950811093668236,
+      "grad_norm": 2.275100035949746,
+      "learning_rate": 2.122861443713987e-06,
+      "loss": 0.8527,
+      "step": 7597
+    },
+    {
+      "epoch": 0.795185766614338,
+      "grad_norm": 1.9482436075243985,
+      "learning_rate": 2.1207736886879406e-06,
+      "loss": 0.8765,
+      "step": 7598
+    },
+    {
+      "epoch": 0.7952904238618524,
+      "grad_norm": 1.9226926868108565,
+      "learning_rate": 2.1186868390009664e-06,
+      "loss": 0.8391,
+      "step": 7599
+    },
+    {
+      "epoch": 0.7953950811093669,
+      "grad_norm": 1.968122748788881,
+      "learning_rate": 2.116600894892855e-06,
+      "loss": 0.8938,
+      "step": 7600
+    },
+    {
+      "epoch": 0.7954997383568813,
+      "grad_norm": 2.0362358868182655,
+      "learning_rate": 2.1145158566032808e-06,
+      "loss": 0.92,
+      "step": 7601
+    },
+    {
+      "epoch": 0.7956043956043956,
+      "grad_norm": 1.8167616269052242,
+      "learning_rate": 2.1124317243718184e-06,
+      "loss": 0.874,
+      "step": 7602
+    },
+    {
+      "epoch": 0.79570905285191,
+      "grad_norm": 2.081440590488677,
+      "learning_rate": 2.1103484984379443e-06,
+      "loss": 0.9316,
+      "step": 7603
+    },
+    {
+      "epoch": 0.7958137100994244,
+      "grad_norm": 2.0790409856006864,
+      "learning_rate": 2.1082661790410197e-06,
+      "loss": 0.8893,
+      "step": 7604
+    },
+    {
+      "epoch": 0.7959183673469388,
+      "grad_norm": 2.0996779895786135,
+      "learning_rate": 2.1061847664203107e-06,
+      "loss": 0.9361,
+      "step": 7605
+    },
+    {
+      "epoch": 0.7960230245944532,
+      "grad_norm": 2.1580716719163386,
+      "learning_rate": 2.1041042608149707e-06,
+      "loss": 0.9383,
+      "step": 7606
+    },
+    {
+      "epoch": 0.7961276818419676,
+      "grad_norm": 1.8816414196677191,
+      "learning_rate": 2.1020246624640595e-06,
+      "loss": 0.7567,
+      "step": 7607
+    },
+    {
+      "epoch": 0.796232339089482,
+      "grad_norm": 2.052842909348786,
+      "learning_rate": 2.0999459716065273e-06,
+      "loss": 0.8624,
+      "step": 7608
+    },
+    {
+      "epoch": 0.7963369963369963,
+      "grad_norm": 2.0109322496007733,
+      "learning_rate": 2.097868188481217e-06,
+      "loss": 0.9112,
+      "step": 7609
+    },
+    {
+      "epoch": 0.7964416535845107,
+      "grad_norm": 1.9850219383542753,
+      "learning_rate": 2.0957913133268726e-06,
+      "loss": 0.7805,
+      "step": 7610
+    },
+    {
+      "epoch": 0.7965463108320251,
+      "grad_norm": 2.257844754431785,
+      "learning_rate": 2.0937153463821257e-06,
+      "loss": 0.8675,
+      "step": 7611
+    },
+    {
+      "epoch": 0.7966509680795395,
+      "grad_norm": 2.002649601742002,
+      "learning_rate": 2.0916402878855168e-06,
+      "loss": 0.8938,
+      "step": 7612
+    },
+    {
+      "epoch": 0.7967556253270539,
+      "grad_norm": 1.921356147737922,
+      "learning_rate": 2.0895661380754715e-06,
+      "loss": 0.8056,
+      "step": 7613
+    },
+    {
+      "epoch": 0.7968602825745683,
+      "grad_norm": 2.2073850950321185,
+      "learning_rate": 2.087492897190312e-06,
+      "loss": 0.9354,
+      "step": 7614
+    },
+    {
+      "epoch": 0.7969649398220827,
+      "grad_norm": 2.3982569594670218,
+      "learning_rate": 2.085420565468256e-06,
+      "loss": 0.9788,
+      "step": 7615
+    },
+    {
+      "epoch": 0.7970695970695971,
+      "grad_norm": 2.3499158850699007,
+      "learning_rate": 2.083349143147422e-06,
+      "loss": 0.9463,
+      "step": 7616
+    },
+    {
+      "epoch": 0.7971742543171114,
+      "grad_norm": 1.784998934231335,
+      "learning_rate": 2.081278630465823e-06,
+      "loss": 0.8517,
+      "step": 7617
+    },
+    {
+      "epoch": 0.7972789115646258,
+      "grad_norm": 2.3333292170699687,
+      "learning_rate": 2.079209027661362e-06,
+      "loss": 0.8067,
+      "step": 7618
+    },
+    {
+      "epoch": 0.7973835688121402,
+      "grad_norm": 2.4308857648334627,
+      "learning_rate": 2.077140334971841e-06,
+      "loss": 0.8981,
+      "step": 7619
+    },
+    {
+      "epoch": 0.7974882260596546,
+      "grad_norm": 2.2155185275133524,
+      "learning_rate": 2.075072552634957e-06,
+      "loss": 0.8724,
+      "step": 7620
+    },
+    {
+      "epoch": 0.797592883307169,
+      "grad_norm": 1.8788752477824127,
+      "learning_rate": 2.0730056808882993e-06,
+      "loss": 0.7483,
+      "step": 7621
+    },
+    {
+      "epoch": 0.7976975405546834,
+      "grad_norm": 2.2866944996543146,
+      "learning_rate": 2.0709397199693616e-06,
+      "loss": 1.0201,
+      "step": 7622
+    },
+    {
+      "epoch": 0.7978021978021979,
+      "grad_norm": 2.2884876205119853,
+      "learning_rate": 2.068874670115524e-06,
+      "loss": 0.8809,
+      "step": 7623
+    },
+    {
+      "epoch": 0.7979068550497121,
+      "grad_norm": 2.2069390747276367,
+      "learning_rate": 2.0668105315640643e-06,
+      "loss": 0.9331,
+      "step": 7624
+    },
+    {
+      "epoch": 0.7980115122972266,
+      "grad_norm": 2.2300750543609316,
+      "learning_rate": 2.0647473045521536e-06,
+      "loss": 0.9861,
+      "step": 7625
+    },
+    {
+      "epoch": 0.798116169544741,
+      "grad_norm": 1.9473189124396104,
+      "learning_rate": 2.0626849893168655e-06,
+      "loss": 0.8974,
+      "step": 7626
+    },
+    {
+      "epoch": 0.7982208267922554,
+      "grad_norm": 2.1241126439462175,
+      "learning_rate": 2.060623586095164e-06,
+      "loss": 0.8733,
+      "step": 7627
+    },
+    {
+      "epoch": 0.7983254840397698,
+      "grad_norm": 2.596050491489067,
+      "learning_rate": 2.058563095123908e-06,
+      "loss": 0.8455,
+      "step": 7628
+    },
+    {
+      "epoch": 0.7984301412872842,
+      "grad_norm": 1.6496184313046733,
+      "learning_rate": 2.056503516639852e-06,
+      "loss": 0.7479,
+      "step": 7629
+    },
+    {
+      "epoch": 0.7985347985347986,
+      "grad_norm": 1.9392866056291445,
+      "learning_rate": 2.054444850879641e-06,
+      "loss": 0.8385,
+      "step": 7630
+    },
+    {
+      "epoch": 0.7986394557823129,
+      "grad_norm": 2.2720725714699754,
+      "learning_rate": 2.052387098079828e-06,
+      "loss": 0.9297,
+      "step": 7631
+    },
+    {
+      "epoch": 0.7987441130298273,
+      "grad_norm": 2.0705720785860238,
+      "learning_rate": 2.050330258476848e-06,
+      "loss": 0.8512,
+      "step": 7632
+    },
+    {
+      "epoch": 0.7988487702773417,
+      "grad_norm": 1.9969354338670204,
+      "learning_rate": 2.048274332307034e-06,
+      "loss": 0.8813,
+      "step": 7633
+    },
+    {
+      "epoch": 0.7989534275248561,
+      "grad_norm": 2.6051722309501764,
+      "learning_rate": 2.0462193198066226e-06,
+      "loss": 0.9048,
+      "step": 7634
+    },
+    {
+      "epoch": 0.7990580847723705,
+      "grad_norm": 2.164629627302996,
+      "learning_rate": 2.0441652212117326e-06,
+      "loss": 0.9597,
+      "step": 7635
+    },
+    {
+      "epoch": 0.7991627420198849,
+      "grad_norm": 1.9912390762348158,
+      "learning_rate": 2.042112036758389e-06,
+      "loss": 0.9058,
+      "step": 7636
+    },
+    {
+      "epoch": 0.7992673992673993,
+      "grad_norm": 1.8720345668097196,
+      "learning_rate": 2.040059766682504e-06,
+      "loss": 0.8318,
+      "step": 7637
+    },
+    {
+      "epoch": 0.7993720565149136,
+      "grad_norm": 2.0710286245998946,
+      "learning_rate": 2.0380084112198893e-06,
+      "loss": 0.849,
+      "step": 7638
+    },
+    {
+      "epoch": 0.799476713762428,
+      "grad_norm": 1.8830717363876028,
+      "learning_rate": 2.0359579706062493e-06,
+      "loss": 0.9041,
+      "step": 7639
+    },
+    {
+      "epoch": 0.7995813710099424,
+      "grad_norm": 1.895209812890259,
+      "learning_rate": 2.0339084450771797e-06,
+      "loss": 0.7978,
+      "step": 7640
+    },
+    {
+      "epoch": 0.7996860282574568,
+      "grad_norm": 1.8023260782135615,
+      "learning_rate": 2.031859834868182e-06,
+      "loss": 0.8613,
+      "step": 7641
+    },
+    {
+      "epoch": 0.7997906855049712,
+      "grad_norm": 2.1045087518839125,
+      "learning_rate": 2.0298121402146386e-06,
+      "loss": 0.8604,
+      "step": 7642
+    },
+    {
+      "epoch": 0.7998953427524856,
+      "grad_norm": 1.7435782505197222,
+      "learning_rate": 2.0277653613518413e-06,
+      "loss": 0.7904,
+      "step": 7643
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 2.393742131042189,
+      "learning_rate": 2.0257194985149653e-06,
+      "loss": 0.8332,
+      "step": 7644
+    },
+    {
+      "epoch": 0.8001046572475143,
+      "grad_norm": 2.0074026574381003,
+      "learning_rate": 2.023674551939081e-06,
+      "loss": 1.0339,
+      "step": 7645
+    },
+    {
+      "epoch": 0.8002093144950287,
+      "grad_norm": 2.1072035338876045,
+      "learning_rate": 2.0216305218591636e-06,
+      "loss": 0.9326,
+      "step": 7646
+    },
+    {
+      "epoch": 0.8003139717425432,
+      "grad_norm": 2.2169694179321766,
+      "learning_rate": 2.019587408510074e-06,
+      "loss": 1.0086,
+      "step": 7647
+    },
+    {
+      "epoch": 0.8004186289900576,
+      "grad_norm": 1.745468429945779,
+      "learning_rate": 2.017545212126568e-06,
+      "loss": 0.7837,
+      "step": 7648
+    },
+    {
+      "epoch": 0.800523286237572,
+      "grad_norm": 2.307042540661392,
+      "learning_rate": 2.015503932943297e-06,
+      "loss": 0.9432,
+      "step": 7649
+    },
+    {
+      "epoch": 0.8006279434850864,
+      "grad_norm": 1.8426257142814524,
+      "learning_rate": 2.0134635711948125e-06,
+      "loss": 0.9432,
+      "step": 7650
+    },
+    {
+      "epoch": 0.8007326007326008,
+      "grad_norm": 1.9616990567522987,
+      "learning_rate": 2.011424127115552e-06,
+      "loss": 0.8726,
+      "step": 7651
+    },
+    {
+      "epoch": 0.8008372579801151,
+      "grad_norm": 2.0819208074157536,
+      "learning_rate": 2.009385600939856e-06,
+      "loss": 0.8719,
+      "step": 7652
+    },
+    {
+      "epoch": 0.8009419152276295,
+      "grad_norm": 2.357873291710539,
+      "learning_rate": 2.0073479929019536e-06,
+      "loss": 1.0302,
+      "step": 7653
+    },
+    {
+      "epoch": 0.8010465724751439,
+      "grad_norm": 2.22508676135223,
+      "learning_rate": 2.005311303235966e-06,
+      "loss": 0.9423,
+      "step": 7654
+    },
+    {
+      "epoch": 0.8011512297226583,
+      "grad_norm": 2.4557652270731616,
+      "learning_rate": 2.00327553217592e-06,
+      "loss": 0.8045,
+      "step": 7655
+    },
+    {
+      "epoch": 0.8012558869701727,
+      "grad_norm": 2.258447659718386,
+      "learning_rate": 2.001240679955727e-06,
+      "loss": 0.8484,
+      "step": 7656
+    },
+    {
+      "epoch": 0.8013605442176871,
+      "grad_norm": 2.0218151709850316,
+      "learning_rate": 1.9992067468091947e-06,
+      "loss": 0.9074,
+      "step": 7657
+    },
+    {
+      "epoch": 0.8014652014652015,
+      "grad_norm": 2.31088112748935,
+      "learning_rate": 1.9971737329700256e-06,
+      "loss": 0.7952,
+      "step": 7658
+    },
+    {
+      "epoch": 0.8015698587127159,
+      "grad_norm": 2.132378499236807,
+      "learning_rate": 1.995141638671816e-06,
+      "loss": 0.8665,
+      "step": 7659
+    },
+    {
+      "epoch": 0.8016745159602302,
+      "grad_norm": 2.1426183845478053,
+      "learning_rate": 1.993110464148059e-06,
+      "loss": 0.9102,
+      "step": 7660
+    },
+    {
+      "epoch": 0.8017791732077446,
+      "grad_norm": 2.1293042574252867,
+      "learning_rate": 1.9910802096321457e-06,
+      "loss": 0.8794,
+      "step": 7661
+    },
+    {
+      "epoch": 0.801883830455259,
+      "grad_norm": 2.1089889655402168,
+      "learning_rate": 1.98905087535735e-06,
+      "loss": 0.9595,
+      "step": 7662
+    },
+    {
+      "epoch": 0.8019884877027734,
+      "grad_norm": 2.0969209074252615,
+      "learning_rate": 1.98702246155685e-06,
+      "loss": 0.9647,
+      "step": 7663
+    },
+    {
+      "epoch": 0.8020931449502878,
+      "grad_norm": 1.992617104506457,
+      "learning_rate": 1.9849949684637103e-06,
+      "loss": 0.8987,
+      "step": 7664
+    },
+    {
+      "epoch": 0.8021978021978022,
+      "grad_norm": 2.633618499233918,
+      "learning_rate": 1.9829683963108992e-06,
+      "loss": 0.9034,
+      "step": 7665
+    },
+    {
+      "epoch": 0.8023024594453166,
+      "grad_norm": 2.2056134263285907,
+      "learning_rate": 1.980942745331271e-06,
+      "loss": 0.8807,
+      "step": 7666
+    },
+    {
+      "epoch": 0.8024071166928309,
+      "grad_norm": 2.068273680645422,
+      "learning_rate": 1.9789180157575795e-06,
+      "loss": 0.8689,
+      "step": 7667
+    },
+    {
+      "epoch": 0.8025117739403453,
+      "grad_norm": 2.019594583927551,
+      "learning_rate": 1.976894207822464e-06,
+      "loss": 0.9953,
+      "step": 7668
+    },
+    {
+      "epoch": 0.8026164311878597,
+      "grad_norm": 2.316889621162141,
+      "learning_rate": 1.9748713217584693e-06,
+      "loss": 0.8515,
+      "step": 7669
+    },
+    {
+      "epoch": 0.8027210884353742,
+      "grad_norm": 2.111794166943864,
+      "learning_rate": 1.972849357798031e-06,
+      "loss": 0.9745,
+      "step": 7670
+    },
+    {
+      "epoch": 0.8028257456828886,
+      "grad_norm": 2.03962760228518,
+      "learning_rate": 1.970828316173474e-06,
+      "loss": 0.8139,
+      "step": 7671
+    },
+    {
+      "epoch": 0.802930402930403,
+      "grad_norm": 2.069040546093496,
+      "learning_rate": 1.9688081971170202e-06,
+      "loss": 0.8368,
+      "step": 7672
+    },
+    {
+      "epoch": 0.8030350601779174,
+      "grad_norm": 2.1304131521617613,
+      "learning_rate": 1.9667890008607827e-06,
+      "loss": 0.8431,
+      "step": 7673
+    },
+    {
+      "epoch": 0.8031397174254317,
+      "grad_norm": 1.95426739187706,
+      "learning_rate": 1.964770727636778e-06,
+      "loss": 0.9034,
+      "step": 7674
+    },
+    {
+      "epoch": 0.8032443746729461,
+      "grad_norm": 2.1064759648577773,
+      "learning_rate": 1.962753377676905e-06,
+      "loss": 0.8432,
+      "step": 7675
+    },
+    {
+      "epoch": 0.8033490319204605,
+      "grad_norm": 1.8850818133792997,
+      "learning_rate": 1.9607369512129615e-06,
+      "loss": 0.8973,
+      "step": 7676
+    },
+    {
+      "epoch": 0.8034536891679749,
+      "grad_norm": 1.5859786709753045,
+      "learning_rate": 1.958721448476637e-06,
+      "loss": 0.7476,
+      "step": 7677
+    },
+    {
+      "epoch": 0.8035583464154893,
+      "grad_norm": 2.220270218255743,
+      "learning_rate": 1.9567068696995227e-06,
+      "loss": 0.9196,
+      "step": 7678
+    },
+    {
+      "epoch": 0.8036630036630037,
+      "grad_norm": 1.6376859017657255,
+      "learning_rate": 1.9546932151130913e-06,
+      "loss": 0.771,
+      "step": 7679
+    },
+    {
+      "epoch": 0.8037676609105181,
+      "grad_norm": 2.1672382842265185,
+      "learning_rate": 1.9526804849487225e-06,
+      "loss": 0.9884,
+      "step": 7680
+    },
+    {
+      "epoch": 0.8038723181580324,
+      "grad_norm": 2.16714258180435,
+      "learning_rate": 1.9506686794376782e-06,
+      "loss": 0.9114,
+      "step": 7681
+    },
+    {
+      "epoch": 0.8039769754055468,
+      "grad_norm": 2.130041323517797,
+      "learning_rate": 1.948657798811121e-06,
+      "loss": 0.9454,
+      "step": 7682
+    },
+    {
+      "epoch": 0.8040816326530612,
+      "grad_norm": 2.0922079549466512,
+      "learning_rate": 1.9466478433001022e-06,
+      "loss": 0.8827,
+      "step": 7683
+    },
+    {
+      "epoch": 0.8041862899005756,
+      "grad_norm": 2.0945418746948197,
+      "learning_rate": 1.9446388131355753e-06,
+      "loss": 0.8897,
+      "step": 7684
+    },
+    {
+      "epoch": 0.80429094714809,
+      "grad_norm": 2.236535687893619,
+      "learning_rate": 1.9426307085483753e-06,
+      "loss": 0.9248,
+      "step": 7685
+    },
+    {
+      "epoch": 0.8043956043956044,
+      "grad_norm": 2.0020235889260647,
+      "learning_rate": 1.9406235297692434e-06,
+      "loss": 0.8902,
+      "step": 7686
+    },
+    {
+      "epoch": 0.8045002616431188,
+      "grad_norm": 2.161723959121442,
+      "learning_rate": 1.9386172770288057e-06,
+      "loss": 0.9059,
+      "step": 7687
+    },
+    {
+      "epoch": 0.8046049188906331,
+      "grad_norm": 2.1277819801638373,
+      "learning_rate": 1.936611950557583e-06,
+      "loss": 0.7741,
+      "step": 7688
+    },
+    {
+      "epoch": 0.8047095761381475,
+      "grad_norm": 1.8553064489672486,
+      "learning_rate": 1.9346075505859953e-06,
+      "loss": 0.8993,
+      "step": 7689
+    },
+    {
+      "epoch": 0.8048142333856619,
+      "grad_norm": 1.7911656465689691,
+      "learning_rate": 1.932604077344351e-06,
+      "loss": 0.7607,
+      "step": 7690
+    },
+    {
+      "epoch": 0.8049188906331763,
+      "grad_norm": 1.6049725585488301,
+      "learning_rate": 1.9306015310628523e-06,
+      "loss": 0.7695,
+      "step": 7691
+    },
+    {
+      "epoch": 0.8050235478806907,
+      "grad_norm": 1.9936774666260564,
+      "learning_rate": 1.9285999119715936e-06,
+      "loss": 0.755,
+      "step": 7692
+    },
+    {
+      "epoch": 0.8051282051282052,
+      "grad_norm": 2.164086612979867,
+      "learning_rate": 1.926599220300569e-06,
+      "loss": 0.9036,
+      "step": 7693
+    },
+    {
+      "epoch": 0.8052328623757196,
+      "grad_norm": 2.1877719623928247,
+      "learning_rate": 1.924599456279659e-06,
+      "loss": 0.9081,
+      "step": 7694
+    },
+    {
+      "epoch": 0.8053375196232339,
+      "grad_norm": 1.9764063147501494,
+      "learning_rate": 1.922600620138645e-06,
+      "loss": 0.7899,
+      "step": 7695
+    },
+    {
+      "epoch": 0.8054421768707483,
+      "grad_norm": 2.096661411005218,
+      "learning_rate": 1.9206027121071937e-06,
+      "loss": 0.8407,
+      "step": 7696
+    },
+    {
+      "epoch": 0.8055468341182627,
+      "grad_norm": 2.2156877709091742,
+      "learning_rate": 1.918605732414869e-06,
+      "loss": 0.924,
+      "step": 7697
+    },
+    {
+      "epoch": 0.8056514913657771,
+      "grad_norm": 1.6524591320676993,
+      "learning_rate": 1.916609681291126e-06,
+      "loss": 0.8249,
+      "step": 7698
+    },
+    {
+      "epoch": 0.8057561486132915,
+      "grad_norm": 2.3280205948736694,
+      "learning_rate": 1.914614558965321e-06,
+      "loss": 0.864,
+      "step": 7699
+    },
+    {
+      "epoch": 0.8058608058608059,
+      "grad_norm": 1.8899662650771822,
+      "learning_rate": 1.9126203656666918e-06,
+      "loss": 0.7827,
+      "step": 7700
+    },
+    {
+      "epoch": 0.8059654631083203,
+      "grad_norm": 1.8294831259412443,
+      "learning_rate": 1.9106271016243792e-06,
+      "loss": 0.7265,
+      "step": 7701
+    },
+    {
+      "epoch": 0.8060701203558347,
+      "grad_norm": 1.8127721118364941,
+      "learning_rate": 1.908634767067409e-06,
+      "loss": 0.7789,
+      "step": 7702
+    },
+    {
+      "epoch": 0.806174777603349,
+      "grad_norm": 2.0601656865089004,
+      "learning_rate": 1.906643362224706e-06,
+      "loss": 0.7856,
+      "step": 7703
+    },
+    {
+      "epoch": 0.8062794348508634,
+      "grad_norm": 2.0340414414029984,
+      "learning_rate": 1.9046528873250902e-06,
+      "loss": 0.8515,
+      "step": 7704
+    },
+    {
+      "epoch": 0.8063840920983778,
+      "grad_norm": 1.9432516065364067,
+      "learning_rate": 1.9026633425972696e-06,
+      "loss": 0.8491,
+      "step": 7705
+    },
+    {
+      "epoch": 0.8064887493458922,
+      "grad_norm": 2.4510874165764838,
+      "learning_rate": 1.9006747282698445e-06,
+      "loss": 0.7151,
+      "step": 7706
+    },
+    {
+      "epoch": 0.8065934065934066,
+      "grad_norm": 2.3967160624954786,
+      "learning_rate": 1.8986870445713112e-06,
+      "loss": 0.8809,
+      "step": 7707
+    },
+    {
+      "epoch": 0.806698063840921,
+      "grad_norm": 2.1255469660957766,
+      "learning_rate": 1.8967002917300614e-06,
+      "loss": 0.918,
+      "step": 7708
+    },
+    {
+      "epoch": 0.8068027210884354,
+      "grad_norm": 2.178668681213783,
+      "learning_rate": 1.8947144699743747e-06,
+      "loss": 0.9926,
+      "step": 7709
+    },
+    {
+      "epoch": 0.8069073783359497,
+      "grad_norm": 2.2070908456686933,
+      "learning_rate": 1.8927295795324275e-06,
+      "loss": 0.8722,
+      "step": 7710
+    },
+    {
+      "epoch": 0.8070120355834641,
+      "grad_norm": 1.817792067239064,
+      "learning_rate": 1.890745620632284e-06,
+      "loss": 0.8888,
+      "step": 7711
+    },
+    {
+      "epoch": 0.8071166928309785,
+      "grad_norm": 2.5554707262103293,
+      "learning_rate": 1.8887625935019072e-06,
+      "loss": 0.9806,
+      "step": 7712
+    },
+    {
+      "epoch": 0.8072213500784929,
+      "grad_norm": 2.304574255962299,
+      "learning_rate": 1.886780498369154e-06,
+      "loss": 0.8317,
+      "step": 7713
+    },
+    {
+      "epoch": 0.8073260073260073,
+      "grad_norm": 2.0339279457346375,
+      "learning_rate": 1.8847993354617689e-06,
+      "loss": 0.9634,
+      "step": 7714
+    },
+    {
+      "epoch": 0.8074306645735218,
+      "grad_norm": 2.0844568689691867,
+      "learning_rate": 1.8828191050073918e-06,
+      "loss": 0.9645,
+      "step": 7715
+    },
+    {
+      "epoch": 0.8075353218210362,
+      "grad_norm": 1.9109270715202566,
+      "learning_rate": 1.8808398072335555e-06,
+      "loss": 0.8214,
+      "step": 7716
+    },
+    {
+      "epoch": 0.8076399790685505,
+      "grad_norm": 1.9339620974128462,
+      "learning_rate": 1.878861442367681e-06,
+      "loss": 0.8987,
+      "step": 7717
+    },
+    {
+      "epoch": 0.8077446363160649,
+      "grad_norm": 1.9397724352401922,
+      "learning_rate": 1.8768840106370922e-06,
+      "loss": 0.8367,
+      "step": 7718
+    },
+    {
+      "epoch": 0.8078492935635793,
+      "grad_norm": 2.1190867449260917,
+      "learning_rate": 1.8749075122689986e-06,
+      "loss": 0.8944,
+      "step": 7719
+    },
+    {
+      "epoch": 0.8079539508110937,
+      "grad_norm": 2.670252322303464,
+      "learning_rate": 1.8729319474905005e-06,
+      "loss": 0.8799,
+      "step": 7720
+    },
+    {
+      "epoch": 0.8080586080586081,
+      "grad_norm": 2.141457200125952,
+      "learning_rate": 1.8709573165286e-06,
+      "loss": 0.8191,
+      "step": 7721
+    },
+    {
+      "epoch": 0.8081632653061225,
+      "grad_norm": 1.7558169293557664,
+      "learning_rate": 1.8689836196101797e-06,
+      "loss": 0.8741,
+      "step": 7722
+    },
+    {
+      "epoch": 0.8082679225536369,
+      "grad_norm": 2.06145311599018,
+      "learning_rate": 1.867010856962027e-06,
+      "loss": 0.9091,
+      "step": 7723
+    },
+    {
+      "epoch": 0.8083725798011512,
+      "grad_norm": 2.1553715349745937,
+      "learning_rate": 1.8650390288108156e-06,
+      "loss": 0.9571,
+      "step": 7724
+    },
+    {
+      "epoch": 0.8084772370486656,
+      "grad_norm": 2.6232290933594307,
+      "learning_rate": 1.8630681353831104e-06,
+      "loss": 0.7693,
+      "step": 7725
+    },
+    {
+      "epoch": 0.80858189429618,
+      "grad_norm": 2.3021150057545627,
+      "learning_rate": 1.8610981769053681e-06,
+      "loss": 0.9316,
+      "step": 7726
+    },
+    {
+      "epoch": 0.8086865515436944,
+      "grad_norm": 1.9431158516385172,
+      "learning_rate": 1.859129153603949e-06,
+      "loss": 0.9173,
+      "step": 7727
+    },
+    {
+      "epoch": 0.8087912087912088,
+      "grad_norm": 2.040241180020073,
+      "learning_rate": 1.8571610657050921e-06,
+      "loss": 0.9225,
+      "step": 7728
+    },
+    {
+      "epoch": 0.8088958660387232,
+      "grad_norm": 2.049568434268225,
+      "learning_rate": 1.855193913434934e-06,
+      "loss": 0.8637,
+      "step": 7729
+    },
+    {
+      "epoch": 0.8090005232862376,
+      "grad_norm": 2.2567330112829866,
+      "learning_rate": 1.8532276970195085e-06,
+      "loss": 0.8094,
+      "step": 7730
+    },
+    {
+      "epoch": 0.8091051805337519,
+      "grad_norm": 2.4898641315793144,
+      "learning_rate": 1.8512624166847338e-06,
+      "loss": 0.9544,
+      "step": 7731
+    },
+    {
+      "epoch": 0.8092098377812663,
+      "grad_norm": 2.087408291721611,
+      "learning_rate": 1.8492980726564292e-06,
+      "loss": 0.8928,
+      "step": 7732
+    },
+    {
+      "epoch": 0.8093144950287807,
+      "grad_norm": 1.9848372835802772,
+      "learning_rate": 1.8473346651602986e-06,
+      "loss": 0.9422,
+      "step": 7733
+    },
+    {
+      "epoch": 0.8094191522762951,
+      "grad_norm": 1.9749206184830705,
+      "learning_rate": 1.8453721944219428e-06,
+      "loss": 0.835,
+      "step": 7734
+    },
+    {
+      "epoch": 0.8095238095238095,
+      "grad_norm": 2.298149888799161,
+      "learning_rate": 1.8434106606668522e-06,
+      "loss": 0.9568,
+      "step": 7735
+    },
+    {
+      "epoch": 0.8096284667713239,
+      "grad_norm": 2.081259887949481,
+      "learning_rate": 1.8414500641204104e-06,
+      "loss": 0.8567,
+      "step": 7736
+    },
+    {
+      "epoch": 0.8097331240188383,
+      "grad_norm": 2.1873810909435294,
+      "learning_rate": 1.8394904050078943e-06,
+      "loss": 0.8909,
+      "step": 7737
+    },
+    {
+      "epoch": 0.8098377812663526,
+      "grad_norm": 2.064239458078798,
+      "learning_rate": 1.8375316835544775e-06,
+      "loss": 0.7535,
+      "step": 7738
+    },
+    {
+      "epoch": 0.809942438513867,
+      "grad_norm": 1.9681755929693125,
+      "learning_rate": 1.8355738999852157e-06,
+      "loss": 0.8752,
+      "step": 7739
+    },
+    {
+      "epoch": 0.8100470957613815,
+      "grad_norm": 1.9419726368546049,
+      "learning_rate": 1.8336170545250653e-06,
+      "loss": 0.9044,
+      "step": 7740
+    },
+    {
+      "epoch": 0.8101517530088959,
+      "grad_norm": 2.1087815522447033,
+      "learning_rate": 1.8316611473988678e-06,
+      "loss": 0.8237,
+      "step": 7741
+    },
+    {
+      "epoch": 0.8102564102564103,
+      "grad_norm": 2.078852602883871,
+      "learning_rate": 1.8297061788313652e-06,
+      "loss": 0.843,
+      "step": 7742
+    },
+    {
+      "epoch": 0.8103610675039247,
+      "grad_norm": 1.951856666967445,
+      "learning_rate": 1.827752149047185e-06,
+      "loss": 0.9578,
+      "step": 7743
+    },
+    {
+      "epoch": 0.8104657247514391,
+      "grad_norm": 1.7953444494690902,
+      "learning_rate": 1.8257990582708508e-06,
+      "loss": 0.7784,
+      "step": 7744
+    },
+    {
+      "epoch": 0.8105703819989535,
+      "grad_norm": 1.9250641329953944,
+      "learning_rate": 1.8238469067267728e-06,
+      "loss": 0.9653,
+      "step": 7745
+    },
+    {
+      "epoch": 0.8106750392464678,
+      "grad_norm": 1.8425811745515865,
+      "learning_rate": 1.8218956946392597e-06,
+      "loss": 0.8461,
+      "step": 7746
+    },
+    {
+      "epoch": 0.8107796964939822,
+      "grad_norm": 1.6927128485168828,
+      "learning_rate": 1.819945422232513e-06,
+      "loss": 0.8375,
+      "step": 7747
+    },
+    {
+      "epoch": 0.8108843537414966,
+      "grad_norm": 1.96120471071044,
+      "learning_rate": 1.8179960897306192e-06,
+      "loss": 0.6927,
+      "step": 7748
+    },
+    {
+      "epoch": 0.810989010989011,
+      "grad_norm": 1.8908478940115092,
+      "learning_rate": 1.8160476973575624e-06,
+      "loss": 0.9127,
+      "step": 7749
+    },
+    {
+      "epoch": 0.8110936682365254,
+      "grad_norm": 2.3740896100862194,
+      "learning_rate": 1.8141002453372126e-06,
+      "loss": 0.8577,
+      "step": 7750
+    },
+    {
+      "epoch": 0.8111983254840398,
+      "grad_norm": 1.9405102048474503,
+      "learning_rate": 1.8121537338933425e-06,
+      "loss": 0.8621,
+      "step": 7751
+    },
+    {
+      "epoch": 0.8113029827315542,
+      "grad_norm": 2.101087698006679,
+      "learning_rate": 1.8102081632496061e-06,
+      "loss": 0.9008,
+      "step": 7752
+    },
+    {
+      "epoch": 0.8114076399790685,
+      "grad_norm": 2.109664241764611,
+      "learning_rate": 1.808263533629554e-06,
+      "loss": 0.8728,
+      "step": 7753
+    },
+    {
+      "epoch": 0.8115122972265829,
+      "grad_norm": 1.957536899983253,
+      "learning_rate": 1.8063198452566288e-06,
+      "loss": 0.9817,
+      "step": 7754
+    },
+    {
+      "epoch": 0.8116169544740973,
+      "grad_norm": 2.29053423643524,
+      "learning_rate": 1.804377098354161e-06,
+      "loss": 0.8997,
+      "step": 7755
+    },
+    {
+      "epoch": 0.8117216117216117,
+      "grad_norm": 1.9515520676387297,
+      "learning_rate": 1.8024352931453791e-06,
+      "loss": 0.9398,
+      "step": 7756
+    },
+    {
+      "epoch": 0.8118262689691261,
+      "grad_norm": 2.2052154737151164,
+      "learning_rate": 1.8004944298534022e-06,
+      "loss": 0.9776,
+      "step": 7757
+    },
+    {
+      "epoch": 0.8119309262166405,
+      "grad_norm": 1.8460361778888326,
+      "learning_rate": 1.798554508701238e-06,
+      "loss": 0.7619,
+      "step": 7758
+    },
+    {
+      "epoch": 0.8120355834641549,
+      "grad_norm": 2.0564089124118703,
+      "learning_rate": 1.796615529911786e-06,
+      "loss": 0.9962,
+      "step": 7759
+    },
+    {
+      "epoch": 0.8121402407116692,
+      "grad_norm": 1.5242392257544974,
+      "learning_rate": 1.7946774937078383e-06,
+      "loss": 0.7455,
+      "step": 7760
+    },
+    {
+      "epoch": 0.8122448979591836,
+      "grad_norm": 1.9373147411960117,
+      "learning_rate": 1.792740400312082e-06,
+      "loss": 0.9294,
+      "step": 7761
+    },
+    {
+      "epoch": 0.812349555206698,
+      "grad_norm": 2.253211453976033,
+      "learning_rate": 1.7908042499470924e-06,
+      "loss": 1.0895,
+      "step": 7762
+    },
+    {
+      "epoch": 0.8124542124542125,
+      "grad_norm": 2.010874372372902,
+      "learning_rate": 1.788869042835335e-06,
+      "loss": 0.7848,
+      "step": 7763
+    },
+    {
+      "epoch": 0.8125588697017269,
+      "grad_norm": 2.1110812558999026,
+      "learning_rate": 1.7869347791991686e-06,
+      "loss": 0.8749,
+      "step": 7764
+    },
+    {
+      "epoch": 0.8126635269492413,
+      "grad_norm": 2.0790033933698506,
+      "learning_rate": 1.7850014592608457e-06,
+      "loss": 0.8357,
+      "step": 7765
+    },
+    {
+      "epoch": 0.8127681841967557,
+      "grad_norm": 2.2764948530690794,
+      "learning_rate": 1.7830690832425114e-06,
+      "loss": 0.85,
+      "step": 7766
+    },
+    {
+      "epoch": 0.81287284144427,
+      "grad_norm": 2.0246844811622617,
+      "learning_rate": 1.781137651366196e-06,
+      "loss": 0.9454,
+      "step": 7767
+    },
+    {
+      "epoch": 0.8129774986917844,
+      "grad_norm": 2.062713254460875,
+      "learning_rate": 1.7792071638538267e-06,
+      "loss": 0.914,
+      "step": 7768
+    },
+    {
+      "epoch": 0.8130821559392988,
+      "grad_norm": 2.060831788514829,
+      "learning_rate": 1.777277620927217e-06,
+      "loss": 0.8909,
+      "step": 7769
+    },
+    {
+      "epoch": 0.8131868131868132,
+      "grad_norm": 2.1960138141055037,
+      "learning_rate": 1.7753490228080795e-06,
+      "loss": 0.8942,
+      "step": 7770
+    },
+    {
+      "epoch": 0.8132914704343276,
+      "grad_norm": 2.019766608559704,
+      "learning_rate": 1.7734213697180137e-06,
+      "loss": 0.7633,
+      "step": 7771
+    },
+    {
+      "epoch": 0.813396127681842,
+      "grad_norm": 1.9474343159482823,
+      "learning_rate": 1.7714946618785056e-06,
+      "loss": 0.8488,
+      "step": 7772
+    },
+    {
+      "epoch": 0.8135007849293564,
+      "grad_norm": 2.1858861212932004,
+      "learning_rate": 1.769568899510945e-06,
+      "loss": 0.8966,
+      "step": 7773
+    },
+    {
+      "epoch": 0.8136054421768707,
+      "grad_norm": 1.8619826191371032,
+      "learning_rate": 1.7676440828366026e-06,
+      "loss": 0.849,
+      "step": 7774
+    },
+    {
+      "epoch": 0.8137100994243851,
+      "grad_norm": 1.882083312231594,
+      "learning_rate": 1.7657202120766414e-06,
+      "loss": 0.851,
+      "step": 7775
+    },
+    {
+      "epoch": 0.8138147566718995,
+      "grad_norm": 2.0412076045520138,
+      "learning_rate": 1.7637972874521237e-06,
+      "loss": 0.711,
+      "step": 7776
+    },
+    {
+      "epoch": 0.8139194139194139,
+      "grad_norm": 2.0941706761569145,
+      "learning_rate": 1.7618753091839924e-06,
+      "loss": 0.904,
+      "step": 7777
+    },
+    {
+      "epoch": 0.8140240711669283,
+      "grad_norm": 2.070231862068319,
+      "learning_rate": 1.75995427749309e-06,
+      "loss": 0.9615,
+      "step": 7778
+    },
+    {
+      "epoch": 0.8141287284144427,
+      "grad_norm": 1.9533497638106823,
+      "learning_rate": 1.758034192600142e-06,
+      "loss": 0.7906,
+      "step": 7779
+    },
+    {
+      "epoch": 0.8142333856619571,
+      "grad_norm": 1.8061715691464177,
+      "learning_rate": 1.7561150547257766e-06,
+      "loss": 0.9816,
+      "step": 7780
+    },
+    {
+      "epoch": 0.8143380429094714,
+      "grad_norm": 1.9499562486471225,
+      "learning_rate": 1.7541968640905005e-06,
+      "loss": 0.8275,
+      "step": 7781
+    },
+    {
+      "epoch": 0.8144427001569858,
+      "grad_norm": 1.9941853271497223,
+      "learning_rate": 1.7522796209147241e-06,
+      "loss": 1.0171,
+      "step": 7782
+    },
+    {
+      "epoch": 0.8145473574045002,
+      "grad_norm": 2.016835021467956,
+      "learning_rate": 1.750363325418738e-06,
+      "loss": 0.9194,
+      "step": 7783
+    },
+    {
+      "epoch": 0.8146520146520146,
+      "grad_norm": 2.349001003678539,
+      "learning_rate": 1.7484479778227281e-06,
+      "loss": 0.8795,
+      "step": 7784
+    },
+    {
+      "epoch": 0.814756671899529,
+      "grad_norm": 2.815561540260484,
+      "learning_rate": 1.7465335783467753e-06,
+      "loss": 0.8893,
+      "step": 7785
+    },
+    {
+      "epoch": 0.8148613291470435,
+      "grad_norm": 1.856879993868537,
+      "learning_rate": 1.7446201272108454e-06,
+      "loss": 0.8485,
+      "step": 7786
+    },
+    {
+      "epoch": 0.8149659863945579,
+      "grad_norm": 1.6972040689665617,
+      "learning_rate": 1.7427076246347984e-06,
+      "loss": 0.8006,
+      "step": 7787
+    },
+    {
+      "epoch": 0.8150706436420723,
+      "grad_norm": 2.4050270886241174,
+      "learning_rate": 1.7407960708383852e-06,
+      "loss": 0.8119,
+      "step": 7788
+    },
+    {
+      "epoch": 0.8151753008895866,
+      "grad_norm": 2.0596208441064747,
+      "learning_rate": 1.7388854660412435e-06,
+      "loss": 0.8787,
+      "step": 7789
+    },
+    {
+      "epoch": 0.815279958137101,
+      "grad_norm": 2.6144498345375675,
+      "learning_rate": 1.7369758104629098e-06,
+      "loss": 1.0528,
+      "step": 7790
+    },
+    {
+      "epoch": 0.8153846153846154,
+      "grad_norm": 1.8301486855272067,
+      "learning_rate": 1.7350671043228072e-06,
+      "loss": 0.8948,
+      "step": 7791
+    },
+    {
+      "epoch": 0.8154892726321298,
+      "grad_norm": 2.2092777762924656,
+      "learning_rate": 1.7331593478402508e-06,
+      "loss": 0.9019,
+      "step": 7792
+    },
+    {
+      "epoch": 0.8155939298796442,
+      "grad_norm": 1.931054386839299,
+      "learning_rate": 1.7312525412344428e-06,
+      "loss": 0.8937,
+      "step": 7793
+    },
+    {
+      "epoch": 0.8156985871271586,
+      "grad_norm": 2.0088421049144327,
+      "learning_rate": 1.7293466847244788e-06,
+      "loss": 0.7165,
+      "step": 7794
+    },
+    {
+      "epoch": 0.815803244374673,
+      "grad_norm": 2.064343587455126,
+      "learning_rate": 1.7274417785293485e-06,
+      "loss": 0.7951,
+      "step": 7795
+    },
+    {
+      "epoch": 0.8159079016221873,
+      "grad_norm": 1.9979352317040366,
+      "learning_rate": 1.7255378228679286e-06,
+      "loss": 0.8228,
+      "step": 7796
+    },
+    {
+      "epoch": 0.8160125588697017,
+      "grad_norm": 2.257708751650851,
+      "learning_rate": 1.7236348179589869e-06,
+      "loss": 0.8863,
+      "step": 7797
+    },
+    {
+      "epoch": 0.8161172161172161,
+      "grad_norm": 2.518173481553307,
+      "learning_rate": 1.72173276402118e-06,
+      "loss": 0.8981,
+      "step": 7798
+    },
+    {
+      "epoch": 0.8162218733647305,
+      "grad_norm": 1.9462944845738346,
+      "learning_rate": 1.7198316612730603e-06,
+      "loss": 0.9347,
+      "step": 7799
+    },
+    {
+      "epoch": 0.8163265306122449,
+      "grad_norm": 2.595705625274348,
+      "learning_rate": 1.7179315099330707e-06,
+      "loss": 0.966,
+      "step": 7800
+    },
+    {
+      "epoch": 0.8164311878597593,
+      "grad_norm": 1.977122273036907,
+      "learning_rate": 1.7160323102195408e-06,
+      "loss": 0.8375,
+      "step": 7801
+    },
+    {
+      "epoch": 0.8165358451072737,
+      "grad_norm": 2.1978666809470386,
+      "learning_rate": 1.7141340623506908e-06,
+      "loss": 0.8184,
+      "step": 7802
+    },
+    {
+      "epoch": 0.816640502354788,
+      "grad_norm": 2.1815414177103523,
+      "learning_rate": 1.7122367665446327e-06,
+      "loss": 0.8825,
+      "step": 7803
+    },
+    {
+      "epoch": 0.8167451596023024,
+      "grad_norm": 1.547128969295642,
+      "learning_rate": 1.7103404230193733e-06,
+      "loss": 0.6754,
+      "step": 7804
+    },
+    {
+      "epoch": 0.8168498168498168,
+      "grad_norm": 1.6815037025349835,
+      "learning_rate": 1.7084450319928037e-06,
+      "loss": 0.6787,
+      "step": 7805
+    },
+    {
+      "epoch": 0.8169544740973312,
+      "grad_norm": 1.959387083414091,
+      "learning_rate": 1.7065505936827097e-06,
+      "loss": 0.972,
+      "step": 7806
+    },
+    {
+      "epoch": 0.8170591313448456,
+      "grad_norm": 2.3459581609655227,
+      "learning_rate": 1.704657108306762e-06,
+      "loss": 0.7988,
+      "step": 7807
+    },
+    {
+      "epoch": 0.81716378859236,
+      "grad_norm": 2.8102944684349676,
+      "learning_rate": 1.7027645760825284e-06,
+      "loss": 0.926,
+      "step": 7808
+    },
+    {
+      "epoch": 0.8172684458398745,
+      "grad_norm": 1.7141239024853219,
+      "learning_rate": 1.7008729972274685e-06,
+      "loss": 0.8889,
+      "step": 7809
+    },
+    {
+      "epoch": 0.8173731030873888,
+      "grad_norm": 1.9586989744589782,
+      "learning_rate": 1.698982371958925e-06,
+      "loss": 0.9195,
+      "step": 7810
+    },
+    {
+      "epoch": 0.8174777603349032,
+      "grad_norm": 2.13663131703756,
+      "learning_rate": 1.697092700494135e-06,
+      "loss": 1.009,
+      "step": 7811
+    },
+    {
+      "epoch": 0.8175824175824176,
+      "grad_norm": 1.9020925116780771,
+      "learning_rate": 1.6952039830502253e-06,
+      "loss": 0.8457,
+      "step": 7812
+    },
+    {
+      "epoch": 0.817687074829932,
+      "grad_norm": 1.9607266851857876,
+      "learning_rate": 1.693316219844211e-06,
+      "loss": 0.8744,
+      "step": 7813
+    },
+    {
+      "epoch": 0.8177917320774464,
+      "grad_norm": 2.1863791159608077,
+      "learning_rate": 1.6914294110930052e-06,
+      "loss": 0.8709,
+      "step": 7814
+    },
+    {
+      "epoch": 0.8178963893249608,
+      "grad_norm": 1.918146066711303,
+      "learning_rate": 1.6895435570134032e-06,
+      "loss": 0.8755,
+      "step": 7815
+    },
+    {
+      "epoch": 0.8180010465724752,
+      "grad_norm": 2.3464466253166925,
+      "learning_rate": 1.687658657822091e-06,
+      "loss": 0.8249,
+      "step": 7816
+    },
+    {
+      "epoch": 0.8181057038199895,
+      "grad_norm": 2.094038559858304,
+      "learning_rate": 1.6857747137356529e-06,
+      "loss": 0.8999,
+      "step": 7817
+    },
+    {
+      "epoch": 0.8182103610675039,
+      "grad_norm": 1.969446534885216,
+      "learning_rate": 1.683891724970551e-06,
+      "loss": 0.7884,
+      "step": 7818
+    },
+    {
+      "epoch": 0.8183150183150183,
+      "grad_norm": 1.9254682173238997,
+      "learning_rate": 1.6820096917431527e-06,
+      "loss": 0.8815,
+      "step": 7819
+    },
+    {
+      "epoch": 0.8184196755625327,
+      "grad_norm": 1.705412210976725,
+      "learning_rate": 1.6801286142697015e-06,
+      "loss": 0.7476,
+      "step": 7820
+    },
+    {
+      "epoch": 0.8185243328100471,
+      "grad_norm": 1.8741055231594557,
+      "learning_rate": 1.67824849276634e-06,
+      "loss": 0.8466,
+      "step": 7821
+    },
+    {
+      "epoch": 0.8186289900575615,
+      "grad_norm": 1.9978999965045052,
+      "learning_rate": 1.6763693274490932e-06,
+      "loss": 0.8856,
+      "step": 7822
+    },
+    {
+      "epoch": 0.8187336473050759,
+      "grad_norm": 2.1962770410951977,
+      "learning_rate": 1.6744911185338874e-06,
+      "loss": 0.8014,
+      "step": 7823
+    },
+    {
+      "epoch": 0.8188383045525902,
+      "grad_norm": 2.5133884425384214,
+      "learning_rate": 1.6726138662365276e-06,
+      "loss": 0.9689,
+      "step": 7824
+    },
+    {
+      "epoch": 0.8189429618001046,
+      "grad_norm": 2.0611491066129197,
+      "learning_rate": 1.6707375707727169e-06,
+      "loss": 0.7664,
+      "step": 7825
+    },
+    {
+      "epoch": 0.819047619047619,
+      "grad_norm": 1.9924960787409718,
+      "learning_rate": 1.6688622323580461e-06,
+      "loss": 0.7497,
+      "step": 7826
+    },
+    {
+      "epoch": 0.8191522762951334,
+      "grad_norm": 2.112698383860386,
+      "learning_rate": 1.6669878512079906e-06,
+      "loss": 0.8948,
+      "step": 7827
+    },
+    {
+      "epoch": 0.8192569335426478,
+      "grad_norm": 2.4115852205279804,
+      "learning_rate": 1.6651144275379261e-06,
+      "loss": 0.9912,
+      "step": 7828
+    },
+    {
+      "epoch": 0.8193615907901622,
+      "grad_norm": 2.141183062004032,
+      "learning_rate": 1.6632419615631112e-06,
+      "loss": 0.8449,
+      "step": 7829
+    },
+    {
+      "epoch": 0.8194662480376766,
+      "grad_norm": 2.3622806603126696,
+      "learning_rate": 1.661370453498694e-06,
+      "loss": 0.9455,
+      "step": 7830
+    },
+    {
+      "epoch": 0.819570905285191,
+      "grad_norm": 2.161674209789048,
+      "learning_rate": 1.659499903559717e-06,
+      "loss": 0.8994,
+      "step": 7831
+    },
+    {
+      "epoch": 0.8196755625327053,
+      "grad_norm": 2.2133213841412993,
+      "learning_rate": 1.6576303119611058e-06,
+      "loss": 0.9306,
+      "step": 7832
+    },
+    {
+      "epoch": 0.8197802197802198,
+      "grad_norm": 2.2348885981045226,
+      "learning_rate": 1.6557616789176844e-06,
+      "loss": 0.7268,
+      "step": 7833
+    },
+    {
+      "epoch": 0.8198848770277342,
+      "grad_norm": 1.9426173933907054,
+      "learning_rate": 1.653894004644163e-06,
+      "loss": 0.8147,
+      "step": 7834
+    },
+    {
+      "epoch": 0.8199895342752486,
+      "grad_norm": 2.4120961582899416,
+      "learning_rate": 1.6520272893551404e-06,
+      "loss": 0.8665,
+      "step": 7835
+    },
+    {
+      "epoch": 0.820094191522763,
+      "grad_norm": 1.7715546995709663,
+      "learning_rate": 1.6501615332651054e-06,
+      "loss": 0.8394,
+      "step": 7836
+    },
+    {
+      "epoch": 0.8201988487702774,
+      "grad_norm": 2.1557330830514867,
+      "learning_rate": 1.648296736588435e-06,
+      "loss": 0.82,
+      "step": 7837
+    },
+    {
+      "epoch": 0.8203035060177918,
+      "grad_norm": 1.9987706186923369,
+      "learning_rate": 1.6464328995394018e-06,
+      "loss": 0.8204,
+      "step": 7838
+    },
+    {
+      "epoch": 0.8204081632653061,
+      "grad_norm": 2.057663150677782,
+      "learning_rate": 1.6445700223321637e-06,
+      "loss": 0.8971,
+      "step": 7839
+    },
+    {
+      "epoch": 0.8205128205128205,
+      "grad_norm": 1.7641921997749053,
+      "learning_rate": 1.642708105180768e-06,
+      "loss": 0.8375,
+      "step": 7840
+    },
+    {
+      "epoch": 0.8206174777603349,
+      "grad_norm": 2.0728485301068957,
+      "learning_rate": 1.6408471482991506e-06,
+      "loss": 0.8569,
+      "step": 7841
+    },
+    {
+      "epoch": 0.8207221350078493,
+      "grad_norm": 2.036559853741599,
+      "learning_rate": 1.6389871519011414e-06,
+      "loss": 0.8177,
+      "step": 7842
+    },
+    {
+      "epoch": 0.8208267922553637,
+      "grad_norm": 2.1297430764924368,
+      "learning_rate": 1.6371281162004604e-06,
+      "loss": 0.8486,
+      "step": 7843
+    },
+    {
+      "epoch": 0.8209314495028781,
+      "grad_norm": 1.920858753141409,
+      "learning_rate": 1.6352700414107113e-06,
+      "loss": 0.9678,
+      "step": 7844
+    },
+    {
+      "epoch": 0.8210361067503925,
+      "grad_norm": 1.977146348689036,
+      "learning_rate": 1.6334129277453935e-06,
+      "loss": 0.8985,
+      "step": 7845
+    },
+    {
+      "epoch": 0.8211407639979068,
+      "grad_norm": 2.220183521723425,
+      "learning_rate": 1.6315567754178896e-06,
+      "loss": 0.7621,
+      "step": 7846
+    },
+    {
+      "epoch": 0.8212454212454212,
+      "grad_norm": 2.1123357019169626,
+      "learning_rate": 1.6297015846414755e-06,
+      "loss": 0.7536,
+      "step": 7847
+    },
+    {
+      "epoch": 0.8213500784929356,
+      "grad_norm": 2.353601092727366,
+      "learning_rate": 1.6278473556293195e-06,
+      "loss": 0.8775,
+      "step": 7848
+    },
+    {
+      "epoch": 0.82145473574045,
+      "grad_norm": 2.14734632695945,
+      "learning_rate": 1.6259940885944737e-06,
+      "loss": 0.8968,
+      "step": 7849
+    },
+    {
+      "epoch": 0.8215593929879644,
+      "grad_norm": 2.0771918291151557,
+      "learning_rate": 1.6241417837498818e-06,
+      "loss": 0.9364,
+      "step": 7850
+    },
+    {
+      "epoch": 0.8216640502354788,
+      "grad_norm": 2.0500743901250504,
+      "learning_rate": 1.6222904413083806e-06,
+      "loss": 0.8666,
+      "step": 7851
+    },
+    {
+      "epoch": 0.8217687074829932,
+      "grad_norm": 2.0916815220393183,
+      "learning_rate": 1.6204400614826886e-06,
+      "loss": 0.7629,
+      "step": 7852
+    },
+    {
+      "epoch": 0.8218733647305075,
+      "grad_norm": 2.0764823972980597,
+      "learning_rate": 1.6185906444854227e-06,
+      "loss": 1.0247,
+      "step": 7853
+    },
+    {
+      "epoch": 0.8219780219780219,
+      "grad_norm": 1.832597003879775,
+      "learning_rate": 1.6167421905290837e-06,
+      "loss": 0.9094,
+      "step": 7854
+    },
+    {
+      "epoch": 0.8220826792255364,
+      "grad_norm": 2.0982201560289813,
+      "learning_rate": 1.6148946998260605e-06,
+      "loss": 1.0225,
+      "step": 7855
+    },
+    {
+      "epoch": 0.8221873364730508,
+      "grad_norm": 1.8679791707883626,
+      "learning_rate": 1.6130481725886327e-06,
+      "loss": 0.854,
+      "step": 7856
+    },
+    {
+      "epoch": 0.8222919937205652,
+      "grad_norm": 2.4377106859096997,
+      "learning_rate": 1.6112026090289744e-06,
+      "loss": 0.9012,
+      "step": 7857
+    },
+    {
+      "epoch": 0.8223966509680796,
+      "grad_norm": 1.9716656960681878,
+      "learning_rate": 1.609358009359142e-06,
+      "loss": 0.8272,
+      "step": 7858
+    },
+    {
+      "epoch": 0.822501308215594,
+      "grad_norm": 2.039719279072701,
+      "learning_rate": 1.6075143737910826e-06,
+      "loss": 0.7955,
+      "step": 7859
+    },
+    {
+      "epoch": 0.8226059654631083,
+      "grad_norm": 2.1817572238574945,
+      "learning_rate": 1.6056717025366364e-06,
+      "loss": 0.8247,
+      "step": 7860
+    },
+    {
+      "epoch": 0.8227106227106227,
+      "grad_norm": 1.9613810519827801,
+      "learning_rate": 1.6038299958075266e-06,
+      "loss": 0.922,
+      "step": 7861
+    },
+    {
+      "epoch": 0.8228152799581371,
+      "grad_norm": 1.964948398942913,
+      "learning_rate": 1.6019892538153725e-06,
+      "loss": 0.7781,
+      "step": 7862
+    },
+    {
+      "epoch": 0.8229199372056515,
+      "grad_norm": 2.13598752292871,
+      "learning_rate": 1.6001494767716786e-06,
+      "loss": 0.7924,
+      "step": 7863
+    },
+    {
+      "epoch": 0.8230245944531659,
+      "grad_norm": 2.204916092833003,
+      "learning_rate": 1.598310664887839e-06,
+      "loss": 1.0455,
+      "step": 7864
+    },
+    {
+      "epoch": 0.8231292517006803,
+      "grad_norm": 2.2186437702248085,
+      "learning_rate": 1.5964728183751343e-06,
+      "loss": 0.8428,
+      "step": 7865
+    },
+    {
+      "epoch": 0.8232339089481947,
+      "grad_norm": 2.478462994063437,
+      "learning_rate": 1.5946359374447374e-06,
+      "loss": 0.8116,
+      "step": 7866
+    },
+    {
+      "epoch": 0.823338566195709,
+      "grad_norm": 1.8826304139905394,
+      "learning_rate": 1.5928000223077123e-06,
+      "loss": 0.9264,
+      "step": 7867
+    },
+    {
+      "epoch": 0.8234432234432234,
+      "grad_norm": 2.166154757400412,
+      "learning_rate": 1.5909650731750048e-06,
+      "loss": 0.8777,
+      "step": 7868
+    },
+    {
+      "epoch": 0.8235478806907378,
+      "grad_norm": 2.6409362779319325,
+      "learning_rate": 1.5891310902574608e-06,
+      "loss": 0.9525,
+      "step": 7869
+    },
+    {
+      "epoch": 0.8236525379382522,
+      "grad_norm": 2.0823720652767244,
+      "learning_rate": 1.587298073765805e-06,
+      "loss": 0.9399,
+      "step": 7870
+    },
+    {
+      "epoch": 0.8237571951857666,
+      "grad_norm": 2.147673418143053,
+      "learning_rate": 1.5854660239106523e-06,
+      "loss": 0.9762,
+      "step": 7871
+    },
+    {
+      "epoch": 0.823861852433281,
+      "grad_norm": 2.000527006452208,
+      "learning_rate": 1.5836349409025143e-06,
+      "loss": 0.7641,
+      "step": 7872
+    },
+    {
+      "epoch": 0.8239665096807954,
+      "grad_norm": 2.4818231228477012,
+      "learning_rate": 1.5818048249517825e-06,
+      "loss": 1.0183,
+      "step": 7873
+    },
+    {
+      "epoch": 0.8240711669283098,
+      "grad_norm": 2.0501868554721265,
+      "learning_rate": 1.5799756762687424e-06,
+      "loss": 0.8242,
+      "step": 7874
+    },
+    {
+      "epoch": 0.8241758241758241,
+      "grad_norm": 2.3433303375111545,
+      "learning_rate": 1.5781474950635633e-06,
+      "loss": 0.8638,
+      "step": 7875
+    },
+    {
+      "epoch": 0.8242804814233385,
+      "grad_norm": 2.0395663531419577,
+      "learning_rate": 1.5763202815463109e-06,
+      "loss": 0.8088,
+      "step": 7876
+    },
+    {
+      "epoch": 0.824385138670853,
+      "grad_norm": 2.037603786377761,
+      "learning_rate": 1.5744940359269357e-06,
+      "loss": 0.9229,
+      "step": 7877
+    },
+    {
+      "epoch": 0.8244897959183674,
+      "grad_norm": 1.8506973012344865,
+      "learning_rate": 1.5726687584152778e-06,
+      "loss": 0.8673,
+      "step": 7878
+    },
+    {
+      "epoch": 0.8245944531658818,
+      "grad_norm": 2.288058667834201,
+      "learning_rate": 1.5708444492210617e-06,
+      "loss": 0.7638,
+      "step": 7879
+    },
+    {
+      "epoch": 0.8246991104133962,
+      "grad_norm": 2.1522314144552337,
+      "learning_rate": 1.5690211085539054e-06,
+      "loss": 0.8846,
+      "step": 7880
+    },
+    {
+      "epoch": 0.8248037676609106,
+      "grad_norm": 1.6233838085258567,
+      "learning_rate": 1.5671987366233166e-06,
+      "loss": 0.7405,
+      "step": 7881
+    },
+    {
+      "epoch": 0.8249084249084249,
+      "grad_norm": 2.0740723600378566,
+      "learning_rate": 1.565377333638688e-06,
+      "loss": 0.9315,
+      "step": 7882
+    },
+    {
+      "epoch": 0.8250130821559393,
+      "grad_norm": 2.2960235649753047,
+      "learning_rate": 1.563556899809302e-06,
+      "loss": 0.9879,
+      "step": 7883
+    },
+    {
+      "epoch": 0.8251177394034537,
+      "grad_norm": 1.7103999518415944,
+      "learning_rate": 1.5617374353443316e-06,
+      "loss": 0.7416,
+      "step": 7884
+    },
+    {
+      "epoch": 0.8252223966509681,
+      "grad_norm": 2.5050104670986597,
+      "learning_rate": 1.5599189404528337e-06,
+      "loss": 0.8611,
+      "step": 7885
+    },
+    {
+      "epoch": 0.8253270538984825,
+      "grad_norm": 2.612738780164187,
+      "learning_rate": 1.5581014153437578e-06,
+      "loss": 0.9615,
+      "step": 7886
+    },
+    {
+      "epoch": 0.8254317111459969,
+      "grad_norm": 1.8339034152233065,
+      "learning_rate": 1.5562848602259473e-06,
+      "loss": 0.9589,
+      "step": 7887
+    },
+    {
+      "epoch": 0.8255363683935113,
+      "grad_norm": 1.9861260916511432,
+      "learning_rate": 1.5544692753081226e-06,
+      "loss": 0.8367,
+      "step": 7888
+    },
+    {
+      "epoch": 0.8256410256410256,
+      "grad_norm": 2.019172682002892,
+      "learning_rate": 1.552654660798899e-06,
+      "loss": 0.8907,
+      "step": 7889
+    },
+    {
+      "epoch": 0.82574568288854,
+      "grad_norm": 2.266069990170753,
+      "learning_rate": 1.5508410169067767e-06,
+      "loss": 0.7255,
+      "step": 7890
+    },
+    {
+      "epoch": 0.8258503401360544,
+      "grad_norm": 1.7448652237490383,
+      "learning_rate": 1.5490283438401533e-06,
+      "loss": 0.8038,
+      "step": 7891
+    },
+    {
+      "epoch": 0.8259549973835688,
+      "grad_norm": 2.236043533397473,
+      "learning_rate": 1.5472166418073042e-06,
+      "loss": 0.8716,
+      "step": 7892
+    },
+    {
+      "epoch": 0.8260596546310832,
+      "grad_norm": 1.7295344167594429,
+      "learning_rate": 1.5454059110163977e-06,
+      "loss": 0.7992,
+      "step": 7893
+    },
+    {
+      "epoch": 0.8261643118785976,
+      "grad_norm": 2.114455364406809,
+      "learning_rate": 1.5435961516754904e-06,
+      "loss": 0.8744,
+      "step": 7894
+    },
+    {
+      "epoch": 0.826268969126112,
+      "grad_norm": 1.7900260474592373,
+      "learning_rate": 1.5417873639925274e-06,
+      "loss": 0.8994,
+      "step": 7895
+    },
+    {
+      "epoch": 0.8263736263736263,
+      "grad_norm": 2.0911021977650135,
+      "learning_rate": 1.5399795481753454e-06,
+      "loss": 0.8086,
+      "step": 7896
+    },
+    {
+      "epoch": 0.8264782836211407,
+      "grad_norm": 2.1517967909735547,
+      "learning_rate": 1.5381727044316641e-06,
+      "loss": 0.9165,
+      "step": 7897
+    },
+    {
+      "epoch": 0.8265829408686551,
+      "grad_norm": 1.864956322145815,
+      "learning_rate": 1.536366832969093e-06,
+      "loss": 0.9554,
+      "step": 7898
+    },
+    {
+      "epoch": 0.8266875981161695,
+      "grad_norm": 2.0999023389103915,
+      "learning_rate": 1.5345619339951267e-06,
+      "loss": 0.884,
+      "step": 7899
+    },
+    {
+      "epoch": 0.826792255363684,
+      "grad_norm": 2.3031128045450577,
+      "learning_rate": 1.5327580077171589e-06,
+      "loss": 0.7399,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8268969126111984,
+      "grad_norm": 1.8424502804616814,
+      "learning_rate": 1.5309550543424612e-06,
+      "loss": 0.7417,
+      "step": 7901
+    },
+    {
+      "epoch": 0.8270015698587128,
+      "grad_norm": 1.9163642835084853,
+      "learning_rate": 1.5291530740781923e-06,
+      "loss": 0.751,
+      "step": 7902
+    },
+    {
+      "epoch": 0.827106227106227,
+      "grad_norm": 1.87018617560281,
+      "learning_rate": 1.5273520671314113e-06,
+      "loss": 0.8395,
+      "step": 7903
+    },
+    {
+      "epoch": 0.8272108843537415,
+      "grad_norm": 1.9411831340122883,
+      "learning_rate": 1.5255520337090534e-06,
+      "loss": 0.8028,
+      "step": 7904
+    },
+    {
+      "epoch": 0.8273155416012559,
+      "grad_norm": 2.4058082707488113,
+      "learning_rate": 1.5237529740179436e-06,
+      "loss": 0.9005,
+      "step": 7905
+    },
+    {
+      "epoch": 0.8274201988487703,
+      "grad_norm": 2.1409745262541287,
+      "learning_rate": 1.5219548882648027e-06,
+      "loss": 0.9353,
+      "step": 7906
+    },
+    {
+      "epoch": 0.8275248560962847,
+      "grad_norm": 1.9577882525040946,
+      "learning_rate": 1.5201577766562314e-06,
+      "loss": 0.8448,
+      "step": 7907
+    },
+    {
+      "epoch": 0.8276295133437991,
+      "grad_norm": 2.0318542534635275,
+      "learning_rate": 1.5183616393987232e-06,
+      "loss": 0.8689,
+      "step": 7908
+    },
+    {
+      "epoch": 0.8277341705913135,
+      "grad_norm": 2.301362720034534,
+      "learning_rate": 1.5165664766986533e-06,
+      "loss": 0.9667,
+      "step": 7909
+    },
+    {
+      "epoch": 0.8278388278388278,
+      "grad_norm": 1.886895169603913,
+      "learning_rate": 1.5147722887622961e-06,
+      "loss": 0.8658,
+      "step": 7910
+    },
+    {
+      "epoch": 0.8279434850863422,
+      "grad_norm": 2.46318780324625,
+      "learning_rate": 1.5129790757958008e-06,
+      "loss": 0.8375,
+      "step": 7911
+    },
+    {
+      "epoch": 0.8280481423338566,
+      "grad_norm": 1.8771844183637372,
+      "learning_rate": 1.5111868380052185e-06,
+      "loss": 0.7701,
+      "step": 7912
+    },
+    {
+      "epoch": 0.828152799581371,
+      "grad_norm": 2.070606556582261,
+      "learning_rate": 1.5093955755964762e-06,
+      "loss": 1.0035,
+      "step": 7913
+    },
+    {
+      "epoch": 0.8282574568288854,
+      "grad_norm": 2.201869626898473,
+      "learning_rate": 1.5076052887753923e-06,
+      "loss": 0.904,
+      "step": 7914
+    },
+    {
+      "epoch": 0.8283621140763998,
+      "grad_norm": 1.996397707124482,
+      "learning_rate": 1.5058159777476788e-06,
+      "loss": 0.7816,
+      "step": 7915
+    },
+    {
+      "epoch": 0.8284667713239142,
+      "grad_norm": 2.0077562325381133,
+      "learning_rate": 1.5040276427189294e-06,
+      "loss": 0.9121,
+      "step": 7916
+    },
+    {
+      "epoch": 0.8285714285714286,
+      "grad_norm": 1.9941521295888824,
+      "learning_rate": 1.502240283894626e-06,
+      "loss": 0.8295,
+      "step": 7917
+    },
+    {
+      "epoch": 0.8286760858189429,
+      "grad_norm": 1.9915709085478066,
+      "learning_rate": 1.500453901480139e-06,
+      "loss": 0.8647,
+      "step": 7918
+    },
+    {
+      "epoch": 0.8287807430664573,
+      "grad_norm": 2.0000071542196145,
+      "learning_rate": 1.498668495680732e-06,
+      "loss": 0.9201,
+      "step": 7919
+    },
+    {
+      "epoch": 0.8288854003139717,
+      "grad_norm": 1.9756711232215005,
+      "learning_rate": 1.4968840667015462e-06,
+      "loss": 0.8459,
+      "step": 7920
+    },
+    {
+      "epoch": 0.8289900575614861,
+      "grad_norm": 1.8528329572690898,
+      "learning_rate": 1.4951006147476211e-06,
+      "loss": 0.7497,
+      "step": 7921
+    },
+    {
+      "epoch": 0.8290947148090005,
+      "grad_norm": 1.782034124098557,
+      "learning_rate": 1.493318140023876e-06,
+      "loss": 0.8792,
+      "step": 7922
+    },
+    {
+      "epoch": 0.829199372056515,
+      "grad_norm": 2.106613209839586,
+      "learning_rate": 1.4915366427351219e-06,
+      "loss": 0.7935,
+      "step": 7923
+    },
+    {
+      "epoch": 0.8293040293040294,
+      "grad_norm": 2.1408753179094857,
+      "learning_rate": 1.4897561230860536e-06,
+      "loss": 1.0341,
+      "step": 7924
+    },
+    {
+      "epoch": 0.8294086865515437,
+      "grad_norm": 2.1278800514184324,
+      "learning_rate": 1.4879765812812607e-06,
+      "loss": 0.8901,
+      "step": 7925
+    },
+    {
+      "epoch": 0.8295133437990581,
+      "grad_norm": 2.1440190302189714,
+      "learning_rate": 1.4861980175252154e-06,
+      "loss": 0.898,
+      "step": 7926
+    },
+    {
+      "epoch": 0.8296180010465725,
+      "grad_norm": 2.0503918898516384,
+      "learning_rate": 1.4844204320222767e-06,
+      "loss": 0.8825,
+      "step": 7927
+    },
+    {
+      "epoch": 0.8297226582940869,
+      "grad_norm": 2.353937197043025,
+      "learning_rate": 1.4826438249766906e-06,
+      "loss": 0.8895,
+      "step": 7928
+    },
+    {
+      "epoch": 0.8298273155416013,
+      "grad_norm": 2.2495600621190586,
+      "learning_rate": 1.4808681965925954e-06,
+      "loss": 0.9815,
+      "step": 7929
+    },
+    {
+      "epoch": 0.8299319727891157,
+      "grad_norm": 2.1503657595480843,
+      "learning_rate": 1.4790935470740175e-06,
+      "loss": 0.9053,
+      "step": 7930
+    },
+    {
+      "epoch": 0.8300366300366301,
+      "grad_norm": 1.8861867201869633,
+      "learning_rate": 1.4773198766248642e-06,
+      "loss": 0.8636,
+      "step": 7931
+    },
+    {
+      "epoch": 0.8301412872841444,
+      "grad_norm": 2.1460191083705906,
+      "learning_rate": 1.4755471854489345e-06,
+      "loss": 1.0363,
+      "step": 7932
+    },
+    {
+      "epoch": 0.8302459445316588,
+      "grad_norm": 2.493276713397057,
+      "learning_rate": 1.4737754737499111e-06,
+      "loss": 0.8972,
+      "step": 7933
+    },
+    {
+      "epoch": 0.8303506017791732,
+      "grad_norm": 1.6858625797981937,
+      "learning_rate": 1.4720047417313731e-06,
+      "loss": 0.7886,
+      "step": 7934
+    },
+    {
+      "epoch": 0.8304552590266876,
+      "grad_norm": 1.8962906371419326,
+      "learning_rate": 1.4702349895967782e-06,
+      "loss": 0.9196,
+      "step": 7935
+    },
+    {
+      "epoch": 0.830559916274202,
+      "grad_norm": 1.9184229473202252,
+      "learning_rate": 1.4684662175494747e-06,
+      "loss": 0.8052,
+      "step": 7936
+    },
+    {
+      "epoch": 0.8306645735217164,
+      "grad_norm": 2.137568048466092,
+      "learning_rate": 1.4666984257926964e-06,
+      "loss": 0.7906,
+      "step": 7937
+    },
+    {
+      "epoch": 0.8307692307692308,
+      "grad_norm": 1.5889031763544768,
+      "learning_rate": 1.4649316145295668e-06,
+      "loss": 0.7409,
+      "step": 7938
+    },
+    {
+      "epoch": 0.8308738880167451,
+      "grad_norm": 1.8404031630051605,
+      "learning_rate": 1.4631657839630996e-06,
+      "loss": 0.7866,
+      "step": 7939
+    },
+    {
+      "epoch": 0.8309785452642595,
+      "grad_norm": 2.286617894950235,
+      "learning_rate": 1.46140093429619e-06,
+      "loss": 0.8095,
+      "step": 7940
+    },
+    {
+      "epoch": 0.8310832025117739,
+      "grad_norm": 2.05496477949122,
+      "learning_rate": 1.4596370657316229e-06,
+      "loss": 0.9575,
+      "step": 7941
+    },
+    {
+      "epoch": 0.8311878597592883,
+      "grad_norm": 2.427977961236888,
+      "learning_rate": 1.4578741784720697e-06,
+      "loss": 0.876,
+      "step": 7942
+    },
+    {
+      "epoch": 0.8312925170068027,
+      "grad_norm": 2.227236132950701,
+      "learning_rate": 1.4561122727200872e-06,
+      "loss": 0.7827,
+      "step": 7943
+    },
+    {
+      "epoch": 0.8313971742543171,
+      "grad_norm": 1.8480415224847573,
+      "learning_rate": 1.454351348678128e-06,
+      "loss": 0.9043,
+      "step": 7944
+    },
+    {
+      "epoch": 0.8315018315018315,
+      "grad_norm": 2.152502773769006,
+      "learning_rate": 1.4525914065485225e-06,
+      "loss": 0.824,
+      "step": 7945
+    },
+    {
+      "epoch": 0.8316064887493458,
+      "grad_norm": 1.9448080461974455,
+      "learning_rate": 1.4508324465334888e-06,
+      "loss": 0.8931,
+      "step": 7946
+    },
+    {
+      "epoch": 0.8317111459968602,
+      "grad_norm": 2.0868231871819733,
+      "learning_rate": 1.449074468835141e-06,
+      "loss": 0.8808,
+      "step": 7947
+    },
+    {
+      "epoch": 0.8318158032443747,
+      "grad_norm": 2.0813903646254786,
+      "learning_rate": 1.4473174736554685e-06,
+      "loss": 0.7902,
+      "step": 7948
+    },
+    {
+      "epoch": 0.8319204604918891,
+      "grad_norm": 2.123099741849126,
+      "learning_rate": 1.4455614611963599e-06,
+      "loss": 0.8902,
+      "step": 7949
+    },
+    {
+      "epoch": 0.8320251177394035,
+      "grad_norm": 1.9932955603081572,
+      "learning_rate": 1.44380643165958e-06,
+      "loss": 0.7715,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8321297749869179,
+      "grad_norm": 2.068162191946537,
+      "learning_rate": 1.4420523852467861e-06,
+      "loss": 0.8567,
+      "step": 7951
+    },
+    {
+      "epoch": 0.8322344322344323,
+      "grad_norm": 1.9814333811554212,
+      "learning_rate": 1.4402993221595197e-06,
+      "loss": 0.8771,
+      "step": 7952
+    },
+    {
+      "epoch": 0.8323390894819466,
+      "grad_norm": 2.2197393374158936,
+      "learning_rate": 1.4385472425992164e-06,
+      "loss": 0.8788,
+      "step": 7953
+    },
+    {
+      "epoch": 0.832443746729461,
+      "grad_norm": 2.375333340064998,
+      "learning_rate": 1.4367961467671875e-06,
+      "loss": 0.9018,
+      "step": 7954
+    },
+    {
+      "epoch": 0.8325484039769754,
+      "grad_norm": 1.9216141945810121,
+      "learning_rate": 1.4350460348646444e-06,
+      "loss": 0.8605,
+      "step": 7955
+    },
+    {
+      "epoch": 0.8326530612244898,
+      "grad_norm": 2.398938782927487,
+      "learning_rate": 1.4332969070926739e-06,
+      "loss": 0.8632,
+      "step": 7956
+    },
+    {
+      "epoch": 0.8327577184720042,
+      "grad_norm": 2.0555051893772056,
+      "learning_rate": 1.431548763652253e-06,
+      "loss": 0.9676,
+      "step": 7957
+    },
+    {
+      "epoch": 0.8328623757195186,
+      "grad_norm": 2.1581306202378276,
+      "learning_rate": 1.429801604744252e-06,
+      "loss": 0.9032,
+      "step": 7958
+    },
+    {
+      "epoch": 0.832967032967033,
+      "grad_norm": 1.897752103573272,
+      "learning_rate": 1.4280554305694205e-06,
+      "loss": 0.8139,
+      "step": 7959
+    },
+    {
+      "epoch": 0.8330716902145474,
+      "grad_norm": 1.9164586231571057,
+      "learning_rate": 1.426310241328397e-06,
+      "loss": 0.794,
+      "step": 7960
+    },
+    {
+      "epoch": 0.8331763474620617,
+      "grad_norm": 2.1626145846140323,
+      "learning_rate": 1.4245660372217062e-06,
+      "loss": 0.9293,
+      "step": 7961
+    },
+    {
+      "epoch": 0.8332810047095761,
+      "grad_norm": 2.1783332161398836,
+      "learning_rate": 1.4228228184497607e-06,
+      "loss": 0.8771,
+      "step": 7962
+    },
+    {
+      "epoch": 0.8333856619570905,
+      "grad_norm": 2.0795885143306276,
+      "learning_rate": 1.4210805852128617e-06,
+      "loss": 0.8529,
+      "step": 7963
+    },
+    {
+      "epoch": 0.8334903192046049,
+      "grad_norm": 2.5330157272730602,
+      "learning_rate": 1.4193393377111964e-06,
+      "loss": 0.8425,
+      "step": 7964
+    },
+    {
+      "epoch": 0.8335949764521193,
+      "grad_norm": 1.8622801123769794,
+      "learning_rate": 1.4175990761448355e-06,
+      "loss": 0.8042,
+      "step": 7965
+    },
+    {
+      "epoch": 0.8336996336996337,
+      "grad_norm": 1.8796275243042533,
+      "learning_rate": 1.4158598007137391e-06,
+      "loss": 0.7952,
+      "step": 7966
+    },
+    {
+      "epoch": 0.8338042909471481,
+      "grad_norm": 2.0693042166079056,
+      "learning_rate": 1.414121511617752e-06,
+      "loss": 0.8121,
+      "step": 7967
+    },
+    {
+      "epoch": 0.8339089481946624,
+      "grad_norm": 2.1022638373991014,
+      "learning_rate": 1.4123842090566108e-06,
+      "loss": 0.915,
+      "step": 7968
+    },
+    {
+      "epoch": 0.8340136054421768,
+      "grad_norm": 1.993990099910452,
+      "learning_rate": 1.4106478932299316e-06,
+      "loss": 0.8766,
+      "step": 7969
+    },
+    {
+      "epoch": 0.8341182626896912,
+      "grad_norm": 1.7756065930189895,
+      "learning_rate": 1.4089125643372237e-06,
+      "loss": 0.777,
+      "step": 7970
+    },
+    {
+      "epoch": 0.8342229199372057,
+      "grad_norm": 2.5496490825925906,
+      "learning_rate": 1.4071782225778751e-06,
+      "loss": 0.8664,
+      "step": 7971
+    },
+    {
+      "epoch": 0.8343275771847201,
+      "grad_norm": 2.2433472242170396,
+      "learning_rate": 1.405444868151168e-06,
+      "loss": 0.9331,
+      "step": 7972
+    },
+    {
+      "epoch": 0.8344322344322345,
+      "grad_norm": 1.7433595951078031,
+      "learning_rate": 1.4037125012562702e-06,
+      "loss": 0.7727,
+      "step": 7973
+    },
+    {
+      "epoch": 0.8345368916797489,
+      "grad_norm": 1.8722606546409677,
+      "learning_rate": 1.401981122092233e-06,
+      "loss": 0.9949,
+      "step": 7974
+    },
+    {
+      "epoch": 0.8346415489272632,
+      "grad_norm": 2.019791859336298,
+      "learning_rate": 1.400250730857996e-06,
+      "loss": 0.8875,
+      "step": 7975
+    },
+    {
+      "epoch": 0.8347462061747776,
+      "grad_norm": 1.984713578640695,
+      "learning_rate": 1.3985213277523802e-06,
+      "loss": 0.8334,
+      "step": 7976
+    },
+    {
+      "epoch": 0.834850863422292,
+      "grad_norm": 1.8860559224768174,
+      "learning_rate": 1.3967929129741032e-06,
+      "loss": 0.8732,
+      "step": 7977
+    },
+    {
+      "epoch": 0.8349555206698064,
+      "grad_norm": 2.4921343619909906,
+      "learning_rate": 1.3950654867217605e-06,
+      "loss": 0.8214,
+      "step": 7978
+    },
+    {
+      "epoch": 0.8350601779173208,
+      "grad_norm": 2.6451975469839133,
+      "learning_rate": 1.3933390491938381e-06,
+      "loss": 1.0114,
+      "step": 7979
+    },
+    {
+      "epoch": 0.8351648351648352,
+      "grad_norm": 2.345016840340204,
+      "learning_rate": 1.3916136005887048e-06,
+      "loss": 0.8941,
+      "step": 7980
+    },
+    {
+      "epoch": 0.8352694924123496,
+      "grad_norm": 2.414100792833315,
+      "learning_rate": 1.3898891411046212e-06,
+      "loss": 0.8546,
+      "step": 7981
+    },
+    {
+      "epoch": 0.8353741496598639,
+      "grad_norm": 1.9845989400524062,
+      "learning_rate": 1.3881656709397272e-06,
+      "loss": 0.8789,
+      "step": 7982
+    },
+    {
+      "epoch": 0.8354788069073783,
+      "grad_norm": 1.7934306576299728,
+      "learning_rate": 1.3864431902920584e-06,
+      "loss": 0.8559,
+      "step": 7983
+    },
+    {
+      "epoch": 0.8355834641548927,
+      "grad_norm": 1.9810087701516985,
+      "learning_rate": 1.3847216993595292e-06,
+      "loss": 0.7373,
+      "step": 7984
+    },
+    {
+      "epoch": 0.8356881214024071,
+      "grad_norm": 1.84000027892072,
+      "learning_rate": 1.3830011983399404e-06,
+      "loss": 0.8423,
+      "step": 7985
+    },
+    {
+      "epoch": 0.8357927786499215,
+      "grad_norm": 1.983947105295788,
+      "learning_rate": 1.381281687430981e-06,
+      "loss": 0.9642,
+      "step": 7986
+    },
+    {
+      "epoch": 0.8358974358974359,
+      "grad_norm": 2.240241092713476,
+      "learning_rate": 1.37956316683023e-06,
+      "loss": 0.8095,
+      "step": 7987
+    },
+    {
+      "epoch": 0.8360020931449503,
+      "grad_norm": 2.328360525402851,
+      "learning_rate": 1.3778456367351455e-06,
+      "loss": 0.9569,
+      "step": 7988
+    },
+    {
+      "epoch": 0.8361067503924646,
+      "grad_norm": 2.290792673993712,
+      "learning_rate": 1.3761290973430753e-06,
+      "loss": 0.8242,
+      "step": 7989
+    },
+    {
+      "epoch": 0.836211407639979,
+      "grad_norm": 1.9941656847351477,
+      "learning_rate": 1.3744135488512556e-06,
+      "loss": 0.9385,
+      "step": 7990
+    },
+    {
+      "epoch": 0.8363160648874934,
+      "grad_norm": 1.9735468832454361,
+      "learning_rate": 1.3726989914568034e-06,
+      "loss": 0.8731,
+      "step": 7991
+    },
+    {
+      "epoch": 0.8364207221350078,
+      "grad_norm": 1.9837606343238872,
+      "learning_rate": 1.3709854253567278e-06,
+      "loss": 0.8264,
+      "step": 7992
+    },
+    {
+      "epoch": 0.8365253793825222,
+      "grad_norm": 1.9088882775380245,
+      "learning_rate": 1.3692728507479214e-06,
+      "loss": 0.8842,
+      "step": 7993
+    },
+    {
+      "epoch": 0.8366300366300367,
+      "grad_norm": 2.119436328203735,
+      "learning_rate": 1.3675612678271588e-06,
+      "loss": 0.906,
+      "step": 7994
+    },
+    {
+      "epoch": 0.8367346938775511,
+      "grad_norm": 2.407719091887164,
+      "learning_rate": 1.365850676791105e-06,
+      "loss": 0.8475,
+      "step": 7995
+    },
+    {
+      "epoch": 0.8368393511250654,
+      "grad_norm": 2.103481136150311,
+      "learning_rate": 1.364141077836314e-06,
+      "loss": 0.8192,
+      "step": 7996
+    },
+    {
+      "epoch": 0.8369440083725798,
+      "grad_norm": 2.1854323710686114,
+      "learning_rate": 1.3624324711592196e-06,
+      "loss": 0.8484,
+      "step": 7997
+    },
+    {
+      "epoch": 0.8370486656200942,
+      "grad_norm": 2.1386964506824895,
+      "learning_rate": 1.3607248569561426e-06,
+      "loss": 0.8292,
+      "step": 7998
+    },
+    {
+      "epoch": 0.8371533228676086,
+      "grad_norm": 2.115583413702666,
+      "learning_rate": 1.359018235423295e-06,
+      "loss": 0.8606,
+      "step": 7999
+    },
+    {
+      "epoch": 0.837257980115123,
+      "grad_norm": 2.238976220893437,
+      "learning_rate": 1.3573126067567688e-06,
+      "loss": 0.8858,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8373626373626374,
+      "grad_norm": 2.049600921572707,
+      "learning_rate": 1.3556079711525439e-06,
+      "loss": 0.8915,
+      "step": 8001
+    },
+    {
+      "epoch": 0.8374672946101518,
+      "grad_norm": 1.8171596729987456,
+      "learning_rate": 1.353904328806488e-06,
+      "loss": 0.8115,
+      "step": 8002
+    },
+    {
+      "epoch": 0.8375719518576662,
+      "grad_norm": 2.399384125655146,
+      "learning_rate": 1.3522016799143522e-06,
+      "loss": 0.9062,
+      "step": 8003
+    },
+    {
+      "epoch": 0.8376766091051805,
+      "grad_norm": 1.7533273781629497,
+      "learning_rate": 1.3505000246717747e-06,
+      "loss": 0.8903,
+      "step": 8004
+    },
+    {
+      "epoch": 0.8377812663526949,
+      "grad_norm": 1.9579932469356014,
+      "learning_rate": 1.348799363274277e-06,
+      "loss": 0.7385,
+      "step": 8005
+    },
+    {
+      "epoch": 0.8378859236002093,
+      "grad_norm": 1.981901250159665,
+      "learning_rate": 1.34709969591727e-06,
+      "loss": 0.8783,
+      "step": 8006
+    },
+    {
+      "epoch": 0.8379905808477237,
+      "grad_norm": 2.167993399995397,
+      "learning_rate": 1.3454010227960513e-06,
+      "loss": 0.9084,
+      "step": 8007
+    },
+    {
+      "epoch": 0.8380952380952381,
+      "grad_norm": 1.9178702073749012,
+      "learning_rate": 1.3437033441057989e-06,
+      "loss": 0.8963,
+      "step": 8008
+    },
+    {
+      "epoch": 0.8381998953427525,
+      "grad_norm": 1.951517186745851,
+      "learning_rate": 1.3420066600415815e-06,
+      "loss": 0.8355,
+      "step": 8009
+    },
+    {
+      "epoch": 0.8383045525902669,
+      "grad_norm": 1.909486949227223,
+      "learning_rate": 1.3403109707983475e-06,
+      "loss": 0.8665,
+      "step": 8010
+    },
+    {
+      "epoch": 0.8384092098377812,
+      "grad_norm": 1.9369839312160841,
+      "learning_rate": 1.338616276570941e-06,
+      "loss": 0.8397,
+      "step": 8011
+    },
+    {
+      "epoch": 0.8385138670852956,
+      "grad_norm": 1.8773973941438988,
+      "learning_rate": 1.336922577554083e-06,
+      "loss": 0.8907,
+      "step": 8012
+    },
+    {
+      "epoch": 0.83861852433281,
+      "grad_norm": 2.116860394094237,
+      "learning_rate": 1.335229873942382e-06,
+      "loss": 0.8874,
+      "step": 8013
+    },
+    {
+      "epoch": 0.8387231815803244,
+      "grad_norm": 2.170506935277406,
+      "learning_rate": 1.3335381659303315e-06,
+      "loss": 0.9886,
+      "step": 8014
+    },
+    {
+      "epoch": 0.8388278388278388,
+      "grad_norm": 1.8806562248296783,
+      "learning_rate": 1.3318474537123138e-06,
+      "loss": 0.7677,
+      "step": 8015
+    },
+    {
+      "epoch": 0.8389324960753533,
+      "grad_norm": 2.407172290628666,
+      "learning_rate": 1.3301577374825992e-06,
+      "loss": 0.7599,
+      "step": 8016
+    },
+    {
+      "epoch": 0.8390371533228677,
+      "grad_norm": 3.184412100368414,
+      "learning_rate": 1.3284690174353364e-06,
+      "loss": 0.8256,
+      "step": 8017
+    },
+    {
+      "epoch": 0.839141810570382,
+      "grad_norm": 1.7731206492806486,
+      "learning_rate": 1.326781293764562e-06,
+      "loss": 0.7793,
+      "step": 8018
+    },
+    {
+      "epoch": 0.8392464678178964,
+      "grad_norm": 1.9962901080318212,
+      "learning_rate": 1.325094566664199e-06,
+      "loss": 0.8381,
+      "step": 8019
+    },
+    {
+      "epoch": 0.8393511250654108,
+      "grad_norm": 2.0858347365200736,
+      "learning_rate": 1.323408836328054e-06,
+      "loss": 0.8617,
+      "step": 8020
+    },
+    {
+      "epoch": 0.8394557823129252,
+      "grad_norm": 1.941127370534017,
+      "learning_rate": 1.3217241029498263e-06,
+      "loss": 0.8268,
+      "step": 8021
+    },
+    {
+      "epoch": 0.8395604395604396,
+      "grad_norm": 1.9438795971577278,
+      "learning_rate": 1.320040366723091e-06,
+      "loss": 0.7854,
+      "step": 8022
+    },
+    {
+      "epoch": 0.839665096807954,
+      "grad_norm": 2.0873160477402175,
+      "learning_rate": 1.3183576278413134e-06,
+      "loss": 0.8417,
+      "step": 8023
+    },
+    {
+      "epoch": 0.8397697540554684,
+      "grad_norm": 1.7491945690304727,
+      "learning_rate": 1.3166758864978424e-06,
+      "loss": 0.8151,
+      "step": 8024
+    },
+    {
+      "epoch": 0.8398744113029827,
+      "grad_norm": 1.9769300150442357,
+      "learning_rate": 1.3149951428859142e-06,
+      "loss": 0.8788,
+      "step": 8025
+    },
+    {
+      "epoch": 0.8399790685504971,
+      "grad_norm": 2.140498082088263,
+      "learning_rate": 1.3133153971986534e-06,
+      "loss": 0.7433,
+      "step": 8026
+    },
+    {
+      "epoch": 0.8400837257980115,
+      "grad_norm": 2.4165371301116045,
+      "learning_rate": 1.3116366496290633e-06,
+      "loss": 0.8785,
+      "step": 8027
+    },
+    {
+      "epoch": 0.8401883830455259,
+      "grad_norm": 1.9222172060976765,
+      "learning_rate": 1.3099589003700341e-06,
+      "loss": 0.9783,
+      "step": 8028
+    },
+    {
+      "epoch": 0.8402930402930403,
+      "grad_norm": 2.3781849035951943,
+      "learning_rate": 1.3082821496143428e-06,
+      "loss": 0.7902,
+      "step": 8029
+    },
+    {
+      "epoch": 0.8403976975405547,
+      "grad_norm": 2.0993772636604304,
+      "learning_rate": 1.3066063975546538e-06,
+      "loss": 0.9355,
+      "step": 8030
+    },
+    {
+      "epoch": 0.8405023547880691,
+      "grad_norm": 2.1721127058441674,
+      "learning_rate": 1.3049316443835137e-06,
+      "loss": 0.7599,
+      "step": 8031
+    },
+    {
+      "epoch": 0.8406070120355834,
+      "grad_norm": 2.531892795002363,
+      "learning_rate": 1.3032578902933546e-06,
+      "loss": 0.8561,
+      "step": 8032
+    },
+    {
+      "epoch": 0.8407116692830978,
+      "grad_norm": 2.2300704327515004,
+      "learning_rate": 1.301585135476492e-06,
+      "loss": 0.8676,
+      "step": 8033
+    },
+    {
+      "epoch": 0.8408163265306122,
+      "grad_norm": 2.1993303961757826,
+      "learning_rate": 1.2999133801251295e-06,
+      "loss": 0.8671,
+      "step": 8034
+    },
+    {
+      "epoch": 0.8409209837781266,
+      "grad_norm": 2.154741893990914,
+      "learning_rate": 1.2982426244313595e-06,
+      "loss": 0.8462,
+      "step": 8035
+    },
+    {
+      "epoch": 0.841025641025641,
+      "grad_norm": 2.129202801530897,
+      "learning_rate": 1.2965728685871525e-06,
+      "loss": 1.0065,
+      "step": 8036
+    },
+    {
+      "epoch": 0.8411302982731554,
+      "grad_norm": 1.8293468950695384,
+      "learning_rate": 1.2949041127843665e-06,
+      "loss": 0.8104,
+      "step": 8037
+    },
+    {
+      "epoch": 0.8412349555206698,
+      "grad_norm": 1.907250964658945,
+      "learning_rate": 1.2932363572147445e-06,
+      "loss": 0.8117,
+      "step": 8038
+    },
+    {
+      "epoch": 0.8413396127681841,
+      "grad_norm": 2.2639632750436793,
+      "learning_rate": 1.2915696020699142e-06,
+      "loss": 0.8617,
+      "step": 8039
+    },
+    {
+      "epoch": 0.8414442700156985,
+      "grad_norm": 2.308249827742698,
+      "learning_rate": 1.2899038475413917e-06,
+      "loss": 0.8622,
+      "step": 8040
+    },
+    {
+      "epoch": 0.841548927263213,
+      "grad_norm": 2.227723639811261,
+      "learning_rate": 1.2882390938205725e-06,
+      "loss": 0.9042,
+      "step": 8041
+    },
+    {
+      "epoch": 0.8416535845107274,
+      "grad_norm": 2.1043886188970413,
+      "learning_rate": 1.2865753410987447e-06,
+      "loss": 0.9361,
+      "step": 8042
+    },
+    {
+      "epoch": 0.8417582417582418,
+      "grad_norm": 1.845500363111109,
+      "learning_rate": 1.2849125895670733e-06,
+      "loss": 0.8998,
+      "step": 8043
+    },
+    {
+      "epoch": 0.8418628990057562,
+      "grad_norm": 2.042113662517905,
+      "learning_rate": 1.283250839416611e-06,
+      "loss": 0.8967,
+      "step": 8044
+    },
+    {
+      "epoch": 0.8419675562532706,
+      "grad_norm": 2.5260072411582986,
+      "learning_rate": 1.2815900908383004e-06,
+      "loss": 0.8395,
+      "step": 8045
+    },
+    {
+      "epoch": 0.842072213500785,
+      "grad_norm": 1.569850831580542,
+      "learning_rate": 1.2799303440229616e-06,
+      "loss": 0.764,
+      "step": 8046
+    },
+    {
+      "epoch": 0.8421768707482993,
+      "grad_norm": 2.164867452931564,
+      "learning_rate": 1.2782715991613048e-06,
+      "loss": 0.8424,
+      "step": 8047
+    },
+    {
+      "epoch": 0.8422815279958137,
+      "grad_norm": 1.713908988310335,
+      "learning_rate": 1.276613856443919e-06,
+      "loss": 0.8094,
+      "step": 8048
+    },
+    {
+      "epoch": 0.8423861852433281,
+      "grad_norm": 1.9683438158526272,
+      "learning_rate": 1.2749571160612872e-06,
+      "loss": 0.8499,
+      "step": 8049
+    },
+    {
+      "epoch": 0.8424908424908425,
+      "grad_norm": 2.9203629814106002,
+      "learning_rate": 1.2733013782037695e-06,
+      "loss": 0.8889,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8425954997383569,
+      "grad_norm": 1.8540479952789417,
+      "learning_rate": 1.2716466430616148e-06,
+      "loss": 0.8076,
+      "step": 8051
+    },
+    {
+      "epoch": 0.8427001569858713,
+      "grad_norm": 1.9814166025936952,
+      "learning_rate": 1.2699929108249565e-06,
+      "loss": 0.8634,
+      "step": 8052
+    },
+    {
+      "epoch": 0.8428048142333857,
+      "grad_norm": 1.9793401850173653,
+      "learning_rate": 1.2683401816838071e-06,
+      "loss": 0.8196,
+      "step": 8053
+    },
+    {
+      "epoch": 0.8429094714809,
+      "grad_norm": 2.267300532522284,
+      "learning_rate": 1.2666884558280745e-06,
+      "loss": 0.7867,
+      "step": 8054
+    },
+    {
+      "epoch": 0.8430141287284144,
+      "grad_norm": 1.789887206030474,
+      "learning_rate": 1.2650377334475416e-06,
+      "loss": 0.7908,
+      "step": 8055
+    },
+    {
+      "epoch": 0.8431187859759288,
+      "grad_norm": 1.881345825048694,
+      "learning_rate": 1.2633880147318822e-06,
+      "loss": 0.8574,
+      "step": 8056
+    },
+    {
+      "epoch": 0.8432234432234432,
+      "grad_norm": 1.9203037592773267,
+      "learning_rate": 1.2617392998706502e-06,
+      "loss": 0.7534,
+      "step": 8057
+    },
+    {
+      "epoch": 0.8433281004709576,
+      "grad_norm": 2.2154089371004573,
+      "learning_rate": 1.260091589053284e-06,
+      "loss": 0.8416,
+      "step": 8058
+    },
+    {
+      "epoch": 0.843432757718472,
+      "grad_norm": 2.1555356685174454,
+      "learning_rate": 1.2584448824691132e-06,
+      "loss": 0.8896,
+      "step": 8059
+    },
+    {
+      "epoch": 0.8435374149659864,
+      "grad_norm": 2.081590502046218,
+      "learning_rate": 1.2567991803073476e-06,
+      "loss": 0.8501,
+      "step": 8060
+    },
+    {
+      "epoch": 0.8436420722135007,
+      "grad_norm": 2.1329378594631203,
+      "learning_rate": 1.2551544827570805e-06,
+      "loss": 0.8398,
+      "step": 8061
+    },
+    {
+      "epoch": 0.8437467294610151,
+      "grad_norm": 2.037326064879042,
+      "learning_rate": 1.2535107900072918e-06,
+      "loss": 0.9452,
+      "step": 8062
+    },
+    {
+      "epoch": 0.8438513867085295,
+      "grad_norm": 1.8536690182768223,
+      "learning_rate": 1.2518681022468414e-06,
+      "loss": 0.841,
+      "step": 8063
+    },
+    {
+      "epoch": 0.843956043956044,
+      "grad_norm": 2.418709398888962,
+      "learning_rate": 1.250226419664483e-06,
+      "loss": 0.8493,
+      "step": 8064
+    },
+    {
+      "epoch": 0.8440607012035584,
+      "grad_norm": 1.9982124764464644,
+      "learning_rate": 1.2485857424488457e-06,
+      "loss": 0.9066,
+      "step": 8065
+    },
+    {
+      "epoch": 0.8441653584510728,
+      "grad_norm": 1.826129673099164,
+      "learning_rate": 1.2469460707884485e-06,
+      "loss": 0.7961,
+      "step": 8066
+    },
+    {
+      "epoch": 0.8442700156985872,
+      "grad_norm": 2.0052685371560126,
+      "learning_rate": 1.2453074048716896e-06,
+      "loss": 0.9495,
+      "step": 8067
+    },
+    {
+      "epoch": 0.8443746729461015,
+      "grad_norm": 2.4774578028328977,
+      "learning_rate": 1.2436697448868572e-06,
+      "loss": 0.8768,
+      "step": 8068
+    },
+    {
+      "epoch": 0.8444793301936159,
+      "grad_norm": 1.9032740646595938,
+      "learning_rate": 1.2420330910221245e-06,
+      "loss": 0.9143,
+      "step": 8069
+    },
+    {
+      "epoch": 0.8445839874411303,
+      "grad_norm": 2.1131379226433524,
+      "learning_rate": 1.240397443465544e-06,
+      "loss": 0.8795,
+      "step": 8070
+    },
+    {
+      "epoch": 0.8446886446886447,
+      "grad_norm": 2.037973685230345,
+      "learning_rate": 1.2387628024050557e-06,
+      "loss": 0.8836,
+      "step": 8071
+    },
+    {
+      "epoch": 0.8447933019361591,
+      "grad_norm": 1.8857034384869027,
+      "learning_rate": 1.237129168028479e-06,
+      "loss": 0.8511,
+      "step": 8072
+    },
+    {
+      "epoch": 0.8448979591836735,
+      "grad_norm": 2.0457364245538807,
+      "learning_rate": 1.2354965405235276e-06,
+      "loss": 0.9149,
+      "step": 8073
+    },
+    {
+      "epoch": 0.8450026164311879,
+      "grad_norm": 2.1833956563172103,
+      "learning_rate": 1.2338649200777908e-06,
+      "loss": 0.7978,
+      "step": 8074
+    },
+    {
+      "epoch": 0.8451072736787022,
+      "grad_norm": 1.95060154540977,
+      "learning_rate": 1.2322343068787456e-06,
+      "loss": 0.8084,
+      "step": 8075
+    },
+    {
+      "epoch": 0.8452119309262166,
+      "grad_norm": 2.051054237471809,
+      "learning_rate": 1.23060470111375e-06,
+      "loss": 0.8513,
+      "step": 8076
+    },
+    {
+      "epoch": 0.845316588173731,
+      "grad_norm": 1.9270136243745513,
+      "learning_rate": 1.2289761029700553e-06,
+      "loss": 0.8409,
+      "step": 8077
+    },
+    {
+      "epoch": 0.8454212454212454,
+      "grad_norm": 2.2645480935646702,
+      "learning_rate": 1.227348512634784e-06,
+      "loss": 0.9181,
+      "step": 8078
+    },
+    {
+      "epoch": 0.8455259026687598,
+      "grad_norm": 2.2076948615617185,
+      "learning_rate": 1.2257219302949541e-06,
+      "loss": 0.9305,
+      "step": 8079
+    },
+    {
+      "epoch": 0.8456305599162742,
+      "grad_norm": 2.511590266125373,
+      "learning_rate": 1.224096356137463e-06,
+      "loss": 0.911,
+      "step": 8080
+    },
+    {
+      "epoch": 0.8457352171637886,
+      "grad_norm": 1.9894212656748564,
+      "learning_rate": 1.2224717903490901e-06,
+      "loss": 0.8961,
+      "step": 8081
+    },
+    {
+      "epoch": 0.8458398744113029,
+      "grad_norm": 2.2954720374673028,
+      "learning_rate": 1.2208482331165007e-06,
+      "loss": 0.8985,
+      "step": 8082
+    },
+    {
+      "epoch": 0.8459445316588173,
+      "grad_norm": 1.9884523540654926,
+      "learning_rate": 1.2192256846262484e-06,
+      "loss": 0.8576,
+      "step": 8083
+    },
+    {
+      "epoch": 0.8460491889063317,
+      "grad_norm": 1.6017827918804268,
+      "learning_rate": 1.2176041450647657e-06,
+      "loss": 0.7318,
+      "step": 8084
+    },
+    {
+      "epoch": 0.8461538461538461,
+      "grad_norm": 2.191831478479857,
+      "learning_rate": 1.215983614618369e-06,
+      "loss": 0.8629,
+      "step": 8085
+    },
+    {
+      "epoch": 0.8462585034013606,
+      "grad_norm": 1.9686856066340168,
+      "learning_rate": 1.2143640934732636e-06,
+      "loss": 0.7544,
+      "step": 8086
+    },
+    {
+      "epoch": 0.846363160648875,
+      "grad_norm": 2.15328965743791,
+      "learning_rate": 1.2127455818155321e-06,
+      "loss": 0.9058,
+      "step": 8087
+    },
+    {
+      "epoch": 0.8464678178963894,
+      "grad_norm": 2.0301113009573046,
+      "learning_rate": 1.2111280798311486e-06,
+      "loss": 0.8852,
+      "step": 8088
+    },
+    {
+      "epoch": 0.8465724751439038,
+      "grad_norm": 2.124293788106227,
+      "learning_rate": 1.209511587705966e-06,
+      "loss": 0.825,
+      "step": 8089
+    },
+    {
+      "epoch": 0.8466771323914181,
+      "grad_norm": 1.8508848742570292,
+      "learning_rate": 1.2078961056257222e-06,
+      "loss": 0.8428,
+      "step": 8090
+    },
+    {
+      "epoch": 0.8467817896389325,
+      "grad_norm": 2.1311794047736337,
+      "learning_rate": 1.2062816337760374e-06,
+      "loss": 0.9301,
+      "step": 8091
+    },
+    {
+      "epoch": 0.8468864468864469,
+      "grad_norm": 2.104243985600498,
+      "learning_rate": 1.2046681723424214e-06,
+      "loss": 0.872,
+      "step": 8092
+    },
+    {
+      "epoch": 0.8469911041339613,
+      "grad_norm": 2.130616894657293,
+      "learning_rate": 1.20305572151026e-06,
+      "loss": 0.7725,
+      "step": 8093
+    },
+    {
+      "epoch": 0.8470957613814757,
+      "grad_norm": 1.9408014352741592,
+      "learning_rate": 1.2014442814648318e-06,
+      "loss": 0.9378,
+      "step": 8094
+    },
+    {
+      "epoch": 0.8472004186289901,
+      "grad_norm": 2.430277353641727,
+      "learning_rate": 1.1998338523912923e-06,
+      "loss": 0.8954,
+      "step": 8095
+    },
+    {
+      "epoch": 0.8473050758765045,
+      "grad_norm": 2.147447871976377,
+      "learning_rate": 1.198224434474683e-06,
+      "loss": 0.9456,
+      "step": 8096
+    },
+    {
+      "epoch": 0.8474097331240188,
+      "grad_norm": 2.150148583677661,
+      "learning_rate": 1.1966160278999273e-06,
+      "loss": 0.845,
+      "step": 8097
+    },
+    {
+      "epoch": 0.8475143903715332,
+      "grad_norm": 1.8999522656749204,
+      "learning_rate": 1.1950086328518383e-06,
+      "loss": 0.9598,
+      "step": 8098
+    },
+    {
+      "epoch": 0.8476190476190476,
+      "grad_norm": 2.22762433278561,
+      "learning_rate": 1.1934022495151064e-06,
+      "loss": 0.9627,
+      "step": 8099
+    },
+    {
+      "epoch": 0.847723704866562,
+      "grad_norm": 1.9018640415303447,
+      "learning_rate": 1.1917968780743094e-06,
+      "loss": 0.9066,
+      "step": 8100
+    },
+    {
+      "epoch": 0.8478283621140764,
+      "grad_norm": 2.2146699983704914,
+      "learning_rate": 1.1901925187139052e-06,
+      "loss": 0.8584,
+      "step": 8101
+    },
+    {
+      "epoch": 0.8479330193615908,
+      "grad_norm": 2.0599972177345305,
+      "learning_rate": 1.1885891716182395e-06,
+      "loss": 0.8889,
+      "step": 8102
+    },
+    {
+      "epoch": 0.8480376766091052,
+      "grad_norm": 2.21986430951051,
+      "learning_rate": 1.1869868369715431e-06,
+      "loss": 0.9083,
+      "step": 8103
+    },
+    {
+      "epoch": 0.8481423338566195,
+      "grad_norm": 2.1417810174460854,
+      "learning_rate": 1.1853855149579251e-06,
+      "loss": 0.8543,
+      "step": 8104
+    },
+    {
+      "epoch": 0.8482469911041339,
+      "grad_norm": 1.7461793353367485,
+      "learning_rate": 1.1837852057613808e-06,
+      "loss": 0.7139,
+      "step": 8105
+    },
+    {
+      "epoch": 0.8483516483516483,
+      "grad_norm": 1.80563643788023,
+      "learning_rate": 1.182185909565785e-06,
+      "loss": 0.8036,
+      "step": 8106
+    },
+    {
+      "epoch": 0.8484563055991627,
+      "grad_norm": 2.3503816379233844,
+      "learning_rate": 1.1805876265549076e-06,
+      "loss": 0.9276,
+      "step": 8107
+    },
+    {
+      "epoch": 0.8485609628466771,
+      "grad_norm": 2.4489921440588605,
+      "learning_rate": 1.1789903569123896e-06,
+      "loss": 0.9568,
+      "step": 8108
+    },
+    {
+      "epoch": 0.8486656200941916,
+      "grad_norm": 1.9034330562968986,
+      "learning_rate": 1.1773941008217627e-06,
+      "loss": 0.8873,
+      "step": 8109
+    },
+    {
+      "epoch": 0.848770277341706,
+      "grad_norm": 2.2375424098728205,
+      "learning_rate": 1.175798858466436e-06,
+      "loss": 0.8962,
+      "step": 8110
+    },
+    {
+      "epoch": 0.8488749345892203,
+      "grad_norm": 2.0427220355559004,
+      "learning_rate": 1.1742046300297084e-06,
+      "loss": 0.8506,
+      "step": 8111
+    },
+    {
+      "epoch": 0.8489795918367347,
+      "grad_norm": 1.9535985727245424,
+      "learning_rate": 1.1726114156947644e-06,
+      "loss": 0.7011,
+      "step": 8112
+    },
+    {
+      "epoch": 0.8490842490842491,
+      "grad_norm": 1.9013330814423128,
+      "learning_rate": 1.171019215644662e-06,
+      "loss": 0.8496,
+      "step": 8113
+    },
+    {
+      "epoch": 0.8491889063317635,
+      "grad_norm": 1.8885673787395196,
+      "learning_rate": 1.1694280300623505e-06,
+      "loss": 0.9879,
+      "step": 8114
+    },
+    {
+      "epoch": 0.8492935635792779,
+      "grad_norm": 1.7156423861186427,
+      "learning_rate": 1.1678378591306604e-06,
+      "loss": 0.812,
+      "step": 8115
+    },
+    {
+      "epoch": 0.8493982208267923,
+      "grad_norm": 2.1750963180245293,
+      "learning_rate": 1.1662487030323021e-06,
+      "loss": 0.9194,
+      "step": 8116
+    },
+    {
+      "epoch": 0.8495028780743067,
+      "grad_norm": 2.0315617464728604,
+      "learning_rate": 1.1646605619498784e-06,
+      "loss": 0.8036,
+      "step": 8117
+    },
+    {
+      "epoch": 0.849607535321821,
+      "grad_norm": 2.1841952937990325,
+      "learning_rate": 1.1630734360658669e-06,
+      "loss": 0.8595,
+      "step": 8118
+    },
+    {
+      "epoch": 0.8497121925693354,
+      "grad_norm": 2.1088635653194627,
+      "learning_rate": 1.161487325562629e-06,
+      "loss": 0.7815,
+      "step": 8119
+    },
+    {
+      "epoch": 0.8498168498168498,
+      "grad_norm": 2.058500389607447,
+      "learning_rate": 1.1599022306224184e-06,
+      "loss": 0.8606,
+      "step": 8120
+    },
+    {
+      "epoch": 0.8499215070643642,
+      "grad_norm": 2.140038653165263,
+      "learning_rate": 1.15831815142736e-06,
+      "loss": 0.9239,
+      "step": 8121
+    },
+    {
+      "epoch": 0.8500261643118786,
+      "grad_norm": 2.0778034483126575,
+      "learning_rate": 1.1567350881594708e-06,
+      "loss": 0.955,
+      "step": 8122
+    },
+    {
+      "epoch": 0.850130821559393,
+      "grad_norm": 1.998286679619626,
+      "learning_rate": 1.1551530410006484e-06,
+      "loss": 0.9129,
+      "step": 8123
+    },
+    {
+      "epoch": 0.8502354788069074,
+      "grad_norm": 1.5817030186016467,
+      "learning_rate": 1.153572010132672e-06,
+      "loss": 0.7925,
+      "step": 8124
+    },
+    {
+      "epoch": 0.8503401360544217,
+      "grad_norm": 1.9794772310958975,
+      "learning_rate": 1.1519919957372029e-06,
+      "loss": 0.9053,
+      "step": 8125
+    },
+    {
+      "epoch": 0.8504447933019361,
+      "grad_norm": 1.8803338170594943,
+      "learning_rate": 1.1504129979957913e-06,
+      "loss": 0.8949,
+      "step": 8126
+    },
+    {
+      "epoch": 0.8505494505494505,
+      "grad_norm": 2.145083879121954,
+      "learning_rate": 1.1488350170898676e-06,
+      "loss": 0.9441,
+      "step": 8127
+    },
+    {
+      "epoch": 0.8506541077969649,
+      "grad_norm": 1.864355430498479,
+      "learning_rate": 1.1472580532007405e-06,
+      "loss": 0.9112,
+      "step": 8128
+    },
+    {
+      "epoch": 0.8507587650444793,
+      "grad_norm": 1.5284392423180004,
+      "learning_rate": 1.1456821065096125e-06,
+      "loss": 0.7273,
+      "step": 8129
+    },
+    {
+      "epoch": 0.8508634222919937,
+      "grad_norm": 1.7736534908958845,
+      "learning_rate": 1.1441071771975576e-06,
+      "loss": 0.746,
+      "step": 8130
+    },
+    {
+      "epoch": 0.8509680795395081,
+      "grad_norm": 2.0415742507935613,
+      "learning_rate": 1.1425332654455422e-06,
+      "loss": 0.7637,
+      "step": 8131
+    },
+    {
+      "epoch": 0.8510727367870226,
+      "grad_norm": 1.9995682557341876,
+      "learning_rate": 1.140960371434411e-06,
+      "loss": 0.8595,
+      "step": 8132
+    },
+    {
+      "epoch": 0.8511773940345368,
+      "grad_norm": 2.134115968088377,
+      "learning_rate": 1.1393884953448931e-06,
+      "loss": 0.9112,
+      "step": 8133
+    },
+    {
+      "epoch": 0.8512820512820513,
+      "grad_norm": 2.0560620276562083,
+      "learning_rate": 1.1378176373575977e-06,
+      "loss": 0.8082,
+      "step": 8134
+    },
+    {
+      "epoch": 0.8513867085295657,
+      "grad_norm": 1.951430060069111,
+      "learning_rate": 1.1362477976530206e-06,
+      "loss": 0.8975,
+      "step": 8135
+    },
+    {
+      "epoch": 0.8514913657770801,
+      "grad_norm": 1.6765488705122318,
+      "learning_rate": 1.1346789764115418e-06,
+      "loss": 0.7827,
+      "step": 8136
+    },
+    {
+      "epoch": 0.8515960230245945,
+      "grad_norm": 2.1566144334273014,
+      "learning_rate": 1.1331111738134193e-06,
+      "loss": 0.8666,
+      "step": 8137
+    },
+    {
+      "epoch": 0.8517006802721089,
+      "grad_norm": 2.324273210903453,
+      "learning_rate": 1.131544390038799e-06,
+      "loss": 0.8484,
+      "step": 8138
+    },
+    {
+      "epoch": 0.8518053375196233,
+      "grad_norm": 2.1363318732510352,
+      "learning_rate": 1.129978625267707e-06,
+      "loss": 0.8032,
+      "step": 8139
+    },
+    {
+      "epoch": 0.8519099947671376,
+      "grad_norm": 1.873627689153686,
+      "learning_rate": 1.1284138796800492e-06,
+      "loss": 0.8898,
+      "step": 8140
+    },
+    {
+      "epoch": 0.852014652014652,
+      "grad_norm": 2.2677845731395205,
+      "learning_rate": 1.1268501534556242e-06,
+      "loss": 0.8408,
+      "step": 8141
+    },
+    {
+      "epoch": 0.8521193092621664,
+      "grad_norm": 2.0835248846493557,
+      "learning_rate": 1.125287446774105e-06,
+      "loss": 0.9684,
+      "step": 8142
+    },
+    {
+      "epoch": 0.8522239665096808,
+      "grad_norm": 2.1770740624500657,
+      "learning_rate": 1.1237257598150487e-06,
+      "loss": 0.8597,
+      "step": 8143
+    },
+    {
+      "epoch": 0.8523286237571952,
+      "grad_norm": 2.1071307255681284,
+      "learning_rate": 1.1221650927578942e-06,
+      "loss": 0.9521,
+      "step": 8144
+    },
+    {
+      "epoch": 0.8524332810047096,
+      "grad_norm": 2.3515348488854944,
+      "learning_rate": 1.1206054457819671e-06,
+      "loss": 0.8748,
+      "step": 8145
+    },
+    {
+      "epoch": 0.852537938252224,
+      "grad_norm": 2.228241072663553,
+      "learning_rate": 1.1190468190664782e-06,
+      "loss": 0.8862,
+      "step": 8146
+    },
+    {
+      "epoch": 0.8526425954997383,
+      "grad_norm": 2.0454265798917985,
+      "learning_rate": 1.1174892127905123e-06,
+      "loss": 0.7851,
+      "step": 8147
+    },
+    {
+      "epoch": 0.8527472527472527,
+      "grad_norm": 2.9906239949974824,
+      "learning_rate": 1.115932627133043e-06,
+      "loss": 0.9877,
+      "step": 8148
+    },
+    {
+      "epoch": 0.8528519099947671,
+      "grad_norm": 1.7183578316341401,
+      "learning_rate": 1.1143770622729232e-06,
+      "loss": 0.8169,
+      "step": 8149
+    },
+    {
+      "epoch": 0.8529565672422815,
+      "grad_norm": 1.9631303418317718,
+      "learning_rate": 1.112822518388893e-06,
+      "loss": 0.9437,
+      "step": 8150
+    },
+    {
+      "epoch": 0.8530612244897959,
+      "grad_norm": 2.2478081957198883,
+      "learning_rate": 1.1112689956595724e-06,
+      "loss": 0.9049,
+      "step": 8151
+    },
+    {
+      "epoch": 0.8531658817373103,
+      "grad_norm": 1.9907598676903082,
+      "learning_rate": 1.109716494263463e-06,
+      "loss": 0.7481,
+      "step": 8152
+    },
+    {
+      "epoch": 0.8532705389848247,
+      "grad_norm": 2.092413746291313,
+      "learning_rate": 1.1081650143789513e-06,
+      "loss": 0.888,
+      "step": 8153
+    },
+    {
+      "epoch": 0.853375196232339,
+      "grad_norm": 2.150281264392465,
+      "learning_rate": 1.1066145561843033e-06,
+      "loss": 0.9098,
+      "step": 8154
+    },
+    {
+      "epoch": 0.8534798534798534,
+      "grad_norm": 1.7025890287640117,
+      "learning_rate": 1.1050651198576713e-06,
+      "loss": 0.7393,
+      "step": 8155
+    },
+    {
+      "epoch": 0.8535845107273679,
+      "grad_norm": 2.165113850117995,
+      "learning_rate": 1.1035167055770901e-06,
+      "loss": 0.8349,
+      "step": 8156
+    },
+    {
+      "epoch": 0.8536891679748823,
+      "grad_norm": 2.5762419911687826,
+      "learning_rate": 1.1019693135204757e-06,
+      "loss": 0.969,
+      "step": 8157
+    },
+    {
+      "epoch": 0.8537938252223967,
+      "grad_norm": 1.9241185785842263,
+      "learning_rate": 1.1004229438656245e-06,
+      "loss": 0.7895,
+      "step": 8158
+    },
+    {
+      "epoch": 0.8538984824699111,
+      "grad_norm": 2.028496874452248,
+      "learning_rate": 1.0988775967902165e-06,
+      "loss": 0.8903,
+      "step": 8159
+    },
+    {
+      "epoch": 0.8540031397174255,
+      "grad_norm": 2.388468483206559,
+      "learning_rate": 1.09733327247182e-06,
+      "loss": 0.773,
+      "step": 8160
+    },
+    {
+      "epoch": 0.8541077969649398,
+      "grad_norm": 1.941083291471243,
+      "learning_rate": 1.0957899710878772e-06,
+      "loss": 0.8528,
+      "step": 8161
+    },
+    {
+      "epoch": 0.8542124542124542,
+      "grad_norm": 2.2068646875053877,
+      "learning_rate": 1.0942476928157175e-06,
+      "loss": 0.8462,
+      "step": 8162
+    },
+    {
+      "epoch": 0.8543171114599686,
+      "grad_norm": 2.071496324715311,
+      "learning_rate": 1.0927064378325503e-06,
+      "loss": 0.9587,
+      "step": 8163
+    },
+    {
+      "epoch": 0.854421768707483,
+      "grad_norm": 2.150573325542622,
+      "learning_rate": 1.0911662063154694e-06,
+      "loss": 0.9465,
+      "step": 8164
+    },
+    {
+      "epoch": 0.8545264259549974,
+      "grad_norm": 1.7662897309607424,
+      "learning_rate": 1.0896269984414543e-06,
+      "loss": 0.8114,
+      "step": 8165
+    },
+    {
+      "epoch": 0.8546310832025118,
+      "grad_norm": 1.7296411837745311,
+      "learning_rate": 1.088088814387359e-06,
+      "loss": 0.7644,
+      "step": 8166
+    },
+    {
+      "epoch": 0.8547357404500262,
+      "grad_norm": 2.7077350662708257,
+      "learning_rate": 1.086551654329927e-06,
+      "loss": 0.6619,
+      "step": 8167
+    },
+    {
+      "epoch": 0.8548403976975405,
+      "grad_norm": 1.7773722774962573,
+      "learning_rate": 1.0850155184457767e-06,
+      "loss": 0.768,
+      "step": 8168
+    },
+    {
+      "epoch": 0.8549450549450549,
+      "grad_norm": 1.913533199941065,
+      "learning_rate": 1.083480406911418e-06,
+      "loss": 0.9215,
+      "step": 8169
+    },
+    {
+      "epoch": 0.8550497121925693,
+      "grad_norm": 2.2851663792939894,
+      "learning_rate": 1.0819463199032354e-06,
+      "loss": 0.9138,
+      "step": 8170
+    },
+    {
+      "epoch": 0.8551543694400837,
+      "grad_norm": 2.434539639739813,
+      "learning_rate": 1.0804132575974979e-06,
+      "loss": 0.7474,
+      "step": 8171
+    },
+    {
+      "epoch": 0.8552590266875981,
+      "grad_norm": 2.220301148902414,
+      "learning_rate": 1.0788812201703614e-06,
+      "loss": 0.9146,
+      "step": 8172
+    },
+    {
+      "epoch": 0.8553636839351125,
+      "grad_norm": 2.412245947837503,
+      "learning_rate": 1.0773502077978571e-06,
+      "loss": 0.9518,
+      "step": 8173
+    },
+    {
+      "epoch": 0.8554683411826269,
+      "grad_norm": 2.0701141368028915,
+      "learning_rate": 1.075820220655901e-06,
+      "loss": 0.7442,
+      "step": 8174
+    },
+    {
+      "epoch": 0.8555729984301413,
+      "grad_norm": 2.0366797846248668,
+      "learning_rate": 1.0742912589202935e-06,
+      "loss": 0.8856,
+      "step": 8175
+    },
+    {
+      "epoch": 0.8556776556776556,
+      "grad_norm": 2.2237305192244743,
+      "learning_rate": 1.0727633227667157e-06,
+      "loss": 0.9214,
+      "step": 8176
+    },
+    {
+      "epoch": 0.85578231292517,
+      "grad_norm": 1.9347821344648546,
+      "learning_rate": 1.0712364123707287e-06,
+      "loss": 0.9748,
+      "step": 8177
+    },
+    {
+      "epoch": 0.8558869701726844,
+      "grad_norm": 2.0653013623994054,
+      "learning_rate": 1.069710527907777e-06,
+      "loss": 0.8951,
+      "step": 8178
+    },
+    {
+      "epoch": 0.8559916274201989,
+      "grad_norm": 2.3214860069724947,
+      "learning_rate": 1.068185669553191e-06,
+      "loss": 0.9846,
+      "step": 8179
+    },
+    {
+      "epoch": 0.8560962846677133,
+      "grad_norm": 2.1905071628526986,
+      "learning_rate": 1.0666618374821759e-06,
+      "loss": 0.7894,
+      "step": 8180
+    },
+    {
+      "epoch": 0.8562009419152277,
+      "grad_norm": 2.1369361625201875,
+      "learning_rate": 1.0651390318698285e-06,
+      "loss": 0.9375,
+      "step": 8181
+    },
+    {
+      "epoch": 0.8563055991627421,
+      "grad_norm": 2.1626356161328935,
+      "learning_rate": 1.0636172528911182e-06,
+      "loss": 0.9148,
+      "step": 8182
+    },
+    {
+      "epoch": 0.8564102564102564,
+      "grad_norm": 2.085183578524157,
+      "learning_rate": 1.0620965007208993e-06,
+      "loss": 0.928,
+      "step": 8183
+    },
+    {
+      "epoch": 0.8565149136577708,
+      "grad_norm": 1.7680046788843726,
+      "learning_rate": 1.0605767755339147e-06,
+      "loss": 0.7463,
+      "step": 8184
+    },
+    {
+      "epoch": 0.8566195709052852,
+      "grad_norm": 1.9582799315088195,
+      "learning_rate": 1.0590580775047798e-06,
+      "loss": 0.9068,
+      "step": 8185
+    },
+    {
+      "epoch": 0.8567242281527996,
+      "grad_norm": 2.0988281526131845,
+      "learning_rate": 1.0575404068079965e-06,
+      "loss": 0.8693,
+      "step": 8186
+    },
+    {
+      "epoch": 0.856828885400314,
+      "grad_norm": 2.192703067041728,
+      "learning_rate": 1.0560237636179493e-06,
+      "loss": 0.8442,
+      "step": 8187
+    },
+    {
+      "epoch": 0.8569335426478284,
+      "grad_norm": 1.8377811369460275,
+      "learning_rate": 1.0545081481089015e-06,
+      "loss": 0.853,
+      "step": 8188
+    },
+    {
+      "epoch": 0.8570381998953428,
+      "grad_norm": 1.9277235332711045,
+      "learning_rate": 1.0529935604550012e-06,
+      "loss": 0.8852,
+      "step": 8189
+    },
+    {
+      "epoch": 0.8571428571428571,
+      "grad_norm": 1.8127579290454636,
+      "learning_rate": 1.0514800008302806e-06,
+      "loss": 0.7326,
+      "step": 8190
+    },
+    {
+      "epoch": 0.8572475143903715,
+      "grad_norm": 2.4923220960574306,
+      "learning_rate": 1.049967469408648e-06,
+      "loss": 1.0099,
+      "step": 8191
+    },
+    {
+      "epoch": 0.8573521716378859,
+      "grad_norm": 2.291069167052195,
+      "learning_rate": 1.0484559663638971e-06,
+      "loss": 0.8967,
+      "step": 8192
+    },
+    {
+      "epoch": 0.8574568288854003,
+      "grad_norm": 2.1378655909099367,
+      "learning_rate": 1.0469454918696998e-06,
+      "loss": 0.9139,
+      "step": 8193
+    },
+    {
+      "epoch": 0.8575614861329147,
+      "grad_norm": 2.0113703844240303,
+      "learning_rate": 1.0454360460996181e-06,
+      "loss": 0.8301,
+      "step": 8194
+    },
+    {
+      "epoch": 0.8576661433804291,
+      "grad_norm": 2.1271770785333066,
+      "learning_rate": 1.0439276292270872e-06,
+      "loss": 0.812,
+      "step": 8195
+    },
+    {
+      "epoch": 0.8577708006279435,
+      "grad_norm": 2.3638152006231903,
+      "learning_rate": 1.042420241425427e-06,
+      "loss": 0.777,
+      "step": 8196
+    },
+    {
+      "epoch": 0.8578754578754578,
+      "grad_norm": 1.9732167242741625,
+      "learning_rate": 1.0409138828678389e-06,
+      "loss": 0.6977,
+      "step": 8197
+    },
+    {
+      "epoch": 0.8579801151229722,
+      "grad_norm": 2.1089360027170083,
+      "learning_rate": 1.0394085537274069e-06,
+      "loss": 0.7946,
+      "step": 8198
+    },
+    {
+      "epoch": 0.8580847723704866,
+      "grad_norm": 1.7844456116204388,
+      "learning_rate": 1.0379042541771e-06,
+      "loss": 0.796,
+      "step": 8199
+    },
+    {
+      "epoch": 0.858189429618001,
+      "grad_norm": 2.136236846911897,
+      "learning_rate": 1.0364009843897615e-06,
+      "loss": 0.8329,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8582940868655154,
+      "grad_norm": 1.86851966011732,
+      "learning_rate": 1.0348987445381208e-06,
+      "loss": 0.774,
+      "step": 8201
+    },
+    {
+      "epoch": 0.8583987441130299,
+      "grad_norm": 2.1299132196721042,
+      "learning_rate": 1.0333975347947866e-06,
+      "loss": 0.9495,
+      "step": 8202
+    },
+    {
+      "epoch": 0.8585034013605443,
+      "grad_norm": 1.785497778491649,
+      "learning_rate": 1.0318973553322553e-06,
+      "loss": 0.8073,
+      "step": 8203
+    },
+    {
+      "epoch": 0.8586080586080586,
+      "grad_norm": 2.259096733636607,
+      "learning_rate": 1.0303982063228978e-06,
+      "loss": 0.8527,
+      "step": 8204
+    },
+    {
+      "epoch": 0.858712715855573,
+      "grad_norm": 2.0636141494036515,
+      "learning_rate": 1.02890008793897e-06,
+      "loss": 0.898,
+      "step": 8205
+    },
+    {
+      "epoch": 0.8588173731030874,
+      "grad_norm": 2.349058809970281,
+      "learning_rate": 1.0274030003526069e-06,
+      "loss": 0.8855,
+      "step": 8206
+    },
+    {
+      "epoch": 0.8589220303506018,
+      "grad_norm": 2.152390907657251,
+      "learning_rate": 1.0259069437358271e-06,
+      "loss": 0.8474,
+      "step": 8207
+    },
+    {
+      "epoch": 0.8590266875981162,
+      "grad_norm": 2.104781303316511,
+      "learning_rate": 1.024411918260535e-06,
+      "loss": 0.8537,
+      "step": 8208
+    },
+    {
+      "epoch": 0.8591313448456306,
+      "grad_norm": 2.002176163454363,
+      "learning_rate": 1.0229179240985089e-06,
+      "loss": 0.8734,
+      "step": 8209
+    },
+    {
+      "epoch": 0.859236002093145,
+      "grad_norm": 2.148962156095097,
+      "learning_rate": 1.021424961421411e-06,
+      "loss": 0.8443,
+      "step": 8210
+    },
+    {
+      "epoch": 0.8593406593406593,
+      "grad_norm": 1.7193849211791452,
+      "learning_rate": 1.0199330304007858e-06,
+      "loss": 0.7636,
+      "step": 8211
+    },
+    {
+      "epoch": 0.8594453165881737,
+      "grad_norm": 2.107603846212653,
+      "learning_rate": 1.0184421312080594e-06,
+      "loss": 0.9227,
+      "step": 8212
+    },
+    {
+      "epoch": 0.8595499738356881,
+      "grad_norm": 1.9252852291664944,
+      "learning_rate": 1.0169522640145412e-06,
+      "loss": 0.9483,
+      "step": 8213
+    },
+    {
+      "epoch": 0.8596546310832025,
+      "grad_norm": 2.226183549251737,
+      "learning_rate": 1.015463428991419e-06,
+      "loss": 1.0013,
+      "step": 8214
+    },
+    {
+      "epoch": 0.8597592883307169,
+      "grad_norm": 2.323773979390847,
+      "learning_rate": 1.013975626309759e-06,
+      "loss": 0.8943,
+      "step": 8215
+    },
+    {
+      "epoch": 0.8598639455782313,
+      "grad_norm": 2.019532286245795,
+      "learning_rate": 1.0124888561405188e-06,
+      "loss": 0.8684,
+      "step": 8216
+    },
+    {
+      "epoch": 0.8599686028257457,
+      "grad_norm": 2.270980325536687,
+      "learning_rate": 1.0110031186545265e-06,
+      "loss": 0.9197,
+      "step": 8217
+    },
+    {
+      "epoch": 0.8600732600732601,
+      "grad_norm": 2.00584994006669,
+      "learning_rate": 1.0095184140225011e-06,
+      "loss": 0.804,
+      "step": 8218
+    },
+    {
+      "epoch": 0.8601779173207744,
+      "grad_norm": 2.059032541958812,
+      "learning_rate": 1.0080347424150349e-06,
+      "loss": 0.9117,
+      "step": 8219
+    },
+    {
+      "epoch": 0.8602825745682888,
+      "grad_norm": 1.8583774944181204,
+      "learning_rate": 1.006552104002605e-06,
+      "loss": 0.8409,
+      "step": 8220
+    },
+    {
+      "epoch": 0.8603872318158032,
+      "grad_norm": 1.7469649900300717,
+      "learning_rate": 1.0050704989555693e-06,
+      "loss": 0.7859,
+      "step": 8221
+    },
+    {
+      "epoch": 0.8604918890633176,
+      "grad_norm": 1.9209381565368364,
+      "learning_rate": 1.0035899274441684e-06,
+      "loss": 0.8993,
+      "step": 8222
+    },
+    {
+      "epoch": 0.860596546310832,
+      "grad_norm": 2.034671367413749,
+      "learning_rate": 1.002110389638521e-06,
+      "loss": 0.8571,
+      "step": 8223
+    },
+    {
+      "epoch": 0.8607012035583465,
+      "grad_norm": 1.9206079541192072,
+      "learning_rate": 1.0006318857086328e-06,
+      "loss": 0.8058,
+      "step": 8224
+    },
+    {
+      "epoch": 0.8608058608058609,
+      "grad_norm": 1.7267084357758915,
+      "learning_rate": 9.991544158243848e-07,
+      "loss": 0.7573,
+      "step": 8225
+    },
+    {
+      "epoch": 0.8609105180533752,
+      "grad_norm": 2.271042408187385,
+      "learning_rate": 9.976779801555404e-07,
+      "loss": 0.9341,
+      "step": 8226
+    },
+    {
+      "epoch": 0.8610151753008896,
+      "grad_norm": 2.194966202864827,
+      "learning_rate": 9.962025788717434e-07,
+      "loss": 0.8549,
+      "step": 8227
+    },
+    {
+      "epoch": 0.861119832548404,
+      "grad_norm": 2.1563970927227016,
+      "learning_rate": 9.94728212142525e-07,
+      "loss": 0.9266,
+      "step": 8228
+    },
+    {
+      "epoch": 0.8612244897959184,
+      "grad_norm": 1.98123025878276,
+      "learning_rate": 9.93254880137291e-07,
+      "loss": 0.8194,
+      "step": 8229
+    },
+    {
+      "epoch": 0.8613291470434328,
+      "grad_norm": 2.027077949956647,
+      "learning_rate": 9.917825830253303e-07,
+      "loss": 0.962,
+      "step": 8230
+    },
+    {
+      "epoch": 0.8614338042909472,
+      "grad_norm": 2.3467824454655966,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.9625,
+      "step": 8231
+    },
+    {
+      "epoch": 0.8615384615384616,
+      "grad_norm": 1.7368296277382502,
+      "learning_rate": 9.888410941577819e-07,
+      "loss": 0.7428,
+      "step": 8232
+    },
+    {
+      "epoch": 0.8616431187859759,
+      "grad_norm": 1.9207232431358994,
+      "learning_rate": 9.87371902740183e-07,
+      "loss": 0.9178,
+      "step": 8233
+    },
+    {
+      "epoch": 0.8617477760334903,
+      "grad_norm": 1.7639576896606057,
+      "learning_rate": 9.859037468918232e-07,
+      "loss": 0.8275,
+      "step": 8234
+    },
+    {
+      "epoch": 0.8618524332810047,
+      "grad_norm": 2.096385523082317,
+      "learning_rate": 9.844366267813965e-07,
+      "loss": 0.8697,
+      "step": 8235
+    },
+    {
+      "epoch": 0.8619570905285191,
+      "grad_norm": 2.1971747126218157,
+      "learning_rate": 9.82970542577475e-07,
+      "loss": 0.86,
+      "step": 8236
+    },
+    {
+      "epoch": 0.8620617477760335,
+      "grad_norm": 2.0670900252205056,
+      "learning_rate": 9.815054944485203e-07,
+      "loss": 0.85,
+      "step": 8237
+    },
+    {
+      "epoch": 0.8621664050235479,
+      "grad_norm": 2.0255958527875766,
+      "learning_rate": 9.800414825628657e-07,
+      "loss": 0.7782,
+      "step": 8238
+    },
+    {
+      "epoch": 0.8622710622710623,
+      "grad_norm": 2.5228783731175466,
+      "learning_rate": 9.78578507088731e-07,
+      "loss": 0.8302,
+      "step": 8239
+    },
+    {
+      "epoch": 0.8623757195185766,
+      "grad_norm": 2.0907415401597547,
+      "learning_rate": 9.771165681942118e-07,
+      "loss": 0.9448,
+      "step": 8240
+    },
+    {
+      "epoch": 0.862480376766091,
+      "grad_norm": 1.795604413521706,
+      "learning_rate": 9.756556660472904e-07,
+      "loss": 0.8402,
+      "step": 8241
+    },
+    {
+      "epoch": 0.8625850340136054,
+      "grad_norm": 2.224964285791813,
+      "learning_rate": 9.741958008158292e-07,
+      "loss": 0.8443,
+      "step": 8242
+    },
+    {
+      "epoch": 0.8626896912611198,
+      "grad_norm": 2.1230855516357847,
+      "learning_rate": 9.727369726675673e-07,
+      "loss": 0.85,
+      "step": 8243
+    },
+    {
+      "epoch": 0.8627943485086342,
+      "grad_norm": 2.047579087438642,
+      "learning_rate": 9.712791817701273e-07,
+      "loss": 0.8912,
+      "step": 8244
+    },
+    {
+      "epoch": 0.8628990057561486,
+      "grad_norm": 1.8823706152629818,
+      "learning_rate": 9.698224282910128e-07,
+      "loss": 0.8377,
+      "step": 8245
+    },
+    {
+      "epoch": 0.863003663003663,
+      "grad_norm": 2.0560768069448714,
+      "learning_rate": 9.68366712397606e-07,
+      "loss": 0.8321,
+      "step": 8246
+    },
+    {
+      "epoch": 0.8631083202511773,
+      "grad_norm": 2.010540058978142,
+      "learning_rate": 9.669120342571747e-07,
+      "loss": 0.8789,
+      "step": 8247
+    },
+    {
+      "epoch": 0.8632129774986917,
+      "grad_norm": 1.78965265727034,
+      "learning_rate": 9.654583940368622e-07,
+      "loss": 0.6758,
+      "step": 8248
+    },
+    {
+      "epoch": 0.8633176347462062,
+      "grad_norm": 2.0314846628821766,
+      "learning_rate": 9.640057919036926e-07,
+      "loss": 0.8291,
+      "step": 8249
+    },
+    {
+      "epoch": 0.8634222919937206,
+      "grad_norm": 2.0004853915166354,
+      "learning_rate": 9.62554228024578e-07,
+      "loss": 0.8815,
+      "step": 8250
+    },
+    {
+      "epoch": 0.863526949241235,
+      "grad_norm": 2.1679537430544205,
+      "learning_rate": 9.611037025663017e-07,
+      "loss": 0.955,
+      "step": 8251
+    },
+    {
+      "epoch": 0.8636316064887494,
+      "grad_norm": 1.8433670228793322,
+      "learning_rate": 9.596542156955357e-07,
+      "loss": 0.8938,
+      "step": 8252
+    },
+    {
+      "epoch": 0.8637362637362638,
+      "grad_norm": 2.1490342455394162,
+      "learning_rate": 9.58205767578827e-07,
+      "loss": 0.8708,
+      "step": 8253
+    },
+    {
+      "epoch": 0.8638409209837782,
+      "grad_norm": 2.258196552431531,
+      "learning_rate": 9.567583583826046e-07,
+      "loss": 0.8351,
+      "step": 8254
+    },
+    {
+      "epoch": 0.8639455782312925,
+      "grad_norm": 2.076287178014026,
+      "learning_rate": 9.553119882731777e-07,
+      "loss": 0.949,
+      "step": 8255
+    },
+    {
+      "epoch": 0.8640502354788069,
+      "grad_norm": 2.247501553759808,
+      "learning_rate": 9.53866657416741e-07,
+      "loss": 0.9345,
+      "step": 8256
+    },
+    {
+      "epoch": 0.8641548927263213,
+      "grad_norm": 2.281215175212065,
+      "learning_rate": 9.524223659793641e-07,
+      "loss": 0.8919,
+      "step": 8257
+    },
+    {
+      "epoch": 0.8642595499738357,
+      "grad_norm": 1.9878173669322587,
+      "learning_rate": 9.509791141269964e-07,
+      "loss": 0.9017,
+      "step": 8258
+    },
+    {
+      "epoch": 0.8643642072213501,
+      "grad_norm": 1.9534688327422565,
+      "learning_rate": 9.495369020254753e-07,
+      "loss": 0.8423,
+      "step": 8259
+    },
+    {
+      "epoch": 0.8644688644688645,
+      "grad_norm": 1.9918692395385151,
+      "learning_rate": 9.48095729840508e-07,
+      "loss": 0.9165,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8645735217163789,
+      "grad_norm": 1.8489436949868145,
+      "learning_rate": 9.466555977376946e-07,
+      "loss": 0.8444,
+      "step": 8261
+    },
+    {
+      "epoch": 0.8646781789638932,
+      "grad_norm": 2.1193645421178093,
+      "learning_rate": 9.452165058825058e-07,
+      "loss": 0.8266,
+      "step": 8262
+    },
+    {
+      "epoch": 0.8647828362114076,
+      "grad_norm": 2.0097760670300975,
+      "learning_rate": 9.437784544402961e-07,
+      "loss": 0.8798,
+      "step": 8263
+    },
+    {
+      "epoch": 0.864887493458922,
+      "grad_norm": 2.107855735727707,
+      "learning_rate": 9.423414435763012e-07,
+      "loss": 0.8519,
+      "step": 8264
+    },
+    {
+      "epoch": 0.8649921507064364,
+      "grad_norm": 1.9743559910054354,
+      "learning_rate": 9.409054734556344e-07,
+      "loss": 0.8584,
+      "step": 8265
+    },
+    {
+      "epoch": 0.8650968079539508,
+      "grad_norm": 2.565003258614491,
+      "learning_rate": 9.39470544243295e-07,
+      "loss": 0.9632,
+      "step": 8266
+    },
+    {
+      "epoch": 0.8652014652014652,
+      "grad_norm": 2.0088418728464643,
+      "learning_rate": 9.380366561041553e-07,
+      "loss": 0.897,
+      "step": 8267
+    },
+    {
+      "epoch": 0.8653061224489796,
+      "grad_norm": 2.102961311337027,
+      "learning_rate": 9.366038092029773e-07,
+      "loss": 0.9862,
+      "step": 8268
+    },
+    {
+      "epoch": 0.8654107796964939,
+      "grad_norm": 2.0365497413831424,
+      "learning_rate": 9.351720037043943e-07,
+      "loss": 0.7902,
+      "step": 8269
+    },
+    {
+      "epoch": 0.8655154369440083,
+      "grad_norm": 2.2180607315989436,
+      "learning_rate": 9.337412397729229e-07,
+      "loss": 0.8182,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8656200941915227,
+      "grad_norm": 2.488823116296402,
+      "learning_rate": 9.323115175729636e-07,
+      "loss": 0.7917,
+      "step": 8271
+    },
+    {
+      "epoch": 0.8657247514390372,
+      "grad_norm": 2.2666525932415844,
+      "learning_rate": 9.308828372687939e-07,
+      "loss": 0.8258,
+      "step": 8272
+    },
+    {
+      "epoch": 0.8658294086865516,
+      "grad_norm": 1.7807084928441752,
+      "learning_rate": 9.294551990245704e-07,
+      "loss": 0.8042,
+      "step": 8273
+    },
+    {
+      "epoch": 0.865934065934066,
+      "grad_norm": 1.886320152024955,
+      "learning_rate": 9.280286030043306e-07,
+      "loss": 0.9426,
+      "step": 8274
+    },
+    {
+      "epoch": 0.8660387231815804,
+      "grad_norm": 2.3063773024197705,
+      "learning_rate": 9.266030493719979e-07,
+      "loss": 0.9073,
+      "step": 8275
+    },
+    {
+      "epoch": 0.8661433804290947,
+      "grad_norm": 2.2479580097764535,
+      "learning_rate": 9.251785382913658e-07,
+      "loss": 0.9288,
+      "step": 8276
+    },
+    {
+      "epoch": 0.8662480376766091,
+      "grad_norm": 2.1411449789130566,
+      "learning_rate": 9.237550699261188e-07,
+      "loss": 0.8874,
+      "step": 8277
+    },
+    {
+      "epoch": 0.8663526949241235,
+      "grad_norm": 1.824544022620309,
+      "learning_rate": 9.223326444398128e-07,
+      "loss": 0.8135,
+      "step": 8278
+    },
+    {
+      "epoch": 0.8664573521716379,
+      "grad_norm": 1.6429551373780529,
+      "learning_rate": 9.20911261995887e-07,
+      "loss": 0.7566,
+      "step": 8279
+    },
+    {
+      "epoch": 0.8665620094191523,
+      "grad_norm": 2.176909472080666,
+      "learning_rate": 9.194909227576642e-07,
+      "loss": 0.9291,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8666666666666667,
+      "grad_norm": 2.093261943588003,
+      "learning_rate": 9.180716268883427e-07,
+      "loss": 0.8321,
+      "step": 8281
+    },
+    {
+      "epoch": 0.8667713239141811,
+      "grad_norm": 2.1603700977335256,
+      "learning_rate": 9.16653374551002e-07,
+      "loss": 0.8882,
+      "step": 8282
+    },
+    {
+      "epoch": 0.8668759811616954,
+      "grad_norm": 1.8687484939968493,
+      "learning_rate": 9.152361659086029e-07,
+      "loss": 0.9273,
+      "step": 8283
+    },
+    {
+      "epoch": 0.8669806384092098,
+      "grad_norm": 2.03676255437104,
+      "learning_rate": 9.138200011239828e-07,
+      "loss": 0.9724,
+      "step": 8284
+    },
+    {
+      "epoch": 0.8670852956567242,
+      "grad_norm": 2.116389446692822,
+      "learning_rate": 9.124048803598639e-07,
+      "loss": 0.9146,
+      "step": 8285
+    },
+    {
+      "epoch": 0.8671899529042386,
+      "grad_norm": 1.919555592655635,
+      "learning_rate": 9.109908037788484e-07,
+      "loss": 0.9482,
+      "step": 8286
+    },
+    {
+      "epoch": 0.867294610151753,
+      "grad_norm": 2.1601273225442625,
+      "learning_rate": 9.095777715434162e-07,
+      "loss": 0.9593,
+      "step": 8287
+    },
+    {
+      "epoch": 0.8673992673992674,
+      "grad_norm": 1.927506321708376,
+      "learning_rate": 9.081657838159253e-07,
+      "loss": 0.8351,
+      "step": 8288
+    },
+    {
+      "epoch": 0.8675039246467818,
+      "grad_norm": 2.0177113831137357,
+      "learning_rate": 9.067548407586146e-07,
+      "loss": 0.889,
+      "step": 8289
+    },
+    {
+      "epoch": 0.8676085818942961,
+      "grad_norm": 2.1523136474810163,
+      "learning_rate": 9.053449425336092e-07,
+      "loss": 0.9081,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8677132391418105,
+      "grad_norm": 1.9842678328945351,
+      "learning_rate": 9.03936089302907e-07,
+      "loss": 0.9047,
+      "step": 8291
+    },
+    {
+      "epoch": 0.8678178963893249,
+      "grad_norm": 1.946904140627659,
+      "learning_rate": 9.025282812283875e-07,
+      "loss": 0.7917,
+      "step": 8292
+    },
+    {
+      "epoch": 0.8679225536368393,
+      "grad_norm": 1.7343881101313423,
+      "learning_rate": 9.011215184718081e-07,
+      "loss": 0.7012,
+      "step": 8293
+    },
+    {
+      "epoch": 0.8680272108843538,
+      "grad_norm": 1.6121923338112008,
+      "learning_rate": 8.997158011948126e-07,
+      "loss": 0.739,
+      "step": 8294
+    },
+    {
+      "epoch": 0.8681318681318682,
+      "grad_norm": 1.7420719829107674,
+      "learning_rate": 8.98311129558922e-07,
+      "loss": 0.6873,
+      "step": 8295
+    },
+    {
+      "epoch": 0.8682365253793826,
+      "grad_norm": 2.1273765607618667,
+      "learning_rate": 8.969075037255326e-07,
+      "loss": 0.9182,
+      "step": 8296
+    },
+    {
+      "epoch": 0.868341182626897,
+      "grad_norm": 1.9827264389330077,
+      "learning_rate": 8.955049238559254e-07,
+      "loss": 0.9137,
+      "step": 8297
+    },
+    {
+      "epoch": 0.8684458398744113,
+      "grad_norm": 1.9931692934159142,
+      "learning_rate": 8.941033901112572e-07,
+      "loss": 0.9219,
+      "step": 8298
+    },
+    {
+      "epoch": 0.8685504971219257,
+      "grad_norm": 2.0187325482134026,
+      "learning_rate": 8.927029026525713e-07,
+      "loss": 0.8923,
+      "step": 8299
+    },
+    {
+      "epoch": 0.8686551543694401,
+      "grad_norm": 2.121244415757878,
+      "learning_rate": 8.913034616407846e-07,
+      "loss": 0.8702,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8687598116169545,
+      "grad_norm": 2.0917927570195145,
+      "learning_rate": 8.899050672366949e-07,
+      "loss": 0.7015,
+      "step": 8301
+    },
+    {
+      "epoch": 0.8688644688644689,
+      "grad_norm": 2.3545018308026564,
+      "learning_rate": 8.885077196009783e-07,
+      "loss": 0.8551,
+      "step": 8302
+    },
+    {
+      "epoch": 0.8689691261119833,
+      "grad_norm": 2.367894373396556,
+      "learning_rate": 8.871114188941987e-07,
+      "loss": 0.8081,
+      "step": 8303
+    },
+    {
+      "epoch": 0.8690737833594977,
+      "grad_norm": 1.9788955523872949,
+      "learning_rate": 8.857161652767887e-07,
+      "loss": 0.83,
+      "step": 8304
+    },
+    {
+      "epoch": 0.869178440607012,
+      "grad_norm": 2.27059813164129,
+      "learning_rate": 8.843219589090701e-07,
+      "loss": 0.8601,
+      "step": 8305
+    },
+    {
+      "epoch": 0.8692830978545264,
+      "grad_norm": 2.3545604185359497,
+      "learning_rate": 8.829287999512371e-07,
+      "loss": 0.8818,
+      "step": 8306
+    },
+    {
+      "epoch": 0.8693877551020408,
+      "grad_norm": 2.122637516854195,
+      "learning_rate": 8.81536688563367e-07,
+      "loss": 0.9421,
+      "step": 8307
+    },
+    {
+      "epoch": 0.8694924123495552,
+      "grad_norm": 2.041713532250219,
+      "learning_rate": 8.801456249054152e-07,
+      "loss": 0.8007,
+      "step": 8308
+    },
+    {
+      "epoch": 0.8695970695970696,
+      "grad_norm": 2.141284640895024,
+      "learning_rate": 8.787556091372207e-07,
+      "loss": 0.7057,
+      "step": 8309
+    },
+    {
+      "epoch": 0.869701726844584,
+      "grad_norm": 2.1933709636305596,
+      "learning_rate": 8.773666414184955e-07,
+      "loss": 0.8779,
+      "step": 8310
+    },
+    {
+      "epoch": 0.8698063840920984,
+      "grad_norm": 2.2346155866799533,
+      "learning_rate": 8.759787219088389e-07,
+      "loss": 0.7949,
+      "step": 8311
+    },
+    {
+      "epoch": 0.8699110413396127,
+      "grad_norm": 2.3392273884858836,
+      "learning_rate": 8.745918507677232e-07,
+      "loss": 0.9422,
+      "step": 8312
+    },
+    {
+      "epoch": 0.8700156985871271,
+      "grad_norm": 1.8380709262710446,
+      "learning_rate": 8.732060281545007e-07,
+      "loss": 0.769,
+      "step": 8313
+    },
+    {
+      "epoch": 0.8701203558346415,
+      "grad_norm": 1.7448127426806292,
+      "learning_rate": 8.718212542284099e-07,
+      "loss": 0.8186,
+      "step": 8314
+    },
+    {
+      "epoch": 0.8702250130821559,
+      "grad_norm": 2.1138633029838867,
+      "learning_rate": 8.704375291485623e-07,
+      "loss": 0.7661,
+      "step": 8315
+    },
+    {
+      "epoch": 0.8703296703296703,
+      "grad_norm": 2.0932322539738784,
+      "learning_rate": 8.690548530739496e-07,
+      "loss": 0.9311,
+      "step": 8316
+    },
+    {
+      "epoch": 0.8704343275771848,
+      "grad_norm": 1.7979421071378918,
+      "learning_rate": 8.676732261634424e-07,
+      "loss": 0.784,
+      "step": 8317
+    },
+    {
+      "epoch": 0.8705389848246992,
+      "grad_norm": 1.9703330311133151,
+      "learning_rate": 8.662926485757961e-07,
+      "loss": 0.8364,
+      "step": 8318
+    },
+    {
+      "epoch": 0.8706436420722135,
+      "grad_norm": 2.3370476834492413,
+      "learning_rate": 8.649131204696392e-07,
+      "loss": 0.9065,
+      "step": 8319
+    },
+    {
+      "epoch": 0.8707482993197279,
+      "grad_norm": 2.404568265608144,
+      "learning_rate": 8.635346420034862e-07,
+      "loss": 0.8264,
+      "step": 8320
+    },
+    {
+      "epoch": 0.8708529565672423,
+      "grad_norm": 1.8537120114233208,
+      "learning_rate": 8.621572133357236e-07,
+      "loss": 0.8729,
+      "step": 8321
+    },
+    {
+      "epoch": 0.8709576138147567,
+      "grad_norm": 2.2930255513118705,
+      "learning_rate": 8.607808346246216e-07,
+      "loss": 0.9316,
+      "step": 8322
+    },
+    {
+      "epoch": 0.8710622710622711,
+      "grad_norm": 2.029189543208393,
+      "learning_rate": 8.594055060283268e-07,
+      "loss": 0.7619,
+      "step": 8323
+    },
+    {
+      "epoch": 0.8711669283097855,
+      "grad_norm": 2.570973752644621,
+      "learning_rate": 8.580312277048708e-07,
+      "loss": 0.9147,
+      "step": 8324
+    },
+    {
+      "epoch": 0.8712715855572999,
+      "grad_norm": 2.0422859206952424,
+      "learning_rate": 8.566579998121593e-07,
+      "loss": 0.9099,
+      "step": 8325
+    },
+    {
+      "epoch": 0.8713762428048142,
+      "grad_norm": 2.76800341990982,
+      "learning_rate": 8.552858225079807e-07,
+      "loss": 0.8457,
+      "step": 8326
+    },
+    {
+      "epoch": 0.8714809000523286,
+      "grad_norm": 2.1054140471943392,
+      "learning_rate": 8.539146959499956e-07,
+      "loss": 0.7575,
+      "step": 8327
+    },
+    {
+      "epoch": 0.871585557299843,
+      "grad_norm": 2.0253201711623565,
+      "learning_rate": 8.525446202957543e-07,
+      "loss": 0.8476,
+      "step": 8328
+    },
+    {
+      "epoch": 0.8716902145473574,
+      "grad_norm": 1.644414039598465,
+      "learning_rate": 8.511755957026813e-07,
+      "loss": 0.6856,
+      "step": 8329
+    },
+    {
+      "epoch": 0.8717948717948718,
+      "grad_norm": 2.278761036199369,
+      "learning_rate": 8.498076223280794e-07,
+      "loss": 0.8249,
+      "step": 8330
+    },
+    {
+      "epoch": 0.8718995290423862,
+      "grad_norm": 2.0636072314286698,
+      "learning_rate": 8.484407003291306e-07,
+      "loss": 0.8947,
+      "step": 8331
+    },
+    {
+      "epoch": 0.8720041862899006,
+      "grad_norm": 2.211711428835568,
+      "learning_rate": 8.47074829862895e-07,
+      "loss": 0.9184,
+      "step": 8332
+    },
+    {
+      "epoch": 0.8721088435374149,
+      "grad_norm": 2.016664180022378,
+      "learning_rate": 8.457100110863192e-07,
+      "loss": 0.8512,
+      "step": 8333
+    },
+    {
+      "epoch": 0.8722135007849293,
+      "grad_norm": 2.0273235658439654,
+      "learning_rate": 8.443462441562211e-07,
+      "loss": 0.9258,
+      "step": 8334
+    },
+    {
+      "epoch": 0.8723181580324437,
+      "grad_norm": 2.1797348714707145,
+      "learning_rate": 8.429835292292987e-07,
+      "loss": 0.9337,
+      "step": 8335
+    },
+    {
+      "epoch": 0.8724228152799581,
+      "grad_norm": 2.128984970826725,
+      "learning_rate": 8.416218664621312e-07,
+      "loss": 0.903,
+      "step": 8336
+    },
+    {
+      "epoch": 0.8725274725274725,
+      "grad_norm": 1.9652208208969064,
+      "learning_rate": 8.402612560111767e-07,
+      "loss": 0.908,
+      "step": 8337
+    },
+    {
+      "epoch": 0.8726321297749869,
+      "grad_norm": 2.0961986353412168,
+      "learning_rate": 8.389016980327746e-07,
+      "loss": 0.9465,
+      "step": 8338
+    },
+    {
+      "epoch": 0.8727367870225013,
+      "grad_norm": 2.5872968306860313,
+      "learning_rate": 8.37543192683139e-07,
+      "loss": 0.9034,
+      "step": 8339
+    },
+    {
+      "epoch": 0.8728414442700158,
+      "grad_norm": 2.443057693441533,
+      "learning_rate": 8.361857401183649e-07,
+      "loss": 0.8729,
+      "step": 8340
+    },
+    {
+      "epoch": 0.87294610151753,
+      "grad_norm": 2.1845773081813635,
+      "learning_rate": 8.348293404944263e-07,
+      "loss": 0.8835,
+      "step": 8341
+    },
+    {
+      "epoch": 0.8730507587650445,
+      "grad_norm": 1.9310878809345682,
+      "learning_rate": 8.334739939671743e-07,
+      "loss": 0.8102,
+      "step": 8342
+    },
+    {
+      "epoch": 0.8731554160125589,
+      "grad_norm": 1.9452518156409584,
+      "learning_rate": 8.321197006923442e-07,
+      "loss": 0.8457,
+      "step": 8343
+    },
+    {
+      "epoch": 0.8732600732600733,
+      "grad_norm": 1.8616667743331106,
+      "learning_rate": 8.307664608255461e-07,
+      "loss": 0.9066,
+      "step": 8344
+    },
+    {
+      "epoch": 0.8733647305075877,
+      "grad_norm": 1.9175403993591587,
+      "learning_rate": 8.294142745222678e-07,
+      "loss": 0.8309,
+      "step": 8345
+    },
+    {
+      "epoch": 0.8734693877551021,
+      "grad_norm": 2.4850953298749996,
+      "learning_rate": 8.280631419378815e-07,
+      "loss": 0.9685,
+      "step": 8346
+    },
+    {
+      "epoch": 0.8735740450026165,
+      "grad_norm": 2.044753493702447,
+      "learning_rate": 8.267130632276321e-07,
+      "loss": 0.9937,
+      "step": 8347
+    },
+    {
+      "epoch": 0.8736787022501308,
+      "grad_norm": 2.3810577763966876,
+      "learning_rate": 8.253640385466499e-07,
+      "loss": 1.0281,
+      "step": 8348
+    },
+    {
+      "epoch": 0.8737833594976452,
+      "grad_norm": 1.963834819571071,
+      "learning_rate": 8.240160680499388e-07,
+      "loss": 0.9345,
+      "step": 8349
+    },
+    {
+      "epoch": 0.8738880167451596,
+      "grad_norm": 1.8751548428811395,
+      "learning_rate": 8.226691518923835e-07,
+      "loss": 0.8952,
+      "step": 8350
+    },
+    {
+      "epoch": 0.873992673992674,
+      "grad_norm": 2.330321107761336,
+      "learning_rate": 8.213232902287438e-07,
+      "loss": 0.899,
+      "step": 8351
+    },
+    {
+      "epoch": 0.8740973312401884,
+      "grad_norm": 2.2633398328522847,
+      "learning_rate": 8.199784832136682e-07,
+      "loss": 0.9012,
+      "step": 8352
+    },
+    {
+      "epoch": 0.8742019884877028,
+      "grad_norm": 1.916777768307915,
+      "learning_rate": 8.18634731001674e-07,
+      "loss": 0.8853,
+      "step": 8353
+    },
+    {
+      "epoch": 0.8743066457352172,
+      "grad_norm": 2.1050944988737412,
+      "learning_rate": 8.172920337471601e-07,
+      "loss": 0.9568,
+      "step": 8354
+    },
+    {
+      "epoch": 0.8744113029827315,
+      "grad_norm": 1.8481452188298315,
+      "learning_rate": 8.159503916044087e-07,
+      "loss": 0.9897,
+      "step": 8355
+    },
+    {
+      "epoch": 0.8745159602302459,
+      "grad_norm": 2.250311788077037,
+      "learning_rate": 8.146098047275741e-07,
+      "loss": 0.8498,
+      "step": 8356
+    },
+    {
+      "epoch": 0.8746206174777603,
+      "grad_norm": 2.3260125318486335,
+      "learning_rate": 8.132702732706954e-07,
+      "loss": 0.9126,
+      "step": 8357
+    },
+    {
+      "epoch": 0.8747252747252747,
+      "grad_norm": 2.2314067384407994,
+      "learning_rate": 8.11931797387685e-07,
+      "loss": 0.8325,
+      "step": 8358
+    },
+    {
+      "epoch": 0.8748299319727891,
+      "grad_norm": 1.9457108277580055,
+      "learning_rate": 8.105943772323377e-07,
+      "loss": 0.9133,
+      "step": 8359
+    },
+    {
+      "epoch": 0.8749345892203035,
+      "grad_norm": 2.0107629993432607,
+      "learning_rate": 8.09258012958325e-07,
+      "loss": 0.8153,
+      "step": 8360
+    },
+    {
+      "epoch": 0.8750392464678179,
+      "grad_norm": 2.2793780471848253,
+      "learning_rate": 8.079227047191962e-07,
+      "loss": 0.9221,
+      "step": 8361
+    },
+    {
+      "epoch": 0.8751439037153322,
+      "grad_norm": 2.123456861211417,
+      "learning_rate": 8.065884526683832e-07,
+      "loss": 0.7958,
+      "step": 8362
+    },
+    {
+      "epoch": 0.8752485609628466,
+      "grad_norm": 1.926502112741626,
+      "learning_rate": 8.052552569591965e-07,
+      "loss": 0.7358,
+      "step": 8363
+    },
+    {
+      "epoch": 0.875353218210361,
+      "grad_norm": 2.0489738249039506,
+      "learning_rate": 8.039231177448192e-07,
+      "loss": 0.8492,
+      "step": 8364
+    },
+    {
+      "epoch": 0.8754578754578755,
+      "grad_norm": 1.8975269132659613,
+      "learning_rate": 8.025920351783189e-07,
+      "loss": 0.8582,
+      "step": 8365
+    },
+    {
+      "epoch": 0.8755625327053899,
+      "grad_norm": 2.256895879823261,
+      "learning_rate": 8.012620094126367e-07,
+      "loss": 0.8729,
+      "step": 8366
+    },
+    {
+      "epoch": 0.8756671899529043,
+      "grad_norm": 2.025845185272999,
+      "learning_rate": 7.999330406005989e-07,
+      "loss": 0.8187,
+      "step": 8367
+    },
+    {
+      "epoch": 0.8757718472004187,
+      "grad_norm": 2.0895848407187922,
+      "learning_rate": 7.986051288949059e-07,
+      "loss": 0.9252,
+      "step": 8368
+    },
+    {
+      "epoch": 0.875876504447933,
+      "grad_norm": 2.139667630918293,
+      "learning_rate": 7.972782744481367e-07,
+      "loss": 0.8566,
+      "step": 8369
+    },
+    {
+      "epoch": 0.8759811616954474,
+      "grad_norm": 2.3232743712934294,
+      "learning_rate": 7.959524774127481e-07,
+      "loss": 0.7616,
+      "step": 8370
+    },
+    {
+      "epoch": 0.8760858189429618,
+      "grad_norm": 2.128576816385248,
+      "learning_rate": 7.946277379410771e-07,
+      "loss": 0.7066,
+      "step": 8371
+    },
+    {
+      "epoch": 0.8761904761904762,
+      "grad_norm": 1.9889905519313846,
+      "learning_rate": 7.933040561853433e-07,
+      "loss": 0.8717,
+      "step": 8372
+    },
+    {
+      "epoch": 0.8762951334379906,
+      "grad_norm": 1.7790368676618327,
+      "learning_rate": 7.919814322976371e-07,
+      "loss": 0.7829,
+      "step": 8373
+    },
+    {
+      "epoch": 0.876399790685505,
+      "grad_norm": 2.191730959271324,
+      "learning_rate": 7.906598664299303e-07,
+      "loss": 0.889,
+      "step": 8374
+    },
+    {
+      "epoch": 0.8765044479330194,
+      "grad_norm": 2.018457235121592,
+      "learning_rate": 7.893393587340725e-07,
+      "loss": 0.9284,
+      "step": 8375
+    },
+    {
+      "epoch": 0.8766091051805337,
+      "grad_norm": 1.844476762441355,
+      "learning_rate": 7.880199093617968e-07,
+      "loss": 0.771,
+      "step": 8376
+    },
+    {
+      "epoch": 0.8767137624280481,
+      "grad_norm": 1.9252327144567856,
+      "learning_rate": 7.867015184647075e-07,
+      "loss": 0.8846,
+      "step": 8377
+    },
+    {
+      "epoch": 0.8768184196755625,
+      "grad_norm": 2.428489369199683,
+      "learning_rate": 7.853841861942912e-07,
+      "loss": 0.8245,
+      "step": 8378
+    },
+    {
+      "epoch": 0.8769230769230769,
+      "grad_norm": 2.1246442895723927,
+      "learning_rate": 7.840679127019124e-07,
+      "loss": 0.9374,
+      "step": 8379
+    },
+    {
+      "epoch": 0.8770277341705913,
+      "grad_norm": 2.0282518178011637,
+      "learning_rate": 7.827526981388112e-07,
+      "loss": 0.7992,
+      "step": 8380
+    },
+    {
+      "epoch": 0.8771323914181057,
+      "grad_norm": 2.2457001610010634,
+      "learning_rate": 7.814385426561099e-07,
+      "loss": 0.8949,
+      "step": 8381
+    },
+    {
+      "epoch": 0.8772370486656201,
+      "grad_norm": 2.358939783286938,
+      "learning_rate": 7.801254464048092e-07,
+      "loss": 0.8137,
+      "step": 8382
+    },
+    {
+      "epoch": 0.8773417059131345,
+      "grad_norm": 1.9086185068244006,
+      "learning_rate": 7.788134095357869e-07,
+      "loss": 0.9158,
+      "step": 8383
+    },
+    {
+      "epoch": 0.8774463631606488,
+      "grad_norm": 1.400948820647384,
+      "learning_rate": 7.775024321997959e-07,
+      "loss": 0.6796,
+      "step": 8384
+    },
+    {
+      "epoch": 0.8775510204081632,
+      "grad_norm": 2.1402975917994134,
+      "learning_rate": 7.761925145474702e-07,
+      "loss": 0.9254,
+      "step": 8385
+    },
+    {
+      "epoch": 0.8776556776556776,
+      "grad_norm": 2.28391791015574,
+      "learning_rate": 7.748836567293238e-07,
+      "loss": 0.8438,
+      "step": 8386
+    },
+    {
+      "epoch": 0.877760334903192,
+      "grad_norm": 2.0399330609262143,
+      "learning_rate": 7.735758588957476e-07,
+      "loss": 0.8036,
+      "step": 8387
+    },
+    {
+      "epoch": 0.8778649921507065,
+      "grad_norm": 1.6399077111346803,
+      "learning_rate": 7.722691211970079e-07,
+      "loss": 0.7449,
+      "step": 8388
+    },
+    {
+      "epoch": 0.8779696493982209,
+      "grad_norm": 2.060963302205363,
+      "learning_rate": 7.709634437832537e-07,
+      "loss": 0.8799,
+      "step": 8389
+    },
+    {
+      "epoch": 0.8780743066457353,
+      "grad_norm": 2.0689600387801446,
+      "learning_rate": 7.696588268045069e-07,
+      "loss": 0.892,
+      "step": 8390
+    },
+    {
+      "epoch": 0.8781789638932496,
+      "grad_norm": 1.8775333765697284,
+      "learning_rate": 7.683552704106756e-07,
+      "loss": 0.7523,
+      "step": 8391
+    },
+    {
+      "epoch": 0.878283621140764,
+      "grad_norm": 2.347297278777117,
+      "learning_rate": 7.670527747515366e-07,
+      "loss": 0.8529,
+      "step": 8392
+    },
+    {
+      "epoch": 0.8783882783882784,
+      "grad_norm": 1.8280093832995228,
+      "learning_rate": 7.657513399767525e-07,
+      "loss": 0.8632,
+      "step": 8393
+    },
+    {
+      "epoch": 0.8784929356357928,
+      "grad_norm": 2.4132747406203694,
+      "learning_rate": 7.644509662358568e-07,
+      "loss": 0.7902,
+      "step": 8394
+    },
+    {
+      "epoch": 0.8785975928833072,
+      "grad_norm": 1.7734274400061258,
+      "learning_rate": 7.631516536782691e-07,
+      "loss": 0.8234,
+      "step": 8395
+    },
+    {
+      "epoch": 0.8787022501308216,
+      "grad_norm": 1.9302295122926418,
+      "learning_rate": 7.61853402453282e-07,
+      "loss": 0.9466,
+      "step": 8396
+    },
+    {
+      "epoch": 0.878806907378336,
+      "grad_norm": 2.017176887079925,
+      "learning_rate": 7.605562127100641e-07,
+      "loss": 0.9422,
+      "step": 8397
+    },
+    {
+      "epoch": 0.8789115646258503,
+      "grad_norm": 2.204693966210214,
+      "learning_rate": 7.592600845976694e-07,
+      "loss": 0.9219,
+      "step": 8398
+    },
+    {
+      "epoch": 0.8790162218733647,
+      "grad_norm": 1.9260542762683226,
+      "learning_rate": 7.579650182650245e-07,
+      "loss": 0.8415,
+      "step": 8399
+    },
+    {
+      "epoch": 0.8791208791208791,
+      "grad_norm": 2.3183132581995025,
+      "learning_rate": 7.566710138609323e-07,
+      "loss": 0.8479,
+      "step": 8400
+    },
+    {
+      "epoch": 0.8792255363683935,
+      "grad_norm": 1.9805826657109464,
+      "learning_rate": 7.55378071534082e-07,
+      "loss": 0.8046,
+      "step": 8401
+    },
+    {
+      "epoch": 0.8793301936159079,
+      "grad_norm": 2.177056885011439,
+      "learning_rate": 7.540861914330299e-07,
+      "loss": 0.9871,
+      "step": 8402
+    },
+    {
+      "epoch": 0.8794348508634223,
+      "grad_norm": 2.287633781357303,
+      "learning_rate": 7.527953737062188e-07,
+      "loss": 0.8662,
+      "step": 8403
+    },
+    {
+      "epoch": 0.8795395081109367,
+      "grad_norm": 2.043158350778613,
+      "learning_rate": 7.515056185019642e-07,
+      "loss": 0.8425,
+      "step": 8404
+    },
+    {
+      "epoch": 0.879644165358451,
+      "grad_norm": 2.4480997698564857,
+      "learning_rate": 7.502169259684633e-07,
+      "loss": 0.9648,
+      "step": 8405
+    },
+    {
+      "epoch": 0.8797488226059654,
+      "grad_norm": 2.4174555351369555,
+      "learning_rate": 7.489292962537887e-07,
+      "loss": 0.8642,
+      "step": 8406
+    },
+    {
+      "epoch": 0.8798534798534798,
+      "grad_norm": 2.360136663855378,
+      "learning_rate": 7.476427295058918e-07,
+      "loss": 0.822,
+      "step": 8407
+    },
+    {
+      "epoch": 0.8799581371009942,
+      "grad_norm": 2.1342092374115054,
+      "learning_rate": 7.463572258726027e-07,
+      "loss": 0.9561,
+      "step": 8408
+    },
+    {
+      "epoch": 0.8800627943485086,
+      "grad_norm": 2.2233381286212315,
+      "learning_rate": 7.45072785501626e-07,
+      "loss": 0.9782,
+      "step": 8409
+    },
+    {
+      "epoch": 0.880167451596023,
+      "grad_norm": 2.0969014014371865,
+      "learning_rate": 7.437894085405484e-07,
+      "loss": 0.9458,
+      "step": 8410
+    },
+    {
+      "epoch": 0.8802721088435375,
+      "grad_norm": 2.1757758424267464,
+      "learning_rate": 7.42507095136833e-07,
+      "loss": 0.8587,
+      "step": 8411
+    },
+    {
+      "epoch": 0.8803767660910518,
+      "grad_norm": 2.306166415311752,
+      "learning_rate": 7.412258454378196e-07,
+      "loss": 0.8882,
+      "step": 8412
+    },
+    {
+      "epoch": 0.8804814233385662,
+      "grad_norm": 2.269362231176697,
+      "learning_rate": 7.39945659590724e-07,
+      "loss": 0.8249,
+      "step": 8413
+    },
+    {
+      "epoch": 0.8805860805860806,
+      "grad_norm": 1.938643465951603,
+      "learning_rate": 7.386665377426438e-07,
+      "loss": 0.8922,
+      "step": 8414
+    },
+    {
+      "epoch": 0.880690737833595,
+      "grad_norm": 2.005191893832761,
+      "learning_rate": 7.37388480040555e-07,
+      "loss": 0.8959,
+      "step": 8415
+    },
+    {
+      "epoch": 0.8807953950811094,
+      "grad_norm": 2.1034084182074713,
+      "learning_rate": 7.361114866313069e-07,
+      "loss": 0.8488,
+      "step": 8416
+    },
+    {
+      "epoch": 0.8809000523286238,
+      "grad_norm": 2.0833541849870536,
+      "learning_rate": 7.348355576616295e-07,
+      "loss": 0.8547,
+      "step": 8417
+    },
+    {
+      "epoch": 0.8810047095761382,
+      "grad_norm": 1.9436409663601768,
+      "learning_rate": 7.33560693278128e-07,
+      "loss": 0.9296,
+      "step": 8418
+    },
+    {
+      "epoch": 0.8811093668236525,
+      "grad_norm": 2.6142542944966416,
+      "learning_rate": 7.322868936272876e-07,
+      "loss": 0.9786,
+      "step": 8419
+    },
+    {
+      "epoch": 0.8812140240711669,
+      "grad_norm": 1.975543273238409,
+      "learning_rate": 7.310141588554709e-07,
+      "loss": 0.8701,
+      "step": 8420
+    },
+    {
+      "epoch": 0.8813186813186813,
+      "grad_norm": 2.175997926219932,
+      "learning_rate": 7.297424891089189e-07,
+      "loss": 0.833,
+      "step": 8421
+    },
+    {
+      "epoch": 0.8814233385661957,
+      "grad_norm": 1.779910295096858,
+      "learning_rate": 7.284718845337469e-07,
+      "loss": 0.8065,
+      "step": 8422
+    },
+    {
+      "epoch": 0.8815279958137101,
+      "grad_norm": 2.323280192485811,
+      "learning_rate": 7.272023452759491e-07,
+      "loss": 0.9152,
+      "step": 8423
+    },
+    {
+      "epoch": 0.8816326530612245,
+      "grad_norm": 1.9712072593281076,
+      "learning_rate": 7.259338714813991e-07,
+      "loss": 0.9558,
+      "step": 8424
+    },
+    {
+      "epoch": 0.8817373103087389,
+      "grad_norm": 2.0403158599975058,
+      "learning_rate": 7.2466646329585e-07,
+      "loss": 0.9015,
+      "step": 8425
+    },
+    {
+      "epoch": 0.8818419675562533,
+      "grad_norm": 2.1242870940274057,
+      "learning_rate": 7.234001208649277e-07,
+      "loss": 0.9321,
+      "step": 8426
+    },
+    {
+      "epoch": 0.8819466248037676,
+      "grad_norm": 2.268858095340938,
+      "learning_rate": 7.221348443341369e-07,
+      "loss": 0.9714,
+      "step": 8427
+    },
+    {
+      "epoch": 0.882051282051282,
+      "grad_norm": 1.995883555744217,
+      "learning_rate": 7.208706338488591e-07,
+      "loss": 0.9395,
+      "step": 8428
+    },
+    {
+      "epoch": 0.8821559392987964,
+      "grad_norm": 2.1098847144251054,
+      "learning_rate": 7.196074895543581e-07,
+      "loss": 0.8725,
+      "step": 8429
+    },
+    {
+      "epoch": 0.8822605965463108,
+      "grad_norm": 2.409353406645249,
+      "learning_rate": 7.183454115957688e-07,
+      "loss": 1.0137,
+      "step": 8430
+    },
+    {
+      "epoch": 0.8823652537938252,
+      "grad_norm": 2.1180455101201408,
+      "learning_rate": 7.170844001181087e-07,
+      "loss": 0.9623,
+      "step": 8431
+    },
+    {
+      "epoch": 0.8824699110413396,
+      "grad_norm": 2.151280329173097,
+      "learning_rate": 7.158244552662674e-07,
+      "loss": 0.9246,
+      "step": 8432
+    },
+    {
+      "epoch": 0.8825745682888541,
+      "grad_norm": 2.0323499167485104,
+      "learning_rate": 7.145655771850179e-07,
+      "loss": 0.9656,
+      "step": 8433
+    },
+    {
+      "epoch": 0.8826792255363684,
+      "grad_norm": 2.282886441190445,
+      "learning_rate": 7.13307766019008e-07,
+      "loss": 0.9472,
+      "step": 8434
+    },
+    {
+      "epoch": 0.8827838827838828,
+      "grad_norm": 2.06209446070617,
+      "learning_rate": 7.120510219127619e-07,
+      "loss": 0.9185,
+      "step": 8435
+    },
+    {
+      "epoch": 0.8828885400313972,
+      "grad_norm": 2.5731722325288855,
+      "learning_rate": 7.107953450106819e-07,
+      "loss": 0.8998,
+      "step": 8436
+    },
+    {
+      "epoch": 0.8829931972789116,
+      "grad_norm": 1.8408399665438568,
+      "learning_rate": 7.095407354570483e-07,
+      "loss": 0.8116,
+      "step": 8437
+    },
+    {
+      "epoch": 0.883097854526426,
+      "grad_norm": 2.0268812460354453,
+      "learning_rate": 7.082871933960156e-07,
+      "loss": 0.8808,
+      "step": 8438
+    },
+    {
+      "epoch": 0.8832025117739404,
+      "grad_norm": 1.6490758747411414,
+      "learning_rate": 7.070347189716231e-07,
+      "loss": 0.7978,
+      "step": 8439
+    },
+    {
+      "epoch": 0.8833071690214548,
+      "grad_norm": 1.8039746990089798,
+      "learning_rate": 7.057833123277779e-07,
+      "loss": 0.8668,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8834118262689691,
+      "grad_norm": 2.0025313820891246,
+      "learning_rate": 7.04532973608274e-07,
+      "loss": 0.8699,
+      "step": 8441
+    },
+    {
+      "epoch": 0.8835164835164835,
+      "grad_norm": 2.013068044107957,
+      "learning_rate": 7.032837029567741e-07,
+      "loss": 0.8394,
+      "step": 8442
+    },
+    {
+      "epoch": 0.8836211407639979,
+      "grad_norm": 1.9791349791116444,
+      "learning_rate": 7.020355005168223e-07,
+      "loss": 0.8135,
+      "step": 8443
+    },
+    {
+      "epoch": 0.8837257980115123,
+      "grad_norm": 1.9761998674343688,
+      "learning_rate": 7.007883664318416e-07,
+      "loss": 0.9571,
+      "step": 8444
+    },
+    {
+      "epoch": 0.8838304552590267,
+      "grad_norm": 2.217869103536607,
+      "learning_rate": 6.995423008451296e-07,
+      "loss": 0.8986,
+      "step": 8445
+    },
+    {
+      "epoch": 0.8839351125065411,
+      "grad_norm": 1.7193627330671324,
+      "learning_rate": 6.982973038998608e-07,
+      "loss": 0.7972,
+      "step": 8446
+    },
+    {
+      "epoch": 0.8840397697540555,
+      "grad_norm": 1.9318398791675795,
+      "learning_rate": 6.970533757390874e-07,
+      "loss": 0.9375,
+      "step": 8447
+    },
+    {
+      "epoch": 0.8841444270015698,
+      "grad_norm": 2.040044194610826,
+      "learning_rate": 6.958105165057405e-07,
+      "loss": 0.9477,
+      "step": 8448
+    },
+    {
+      "epoch": 0.8842490842490842,
+      "grad_norm": 2.3099661723992795,
+      "learning_rate": 6.945687263426259e-07,
+      "loss": 0.996,
+      "step": 8449
+    },
+    {
+      "epoch": 0.8843537414965986,
+      "grad_norm": 2.3098910608338126,
+      "learning_rate": 6.933280053924307e-07,
+      "loss": 0.8932,
+      "step": 8450
+    },
+    {
+      "epoch": 0.884458398744113,
+      "grad_norm": 2.0083350641115847,
+      "learning_rate": 6.920883537977141e-07,
+      "loss": 0.8619,
+      "step": 8451
+    },
+    {
+      "epoch": 0.8845630559916274,
+      "grad_norm": 2.17092705026124,
+      "learning_rate": 6.908497717009132e-07,
+      "loss": 0.8771,
+      "step": 8452
+    },
+    {
+      "epoch": 0.8846677132391418,
+      "grad_norm": 1.9588841874329892,
+      "learning_rate": 6.896122592443466e-07,
+      "loss": 0.843,
+      "step": 8453
+    },
+    {
+      "epoch": 0.8847723704866562,
+      "grad_norm": 2.06887109912915,
+      "learning_rate": 6.883758165702048e-07,
+      "loss": 0.9597,
+      "step": 8454
+    },
+    {
+      "epoch": 0.8848770277341705,
+      "grad_norm": 2.0239969794410007,
+      "learning_rate": 6.871404438205598e-07,
+      "loss": 0.8352,
+      "step": 8455
+    },
+    {
+      "epoch": 0.884981684981685,
+      "grad_norm": 2.2891549492476795,
+      "learning_rate": 6.859061411373557e-07,
+      "loss": 0.8596,
+      "step": 8456
+    },
+    {
+      "epoch": 0.8850863422291994,
+      "grad_norm": 2.134710339429134,
+      "learning_rate": 6.846729086624159e-07,
+      "loss": 0.7921,
+      "step": 8457
+    },
+    {
+      "epoch": 0.8851909994767138,
+      "grad_norm": 2.021668403658534,
+      "learning_rate": 6.834407465374438e-07,
+      "loss": 0.9,
+      "step": 8458
+    },
+    {
+      "epoch": 0.8852956567242282,
+      "grad_norm": 1.6910079970551906,
+      "learning_rate": 6.82209654904018e-07,
+      "loss": 0.7481,
+      "step": 8459
+    },
+    {
+      "epoch": 0.8854003139717426,
+      "grad_norm": 2.2018914810401933,
+      "learning_rate": 6.809796339035923e-07,
+      "loss": 0.915,
+      "step": 8460
+    },
+    {
+      "epoch": 0.885504971219257,
+      "grad_norm": 1.6986887147965564,
+      "learning_rate": 6.797506836774981e-07,
+      "loss": 0.7935,
+      "step": 8461
+    },
+    {
+      "epoch": 0.8856096284667713,
+      "grad_norm": 1.9522509649243391,
+      "learning_rate": 6.785228043669423e-07,
+      "loss": 0.8457,
+      "step": 8462
+    },
+    {
+      "epoch": 0.8857142857142857,
+      "grad_norm": 2.5679630478892643,
+      "learning_rate": 6.772959961130154e-07,
+      "loss": 0.9314,
+      "step": 8463
+    },
+    {
+      "epoch": 0.8858189429618001,
+      "grad_norm": 1.8880060673610959,
+      "learning_rate": 6.76070259056677e-07,
+      "loss": 0.8698,
+      "step": 8464
+    },
+    {
+      "epoch": 0.8859236002093145,
+      "grad_norm": 2.0499745907035742,
+      "learning_rate": 6.748455933387666e-07,
+      "loss": 0.8072,
+      "step": 8465
+    },
+    {
+      "epoch": 0.8860282574568289,
+      "grad_norm": 2.0849975990310283,
+      "learning_rate": 6.736219991000015e-07,
+      "loss": 0.8479,
+      "step": 8466
+    },
+    {
+      "epoch": 0.8861329147043433,
+      "grad_norm": 2.285861980076164,
+      "learning_rate": 6.72399476480975e-07,
+      "loss": 0.8301,
+      "step": 8467
+    },
+    {
+      "epoch": 0.8862375719518577,
+      "grad_norm": 2.0055314886862146,
+      "learning_rate": 6.711780256221578e-07,
+      "loss": 0.9137,
+      "step": 8468
+    },
+    {
+      "epoch": 0.8863422291993721,
+      "grad_norm": 2.3060763965788507,
+      "learning_rate": 6.699576466638991e-07,
+      "loss": 0.9079,
+      "step": 8469
+    },
+    {
+      "epoch": 0.8864468864468864,
+      "grad_norm": 2.007322337516734,
+      "learning_rate": 6.687383397464187e-07,
+      "loss": 0.8655,
+      "step": 8470
+    },
+    {
+      "epoch": 0.8865515436944008,
+      "grad_norm": 2.0833706920548356,
+      "learning_rate": 6.67520105009819e-07,
+      "loss": 0.783,
+      "step": 8471
+    },
+    {
+      "epoch": 0.8866562009419152,
+      "grad_norm": 2.0582965508334468,
+      "learning_rate": 6.663029425940804e-07,
+      "loss": 0.9511,
+      "step": 8472
+    },
+    {
+      "epoch": 0.8867608581894296,
+      "grad_norm": 2.2808925451077933,
+      "learning_rate": 6.650868526390541e-07,
+      "loss": 0.8643,
+      "step": 8473
+    },
+    {
+      "epoch": 0.886865515436944,
+      "grad_norm": 1.946899059032306,
+      "learning_rate": 6.63871835284472e-07,
+      "loss": 0.9066,
+      "step": 8474
+    },
+    {
+      "epoch": 0.8869701726844584,
+      "grad_norm": 2.068214617975085,
+      "learning_rate": 6.626578906699421e-07,
+      "loss": 0.8209,
+      "step": 8475
+    },
+    {
+      "epoch": 0.8870748299319728,
+      "grad_norm": 2.093189904582817,
+      "learning_rate": 6.614450189349497e-07,
+      "loss": 0.9761,
+      "step": 8476
+    },
+    {
+      "epoch": 0.8871794871794871,
+      "grad_norm": 2.1709216272284424,
+      "learning_rate": 6.602332202188544e-07,
+      "loss": 0.9507,
+      "step": 8477
+    },
+    {
+      "epoch": 0.8872841444270015,
+      "grad_norm": 1.9156403618097797,
+      "learning_rate": 6.590224946608981e-07,
+      "loss": 0.7868,
+      "step": 8478
+    },
+    {
+      "epoch": 0.887388801674516,
+      "grad_norm": 1.8763348533122814,
+      "learning_rate": 6.578128424001929e-07,
+      "loss": 0.7929,
+      "step": 8479
+    },
+    {
+      "epoch": 0.8874934589220304,
+      "grad_norm": 2.1596639833953897,
+      "learning_rate": 6.566042635757297e-07,
+      "loss": 0.9342,
+      "step": 8480
+    },
+    {
+      "epoch": 0.8875981161695448,
+      "grad_norm": 2.25994957436867,
+      "learning_rate": 6.553967583263776e-07,
+      "loss": 0.8573,
+      "step": 8481
+    },
+    {
+      "epoch": 0.8877027734170592,
+      "grad_norm": 2.171303606502574,
+      "learning_rate": 6.541903267908823e-07,
+      "loss": 0.8721,
+      "step": 8482
+    },
+    {
+      "epoch": 0.8878074306645736,
+      "grad_norm": 2.1186511007584454,
+      "learning_rate": 6.529849691078638e-07,
+      "loss": 0.8509,
+      "step": 8483
+    },
+    {
+      "epoch": 0.8879120879120879,
+      "grad_norm": 1.9924052169713273,
+      "learning_rate": 6.517806854158204e-07,
+      "loss": 0.8332,
+      "step": 8484
+    },
+    {
+      "epoch": 0.8880167451596023,
+      "grad_norm": 1.8798641457276724,
+      "learning_rate": 6.50577475853128e-07,
+      "loss": 0.9086,
+      "step": 8485
+    },
+    {
+      "epoch": 0.8881214024071167,
+      "grad_norm": 2.030261354667674,
+      "learning_rate": 6.493753405580349e-07,
+      "loss": 0.8013,
+      "step": 8486
+    },
+    {
+      "epoch": 0.8882260596546311,
+      "grad_norm": 1.9482170795471423,
+      "learning_rate": 6.48174279668673e-07,
+      "loss": 0.8105,
+      "step": 8487
+    },
+    {
+      "epoch": 0.8883307169021455,
+      "grad_norm": 2.1213839881968064,
+      "learning_rate": 6.46974293323045e-07,
+      "loss": 0.8659,
+      "step": 8488
+    },
+    {
+      "epoch": 0.8884353741496599,
+      "grad_norm": 1.7682685942041678,
+      "learning_rate": 6.457753816590307e-07,
+      "loss": 0.8687,
+      "step": 8489
+    },
+    {
+      "epoch": 0.8885400313971743,
+      "grad_norm": 2.0809007657182397,
+      "learning_rate": 6.445775448143854e-07,
+      "loss": 0.8524,
+      "step": 8490
+    },
+    {
+      "epoch": 0.8886446886446886,
+      "grad_norm": 2.2455986826973278,
+      "learning_rate": 6.433807829267491e-07,
+      "loss": 0.9529,
+      "step": 8491
+    },
+    {
+      "epoch": 0.888749345892203,
+      "grad_norm": 1.9497228881367459,
+      "learning_rate": 6.421850961336262e-07,
+      "loss": 0.9028,
+      "step": 8492
+    },
+    {
+      "epoch": 0.8888540031397174,
+      "grad_norm": 1.7807386421204503,
+      "learning_rate": 6.409904845724068e-07,
+      "loss": 0.8272,
+      "step": 8493
+    },
+    {
+      "epoch": 0.8889586603872318,
+      "grad_norm": 1.9539951561107944,
+      "learning_rate": 6.397969483803546e-07,
+      "loss": 0.8359,
+      "step": 8494
+    },
+    {
+      "epoch": 0.8890633176347462,
+      "grad_norm": 1.8823048125200246,
+      "learning_rate": 6.386044876946073e-07,
+      "loss": 0.9503,
+      "step": 8495
+    },
+    {
+      "epoch": 0.8891679748822606,
+      "grad_norm": 2.013354508723251,
+      "learning_rate": 6.374131026521813e-07,
+      "loss": 0.8966,
+      "step": 8496
+    },
+    {
+      "epoch": 0.889272632129775,
+      "grad_norm": 2.187845858922677,
+      "learning_rate": 6.362227933899701e-07,
+      "loss": 0.8339,
+      "step": 8497
+    },
+    {
+      "epoch": 0.8893772893772893,
+      "grad_norm": 2.1954224044761563,
+      "learning_rate": 6.350335600447433e-07,
+      "loss": 0.9413,
+      "step": 8498
+    },
+    {
+      "epoch": 0.8894819466248037,
+      "grad_norm": 2.1320897727273165,
+      "learning_rate": 6.338454027531449e-07,
+      "loss": 0.952,
+      "step": 8499
+    },
+    {
+      "epoch": 0.8895866038723181,
+      "grad_norm": 2.336668103416007,
+      "learning_rate": 6.326583216516957e-07,
+      "loss": 0.8382,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8896912611198325,
+      "grad_norm": 1.8795899973747725,
+      "learning_rate": 6.314723168767945e-07,
+      "loss": 0.8066,
+      "step": 8501
+    },
+    {
+      "epoch": 0.889795918367347,
+      "grad_norm": 1.9308760249674062,
+      "learning_rate": 6.302873885647176e-07,
+      "loss": 0.8645,
+      "step": 8502
+    },
+    {
+      "epoch": 0.8899005756148614,
+      "grad_norm": 1.9444834126204222,
+      "learning_rate": 6.291035368516141e-07,
+      "loss": 0.7657,
+      "step": 8503
+    },
+    {
+      "epoch": 0.8900052328623758,
+      "grad_norm": 2.1736958135133593,
+      "learning_rate": 6.279207618735117e-07,
+      "loss": 1.0026,
+      "step": 8504
+    },
+    {
+      "epoch": 0.8901098901098901,
+      "grad_norm": 2.1650434307122093,
+      "learning_rate": 6.267390637663107e-07,
+      "loss": 0.8737,
+      "step": 8505
+    },
+    {
+      "epoch": 0.8902145473574045,
+      "grad_norm": 2.5479226003532167,
+      "learning_rate": 6.255584426657946e-07,
+      "loss": 0.8008,
+      "step": 8506
+    },
+    {
+      "epoch": 0.8903192046049189,
+      "grad_norm": 1.8553494681317815,
+      "learning_rate": 6.243788987076172e-07,
+      "loss": 0.8209,
+      "step": 8507
+    },
+    {
+      "epoch": 0.8904238618524333,
+      "grad_norm": 1.9483177698300278,
+      "learning_rate": 6.232004320273111e-07,
+      "loss": 0.8308,
+      "step": 8508
+    },
+    {
+      "epoch": 0.8905285190999477,
+      "grad_norm": 1.6634174156291366,
+      "learning_rate": 6.220230427602814e-07,
+      "loss": 0.827,
+      "step": 8509
+    },
+    {
+      "epoch": 0.8906331763474621,
+      "grad_norm": 2.1685853413646323,
+      "learning_rate": 6.208467310418165e-07,
+      "loss": 0.9,
+      "step": 8510
+    },
+    {
+      "epoch": 0.8907378335949765,
+      "grad_norm": 1.9637945475462335,
+      "learning_rate": 6.196714970070772e-07,
+      "loss": 0.8718,
+      "step": 8511
+    },
+    {
+      "epoch": 0.8908424908424909,
+      "grad_norm": 1.96490255215975,
+      "learning_rate": 6.184973407910977e-07,
+      "loss": 0.8939,
+      "step": 8512
+    },
+    {
+      "epoch": 0.8909471480900052,
+      "grad_norm": 2.101746073662847,
+      "learning_rate": 6.173242625287934e-07,
+      "loss": 0.8335,
+      "step": 8513
+    },
+    {
+      "epoch": 0.8910518053375196,
+      "grad_norm": 2.106827762594044,
+      "learning_rate": 6.161522623549509e-07,
+      "loss": 0.8706,
+      "step": 8514
+    },
+    {
+      "epoch": 0.891156462585034,
+      "grad_norm": 2.1039813494995157,
+      "learning_rate": 6.14981340404236e-07,
+      "loss": 0.8625,
+      "step": 8515
+    },
+    {
+      "epoch": 0.8912611198325484,
+      "grad_norm": 2.1108841193616983,
+      "learning_rate": 6.138114968111918e-07,
+      "loss": 0.9328,
+      "step": 8516
+    },
+    {
+      "epoch": 0.8913657770800628,
+      "grad_norm": 2.3087652071285567,
+      "learning_rate": 6.126427317102335e-07,
+      "loss": 0.9118,
+      "step": 8517
+    },
+    {
+      "epoch": 0.8914704343275772,
+      "grad_norm": 2.2400146974772515,
+      "learning_rate": 6.114750452356555e-07,
+      "loss": 0.9425,
+      "step": 8518
+    },
+    {
+      "epoch": 0.8915750915750916,
+      "grad_norm": 2.311746241511402,
+      "learning_rate": 6.103084375216273e-07,
+      "loss": 0.9304,
+      "step": 8519
+    },
+    {
+      "epoch": 0.8916797488226059,
+      "grad_norm": 2.1638223427184586,
+      "learning_rate": 6.091429087021938e-07,
+      "loss": 0.9901,
+      "step": 8520
+    },
+    {
+      "epoch": 0.8917844060701203,
+      "grad_norm": 1.9528700522695477,
+      "learning_rate": 6.079784589112792e-07,
+      "loss": 0.7986,
+      "step": 8521
+    },
+    {
+      "epoch": 0.8918890633176347,
+      "grad_norm": 1.6876998724103345,
+      "learning_rate": 6.068150882826785e-07,
+      "loss": 0.7391,
+      "step": 8522
+    },
+    {
+      "epoch": 0.8919937205651491,
+      "grad_norm": 1.8784697035517413,
+      "learning_rate": 6.056527969500659e-07,
+      "loss": 0.7807,
+      "step": 8523
+    },
+    {
+      "epoch": 0.8920983778126635,
+      "grad_norm": 1.7297614397650942,
+      "learning_rate": 6.044915850469912e-07,
+      "loss": 0.7549,
+      "step": 8524
+    },
+    {
+      "epoch": 0.892203035060178,
+      "grad_norm": 2.148635473847736,
+      "learning_rate": 6.033314527068801e-07,
+      "loss": 0.901,
+      "step": 8525
+    },
+    {
+      "epoch": 0.8923076923076924,
+      "grad_norm": 2.409030290338374,
+      "learning_rate": 6.021724000630347e-07,
+      "loss": 0.8012,
+      "step": 8526
+    },
+    {
+      "epoch": 0.8924123495552067,
+      "grad_norm": 2.035224893111058,
+      "learning_rate": 6.010144272486307e-07,
+      "loss": 0.8819,
+      "step": 8527
+    },
+    {
+      "epoch": 0.8925170068027211,
+      "grad_norm": 1.9861692794496122,
+      "learning_rate": 5.998575343967238e-07,
+      "loss": 0.8553,
+      "step": 8528
+    },
+    {
+      "epoch": 0.8926216640502355,
+      "grad_norm": 2.139461496658353,
+      "learning_rate": 5.987017216402402e-07,
+      "loss": 0.8632,
+      "step": 8529
+    },
+    {
+      "epoch": 0.8927263212977499,
+      "grad_norm": 2.3325515767132985,
+      "learning_rate": 5.9754698911199e-07,
+      "loss": 0.8565,
+      "step": 8530
+    },
+    {
+      "epoch": 0.8928309785452643,
+      "grad_norm": 2.1213813372529686,
+      "learning_rate": 5.963933369446506e-07,
+      "loss": 0.9007,
+      "step": 8531
+    },
+    {
+      "epoch": 0.8929356357927787,
+      "grad_norm": 1.9716547476429467,
+      "learning_rate": 5.952407652707803e-07,
+      "loss": 0.801,
+      "step": 8532
+    },
+    {
+      "epoch": 0.8930402930402931,
+      "grad_norm": 2.0956419653603167,
+      "learning_rate": 5.940892742228111e-07,
+      "loss": 0.9066,
+      "step": 8533
+    },
+    {
+      "epoch": 0.8931449502878074,
+      "grad_norm": 2.1015997554130053,
+      "learning_rate": 5.929388639330514e-07,
+      "loss": 0.8079,
+      "step": 8534
+    },
+    {
+      "epoch": 0.8932496075353218,
+      "grad_norm": 2.338930939945522,
+      "learning_rate": 5.917895345336866e-07,
+      "loss": 0.9304,
+      "step": 8535
+    },
+    {
+      "epoch": 0.8933542647828362,
+      "grad_norm": 2.370725798752193,
+      "learning_rate": 5.906412861567768e-07,
+      "loss": 0.9179,
+      "step": 8536
+    },
+    {
+      "epoch": 0.8934589220303506,
+      "grad_norm": 2.212434822841478,
+      "learning_rate": 5.894941189342585e-07,
+      "loss": 0.9368,
+      "step": 8537
+    },
+    {
+      "epoch": 0.893563579277865,
+      "grad_norm": 1.7221383035811781,
+      "learning_rate": 5.88348032997943e-07,
+      "loss": 0.818,
+      "step": 8538
+    },
+    {
+      "epoch": 0.8936682365253794,
+      "grad_norm": 2.160032789570768,
+      "learning_rate": 5.872030284795161e-07,
+      "loss": 0.8613,
+      "step": 8539
+    },
+    {
+      "epoch": 0.8937728937728938,
+      "grad_norm": 2.0833805352848525,
+      "learning_rate": 5.860591055105446e-07,
+      "loss": 0.8991,
+      "step": 8540
+    },
+    {
+      "epoch": 0.8938775510204081,
+      "grad_norm": 2.386495900806836,
+      "learning_rate": 5.849162642224659e-07,
+      "loss": 0.973,
+      "step": 8541
+    },
+    {
+      "epoch": 0.8939822082679225,
+      "grad_norm": 2.127775483833274,
+      "learning_rate": 5.837745047465959e-07,
+      "loss": 0.8681,
+      "step": 8542
+    },
+    {
+      "epoch": 0.8940868655154369,
+      "grad_norm": 2.1017239869056916,
+      "learning_rate": 5.826338272141219e-07,
+      "loss": 0.8879,
+      "step": 8543
+    },
+    {
+      "epoch": 0.8941915227629513,
+      "grad_norm": 2.470116877242277,
+      "learning_rate": 5.814942317561134e-07,
+      "loss": 0.6756,
+      "step": 8544
+    },
+    {
+      "epoch": 0.8942961800104657,
+      "grad_norm": 2.4032567630871706,
+      "learning_rate": 5.803557185035092e-07,
+      "loss": 0.9041,
+      "step": 8545
+    },
+    {
+      "epoch": 0.8944008372579801,
+      "grad_norm": 1.847332668697746,
+      "learning_rate": 5.792182875871299e-07,
+      "loss": 0.8516,
+      "step": 8546
+    },
+    {
+      "epoch": 0.8945054945054945,
+      "grad_norm": 2.2507894430237214,
+      "learning_rate": 5.78081939137668e-07,
+      "loss": 0.862,
+      "step": 8547
+    },
+    {
+      "epoch": 0.8946101517530088,
+      "grad_norm": 2.233081933735458,
+      "learning_rate": 5.769466732856899e-07,
+      "loss": 0.955,
+      "step": 8548
+    },
+    {
+      "epoch": 0.8947148090005232,
+      "grad_norm": 2.158698856453009,
+      "learning_rate": 5.758124901616424e-07,
+      "loss": 0.8542,
+      "step": 8549
+    },
+    {
+      "epoch": 0.8948194662480377,
+      "grad_norm": 2.3430196102926213,
+      "learning_rate": 5.746793898958458e-07,
+      "loss": 0.9469,
+      "step": 8550
+    },
+    {
+      "epoch": 0.8949241234955521,
+      "grad_norm": 2.313609302497972,
+      "learning_rate": 5.735473726184937e-07,
+      "loss": 0.859,
+      "step": 8551
+    },
+    {
+      "epoch": 0.8950287807430665,
+      "grad_norm": 2.2653809195446235,
+      "learning_rate": 5.724164384596576e-07,
+      "loss": 0.8801,
+      "step": 8552
+    },
+    {
+      "epoch": 0.8951334379905809,
+      "grad_norm": 1.7697838567450326,
+      "learning_rate": 5.712865875492835e-07,
+      "loss": 0.7973,
+      "step": 8553
+    },
+    {
+      "epoch": 0.8952380952380953,
+      "grad_norm": 2.25827405042216,
+      "learning_rate": 5.701578200171942e-07,
+      "loss": 0.845,
+      "step": 8554
+    },
+    {
+      "epoch": 0.8953427524856097,
+      "grad_norm": 2.3169595444377893,
+      "learning_rate": 5.690301359930883e-07,
+      "loss": 0.8851,
+      "step": 8555
+    },
+    {
+      "epoch": 0.895447409733124,
+      "grad_norm": 2.1775535449606003,
+      "learning_rate": 5.679035356065399e-07,
+      "loss": 0.7539,
+      "step": 8556
+    },
+    {
+      "epoch": 0.8955520669806384,
+      "grad_norm": 1.827925306636399,
+      "learning_rate": 5.667780189869954e-07,
+      "loss": 0.8239,
+      "step": 8557
+    },
+    {
+      "epoch": 0.8956567242281528,
+      "grad_norm": 2.042695270687885,
+      "learning_rate": 5.656535862637779e-07,
+      "loss": 0.8298,
+      "step": 8558
+    },
+    {
+      "epoch": 0.8957613814756672,
+      "grad_norm": 1.8663790088557182,
+      "learning_rate": 5.64530237566091e-07,
+      "loss": 0.8654,
+      "step": 8559
+    },
+    {
+      "epoch": 0.8958660387231816,
+      "grad_norm": 2.3551080593344285,
+      "learning_rate": 5.634079730230069e-07,
+      "loss": 0.8198,
+      "step": 8560
+    },
+    {
+      "epoch": 0.895970695970696,
+      "grad_norm": 2.085971922826618,
+      "learning_rate": 5.622867927634768e-07,
+      "loss": 0.9079,
+      "step": 8561
+    },
+    {
+      "epoch": 0.8960753532182104,
+      "grad_norm": 2.1241341765259976,
+      "learning_rate": 5.611666969163243e-07,
+      "loss": 0.8598,
+      "step": 8562
+    },
+    {
+      "epoch": 0.8961800104657247,
+      "grad_norm": 2.0606089660513596,
+      "learning_rate": 5.600476856102543e-07,
+      "loss": 0.9451,
+      "step": 8563
+    },
+    {
+      "epoch": 0.8962846677132391,
+      "grad_norm": 2.0689539964279797,
+      "learning_rate": 5.589297589738418e-07,
+      "loss": 0.9374,
+      "step": 8564
+    },
+    {
+      "epoch": 0.8963893249607535,
+      "grad_norm": 1.7677701894726472,
+      "learning_rate": 5.578129171355407e-07,
+      "loss": 0.6829,
+      "step": 8565
+    },
+    {
+      "epoch": 0.8964939822082679,
+      "grad_norm": 2.0461327983112727,
+      "learning_rate": 5.56697160223676e-07,
+      "loss": 0.8308,
+      "step": 8566
+    },
+    {
+      "epoch": 0.8965986394557823,
+      "grad_norm": 2.2030981092811928,
+      "learning_rate": 5.555824883664496e-07,
+      "loss": 0.8447,
+      "step": 8567
+    },
+    {
+      "epoch": 0.8967032967032967,
+      "grad_norm": 2.165264235397933,
+      "learning_rate": 5.544689016919425e-07,
+      "loss": 0.8813,
+      "step": 8568
+    },
+    {
+      "epoch": 0.8968079539508111,
+      "grad_norm": 2.040145311512098,
+      "learning_rate": 5.533564003281067e-07,
+      "loss": 0.941,
+      "step": 8569
+    },
+    {
+      "epoch": 0.8969126111983254,
+      "grad_norm": 2.139615944092077,
+      "learning_rate": 5.52244984402771e-07,
+      "loss": 0.8565,
+      "step": 8570
+    },
+    {
+      "epoch": 0.8970172684458398,
+      "grad_norm": 2.000369105386169,
+      "learning_rate": 5.51134654043638e-07,
+      "loss": 0.9503,
+      "step": 8571
+    },
+    {
+      "epoch": 0.8971219256933542,
+      "grad_norm": 2.3047482566850266,
+      "learning_rate": 5.500254093782886e-07,
+      "loss": 0.9249,
+      "step": 8572
+    },
+    {
+      "epoch": 0.8972265829408687,
+      "grad_norm": 2.357051052649448,
+      "learning_rate": 5.489172505341756e-07,
+      "loss": 0.8812,
+      "step": 8573
+    },
+    {
+      "epoch": 0.8973312401883831,
+      "grad_norm": 1.7124549432994773,
+      "learning_rate": 5.478101776386324e-07,
+      "loss": 0.8168,
+      "step": 8574
+    },
+    {
+      "epoch": 0.8974358974358975,
+      "grad_norm": 2.081260844225879,
+      "learning_rate": 5.467041908188608e-07,
+      "loss": 0.9044,
+      "step": 8575
+    },
+    {
+      "epoch": 0.8975405546834119,
+      "grad_norm": 2.3845487096865634,
+      "learning_rate": 5.455992902019413e-07,
+      "loss": 0.9173,
+      "step": 8576
+    },
+    {
+      "epoch": 0.8976452119309262,
+      "grad_norm": 2.236894173393846,
+      "learning_rate": 5.44495475914828e-07,
+      "loss": 0.902,
+      "step": 8577
+    },
+    {
+      "epoch": 0.8977498691784406,
+      "grad_norm": 1.8686621081953942,
+      "learning_rate": 5.433927480843537e-07,
+      "loss": 0.8863,
+      "step": 8578
+    },
+    {
+      "epoch": 0.897854526425955,
+      "grad_norm": 2.0220966066928128,
+      "learning_rate": 5.422911068372227e-07,
+      "loss": 0.8258,
+      "step": 8579
+    },
+    {
+      "epoch": 0.8979591836734694,
+      "grad_norm": 2.407389734858554,
+      "learning_rate": 5.411905523000172e-07,
+      "loss": 0.8185,
+      "step": 8580
+    },
+    {
+      "epoch": 0.8980638409209838,
+      "grad_norm": 2.180556601813161,
+      "learning_rate": 5.400910845991925e-07,
+      "loss": 0.7839,
+      "step": 8581
+    },
+    {
+      "epoch": 0.8981684981684982,
+      "grad_norm": 2.2606617621305434,
+      "learning_rate": 5.389927038610776e-07,
+      "loss": 0.8516,
+      "step": 8582
+    },
+    {
+      "epoch": 0.8982731554160126,
+      "grad_norm": 1.780676054406007,
+      "learning_rate": 5.378954102118827e-07,
+      "loss": 0.7737,
+      "step": 8583
+    },
+    {
+      "epoch": 0.8983778126635269,
+      "grad_norm": 2.0946716817061386,
+      "learning_rate": 5.367992037776859e-07,
+      "loss": 0.9084,
+      "step": 8584
+    },
+    {
+      "epoch": 0.8984824699110413,
+      "grad_norm": 1.91970079060269,
+      "learning_rate": 5.357040846844452e-07,
+      "loss": 0.7988,
+      "step": 8585
+    },
+    {
+      "epoch": 0.8985871271585557,
+      "grad_norm": 1.9948211908466287,
+      "learning_rate": 5.346100530579911e-07,
+      "loss": 0.8677,
+      "step": 8586
+    },
+    {
+      "epoch": 0.8986917844060701,
+      "grad_norm": 2.1808113443571493,
+      "learning_rate": 5.335171090240287e-07,
+      "loss": 0.9124,
+      "step": 8587
+    },
+    {
+      "epoch": 0.8987964416535845,
+      "grad_norm": 1.9339613369467432,
+      "learning_rate": 5.324252527081408e-07,
+      "loss": 0.8807,
+      "step": 8588
+    },
+    {
+      "epoch": 0.8989010989010989,
+      "grad_norm": 1.8243315084924947,
+      "learning_rate": 5.31334484235786e-07,
+      "loss": 0.7639,
+      "step": 8589
+    },
+    {
+      "epoch": 0.8990057561486133,
+      "grad_norm": 1.8725702372061233,
+      "learning_rate": 5.302448037322938e-07,
+      "loss": 0.7613,
+      "step": 8590
+    },
+    {
+      "epoch": 0.8991104133961276,
+      "grad_norm": 2.267799187028563,
+      "learning_rate": 5.29156211322871e-07,
+      "loss": 0.8779,
+      "step": 8591
+    },
+    {
+      "epoch": 0.899215070643642,
+      "grad_norm": 2.147263551468828,
+      "learning_rate": 5.280687071325974e-07,
+      "loss": 0.7696,
+      "step": 8592
+    },
+    {
+      "epoch": 0.8993197278911564,
+      "grad_norm": 2.2925736831888557,
+      "learning_rate": 5.269822912864319e-07,
+      "loss": 0.8893,
+      "step": 8593
+    },
+    {
+      "epoch": 0.8994243851386708,
+      "grad_norm": 2.1055062293520415,
+      "learning_rate": 5.258969639092049e-07,
+      "loss": 0.8334,
+      "step": 8594
+    },
+    {
+      "epoch": 0.8995290423861853,
+      "grad_norm": 2.361570188646754,
+      "learning_rate": 5.248127251256241e-07,
+      "loss": 0.8185,
+      "step": 8595
+    },
+    {
+      "epoch": 0.8996336996336997,
+      "grad_norm": 1.937369636411695,
+      "learning_rate": 5.237295750602667e-07,
+      "loss": 0.8652,
+      "step": 8596
+    },
+    {
+      "epoch": 0.8997383568812141,
+      "grad_norm": 2.090179273646795,
+      "learning_rate": 5.22647513837592e-07,
+      "loss": 0.8299,
+      "step": 8597
+    },
+    {
+      "epoch": 0.8998430141287285,
+      "grad_norm": 1.984663581897822,
+      "learning_rate": 5.215665415819315e-07,
+      "loss": 0.9113,
+      "step": 8598
+    },
+    {
+      "epoch": 0.8999476713762428,
+      "grad_norm": 2.3235181239813163,
+      "learning_rate": 5.204866584174906e-07,
+      "loss": 0.9133,
+      "step": 8599
+    },
+    {
+      "epoch": 0.9000523286237572,
+      "grad_norm": 2.0441409129727797,
+      "learning_rate": 5.194078644683498e-07,
+      "loss": 0.943,
+      "step": 8600
+    },
+    {
+      "epoch": 0.9001569858712716,
+      "grad_norm": 1.8067996809796987,
+      "learning_rate": 5.183301598584633e-07,
+      "loss": 0.8274,
+      "step": 8601
+    },
+    {
+      "epoch": 0.900261643118786,
+      "grad_norm": 2.1626365258620486,
+      "learning_rate": 5.172535447116634e-07,
+      "loss": 0.8934,
+      "step": 8602
+    },
+    {
+      "epoch": 0.9003663003663004,
+      "grad_norm": 2.2909402751050814,
+      "learning_rate": 5.161780191516552e-07,
+      "loss": 0.8364,
+      "step": 8603
+    },
+    {
+      "epoch": 0.9004709576138148,
+      "grad_norm": 1.9983348573884885,
+      "learning_rate": 5.151035833020179e-07,
+      "loss": 0.862,
+      "step": 8604
+    },
+    {
+      "epoch": 0.9005756148613292,
+      "grad_norm": 2.202331715706061,
+      "learning_rate": 5.14030237286206e-07,
+      "loss": 0.8951,
+      "step": 8605
+    },
+    {
+      "epoch": 0.9006802721088435,
+      "grad_norm": 1.838553460141115,
+      "learning_rate": 5.129579812275498e-07,
+      "loss": 0.7951,
+      "step": 8606
+    },
+    {
+      "epoch": 0.9007849293563579,
+      "grad_norm": 2.2012009750979735,
+      "learning_rate": 5.118868152492551e-07,
+      "loss": 0.6918,
+      "step": 8607
+    },
+    {
+      "epoch": 0.9008895866038723,
+      "grad_norm": 1.8524490674209988,
+      "learning_rate": 5.10816739474399e-07,
+      "loss": 0.8473,
+      "step": 8608
+    },
+    {
+      "epoch": 0.9009942438513867,
+      "grad_norm": 2.3959923668906518,
+      "learning_rate": 5.097477540259365e-07,
+      "loss": 0.8172,
+      "step": 8609
+    },
+    {
+      "epoch": 0.9010989010989011,
+      "grad_norm": 1.9678387832442228,
+      "learning_rate": 5.086798590266961e-07,
+      "loss": 0.9161,
+      "step": 8610
+    },
+    {
+      "epoch": 0.9012035583464155,
+      "grad_norm": 1.9199312869676968,
+      "learning_rate": 5.076130545993796e-07,
+      "loss": 0.7752,
+      "step": 8611
+    },
+    {
+      "epoch": 0.9013082155939299,
+      "grad_norm": 2.7222834602307215,
+      "learning_rate": 5.065473408665678e-07,
+      "loss": 0.9413,
+      "step": 8612
+    },
+    {
+      "epoch": 0.9014128728414442,
+      "grad_norm": 1.9979574681439125,
+      "learning_rate": 5.054827179507115e-07,
+      "loss": 0.7943,
+      "step": 8613
+    },
+    {
+      "epoch": 0.9015175300889586,
+      "grad_norm": 2.206392186358271,
+      "learning_rate": 5.044191859741365e-07,
+      "loss": 0.8969,
+      "step": 8614
+    },
+    {
+      "epoch": 0.901622187336473,
+      "grad_norm": 2.1430840266190376,
+      "learning_rate": 5.033567450590482e-07,
+      "loss": 0.8642,
+      "step": 8615
+    },
+    {
+      "epoch": 0.9017268445839874,
+      "grad_norm": 2.017404088868683,
+      "learning_rate": 5.022953953275201e-07,
+      "loss": 0.9434,
+      "step": 8616
+    },
+    {
+      "epoch": 0.9018315018315018,
+      "grad_norm": 2.0133723850020346,
+      "learning_rate": 5.012351369015067e-07,
+      "loss": 0.7414,
+      "step": 8617
+    },
+    {
+      "epoch": 0.9019361590790163,
+      "grad_norm": 2.046862624034036,
+      "learning_rate": 5.001759699028319e-07,
+      "loss": 0.9387,
+      "step": 8618
+    },
+    {
+      "epoch": 0.9020408163265307,
+      "grad_norm": 2.0907637713335987,
+      "learning_rate": 4.99117894453196e-07,
+      "loss": 0.7797,
+      "step": 8619
+    },
+    {
+      "epoch": 0.902145473574045,
+      "grad_norm": 2.013207011952914,
+      "learning_rate": 4.980609106741729e-07,
+      "loss": 0.9043,
+      "step": 8620
+    },
+    {
+      "epoch": 0.9022501308215594,
+      "grad_norm": 1.8844202484567696,
+      "learning_rate": 4.970050186872155e-07,
+      "loss": 0.879,
+      "step": 8621
+    },
+    {
+      "epoch": 0.9023547880690738,
+      "grad_norm": 2.3362225206127696,
+      "learning_rate": 4.959502186136456e-07,
+      "loss": 0.8738,
+      "step": 8622
+    },
+    {
+      "epoch": 0.9024594453165882,
+      "grad_norm": 2.0248898990471367,
+      "learning_rate": 4.948965105746617e-07,
+      "loss": 0.8554,
+      "step": 8623
+    },
+    {
+      "epoch": 0.9025641025641026,
+      "grad_norm": 1.8717907789505432,
+      "learning_rate": 4.938438946913382e-07,
+      "loss": 0.8623,
+      "step": 8624
+    },
+    {
+      "epoch": 0.902668759811617,
+      "grad_norm": 2.1406021198538188,
+      "learning_rate": 4.927923710846228e-07,
+      "loss": 0.8994,
+      "step": 8625
+    },
+    {
+      "epoch": 0.9027734170591314,
+      "grad_norm": 2.105701972491203,
+      "learning_rate": 4.917419398753342e-07,
+      "loss": 0.8342,
+      "step": 8626
+    },
+    {
+      "epoch": 0.9028780743066457,
+      "grad_norm": 2.0643422231373707,
+      "learning_rate": 4.906926011841751e-07,
+      "loss": 0.9147,
+      "step": 8627
+    },
+    {
+      "epoch": 0.9029827315541601,
+      "grad_norm": 1.9814741563458926,
+      "learning_rate": 4.89644355131712e-07,
+      "loss": 0.8054,
+      "step": 8628
+    },
+    {
+      "epoch": 0.9030873888016745,
+      "grad_norm": 2.1051194899963317,
+      "learning_rate": 4.885972018383933e-07,
+      "loss": 0.9514,
+      "step": 8629
+    },
+    {
+      "epoch": 0.9031920460491889,
+      "grad_norm": 2.459613012959799,
+      "learning_rate": 4.87551141424536e-07,
+      "loss": 0.9572,
+      "step": 8630
+    },
+    {
+      "epoch": 0.9032967032967033,
+      "grad_norm": 2.629282062410733,
+      "learning_rate": 4.865061740103361e-07,
+      "loss": 0.8158,
+      "step": 8631
+    },
+    {
+      "epoch": 0.9034013605442177,
+      "grad_norm": 2.0103071109487853,
+      "learning_rate": 4.854622997158643e-07,
+      "loss": 0.8943,
+      "step": 8632
+    },
+    {
+      "epoch": 0.9035060177917321,
+      "grad_norm": 1.9239594017345965,
+      "learning_rate": 4.844195186610622e-07,
+      "loss": 0.8074,
+      "step": 8633
+    },
+    {
+      "epoch": 0.9036106750392464,
+      "grad_norm": 2.1402627344081875,
+      "learning_rate": 4.833778309657467e-07,
+      "loss": 0.8907,
+      "step": 8634
+    },
+    {
+      "epoch": 0.9037153322867608,
+      "grad_norm": 2.137809328262573,
+      "learning_rate": 4.823372367496104e-07,
+      "loss": 0.8235,
+      "step": 8635
+    },
+    {
+      "epoch": 0.9038199895342752,
+      "grad_norm": 2.0550886701913593,
+      "learning_rate": 4.812977361322201e-07,
+      "loss": 0.8927,
+      "step": 8636
+    },
+    {
+      "epoch": 0.9039246467817896,
+      "grad_norm": 1.9128522746883567,
+      "learning_rate": 4.802593292330171e-07,
+      "loss": 0.7968,
+      "step": 8637
+    },
+    {
+      "epoch": 0.904029304029304,
+      "grad_norm": 1.9358701087954804,
+      "learning_rate": 4.792220161713157e-07,
+      "loss": 0.8978,
+      "step": 8638
+    },
+    {
+      "epoch": 0.9041339612768184,
+      "grad_norm": 2.291398568641111,
+      "learning_rate": 4.78185797066304e-07,
+      "loss": 0.8892,
+      "step": 8639
+    },
+    {
+      "epoch": 0.9042386185243328,
+      "grad_norm": 2.4919196805707418,
+      "learning_rate": 4.771506720370466e-07,
+      "loss": 0.9424,
+      "step": 8640
+    },
+    {
+      "epoch": 0.9043432757718473,
+      "grad_norm": 1.6894319599246974,
+      "learning_rate": 4.761166412024831e-07,
+      "loss": 0.7277,
+      "step": 8641
+    },
+    {
+      "epoch": 0.9044479330193615,
+      "grad_norm": 1.9107521958807865,
+      "learning_rate": 4.7508370468142496e-07,
+      "loss": 0.885,
+      "step": 8642
+    },
+    {
+      "epoch": 0.904552590266876,
+      "grad_norm": 2.0614995865460863,
+      "learning_rate": 4.7405186259255833e-07,
+      "loss": 0.7822,
+      "step": 8643
+    },
+    {
+      "epoch": 0.9046572475143904,
+      "grad_norm": 1.7924039712471196,
+      "learning_rate": 4.7302111505444284e-07,
+      "loss": 0.9039,
+      "step": 8644
+    },
+    {
+      "epoch": 0.9047619047619048,
+      "grad_norm": 2.249557355888428,
+      "learning_rate": 4.719914621855137e-07,
+      "loss": 0.8715,
+      "step": 8645
+    },
+    {
+      "epoch": 0.9048665620094192,
+      "grad_norm": 2.1203352917619345,
+      "learning_rate": 4.7096290410408196e-07,
+      "loss": 0.8193,
+      "step": 8646
+    },
+    {
+      "epoch": 0.9049712192569336,
+      "grad_norm": 2.115446838243104,
+      "learning_rate": 4.6993544092832965e-07,
+      "loss": 0.7967,
+      "step": 8647
+    },
+    {
+      "epoch": 0.905075876504448,
+      "grad_norm": 1.6513015444716435,
+      "learning_rate": 4.689090727763157e-07,
+      "loss": 0.6944,
+      "step": 8648
+    },
+    {
+      "epoch": 0.9051805337519623,
+      "grad_norm": 2.1211276289299272,
+      "learning_rate": 4.6788379976596797e-07,
+      "loss": 0.8097,
+      "step": 8649
+    },
+    {
+      "epoch": 0.9052851909994767,
+      "grad_norm": 1.7872427675425049,
+      "learning_rate": 4.668596220150967e-07,
+      "loss": 0.742,
+      "step": 8650
+    },
+    {
+      "epoch": 0.9053898482469911,
+      "grad_norm": 1.898405645689105,
+      "learning_rate": 4.658365396413811e-07,
+      "loss": 0.6804,
+      "step": 8651
+    },
+    {
+      "epoch": 0.9054945054945055,
+      "grad_norm": 2.2484345299825756,
+      "learning_rate": 4.6481455276237485e-07,
+      "loss": 0.9168,
+      "step": 8652
+    },
+    {
+      "epoch": 0.9055991627420199,
+      "grad_norm": 1.8890310215611323,
+      "learning_rate": 4.637936614955063e-07,
+      "loss": 0.8853,
+      "step": 8653
+    },
+    {
+      "epoch": 0.9057038199895343,
+      "grad_norm": 2.116084072039392,
+      "learning_rate": 4.6277386595807717e-07,
+      "loss": 0.9191,
+      "step": 8654
+    },
+    {
+      "epoch": 0.9058084772370487,
+      "grad_norm": 1.8431218097882023,
+      "learning_rate": 4.617551662672659e-07,
+      "loss": 0.8677,
+      "step": 8655
+    },
+    {
+      "epoch": 0.905913134484563,
+      "grad_norm": 2.3481517816362905,
+      "learning_rate": 4.6073756254012224e-07,
+      "loss": 0.8765,
+      "step": 8656
+    },
+    {
+      "epoch": 0.9060177917320774,
+      "grad_norm": 2.0817117165028898,
+      "learning_rate": 4.597210548935693e-07,
+      "loss": 0.8923,
+      "step": 8657
+    },
+    {
+      "epoch": 0.9061224489795918,
+      "grad_norm": 2.2883970597196366,
+      "learning_rate": 4.5870564344440815e-07,
+      "loss": 0.914,
+      "step": 8658
+    },
+    {
+      "epoch": 0.9062271062271062,
+      "grad_norm": 1.83618675401989,
+      "learning_rate": 4.576913283093098e-07,
+      "loss": 0.765,
+      "step": 8659
+    },
+    {
+      "epoch": 0.9063317634746206,
+      "grad_norm": 1.989383983995067,
+      "learning_rate": 4.566781096048234e-07,
+      "loss": 0.7973,
+      "step": 8660
+    },
+    {
+      "epoch": 0.906436420722135,
+      "grad_norm": 2.3861984180984344,
+      "learning_rate": 4.5566598744736903e-07,
+      "loss": 0.9352,
+      "step": 8661
+    },
+    {
+      "epoch": 0.9065410779696494,
+      "grad_norm": 2.0378146649226068,
+      "learning_rate": 4.5465496195324035e-07,
+      "loss": 0.8419,
+      "step": 8662
+    },
+    {
+      "epoch": 0.9066457352171637,
+      "grad_norm": 2.392679580789577,
+      "learning_rate": 4.5364503323860666e-07,
+      "loss": 0.849,
+      "step": 8663
+    },
+    {
+      "epoch": 0.9067503924646781,
+      "grad_norm": 1.9701786010249667,
+      "learning_rate": 4.526362014195107e-07,
+      "loss": 0.9152,
+      "step": 8664
+    },
+    {
+      "epoch": 0.9068550497121926,
+      "grad_norm": 2.2953866617085827,
+      "learning_rate": 4.516284666118709e-07,
+      "loss": 0.9003,
+      "step": 8665
+    },
+    {
+      "epoch": 0.906959706959707,
+      "grad_norm": 1.9748169384630196,
+      "learning_rate": 4.506218289314757e-07,
+      "loss": 0.8263,
+      "step": 8666
+    },
+    {
+      "epoch": 0.9070643642072214,
+      "grad_norm": 1.9629754135097301,
+      "learning_rate": 4.496162884939914e-07,
+      "loss": 0.934,
+      "step": 8667
+    },
+    {
+      "epoch": 0.9071690214547358,
+      "grad_norm": 2.008476459907381,
+      "learning_rate": 4.4861184541495685e-07,
+      "loss": 0.8925,
+      "step": 8668
+    },
+    {
+      "epoch": 0.9072736787022502,
+      "grad_norm": 1.7441402259407277,
+      "learning_rate": 4.4760849980978184e-07,
+      "loss": 0.8045,
+      "step": 8669
+    },
+    {
+      "epoch": 0.9073783359497645,
+      "grad_norm": 2.085568340205773,
+      "learning_rate": 4.466062517937575e-07,
+      "loss": 0.8026,
+      "step": 8670
+    },
+    {
+      "epoch": 0.9074829931972789,
+      "grad_norm": 2.1982232449226977,
+      "learning_rate": 4.456051014820406e-07,
+      "loss": 0.8355,
+      "step": 8671
+    },
+    {
+      "epoch": 0.9075876504447933,
+      "grad_norm": 2.1354303926988765,
+      "learning_rate": 4.4460504898966693e-07,
+      "loss": 0.8294,
+      "step": 8672
+    },
+    {
+      "epoch": 0.9076923076923077,
+      "grad_norm": 2.034301795378783,
+      "learning_rate": 4.4360609443154233e-07,
+      "loss": 0.9312,
+      "step": 8673
+    },
+    {
+      "epoch": 0.9077969649398221,
+      "grad_norm": 2.355315349345066,
+      "learning_rate": 4.426082379224517e-07,
+      "loss": 0.9095,
+      "step": 8674
+    },
+    {
+      "epoch": 0.9079016221873365,
+      "grad_norm": 1.9806257600484884,
+      "learning_rate": 4.4161147957704767e-07,
+      "loss": 0.7933,
+      "step": 8675
+    },
+    {
+      "epoch": 0.9080062794348509,
+      "grad_norm": 2.2251811085208804,
+      "learning_rate": 4.406158195098642e-07,
+      "loss": 0.9325,
+      "step": 8676
+    },
+    {
+      "epoch": 0.9081109366823652,
+      "grad_norm": 2.2630252661426575,
+      "learning_rate": 4.39621257835301e-07,
+      "loss": 0.8204,
+      "step": 8677
+    },
+    {
+      "epoch": 0.9082155939298796,
+      "grad_norm": 2.013858317073263,
+      "learning_rate": 4.386277946676343e-07,
+      "loss": 0.8767,
+      "step": 8678
+    },
+    {
+      "epoch": 0.908320251177394,
+      "grad_norm": 2.051873546111938,
+      "learning_rate": 4.376354301210184e-07,
+      "loss": 0.9409,
+      "step": 8679
+    },
+    {
+      "epoch": 0.9084249084249084,
+      "grad_norm": 2.13004369241599,
+      "learning_rate": 4.3664416430947766e-07,
+      "loss": 0.8222,
+      "step": 8680
+    },
+    {
+      "epoch": 0.9085295656724228,
+      "grad_norm": 2.0442840076206705,
+      "learning_rate": 4.356539973469087e-07,
+      "loss": 0.9183,
+      "step": 8681
+    },
+    {
+      "epoch": 0.9086342229199372,
+      "grad_norm": 1.7421295502754213,
+      "learning_rate": 4.346649293470839e-07,
+      "loss": 0.8429,
+      "step": 8682
+    },
+    {
+      "epoch": 0.9087388801674516,
+      "grad_norm": 2.3968867260487117,
+      "learning_rate": 4.33676960423649e-07,
+      "loss": 0.9008,
+      "step": 8683
+    },
+    {
+      "epoch": 0.908843537414966,
+      "grad_norm": 1.9127124116543281,
+      "learning_rate": 4.326900906901232e-07,
+      "loss": 0.7683,
+      "step": 8684
+    },
+    {
+      "epoch": 0.9089481946624803,
+      "grad_norm": 1.95961808035582,
+      "learning_rate": 4.317043202599036e-07,
+      "loss": 0.8165,
+      "step": 8685
+    },
+    {
+      "epoch": 0.9090528519099947,
+      "grad_norm": 1.9844531465124018,
+      "learning_rate": 4.30719649246254e-07,
+      "loss": 0.8512,
+      "step": 8686
+    },
+    {
+      "epoch": 0.9091575091575091,
+      "grad_norm": 1.9622621837590426,
+      "learning_rate": 4.297360777623161e-07,
+      "loss": 0.7713,
+      "step": 8687
+    },
+    {
+      "epoch": 0.9092621664050236,
+      "grad_norm": 1.8041201019218724,
+      "learning_rate": 4.287536059211017e-07,
+      "loss": 0.8295,
+      "step": 8688
+    },
+    {
+      "epoch": 0.909366823652538,
+      "grad_norm": 2.337575130743714,
+      "learning_rate": 4.277722338355028e-07,
+      "loss": 0.8251,
+      "step": 8689
+    },
+    {
+      "epoch": 0.9094714809000524,
+      "grad_norm": 2.0458103819689866,
+      "learning_rate": 4.267919616182792e-07,
+      "loss": 0.8311,
+      "step": 8690
+    },
+    {
+      "epoch": 0.9095761381475668,
+      "grad_norm": 1.8361080971407697,
+      "learning_rate": 4.2581278938206626e-07,
+      "loss": 0.7123,
+      "step": 8691
+    },
+    {
+      "epoch": 0.9096807953950811,
+      "grad_norm": 2.1827141207863607,
+      "learning_rate": 4.2483471723937075e-07,
+      "loss": 0.9557,
+      "step": 8692
+    },
+    {
+      "epoch": 0.9097854526425955,
+      "grad_norm": 1.7316495003042163,
+      "learning_rate": 4.2385774530257717e-07,
+      "loss": 0.7874,
+      "step": 8693
+    },
+    {
+      "epoch": 0.9098901098901099,
+      "grad_norm": 2.1643266776521997,
+      "learning_rate": 4.2288187368394353e-07,
+      "loss": 0.7739,
+      "step": 8694
+    },
+    {
+      "epoch": 0.9099947671376243,
+      "grad_norm": 1.6984430830527935,
+      "learning_rate": 4.219071024955967e-07,
+      "loss": 0.781,
+      "step": 8695
+    },
+    {
+      "epoch": 0.9100994243851387,
+      "grad_norm": 1.9916952789284774,
+      "learning_rate": 4.209334318495406e-07,
+      "loss": 0.8571,
+      "step": 8696
+    },
+    {
+      "epoch": 0.9102040816326531,
+      "grad_norm": 2.02650344774727,
+      "learning_rate": 4.1996086185765115e-07,
+      "loss": 0.8146,
+      "step": 8697
+    },
+    {
+      "epoch": 0.9103087388801675,
+      "grad_norm": 2.0774078712351667,
+      "learning_rate": 4.1898939263168127e-07,
+      "loss": 0.8396,
+      "step": 8698
+    },
+    {
+      "epoch": 0.9104133961276818,
+      "grad_norm": 2.171351968437315,
+      "learning_rate": 4.1801902428325156e-07,
+      "loss": 0.9199,
+      "step": 8699
+    },
+    {
+      "epoch": 0.9105180533751962,
+      "grad_norm": 2.0626628160103855,
+      "learning_rate": 4.1704975692386186e-07,
+      "loss": 0.8262,
+      "step": 8700
+    },
+    {
+      "epoch": 0.9106227106227106,
+      "grad_norm": 2.2474029662554256,
+      "learning_rate": 4.160815906648796e-07,
+      "loss": 0.8491,
+      "step": 8701
+    },
+    {
+      "epoch": 0.910727367870225,
+      "grad_norm": 1.8343930692503205,
+      "learning_rate": 4.1511452561755263e-07,
+      "loss": 0.9191,
+      "step": 8702
+    },
+    {
+      "epoch": 0.9108320251177394,
+      "grad_norm": 1.919914539345838,
+      "learning_rate": 4.1414856189299635e-07,
+      "loss": 0.8064,
+      "step": 8703
+    },
+    {
+      "epoch": 0.9109366823652538,
+      "grad_norm": 2.2697378486199185,
+      "learning_rate": 4.131836996022043e-07,
+      "loss": 0.8934,
+      "step": 8704
+    },
+    {
+      "epoch": 0.9110413396127682,
+      "grad_norm": 2.1565994352606648,
+      "learning_rate": 4.122199388560388e-07,
+      "loss": 0.8283,
+      "step": 8705
+    },
+    {
+      "epoch": 0.9111459968602825,
+      "grad_norm": 2.0593897209824243,
+      "learning_rate": 4.1125727976523923e-07,
+      "loss": 0.8751,
+      "step": 8706
+    },
+    {
+      "epoch": 0.9112506541077969,
+      "grad_norm": 2.633616179719094,
+      "learning_rate": 4.1029572244041585e-07,
+      "loss": 0.8806,
+      "step": 8707
+    },
+    {
+      "epoch": 0.9113553113553113,
+      "grad_norm": 2.217152686708234,
+      "learning_rate": 4.0933526699205475e-07,
+      "loss": 0.911,
+      "step": 8708
+    },
+    {
+      "epoch": 0.9114599686028257,
+      "grad_norm": 2.166895554817742,
+      "learning_rate": 4.083759135305132e-07,
+      "loss": 0.7358,
+      "step": 8709
+    },
+    {
+      "epoch": 0.9115646258503401,
+      "grad_norm": 2.0236710752805713,
+      "learning_rate": 4.074176621660242e-07,
+      "loss": 0.8996,
+      "step": 8710
+    },
+    {
+      "epoch": 0.9116692830978546,
+      "grad_norm": 2.2962325093594607,
+      "learning_rate": 4.0646051300869295e-07,
+      "loss": 0.9279,
+      "step": 8711
+    },
+    {
+      "epoch": 0.911773940345369,
+      "grad_norm": 2.042939432062761,
+      "learning_rate": 4.0550446616849484e-07,
+      "loss": 0.8324,
+      "step": 8712
+    },
+    {
+      "epoch": 0.9118785975928833,
+      "grad_norm": 2.428545782174381,
+      "learning_rate": 4.0454952175528527e-07,
+      "loss": 0.8856,
+      "step": 8713
+    },
+    {
+      "epoch": 0.9119832548403977,
+      "grad_norm": 2.2988669806643784,
+      "learning_rate": 4.0359567987878655e-07,
+      "loss": 0.8653,
+      "step": 8714
+    },
+    {
+      "epoch": 0.9120879120879121,
+      "grad_norm": 1.9821777641406835,
+      "learning_rate": 4.026429406485988e-07,
+      "loss": 0.9642,
+      "step": 8715
+    },
+    {
+      "epoch": 0.9121925693354265,
+      "grad_norm": 2.301086960995479,
+      "learning_rate": 4.016913041741921e-07,
+      "loss": 0.8863,
+      "step": 8716
+    },
+    {
+      "epoch": 0.9122972265829409,
+      "grad_norm": 2.5204244955883515,
+      "learning_rate": 4.0074077056491355e-07,
+      "loss": 0.8635,
+      "step": 8717
+    },
+    {
+      "epoch": 0.9124018838304553,
+      "grad_norm": 1.8160765406116912,
+      "learning_rate": 3.997913399299791e-07,
+      "loss": 0.7991,
+      "step": 8718
+    },
+    {
+      "epoch": 0.9125065410779697,
+      "grad_norm": 1.934088855695629,
+      "learning_rate": 3.9884301237848255e-07,
+      "loss": 0.9356,
+      "step": 8719
+    },
+    {
+      "epoch": 0.912611198325484,
+      "grad_norm": 2.2496984791324177,
+      "learning_rate": 3.978957880193868e-07,
+      "loss": 0.8782,
+      "step": 8720
+    },
+    {
+      "epoch": 0.9127158555729984,
+      "grad_norm": 2.3632313366350703,
+      "learning_rate": 3.969496669615314e-07,
+      "loss": 0.8665,
+      "step": 8721
+    },
+    {
+      "epoch": 0.9128205128205128,
+      "grad_norm": 2.272870516424909,
+      "learning_rate": 3.9600464931362494e-07,
+      "loss": 0.973,
+      "step": 8722
+    },
+    {
+      "epoch": 0.9129251700680272,
+      "grad_norm": 2.146882820843765,
+      "learning_rate": 3.9506073518425504e-07,
+      "loss": 0.9286,
+      "step": 8723
+    },
+    {
+      "epoch": 0.9130298273155416,
+      "grad_norm": 2.0647435155269576,
+      "learning_rate": 3.9411792468187716e-07,
+      "loss": 0.8374,
+      "step": 8724
+    },
+    {
+      "epoch": 0.913134484563056,
+      "grad_norm": 2.059170164055368,
+      "learning_rate": 3.931762179148235e-07,
+      "loss": 0.8575,
+      "step": 8725
+    },
+    {
+      "epoch": 0.9132391418105704,
+      "grad_norm": 1.9573837147170599,
+      "learning_rate": 3.9223561499129647e-07,
+      "loss": 0.8199,
+      "step": 8726
+    },
+    {
+      "epoch": 0.9133437990580848,
+      "grad_norm": 2.1437953288885017,
+      "learning_rate": 3.91296116019374e-07,
+      "loss": 0.7705,
+      "step": 8727
+    },
+    {
+      "epoch": 0.9134484563055991,
+      "grad_norm": 1.858952500811545,
+      "learning_rate": 3.903577211070075e-07,
+      "loss": 0.825,
+      "step": 8728
+    },
+    {
+      "epoch": 0.9135531135531135,
+      "grad_norm": 2.160108943394208,
+      "learning_rate": 3.894204303620197e-07,
+      "loss": 0.8654,
+      "step": 8729
+    },
+    {
+      "epoch": 0.9136577708006279,
+      "grad_norm": 2.1084513039063117,
+      "learning_rate": 3.884842438921077e-07,
+      "loss": 1.033,
+      "step": 8730
+    },
+    {
+      "epoch": 0.9137624280481423,
+      "grad_norm": 2.7878592333589607,
+      "learning_rate": 3.875491618048388e-07,
+      "loss": 0.9281,
+      "step": 8731
+    },
+    {
+      "epoch": 0.9138670852956567,
+      "grad_norm": 1.991323149670702,
+      "learning_rate": 3.866151842076593e-07,
+      "loss": 0.9283,
+      "step": 8732
+    },
+    {
+      "epoch": 0.9139717425431712,
+      "grad_norm": 2.2833156721267507,
+      "learning_rate": 3.8568231120788334e-07,
+      "loss": 0.9989,
+      "step": 8733
+    },
+    {
+      "epoch": 0.9140763997906856,
+      "grad_norm": 1.9572296751214142,
+      "learning_rate": 3.8475054291270077e-07,
+      "loss": 0.7219,
+      "step": 8734
+    },
+    {
+      "epoch": 0.9141810570381999,
+      "grad_norm": 2.2701248411948085,
+      "learning_rate": 3.838198794291714e-07,
+      "loss": 0.911,
+      "step": 8735
+    },
+    {
+      "epoch": 0.9142857142857143,
+      "grad_norm": 1.924604288549798,
+      "learning_rate": 3.8289032086423095e-07,
+      "loss": 0.9511,
+      "step": 8736
+    },
+    {
+      "epoch": 0.9143903715332287,
+      "grad_norm": 2.141850749754829,
+      "learning_rate": 3.8196186732469056e-07,
+      "loss": 0.8538,
+      "step": 8737
+    },
+    {
+      "epoch": 0.9144950287807431,
+      "grad_norm": 1.8871726093707848,
+      "learning_rate": 3.810345189172293e-07,
+      "loss": 0.8743,
+      "step": 8738
+    },
+    {
+      "epoch": 0.9145996860282575,
+      "grad_norm": 2.345605190515875,
+      "learning_rate": 3.80108275748402e-07,
+      "loss": 0.9693,
+      "step": 8739
+    },
+    {
+      "epoch": 0.9147043432757719,
+      "grad_norm": 1.8872898192305327,
+      "learning_rate": 3.791831379246347e-07,
+      "loss": 0.8189,
+      "step": 8740
+    },
+    {
+      "epoch": 0.9148090005232863,
+      "grad_norm": 2.146002787179079,
+      "learning_rate": 3.7825910555222666e-07,
+      "loss": 0.8324,
+      "step": 8741
+    },
+    {
+      "epoch": 0.9149136577708006,
+      "grad_norm": 2.085783356155515,
+      "learning_rate": 3.773361787373553e-07,
+      "loss": 0.8937,
+      "step": 8742
+    },
+    {
+      "epoch": 0.915018315018315,
+      "grad_norm": 2.383660628094254,
+      "learning_rate": 3.7641435758606347e-07,
+      "loss": 0.9274,
+      "step": 8743
+    },
+    {
+      "epoch": 0.9151229722658294,
+      "grad_norm": 1.887353522221794,
+      "learning_rate": 3.754936422042699e-07,
+      "loss": 0.8397,
+      "step": 8744
+    },
+    {
+      "epoch": 0.9152276295133438,
+      "grad_norm": 2.1935449423146656,
+      "learning_rate": 3.745740326977687e-07,
+      "loss": 0.8892,
+      "step": 8745
+    },
+    {
+      "epoch": 0.9153322867608582,
+      "grad_norm": 1.9262362500423253,
+      "learning_rate": 3.7365552917222213e-07,
+      "loss": 0.8224,
+      "step": 8746
+    },
+    {
+      "epoch": 0.9154369440083726,
+      "grad_norm": 2.0505509798208417,
+      "learning_rate": 3.7273813173317243e-07,
+      "loss": 0.8285,
+      "step": 8747
+    },
+    {
+      "epoch": 0.915541601255887,
+      "grad_norm": 2.3324333915301216,
+      "learning_rate": 3.718218404860263e-07,
+      "loss": 0.8936,
+      "step": 8748
+    },
+    {
+      "epoch": 0.9156462585034013,
+      "grad_norm": 2.031276939093479,
+      "learning_rate": 3.7090665553607076e-07,
+      "loss": 0.7884,
+      "step": 8749
+    },
+    {
+      "epoch": 0.9157509157509157,
+      "grad_norm": 2.2510972837867875,
+      "learning_rate": 3.6999257698845825e-07,
+      "loss": 0.9864,
+      "step": 8750
+    },
+    {
+      "epoch": 0.9158555729984301,
+      "grad_norm": 2.3120088138033257,
+      "learning_rate": 3.690796049482226e-07,
+      "loss": 0.8674,
+      "step": 8751
+    },
+    {
+      "epoch": 0.9159602302459445,
+      "grad_norm": 2.5331940368689367,
+      "learning_rate": 3.681677395202632e-07,
+      "loss": 0.8305,
+      "step": 8752
+    },
+    {
+      "epoch": 0.9160648874934589,
+      "grad_norm": 2.134818511890312,
+      "learning_rate": 3.6725698080935626e-07,
+      "loss": 0.8018,
+      "step": 8753
+    },
+    {
+      "epoch": 0.9161695447409733,
+      "grad_norm": 2.2432740685775148,
+      "learning_rate": 3.6634732892015025e-07,
+      "loss": 0.9043,
+      "step": 8754
+    },
+    {
+      "epoch": 0.9162742019884877,
+      "grad_norm": 1.9549596052281204,
+      "learning_rate": 3.654387839571649e-07,
+      "loss": 0.9349,
+      "step": 8755
+    },
+    {
+      "epoch": 0.916378859236002,
+      "grad_norm": 2.0607140029707605,
+      "learning_rate": 3.645313460247957e-07,
+      "loss": 0.8228,
+      "step": 8756
+    },
+    {
+      "epoch": 0.9164835164835164,
+      "grad_norm": 1.9888703343791325,
+      "learning_rate": 3.6362501522730797e-07,
+      "loss": 0.7932,
+      "step": 8757
+    },
+    {
+      "epoch": 0.9165881737310309,
+      "grad_norm": 2.1547284667729314,
+      "learning_rate": 3.6271979166884076e-07,
+      "loss": 0.8284,
+      "step": 8758
+    },
+    {
+      "epoch": 0.9166928309785453,
+      "grad_norm": 2.3830343538126315,
+      "learning_rate": 3.6181567545340634e-07,
+      "loss": 0.8717,
+      "step": 8759
+    },
+    {
+      "epoch": 0.9167974882260597,
+      "grad_norm": 2.7710627812990074,
+      "learning_rate": 3.6091266668488835e-07,
+      "loss": 0.7445,
+      "step": 8760
+    },
+    {
+      "epoch": 0.9169021454735741,
+      "grad_norm": 1.8300014783449674,
+      "learning_rate": 3.6001076546704595e-07,
+      "loss": 0.92,
+      "step": 8761
+    },
+    {
+      "epoch": 0.9170068027210885,
+      "grad_norm": 2.035387613216914,
+      "learning_rate": 3.591099719035096e-07,
+      "loss": 0.7801,
+      "step": 8762
+    },
+    {
+      "epoch": 0.9171114599686028,
+      "grad_norm": 2.104498222120233,
+      "learning_rate": 3.582102860977821e-07,
+      "loss": 0.8691,
+      "step": 8763
+    },
+    {
+      "epoch": 0.9172161172161172,
+      "grad_norm": 2.083491104776923,
+      "learning_rate": 3.5731170815323733e-07,
+      "loss": 0.918,
+      "step": 8764
+    },
+    {
+      "epoch": 0.9173207744636316,
+      "grad_norm": 2.0988470418475984,
+      "learning_rate": 3.5641423817312503e-07,
+      "loss": 0.8197,
+      "step": 8765
+    },
+    {
+      "epoch": 0.917425431711146,
+      "grad_norm": 2.3323620083693677,
+      "learning_rate": 3.555178762605671e-07,
+      "loss": 0.9047,
+      "step": 8766
+    },
+    {
+      "epoch": 0.9175300889586604,
+      "grad_norm": 2.235958503717904,
+      "learning_rate": 3.546226225185567e-07,
+      "loss": 0.8173,
+      "step": 8767
+    },
+    {
+      "epoch": 0.9176347462061748,
+      "grad_norm": 2.3113114359456826,
+      "learning_rate": 3.537284770499605e-07,
+      "loss": 0.9114,
+      "step": 8768
+    },
+    {
+      "epoch": 0.9177394034536892,
+      "grad_norm": 2.095066016594399,
+      "learning_rate": 3.528354399575151e-07,
+      "loss": 1.0252,
+      "step": 8769
+    },
+    {
+      "epoch": 0.9178440607012036,
+      "grad_norm": 1.9933094078470959,
+      "learning_rate": 3.519435113438352e-07,
+      "loss": 0.8784,
+      "step": 8770
+    },
+    {
+      "epoch": 0.9179487179487179,
+      "grad_norm": 2.0786146110365746,
+      "learning_rate": 3.510526913114065e-07,
+      "loss": 0.8294,
+      "step": 8771
+    },
+    {
+      "epoch": 0.9180533751962323,
+      "grad_norm": 2.195402968074223,
+      "learning_rate": 3.5016297996258276e-07,
+      "loss": 0.9349,
+      "step": 8772
+    },
+    {
+      "epoch": 0.9181580324437467,
+      "grad_norm": 2.05314297727512,
+      "learning_rate": 3.4927437739959547e-07,
+      "loss": 0.7679,
+      "step": 8773
+    },
+    {
+      "epoch": 0.9182626896912611,
+      "grad_norm": 2.244479903766985,
+      "learning_rate": 3.483868837245452e-07,
+      "loss": 0.929,
+      "step": 8774
+    },
+    {
+      "epoch": 0.9183673469387755,
+      "grad_norm": 2.015140547965409,
+      "learning_rate": 3.4750049903940817e-07,
+      "loss": 0.8448,
+      "step": 8775
+    },
+    {
+      "epoch": 0.9184720041862899,
+      "grad_norm": 1.8382679008968412,
+      "learning_rate": 3.466152234460318e-07,
+      "loss": 0.751,
+      "step": 8776
+    },
+    {
+      "epoch": 0.9185766614338043,
+      "grad_norm": 2.205165672351134,
+      "learning_rate": 3.4573105704613586e-07,
+      "loss": 0.9405,
+      "step": 8777
+    },
+    {
+      "epoch": 0.9186813186813186,
+      "grad_norm": 1.821503436915656,
+      "learning_rate": 3.4484799994131345e-07,
+      "loss": 0.808,
+      "step": 8778
+    },
+    {
+      "epoch": 0.918785975928833,
+      "grad_norm": 2.259936480939421,
+      "learning_rate": 3.439660522330268e-07,
+      "loss": 0.8702,
+      "step": 8779
+    },
+    {
+      "epoch": 0.9188906331763474,
+      "grad_norm": 2.0808618852041847,
+      "learning_rate": 3.43085214022616e-07,
+      "loss": 0.7654,
+      "step": 8780
+    },
+    {
+      "epoch": 0.9189952904238619,
+      "grad_norm": 1.970573498767891,
+      "learning_rate": 3.4220548541129217e-07,
+      "loss": 0.8929,
+      "step": 8781
+    },
+    {
+      "epoch": 0.9190999476713763,
+      "grad_norm": 1.9944791324781226,
+      "learning_rate": 3.413268665001357e-07,
+      "loss": 0.8615,
+      "step": 8782
+    },
+    {
+      "epoch": 0.9192046049188907,
+      "grad_norm": 1.9279258269514712,
+      "learning_rate": 3.404493573901024e-07,
+      "loss": 0.7665,
+      "step": 8783
+    },
+    {
+      "epoch": 0.9193092621664051,
+      "grad_norm": 2.180216746439435,
+      "learning_rate": 3.3957295818201954e-07,
+      "loss": 0.9315,
+      "step": 8784
+    },
+    {
+      "epoch": 0.9194139194139194,
+      "grad_norm": 1.9619360610575096,
+      "learning_rate": 3.3869766897658753e-07,
+      "loss": 0.878,
+      "step": 8785
+    },
+    {
+      "epoch": 0.9195185766614338,
+      "grad_norm": 2.030802148720462,
+      "learning_rate": 3.378234898743804e-07,
+      "loss": 0.73,
+      "step": 8786
+    },
+    {
+      "epoch": 0.9196232339089482,
+      "grad_norm": 2.1100710529585625,
+      "learning_rate": 3.369504209758401e-07,
+      "loss": 0.8821,
+      "step": 8787
+    },
+    {
+      "epoch": 0.9197278911564626,
+      "grad_norm": 2.2642436944668822,
+      "learning_rate": 3.3607846238128517e-07,
+      "loss": 0.8657,
+      "step": 8788
+    },
+    {
+      "epoch": 0.919832548403977,
+      "grad_norm": 1.8980867961682124,
+      "learning_rate": 3.352076141909055e-07,
+      "loss": 0.7698,
+      "step": 8789
+    },
+    {
+      "epoch": 0.9199372056514914,
+      "grad_norm": 2.116331603985098,
+      "learning_rate": 3.3433787650476444e-07,
+      "loss": 0.7825,
+      "step": 8790
+    },
+    {
+      "epoch": 0.9200418628990058,
+      "grad_norm": 2.016895335801583,
+      "learning_rate": 3.3346924942279645e-07,
+      "loss": 0.8306,
+      "step": 8791
+    },
+    {
+      "epoch": 0.9201465201465201,
+      "grad_norm": 2.1969878425950435,
+      "learning_rate": 3.3260173304480724e-07,
+      "loss": 0.9345,
+      "step": 8792
+    },
+    {
+      "epoch": 0.9202511773940345,
+      "grad_norm": 2.4238866705980904,
+      "learning_rate": 3.317353274704749e-07,
+      "loss": 0.8095,
+      "step": 8793
+    },
+    {
+      "epoch": 0.9203558346415489,
+      "grad_norm": 2.061511830200133,
+      "learning_rate": 3.3087003279935527e-07,
+      "loss": 0.9398,
+      "step": 8794
+    },
+    {
+      "epoch": 0.9204604918890633,
+      "grad_norm": 1.9785156370811716,
+      "learning_rate": 3.3000584913087e-07,
+      "loss": 0.8761,
+      "step": 8795
+    },
+    {
+      "epoch": 0.9205651491365777,
+      "grad_norm": 2.6239846193899687,
+      "learning_rate": 3.2914277656431405e-07,
+      "loss": 0.9125,
+      "step": 8796
+    },
+    {
+      "epoch": 0.9206698063840921,
+      "grad_norm": 2.098352277907113,
+      "learning_rate": 3.282808151988603e-07,
+      "loss": 0.7692,
+      "step": 8797
+    },
+    {
+      "epoch": 0.9207744636316065,
+      "grad_norm": 1.9428905633278999,
+      "learning_rate": 3.2741996513354637e-07,
+      "loss": 0.8322,
+      "step": 8798
+    },
+    {
+      "epoch": 0.9208791208791208,
+      "grad_norm": 1.783743284332841,
+      "learning_rate": 3.265602264672862e-07,
+      "loss": 0.7968,
+      "step": 8799
+    },
+    {
+      "epoch": 0.9209837781266352,
+      "grad_norm": 1.9603968225564672,
+      "learning_rate": 3.257015992988677e-07,
+      "loss": 0.9033,
+      "step": 8800
+    },
+    {
+      "epoch": 0.9210884353741496,
+      "grad_norm": 1.9634203307317444,
+      "learning_rate": 3.248440837269462e-07,
+      "loss": 0.9142,
+      "step": 8801
+    },
+    {
+      "epoch": 0.921193092621664,
+      "grad_norm": 1.8398499107034954,
+      "learning_rate": 3.239876798500541e-07,
+      "loss": 0.8375,
+      "step": 8802
+    },
+    {
+      "epoch": 0.9212977498691785,
+      "grad_norm": 2.113545071381351,
+      "learning_rate": 3.2313238776659037e-07,
+      "loss": 0.842,
+      "step": 8803
+    },
+    {
+      "epoch": 0.9214024071166929,
+      "grad_norm": 2.0202953316393164,
+      "learning_rate": 3.222782075748354e-07,
+      "loss": 0.7569,
+      "step": 8804
+    },
+    {
+      "epoch": 0.9215070643642073,
+      "grad_norm": 2.1413122563788227,
+      "learning_rate": 3.2142513937293064e-07,
+      "loss": 0.8861,
+      "step": 8805
+    },
+    {
+      "epoch": 0.9216117216117216,
+      "grad_norm": 2.235823853768752,
+      "learning_rate": 3.2057318325889877e-07,
+      "loss": 0.9496,
+      "step": 8806
+    },
+    {
+      "epoch": 0.921716378859236,
+      "grad_norm": 2.16956827769149,
+      "learning_rate": 3.1972233933063144e-07,
+      "loss": 0.8859,
+      "step": 8807
+    },
+    {
+      "epoch": 0.9218210361067504,
+      "grad_norm": 2.1026176075046874,
+      "learning_rate": 3.1887260768588946e-07,
+      "loss": 0.8599,
+      "step": 8808
+    },
+    {
+      "epoch": 0.9219256933542648,
+      "grad_norm": 1.8287944523325435,
+      "learning_rate": 3.180239884223124e-07,
+      "loss": 0.8797,
+      "step": 8809
+    },
+    {
+      "epoch": 0.9220303506017792,
+      "grad_norm": 2.0117907483069764,
+      "learning_rate": 3.1717648163740567e-07,
+      "loss": 0.9159,
+      "step": 8810
+    },
+    {
+      "epoch": 0.9221350078492936,
+      "grad_norm": 2.1179998673763234,
+      "learning_rate": 3.163300874285513e-07,
+      "loss": 0.7744,
+      "step": 8811
+    },
+    {
+      "epoch": 0.922239665096808,
+      "grad_norm": 2.2470761723344452,
+      "learning_rate": 3.154848058929982e-07,
+      "loss": 0.9273,
+      "step": 8812
+    },
+    {
+      "epoch": 0.9223443223443224,
+      "grad_norm": 1.881792495195148,
+      "learning_rate": 3.146406371278754e-07,
+      "loss": 0.8834,
+      "step": 8813
+    },
+    {
+      "epoch": 0.9224489795918367,
+      "grad_norm": 1.979370856312482,
+      "learning_rate": 3.1379758123017636e-07,
+      "loss": 0.8525,
+      "step": 8814
+    },
+    {
+      "epoch": 0.9225536368393511,
+      "grad_norm": 2.139478294351095,
+      "learning_rate": 3.1295563829677247e-07,
+      "loss": 0.8343,
+      "step": 8815
+    },
+    {
+      "epoch": 0.9226582940868655,
+      "grad_norm": 2.1238442720838013,
+      "learning_rate": 3.121148084244041e-07,
+      "loss": 1.0092,
+      "step": 8816
+    },
+    {
+      "epoch": 0.9227629513343799,
+      "grad_norm": 1.7314699596986751,
+      "learning_rate": 3.1127509170968295e-07,
+      "loss": 0.8082,
+      "step": 8817
+    },
+    {
+      "epoch": 0.9228676085818943,
+      "grad_norm": 2.012711651464119,
+      "learning_rate": 3.10436488249094e-07,
+      "loss": 0.8154,
+      "step": 8818
+    },
+    {
+      "epoch": 0.9229722658294087,
+      "grad_norm": 2.611942421259733,
+      "learning_rate": 3.0959899813899576e-07,
+      "loss": 0.6697,
+      "step": 8819
+    },
+    {
+      "epoch": 0.9230769230769231,
+      "grad_norm": 2.2358990969751193,
+      "learning_rate": 3.0876262147561784e-07,
+      "loss": 0.9779,
+      "step": 8820
+    },
+    {
+      "epoch": 0.9231815803244374,
+      "grad_norm": 2.303016822984439,
+      "learning_rate": 3.0792735835506015e-07,
+      "loss": 0.7672,
+      "step": 8821
+    },
+    {
+      "epoch": 0.9232862375719518,
+      "grad_norm": 2.125149271133452,
+      "learning_rate": 3.0709320887329697e-07,
+      "loss": 0.8014,
+      "step": 8822
+    },
+    {
+      "epoch": 0.9233908948194662,
+      "grad_norm": 1.825121847185592,
+      "learning_rate": 3.0626017312617274e-07,
+      "loss": 0.841,
+      "step": 8823
+    },
+    {
+      "epoch": 0.9234955520669806,
+      "grad_norm": 1.8166108344674616,
+      "learning_rate": 3.0542825120940757e-07,
+      "loss": 0.9354,
+      "step": 8824
+    },
+    {
+      "epoch": 0.923600209314495,
+      "grad_norm": 2.1961545359673376,
+      "learning_rate": 3.045974432185883e-07,
+      "loss": 0.9738,
+      "step": 8825
+    },
+    {
+      "epoch": 0.9237048665620095,
+      "grad_norm": 2.3971264132457257,
+      "learning_rate": 3.0376774924917863e-07,
+      "loss": 1.0354,
+      "step": 8826
+    },
+    {
+      "epoch": 0.9238095238095239,
+      "grad_norm": 2.021091464574574,
+      "learning_rate": 3.029391693965089e-07,
+      "loss": 0.7179,
+      "step": 8827
+    },
+    {
+      "epoch": 0.9239141810570382,
+      "grad_norm": 2.02476604473955,
+      "learning_rate": 3.0211170375578747e-07,
+      "loss": 0.9331,
+      "step": 8828
+    },
+    {
+      "epoch": 0.9240188383045526,
+      "grad_norm": 2.2717440474645407,
+      "learning_rate": 3.0128535242209155e-07,
+      "loss": 0.8373,
+      "step": 8829
+    },
+    {
+      "epoch": 0.924123495552067,
+      "grad_norm": 2.419214715389792,
+      "learning_rate": 3.004601154903697e-07,
+      "loss": 0.9407,
+      "step": 8830
+    },
+    {
+      "epoch": 0.9242281527995814,
+      "grad_norm": 2.0851387695815573,
+      "learning_rate": 2.9963599305544267e-07,
+      "loss": 0.9082,
+      "step": 8831
+    },
+    {
+      "epoch": 0.9243328100470958,
+      "grad_norm": 2.2246862727059207,
+      "learning_rate": 2.9881298521200477e-07,
+      "loss": 0.8642,
+      "step": 8832
+    },
+    {
+      "epoch": 0.9244374672946102,
+      "grad_norm": 2.3566781906040704,
+      "learning_rate": 2.9799109205462254e-07,
+      "loss": 0.9774,
+      "step": 8833
+    },
+    {
+      "epoch": 0.9245421245421246,
+      "grad_norm": 2.2258715574619967,
+      "learning_rate": 2.971703136777315e-07,
+      "loss": 0.9839,
+      "step": 8834
+    },
+    {
+      "epoch": 0.9246467817896389,
+      "grad_norm": 2.37543036241895,
+      "learning_rate": 2.963506501756419e-07,
+      "loss": 0.8264,
+      "step": 8835
+    },
+    {
+      "epoch": 0.9247514390371533,
+      "grad_norm": 2.1881079550525584,
+      "learning_rate": 2.955321016425328e-07,
+      "loss": 0.8516,
+      "step": 8836
+    },
+    {
+      "epoch": 0.9248560962846677,
+      "grad_norm": 2.0859708402910884,
+      "learning_rate": 2.947146681724578e-07,
+      "loss": 0.8664,
+      "step": 8837
+    },
+    {
+      "epoch": 0.9249607535321821,
+      "grad_norm": 2.1999782584393506,
+      "learning_rate": 2.938983498593428e-07,
+      "loss": 0.8514,
+      "step": 8838
+    },
+    {
+      "epoch": 0.9250654107796965,
+      "grad_norm": 2.1661333998629333,
+      "learning_rate": 2.930831467969841e-07,
+      "loss": 0.8986,
+      "step": 8839
+    },
+    {
+      "epoch": 0.9251700680272109,
+      "grad_norm": 2.4097416377658254,
+      "learning_rate": 2.922690590790478e-07,
+      "loss": 0.9982,
+      "step": 8840
+    },
+    {
+      "epoch": 0.9252747252747253,
+      "grad_norm": 1.7688846225655979,
+      "learning_rate": 2.91456086799079e-07,
+      "loss": 0.7996,
+      "step": 8841
+    },
+    {
+      "epoch": 0.9253793825222396,
+      "grad_norm": 2.3499275256087273,
+      "learning_rate": 2.906442300504841e-07,
+      "loss": 0.9031,
+      "step": 8842
+    },
+    {
+      "epoch": 0.925484039769754,
+      "grad_norm": 2.0100308955005257,
+      "learning_rate": 2.8983348892655303e-07,
+      "loss": 0.9804,
+      "step": 8843
+    },
+    {
+      "epoch": 0.9255886970172684,
+      "grad_norm": 2.6234232462180898,
+      "learning_rate": 2.8902386352043675e-07,
+      "loss": 0.8138,
+      "step": 8844
+    },
+    {
+      "epoch": 0.9256933542647828,
+      "grad_norm": 2.3067667180043174,
+      "learning_rate": 2.8821535392516644e-07,
+      "loss": 0.8897,
+      "step": 8845
+    },
+    {
+      "epoch": 0.9257980115122972,
+      "grad_norm": 2.275440873262541,
+      "learning_rate": 2.874079602336377e-07,
+      "loss": 0.8794,
+      "step": 8846
+    },
+    {
+      "epoch": 0.9259026687598116,
+      "grad_norm": 2.5876415121542107,
+      "learning_rate": 2.866016825386253e-07,
+      "loss": 0.8667,
+      "step": 8847
+    },
+    {
+      "epoch": 0.926007326007326,
+      "grad_norm": 1.8078289776282361,
+      "learning_rate": 2.857965209327695e-07,
+      "loss": 0.8499,
+      "step": 8848
+    },
+    {
+      "epoch": 0.9261119832548403,
+      "grad_norm": 1.8051645575388477,
+      "learning_rate": 2.8499247550858753e-07,
+      "loss": 0.8481,
+      "step": 8849
+    },
+    {
+      "epoch": 0.9262166405023547,
+      "grad_norm": 2.1114047838811127,
+      "learning_rate": 2.841895463584643e-07,
+      "loss": 0.8574,
+      "step": 8850
+    },
+    {
+      "epoch": 0.9263212977498692,
+      "grad_norm": 2.221883898704795,
+      "learning_rate": 2.833877335746571e-07,
+      "loss": 0.9177,
+      "step": 8851
+    },
+    {
+      "epoch": 0.9264259549973836,
+      "grad_norm": 2.049867364566891,
+      "learning_rate": 2.8258703724929783e-07,
+      "loss": 0.8104,
+      "step": 8852
+    },
+    {
+      "epoch": 0.926530612244898,
+      "grad_norm": 2.0227819344438682,
+      "learning_rate": 2.817874574743873e-07,
+      "loss": 0.9001,
+      "step": 8853
+    },
+    {
+      "epoch": 0.9266352694924124,
+      "grad_norm": 2.154423868042412,
+      "learning_rate": 2.8098899434179874e-07,
+      "loss": 0.8108,
+      "step": 8854
+    },
+    {
+      "epoch": 0.9267399267399268,
+      "grad_norm": 2.2633954827114424,
+      "learning_rate": 2.8019164794327756e-07,
+      "loss": 0.8926,
+      "step": 8855
+    },
+    {
+      "epoch": 0.9268445839874412,
+      "grad_norm": 2.122598963752719,
+      "learning_rate": 2.793954183704384e-07,
+      "loss": 0.8493,
+      "step": 8856
+    },
+    {
+      "epoch": 0.9269492412349555,
+      "grad_norm": 2.2463946823059646,
+      "learning_rate": 2.7860030571477237e-07,
+      "loss": 0.9276,
+      "step": 8857
+    },
+    {
+      "epoch": 0.9270538984824699,
+      "grad_norm": 1.7833875668973516,
+      "learning_rate": 2.778063100676387e-07,
+      "loss": 0.7611,
+      "step": 8858
+    },
+    {
+      "epoch": 0.9271585557299843,
+      "grad_norm": 2.1337740367922007,
+      "learning_rate": 2.7701343152026883e-07,
+      "loss": 0.9961,
+      "step": 8859
+    },
+    {
+      "epoch": 0.9272632129774987,
+      "grad_norm": 1.6322981589903927,
+      "learning_rate": 2.7622167016376767e-07,
+      "loss": 0.7977,
+      "step": 8860
+    },
+    {
+      "epoch": 0.9273678702250131,
+      "grad_norm": 1.9024737473287638,
+      "learning_rate": 2.754310260891058e-07,
+      "loss": 0.8432,
+      "step": 8861
+    },
+    {
+      "epoch": 0.9274725274725275,
+      "grad_norm": 1.8617079620794417,
+      "learning_rate": 2.746414993871349e-07,
+      "loss": 0.7266,
+      "step": 8862
+    },
+    {
+      "epoch": 0.9275771847200419,
+      "grad_norm": 2.230787630710201,
+      "learning_rate": 2.738530901485714e-07,
+      "loss": 0.8972,
+      "step": 8863
+    },
+    {
+      "epoch": 0.9276818419675562,
+      "grad_norm": 1.9515071420546217,
+      "learning_rate": 2.7306579846400396e-07,
+      "loss": 0.7411,
+      "step": 8864
+    },
+    {
+      "epoch": 0.9277864992150706,
+      "grad_norm": 2.0010638403952616,
+      "learning_rate": 2.7227962442389453e-07,
+      "loss": 0.7764,
+      "step": 8865
+    },
+    {
+      "epoch": 0.927891156462585,
+      "grad_norm": 2.269353148166856,
+      "learning_rate": 2.7149456811857546e-07,
+      "loss": 0.8705,
+      "step": 8866
+    },
+    {
+      "epoch": 0.9279958137100994,
+      "grad_norm": 2.1031934409220963,
+      "learning_rate": 2.7071062963825444e-07,
+      "loss": 0.8417,
+      "step": 8867
+    },
+    {
+      "epoch": 0.9281004709576138,
+      "grad_norm": 2.015166756657374,
+      "learning_rate": 2.699278090730051e-07,
+      "loss": 0.7953,
+      "step": 8868
+    },
+    {
+      "epoch": 0.9282051282051282,
+      "grad_norm": 2.3657122687519143,
+      "learning_rate": 2.6914610651277427e-07,
+      "loss": 0.8704,
+      "step": 8869
+    },
+    {
+      "epoch": 0.9283097854526426,
+      "grad_norm": 1.7952034748061156,
+      "learning_rate": 2.683655220473824e-07,
+      "loss": 0.8351,
+      "step": 8870
+    },
+    {
+      "epoch": 0.9284144427001569,
+      "grad_norm": 2.261515967889273,
+      "learning_rate": 2.675860557665211e-07,
+      "loss": 0.8985,
+      "step": 8871
+    },
+    {
+      "epoch": 0.9285190999476713,
+      "grad_norm": 1.8641043302959848,
+      "learning_rate": 2.6680770775975196e-07,
+      "loss": 0.8101,
+      "step": 8872
+    },
+    {
+      "epoch": 0.9286237571951858,
+      "grad_norm": 1.9479323389296166,
+      "learning_rate": 2.6603047811650795e-07,
+      "loss": 0.8856,
+      "step": 8873
+    },
+    {
+      "epoch": 0.9287284144427002,
+      "grad_norm": 2.3595217516180487,
+      "learning_rate": 2.652543669260932e-07,
+      "loss": 0.8375,
+      "step": 8874
+    },
+    {
+      "epoch": 0.9288330716902146,
+      "grad_norm": 2.2492688188141723,
+      "learning_rate": 2.644793742776874e-07,
+      "loss": 0.7974,
+      "step": 8875
+    },
+    {
+      "epoch": 0.928937728937729,
+      "grad_norm": 1.5664487634193003,
+      "learning_rate": 2.637055002603373e-07,
+      "loss": 0.7147,
+      "step": 8876
+    },
+    {
+      "epoch": 0.9290423861852434,
+      "grad_norm": 1.6411930077631471,
+      "learning_rate": 2.6293274496296263e-07,
+      "loss": 0.7573,
+      "step": 8877
+    },
+    {
+      "epoch": 0.9291470434327577,
+      "grad_norm": 2.2388975244958655,
+      "learning_rate": 2.621611084743558e-07,
+      "loss": 0.8901,
+      "step": 8878
+    },
+    {
+      "epoch": 0.9292517006802721,
+      "grad_norm": 1.8373181670762098,
+      "learning_rate": 2.6139059088317707e-07,
+      "loss": 0.7095,
+      "step": 8879
+    },
+    {
+      "epoch": 0.9293563579277865,
+      "grad_norm": 2.0228887964257214,
+      "learning_rate": 2.606211922779611e-07,
+      "loss": 0.7826,
+      "step": 8880
+    },
+    {
+      "epoch": 0.9294610151753009,
+      "grad_norm": 2.582925357687833,
+      "learning_rate": 2.59852912747115e-07,
+      "loss": 0.9785,
+      "step": 8881
+    },
+    {
+      "epoch": 0.9295656724228153,
+      "grad_norm": 2.186974123773659,
+      "learning_rate": 2.5908575237891476e-07,
+      "loss": 0.8608,
+      "step": 8882
+    },
+    {
+      "epoch": 0.9296703296703297,
+      "grad_norm": 1.5660897431908913,
+      "learning_rate": 2.5831971126150767e-07,
+      "loss": 0.6913,
+      "step": 8883
+    },
+    {
+      "epoch": 0.9297749869178441,
+      "grad_norm": 2.4094802073679755,
+      "learning_rate": 2.575547894829145e-07,
+      "loss": 0.9072,
+      "step": 8884
+    },
+    {
+      "epoch": 0.9298796441653584,
+      "grad_norm": 2.2742005617782755,
+      "learning_rate": 2.567909871310259e-07,
+      "loss": 0.9929,
+      "step": 8885
+    },
+    {
+      "epoch": 0.9299843014128728,
+      "grad_norm": 2.4008864387506677,
+      "learning_rate": 2.5602830429360516e-07,
+      "loss": 0.8501,
+      "step": 8886
+    },
+    {
+      "epoch": 0.9300889586603872,
+      "grad_norm": 1.9539433575898302,
+      "learning_rate": 2.552667410582843e-07,
+      "loss": 0.8582,
+      "step": 8887
+    },
+    {
+      "epoch": 0.9301936159079016,
+      "grad_norm": 2.0416965429628635,
+      "learning_rate": 2.5450629751257117e-07,
+      "loss": 0.8709,
+      "step": 8888
+    },
+    {
+      "epoch": 0.930298273155416,
+      "grad_norm": 2.506235715702831,
+      "learning_rate": 2.5374697374383806e-07,
+      "loss": 1.0232,
+      "step": 8889
+    },
+    {
+      "epoch": 0.9304029304029304,
+      "grad_norm": 1.99806566122585,
+      "learning_rate": 2.529887698393374e-07,
+      "loss": 0.8371,
+      "step": 8890
+    },
+    {
+      "epoch": 0.9305075876504448,
+      "grad_norm": 1.9042092264811785,
+      "learning_rate": 2.522316858861862e-07,
+      "loss": 0.855,
+      "step": 8891
+    },
+    {
+      "epoch": 0.9306122448979591,
+      "grad_norm": 2.038381656157126,
+      "learning_rate": 2.514757219713737e-07,
+      "loss": 0.8411,
+      "step": 8892
+    },
+    {
+      "epoch": 0.9307169021454735,
+      "grad_norm": 2.340813532948446,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.923,
+      "step": 8893
+    },
+    {
+      "epoch": 0.9308215593929879,
+      "grad_norm": 2.0448666923378265,
+      "learning_rate": 2.499671546040894e-07,
+      "loss": 0.8978,
+      "step": 8894
+    },
+    {
+      "epoch": 0.9309262166405023,
+      "grad_norm": 2.199976902362381,
+      "learning_rate": 2.492145513249533e-07,
+      "loss": 0.961,
+      "step": 8895
+    },
+    {
+      "epoch": 0.9310308738880168,
+      "grad_norm": 2.0776211466361274,
+      "learning_rate": 2.4846306843083315e-07,
+      "loss": 0.7628,
+      "step": 8896
+    },
+    {
+      "epoch": 0.9311355311355312,
+      "grad_norm": 1.957550937998734,
+      "learning_rate": 2.477127060080753e-07,
+      "loss": 0.8212,
+      "step": 8897
+    },
+    {
+      "epoch": 0.9312401883830456,
+      "grad_norm": 1.925782850036767,
+      "learning_rate": 2.469634641428964e-07,
+      "loss": 0.8907,
+      "step": 8898
+    },
+    {
+      "epoch": 0.93134484563056,
+      "grad_norm": 2.1162098484306533,
+      "learning_rate": 2.462153429213865e-07,
+      "loss": 0.8915,
+      "step": 8899
+    },
+    {
+      "epoch": 0.9314495028780743,
+      "grad_norm": 1.9771946100363844,
+      "learning_rate": 2.454683424295068e-07,
+      "loss": 0.742,
+      "step": 8900
+    },
+    {
+      "epoch": 0.9315541601255887,
+      "grad_norm": 1.9810077999238465,
+      "learning_rate": 2.4472246275308976e-07,
+      "loss": 0.9926,
+      "step": 8901
+    },
+    {
+      "epoch": 0.9316588173731031,
+      "grad_norm": 1.9014151842209916,
+      "learning_rate": 2.4397770397783794e-07,
+      "loss": 0.8555,
+      "step": 8902
+    },
+    {
+      "epoch": 0.9317634746206175,
+      "grad_norm": 1.9053437153678374,
+      "learning_rate": 2.4323406618932623e-07,
+      "loss": 0.8252,
+      "step": 8903
+    },
+    {
+      "epoch": 0.9318681318681319,
+      "grad_norm": 2.1866635447476193,
+      "learning_rate": 2.4249154947299734e-07,
+      "loss": 0.8793,
+      "step": 8904
+    },
+    {
+      "epoch": 0.9319727891156463,
+      "grad_norm": 2.162217503848748,
+      "learning_rate": 2.4175015391417087e-07,
+      "loss": 0.8766,
+      "step": 8905
+    },
+    {
+      "epoch": 0.9320774463631607,
+      "grad_norm": 2.3164940465618886,
+      "learning_rate": 2.410098795980342e-07,
+      "loss": 0.9261,
+      "step": 8906
+    },
+    {
+      "epoch": 0.932182103610675,
+      "grad_norm": 1.968520269068906,
+      "learning_rate": 2.40270726609646e-07,
+      "loss": 0.879,
+      "step": 8907
+    },
+    {
+      "epoch": 0.9322867608581894,
+      "grad_norm": 2.3833618329222865,
+      "learning_rate": 2.3953269503393496e-07,
+      "loss": 0.9269,
+      "step": 8908
+    },
+    {
+      "epoch": 0.9323914181057038,
+      "grad_norm": 2.403516866391237,
+      "learning_rate": 2.387957849557032e-07,
+      "loss": 0.8012,
+      "step": 8909
+    },
+    {
+      "epoch": 0.9324960753532182,
+      "grad_norm": 2.0016102230175528,
+      "learning_rate": 2.380599964596264e-07,
+      "loss": 0.8501,
+      "step": 8910
+    },
+    {
+      "epoch": 0.9326007326007326,
+      "grad_norm": 2.0012188366238703,
+      "learning_rate": 2.3732532963024468e-07,
+      "loss": 0.9091,
+      "step": 8911
+    },
+    {
+      "epoch": 0.932705389848247,
+      "grad_norm": 2.332974869714162,
+      "learning_rate": 2.3659178455197274e-07,
+      "loss": 0.8408,
+      "step": 8912
+    },
+    {
+      "epoch": 0.9328100470957614,
+      "grad_norm": 2.2463160819031587,
+      "learning_rate": 2.358593613090987e-07,
+      "loss": 0.8569,
+      "step": 8913
+    },
+    {
+      "epoch": 0.9329147043432757,
+      "grad_norm": 1.8053639500260543,
+      "learning_rate": 2.3512805998577638e-07,
+      "loss": 0.7705,
+      "step": 8914
+    },
+    {
+      "epoch": 0.9330193615907901,
+      "grad_norm": 2.2853468515179345,
+      "learning_rate": 2.343978806660363e-07,
+      "loss": 0.8963,
+      "step": 8915
+    },
+    {
+      "epoch": 0.9331240188383045,
+      "grad_norm": 2.0353226627808616,
+      "learning_rate": 2.3366882343377695e-07,
+      "loss": 0.8253,
+      "step": 8916
+    },
+    {
+      "epoch": 0.9332286760858189,
+      "grad_norm": 2.312638823803725,
+      "learning_rate": 2.3294088837276796e-07,
+      "loss": 0.8579,
+      "step": 8917
+    },
+    {
+      "epoch": 0.9333333333333333,
+      "grad_norm": 1.9526099594306408,
+      "learning_rate": 2.322140755666491e-07,
+      "loss": 0.812,
+      "step": 8918
+    },
+    {
+      "epoch": 0.9334379905808478,
+      "grad_norm": 2.0469902618255724,
+      "learning_rate": 2.3148838509893579e-07,
+      "loss": 0.8683,
+      "step": 8919
+    },
+    {
+      "epoch": 0.9335426478283622,
+      "grad_norm": 1.880159180549677,
+      "learning_rate": 2.3076381705301022e-07,
+      "loss": 0.9037,
+      "step": 8920
+    },
+    {
+      "epoch": 0.9336473050758765,
+      "grad_norm": 1.5969199873906468,
+      "learning_rate": 2.3004037151212576e-07,
+      "loss": 0.7873,
+      "step": 8921
+    },
+    {
+      "epoch": 0.9337519623233909,
+      "grad_norm": 1.9933441261897502,
+      "learning_rate": 2.293180485594093e-07,
+      "loss": 0.9562,
+      "step": 8922
+    },
+    {
+      "epoch": 0.9338566195709053,
+      "grad_norm": 1.9912633475021717,
+      "learning_rate": 2.2859684827785444e-07,
+      "loss": 0.8938,
+      "step": 8923
+    },
+    {
+      "epoch": 0.9339612768184197,
+      "grad_norm": 1.7915205291719956,
+      "learning_rate": 2.2787677075033266e-07,
+      "loss": 0.797,
+      "step": 8924
+    },
+    {
+      "epoch": 0.9340659340659341,
+      "grad_norm": 1.9232604600710044,
+      "learning_rate": 2.2715781605957886e-07,
+      "loss": 0.8286,
+      "step": 8925
+    },
+    {
+      "epoch": 0.9341705913134485,
+      "grad_norm": 2.106021331701821,
+      "learning_rate": 2.2643998428820368e-07,
+      "loss": 0.7703,
+      "step": 8926
+    },
+    {
+      "epoch": 0.9342752485609629,
+      "grad_norm": 1.880171483367633,
+      "learning_rate": 2.2572327551868778e-07,
+      "loss": 0.8119,
+      "step": 8927
+    },
+    {
+      "epoch": 0.9343799058084772,
+      "grad_norm": 2.318935123745428,
+      "learning_rate": 2.2500768983338085e-07,
+      "loss": 0.9226,
+      "step": 8928
+    },
+    {
+      "epoch": 0.9344845630559916,
+      "grad_norm": 1.6832835230122096,
+      "learning_rate": 2.2429322731450708e-07,
+      "loss": 0.7947,
+      "step": 8929
+    },
+    {
+      "epoch": 0.934589220303506,
+      "grad_norm": 2.1120628902234615,
+      "learning_rate": 2.2357988804415976e-07,
+      "loss": 0.9599,
+      "step": 8930
+    },
+    {
+      "epoch": 0.9346938775510204,
+      "grad_norm": 2.0233896081392504,
+      "learning_rate": 2.2286767210430105e-07,
+      "loss": 0.8827,
+      "step": 8931
+    },
+    {
+      "epoch": 0.9347985347985348,
+      "grad_norm": 2.040817721869765,
+      "learning_rate": 2.2215657957676774e-07,
+      "loss": 0.9144,
+      "step": 8932
+    },
+    {
+      "epoch": 0.9349031920460492,
+      "grad_norm": 1.922624595035762,
+      "learning_rate": 2.2144661054326443e-07,
+      "loss": 0.8159,
+      "step": 8933
+    },
+    {
+      "epoch": 0.9350078492935636,
+      "grad_norm": 1.6714847110688713,
+      "learning_rate": 2.2073776508537038e-07,
+      "loss": 0.7365,
+      "step": 8934
+    },
+    {
+      "epoch": 0.9351125065410779,
+      "grad_norm": 2.1817225705896393,
+      "learning_rate": 2.2003004328452926e-07,
+      "loss": 0.7756,
+      "step": 8935
+    },
+    {
+      "epoch": 0.9352171637885923,
+      "grad_norm": 2.173883426438636,
+      "learning_rate": 2.1932344522206494e-07,
+      "loss": 0.8909,
+      "step": 8936
+    },
+    {
+      "epoch": 0.9353218210361067,
+      "grad_norm": 2.029134530081249,
+      "learning_rate": 2.186179709791625e-07,
+      "loss": 0.9164,
+      "step": 8937
+    },
+    {
+      "epoch": 0.9354264782836211,
+      "grad_norm": 2.020431951662609,
+      "learning_rate": 2.1791362063688481e-07,
+      "loss": 0.8419,
+      "step": 8938
+    },
+    {
+      "epoch": 0.9355311355311355,
+      "grad_norm": 2.1073518757776664,
+      "learning_rate": 2.1721039427616164e-07,
+      "loss": 0.8759,
+      "step": 8939
+    },
+    {
+      "epoch": 0.9356357927786499,
+      "grad_norm": 1.976379251020562,
+      "learning_rate": 2.1650829197779722e-07,
+      "loss": 0.9092,
+      "step": 8940
+    },
+    {
+      "epoch": 0.9357404500261643,
+      "grad_norm": 2.2148390374272737,
+      "learning_rate": 2.1580731382246255e-07,
+      "loss": 0.8463,
+      "step": 8941
+    },
+    {
+      "epoch": 0.9358451072736788,
+      "grad_norm": 2.3658272163282734,
+      "learning_rate": 2.1510745989070215e-07,
+      "loss": 0.8825,
+      "step": 8942
+    },
+    {
+      "epoch": 0.935949764521193,
+      "grad_norm": 2.1121430732937596,
+      "learning_rate": 2.144087302629305e-07,
+      "loss": 0.7826,
+      "step": 8943
+    },
+    {
+      "epoch": 0.9360544217687075,
+      "grad_norm": 1.725035272587334,
+      "learning_rate": 2.1371112501943348e-07,
+      "loss": 0.7282,
+      "step": 8944
+    },
+    {
+      "epoch": 0.9361590790162219,
+      "grad_norm": 1.8745605871155686,
+      "learning_rate": 2.130146442403669e-07,
+      "loss": 0.864,
+      "step": 8945
+    },
+    {
+      "epoch": 0.9362637362637363,
+      "grad_norm": 2.4987700519915994,
+      "learning_rate": 2.1231928800575897e-07,
+      "loss": 0.9905,
+      "step": 8946
+    },
+    {
+      "epoch": 0.9363683935112507,
+      "grad_norm": 1.7532505918821066,
+      "learning_rate": 2.1162505639550468e-07,
+      "loss": 0.7843,
+      "step": 8947
+    },
+    {
+      "epoch": 0.9364730507587651,
+      "grad_norm": 2.1811207939897783,
+      "learning_rate": 2.109319494893758e-07,
+      "loss": 0.9079,
+      "step": 8948
+    },
+    {
+      "epoch": 0.9365777080062795,
+      "grad_norm": 2.0571665531615944,
+      "learning_rate": 2.1023996736700968e-07,
+      "loss": 0.7592,
+      "step": 8949
+    },
+    {
+      "epoch": 0.9366823652537938,
+      "grad_norm": 1.9908145189339748,
+      "learning_rate": 2.0954911010791834e-07,
+      "loss": 0.8919,
+      "step": 8950
+    },
+    {
+      "epoch": 0.9367870225013082,
+      "grad_norm": 1.879635801498437,
+      "learning_rate": 2.0885937779148158e-07,
+      "loss": 0.756,
+      "step": 8951
+    },
+    {
+      "epoch": 0.9368916797488226,
+      "grad_norm": 2.0278954913132843,
+      "learning_rate": 2.0817077049694934e-07,
+      "loss": 0.8411,
+      "step": 8952
+    },
+    {
+      "epoch": 0.936996336996337,
+      "grad_norm": 2.1053845951642396,
+      "learning_rate": 2.074832883034461e-07,
+      "loss": 0.8745,
+      "step": 8953
+    },
+    {
+      "epoch": 0.9371009942438514,
+      "grad_norm": 1.9499721033329678,
+      "learning_rate": 2.0679693128996535e-07,
+      "loss": 0.8599,
+      "step": 8954
+    },
+    {
+      "epoch": 0.9372056514913658,
+      "grad_norm": 2.2937105767755996,
+      "learning_rate": 2.061116995353707e-07,
+      "loss": 0.9106,
+      "step": 8955
+    },
+    {
+      "epoch": 0.9373103087388802,
+      "grad_norm": 2.0231281748039085,
+      "learning_rate": 2.0542759311839468e-07,
+      "loss": 0.8773,
+      "step": 8956
+    },
+    {
+      "epoch": 0.9374149659863945,
+      "grad_norm": 2.0532732484680594,
+      "learning_rate": 2.047446121176444e-07,
+      "loss": 0.939,
+      "step": 8957
+    },
+    {
+      "epoch": 0.9375196232339089,
+      "grad_norm": 1.8426241069013005,
+      "learning_rate": 2.040627566115949e-07,
+      "loss": 0.8707,
+      "step": 8958
+    },
+    {
+      "epoch": 0.9376242804814233,
+      "grad_norm": 1.9657462776989114,
+      "learning_rate": 2.0338202667859462e-07,
+      "loss": 0.9519,
+      "step": 8959
+    },
+    {
+      "epoch": 0.9377289377289377,
+      "grad_norm": 2.2777404864137556,
+      "learning_rate": 2.0270242239685768e-07,
+      "loss": 0.8969,
+      "step": 8960
+    },
+    {
+      "epoch": 0.9378335949764521,
+      "grad_norm": 2.080680228339346,
+      "learning_rate": 2.0202394384447376e-07,
+      "loss": 0.9103,
+      "step": 8961
+    },
+    {
+      "epoch": 0.9379382522239665,
+      "grad_norm": 2.301306982738778,
+      "learning_rate": 2.0134659109940057e-07,
+      "loss": 0.9517,
+      "step": 8962
+    },
+    {
+      "epoch": 0.938042909471481,
+      "grad_norm": 2.1571228068534767,
+      "learning_rate": 2.0067036423946916e-07,
+      "loss": 0.9849,
+      "step": 8963
+    },
+    {
+      "epoch": 0.9381475667189952,
+      "grad_norm": 1.9252681696412213,
+      "learning_rate": 1.999952633423785e-07,
+      "loss": 0.7812,
+      "step": 8964
+    },
+    {
+      "epoch": 0.9382522239665096,
+      "grad_norm": 1.9760309458462875,
+      "learning_rate": 1.9932128848569875e-07,
+      "loss": 0.8616,
+      "step": 8965
+    },
+    {
+      "epoch": 0.938356881214024,
+      "grad_norm": 2.4624548411289324,
+      "learning_rate": 1.9864843974687133e-07,
+      "loss": 0.9182,
+      "step": 8966
+    },
+    {
+      "epoch": 0.9384615384615385,
+      "grad_norm": 2.0682795924841653,
+      "learning_rate": 1.9797671720320543e-07,
+      "loss": 0.9754,
+      "step": 8967
+    },
+    {
+      "epoch": 0.9385661957090529,
+      "grad_norm": 2.014284349621675,
+      "learning_rate": 1.9730612093188716e-07,
+      "loss": 0.8685,
+      "step": 8968
+    },
+    {
+      "epoch": 0.9386708529565673,
+      "grad_norm": 2.4120062967864513,
+      "learning_rate": 1.966366510099682e-07,
+      "loss": 0.9244,
+      "step": 8969
+    },
+    {
+      "epoch": 0.9387755102040817,
+      "grad_norm": 2.3280174180689035,
+      "learning_rate": 1.9596830751437034e-07,
+      "loss": 0.9299,
+      "step": 8970
+    },
+    {
+      "epoch": 0.938880167451596,
+      "grad_norm": 2.074408047932079,
+      "learning_rate": 1.9530109052188883e-07,
+      "loss": 0.9401,
+      "step": 8971
+    },
+    {
+      "epoch": 0.9389848246991104,
+      "grad_norm": 1.7420567087126304,
+      "learning_rate": 1.946350001091879e-07,
+      "loss": 0.669,
+      "step": 8972
+    },
+    {
+      "epoch": 0.9390894819466248,
+      "grad_norm": 2.2410970937411743,
+      "learning_rate": 1.9397003635280408e-07,
+      "loss": 0.8553,
+      "step": 8973
+    },
+    {
+      "epoch": 0.9391941391941392,
+      "grad_norm": 1.9769963769064203,
+      "learning_rate": 1.9330619932914184e-07,
+      "loss": 0.7936,
+      "step": 8974
+    },
+    {
+      "epoch": 0.9392987964416536,
+      "grad_norm": 2.37273596450195,
+      "learning_rate": 1.9264348911447795e-07,
+      "loss": 0.7941,
+      "step": 8975
+    },
+    {
+      "epoch": 0.939403453689168,
+      "grad_norm": 2.0813158115777854,
+      "learning_rate": 1.91981905784957e-07,
+      "loss": 0.9177,
+      "step": 8976
+    },
+    {
+      "epoch": 0.9395081109366824,
+      "grad_norm": 2.290897372588027,
+      "learning_rate": 1.9132144941659935e-07,
+      "loss": 0.8598,
+      "step": 8977
+    },
+    {
+      "epoch": 0.9396127681841967,
+      "grad_norm": 2.2230141215809534,
+      "learning_rate": 1.9066212008529095e-07,
+      "loss": 0.9614,
+      "step": 8978
+    },
+    {
+      "epoch": 0.9397174254317111,
+      "grad_norm": 1.9411269671666274,
+      "learning_rate": 1.9000391786679118e-07,
+      "loss": 0.8767,
+      "step": 8979
+    },
+    {
+      "epoch": 0.9398220826792255,
+      "grad_norm": 1.9684623940624517,
+      "learning_rate": 1.8934684283672844e-07,
+      "loss": 0.7992,
+      "step": 8980
+    },
+    {
+      "epoch": 0.9399267399267399,
+      "grad_norm": 2.2097045916457687,
+      "learning_rate": 1.886908950706001e-07,
+      "loss": 0.7911,
+      "step": 8981
+    },
+    {
+      "epoch": 0.9400313971742543,
+      "grad_norm": 2.2051242891982823,
+      "learning_rate": 1.8803607464377816e-07,
+      "loss": 0.8268,
+      "step": 8982
+    },
+    {
+      "epoch": 0.9401360544217687,
+      "grad_norm": 2.0731015805325947,
+      "learning_rate": 1.8738238163150346e-07,
+      "loss": 0.9092,
+      "step": 8983
+    },
+    {
+      "epoch": 0.9402407116692831,
+      "grad_norm": 2.0982860002338235,
+      "learning_rate": 1.8672981610888484e-07,
+      "loss": 0.9503,
+      "step": 8984
+    },
+    {
+      "epoch": 0.9403453689167975,
+      "grad_norm": 2.0377487369557055,
+      "learning_rate": 1.8607837815090345e-07,
+      "loss": 0.9304,
+      "step": 8985
+    },
+    {
+      "epoch": 0.9404500261643118,
+      "grad_norm": 2.2351989370813543,
+      "learning_rate": 1.8542806783241053e-07,
+      "loss": 0.8992,
+      "step": 8986
+    },
+    {
+      "epoch": 0.9405546834118262,
+      "grad_norm": 2.0623835431044477,
+      "learning_rate": 1.8477888522812847e-07,
+      "loss": 0.9759,
+      "step": 8987
+    },
+    {
+      "epoch": 0.9406593406593406,
+      "grad_norm": 2.2769860510032776,
+      "learning_rate": 1.8413083041265213e-07,
+      "loss": 0.8573,
+      "step": 8988
+    },
+    {
+      "epoch": 0.940763997906855,
+      "grad_norm": 1.9523228068902658,
+      "learning_rate": 1.8348390346044187e-07,
+      "loss": 0.7817,
+      "step": 8989
+    },
+    {
+      "epoch": 0.9408686551543695,
+      "grad_norm": 2.18666471676111,
+      "learning_rate": 1.828381044458305e-07,
+      "loss": 0.9009,
+      "step": 8990
+    },
+    {
+      "epoch": 0.9409733124018839,
+      "grad_norm": 1.9241205019311667,
+      "learning_rate": 1.8219343344302198e-07,
+      "loss": 0.8631,
+      "step": 8991
+    },
+    {
+      "epoch": 0.9410779696493983,
+      "grad_norm": 2.04039821239051,
+      "learning_rate": 1.815498905260915e-07,
+      "loss": 0.8371,
+      "step": 8992
+    },
+    {
+      "epoch": 0.9411826268969126,
+      "grad_norm": 2.3419236200201516,
+      "learning_rate": 1.8090747576898217e-07,
+      "loss": 0.8494,
+      "step": 8993
+    },
+    {
+      "epoch": 0.941287284144427,
+      "grad_norm": 2.1610439047997145,
+      "learning_rate": 1.8026618924551041e-07,
+      "loss": 0.8086,
+      "step": 8994
+    },
+    {
+      "epoch": 0.9413919413919414,
+      "grad_norm": 1.9220224743991272,
+      "learning_rate": 1.7962603102935848e-07,
+      "loss": 0.8518,
+      "step": 8995
+    },
+    {
+      "epoch": 0.9414965986394558,
+      "grad_norm": 2.0058482503008763,
+      "learning_rate": 1.789870011940842e-07,
+      "loss": 0.7359,
+      "step": 8996
+    },
+    {
+      "epoch": 0.9416012558869702,
+      "grad_norm": 2.1650612466125145,
+      "learning_rate": 1.7834909981311321e-07,
+      "loss": 0.9187,
+      "step": 8997
+    },
+    {
+      "epoch": 0.9417059131344846,
+      "grad_norm": 2.1222322501737008,
+      "learning_rate": 1.7771232695974028e-07,
+      "loss": 0.9376,
+      "step": 8998
+    },
+    {
+      "epoch": 0.941810570381999,
+      "grad_norm": 1.8756795331701073,
+      "learning_rate": 1.7707668270713463e-07,
+      "loss": 0.8314,
+      "step": 8999
+    },
+    {
+      "epoch": 0.9419152276295133,
+      "grad_norm": 2.0250266930245373,
+      "learning_rate": 1.7644216712832897e-07,
+      "loss": 0.8112,
+      "step": 9000
+    },
+    {
+      "epoch": 0.9420198848770277,
+      "grad_norm": 1.8965489062179972,
+      "learning_rate": 1.7580878029623382e-07,
+      "loss": 0.8542,
+      "step": 9001
+    },
+    {
+      "epoch": 0.9421245421245421,
+      "grad_norm": 2.0182139408122657,
+      "learning_rate": 1.751765222836266e-07,
+      "loss": 0.9074,
+      "step": 9002
+    },
+    {
+      "epoch": 0.9422291993720565,
+      "grad_norm": 1.915756504936258,
+      "learning_rate": 1.7454539316315356e-07,
+      "loss": 0.9071,
+      "step": 9003
+    },
+    {
+      "epoch": 0.9423338566195709,
+      "grad_norm": 1.8890927536983408,
+      "learning_rate": 1.7391539300733118e-07,
+      "loss": 0.7825,
+      "step": 9004
+    },
+    {
+      "epoch": 0.9424385138670853,
+      "grad_norm": 2.0792217263743016,
+      "learning_rate": 1.7328652188855044e-07,
+      "loss": 0.8445,
+      "step": 9005
+    },
+    {
+      "epoch": 0.9425431711145997,
+      "grad_norm": 2.2281920177680767,
+      "learning_rate": 1.726587798790702e-07,
+      "loss": 0.7547,
+      "step": 9006
+    },
+    {
+      "epoch": 0.942647828362114,
+      "grad_norm": 2.3393383954988853,
+      "learning_rate": 1.7203216705101834e-07,
+      "loss": 0.8758,
+      "step": 9007
+    },
+    {
+      "epoch": 0.9427524856096284,
+      "grad_norm": 2.025231383373048,
+      "learning_rate": 1.714066834763939e-07,
+      "loss": 0.8952,
+      "step": 9008
+    },
+    {
+      "epoch": 0.9428571428571428,
+      "grad_norm": 1.9231761232116151,
+      "learning_rate": 1.7078232922706495e-07,
+      "loss": 0.8794,
+      "step": 9009
+    },
+    {
+      "epoch": 0.9429618001046572,
+      "grad_norm": 2.55540070865077,
+      "learning_rate": 1.70159104374773e-07,
+      "loss": 0.8213,
+      "step": 9010
+    },
+    {
+      "epoch": 0.9430664573521716,
+      "grad_norm": 2.2864462638812113,
+      "learning_rate": 1.6953700899112746e-07,
+      "loss": 0.9661,
+      "step": 9011
+    },
+    {
+      "epoch": 0.9431711145996861,
+      "grad_norm": 2.022644471342003,
+      "learning_rate": 1.689160431476078e-07,
+      "loss": 0.7653,
+      "step": 9012
+    },
+    {
+      "epoch": 0.9432757718472005,
+      "grad_norm": 1.9668619711190107,
+      "learning_rate": 1.682962069155636e-07,
+      "loss": 0.8767,
+      "step": 9013
+    },
+    {
+      "epoch": 0.9433804290947148,
+      "grad_norm": 2.125692202668173,
+      "learning_rate": 1.6767750036621677e-07,
+      "loss": 0.8827,
+      "step": 9014
+    },
+    {
+      "epoch": 0.9434850863422292,
+      "grad_norm": 1.9970347338306567,
+      "learning_rate": 1.6705992357065713e-07,
+      "loss": 0.8181,
+      "step": 9015
+    },
+    {
+      "epoch": 0.9435897435897436,
+      "grad_norm": 2.1417706192896504,
+      "learning_rate": 1.6644347659984571e-07,
+      "loss": 0.8472,
+      "step": 9016
+    },
+    {
+      "epoch": 0.943694400837258,
+      "grad_norm": 2.4107829527631517,
+      "learning_rate": 1.6582815952461474e-07,
+      "loss": 0.9227,
+      "step": 9017
+    },
+    {
+      "epoch": 0.9437990580847724,
+      "grad_norm": 2.272587316891678,
+      "learning_rate": 1.6521397241566317e-07,
+      "loss": 0.7621,
+      "step": 9018
+    },
+    {
+      "epoch": 0.9439037153322868,
+      "grad_norm": 1.9616247365575001,
+      "learning_rate": 1.6460091534356238e-07,
+      "loss": 0.8642,
+      "step": 9019
+    },
+    {
+      "epoch": 0.9440083725798012,
+      "grad_norm": 2.213024695754038,
+      "learning_rate": 1.6398898837875598e-07,
+      "loss": 0.8659,
+      "step": 9020
+    },
+    {
+      "epoch": 0.9441130298273155,
+      "grad_norm": 2.18849504081884,
+      "learning_rate": 1.6337819159155556e-07,
+      "loss": 0.8914,
+      "step": 9021
+    },
+    {
+      "epoch": 0.9442176870748299,
+      "grad_norm": 1.9962396515971108,
+      "learning_rate": 1.6276852505213937e-07,
+      "loss": 0.8487,
+      "step": 9022
+    },
+    {
+      "epoch": 0.9443223443223443,
+      "grad_norm": 1.60529112595556,
+      "learning_rate": 1.621599888305636e-07,
+      "loss": 0.8027,
+      "step": 9023
+    },
+    {
+      "epoch": 0.9444270015698587,
+      "grad_norm": 2.3647898369898535,
+      "learning_rate": 1.615525829967479e-07,
+      "loss": 0.8755,
+      "step": 9024
+    },
+    {
+      "epoch": 0.9445316588173731,
+      "grad_norm": 2.199820408495756,
+      "learning_rate": 1.6094630762048535e-07,
+      "loss": 0.9419,
+      "step": 9025
+    },
+    {
+      "epoch": 0.9446363160648875,
+      "grad_norm": 2.1623987139154295,
+      "learning_rate": 1.6034116277143795e-07,
+      "loss": 0.9243,
+      "step": 9026
+    },
+    {
+      "epoch": 0.9447409733124019,
+      "grad_norm": 2.223255861105165,
+      "learning_rate": 1.5973714851913792e-07,
+      "loss": 0.9347,
+      "step": 9027
+    },
+    {
+      "epoch": 0.9448456305599163,
+      "grad_norm": 2.227214544210135,
+      "learning_rate": 1.5913426493298745e-07,
+      "loss": 0.8761,
+      "step": 9028
+    },
+    {
+      "epoch": 0.9449502878074306,
+      "grad_norm": 2.0771281382507465,
+      "learning_rate": 1.5853251208225895e-07,
+      "loss": 0.8977,
+      "step": 9029
+    },
+    {
+      "epoch": 0.945054945054945,
+      "grad_norm": 2.1250598064265716,
+      "learning_rate": 1.57931890036096e-07,
+      "loss": 0.8215,
+      "step": 9030
+    },
+    {
+      "epoch": 0.9451596023024594,
+      "grad_norm": 1.906429846009798,
+      "learning_rate": 1.5733239886351114e-07,
+      "loss": 0.8799,
+      "step": 9031
+    },
+    {
+      "epoch": 0.9452642595499738,
+      "grad_norm": 2.0717540052678385,
+      "learning_rate": 1.567340386333871e-07,
+      "loss": 0.8241,
+      "step": 9032
+    },
+    {
+      "epoch": 0.9453689167974882,
+      "grad_norm": 2.2358749882717532,
+      "learning_rate": 1.5613680941447663e-07,
+      "loss": 0.878,
+      "step": 9033
+    },
+    {
+      "epoch": 0.9454735740450027,
+      "grad_norm": 2.3438037978057498,
+      "learning_rate": 1.555407112754015e-07,
+      "loss": 0.9401,
+      "step": 9034
+    },
+    {
+      "epoch": 0.9455782312925171,
+      "grad_norm": 1.8342854640981783,
+      "learning_rate": 1.549457442846558e-07,
+      "loss": 0.9056,
+      "step": 9035
+    },
+    {
+      "epoch": 0.9456828885400314,
+      "grad_norm": 1.9116944068383945,
+      "learning_rate": 1.5435190851060156e-07,
+      "loss": 0.7609,
+      "step": 9036
+    },
+    {
+      "epoch": 0.9457875457875458,
+      "grad_norm": 1.7292150071570478,
+      "learning_rate": 1.5375920402147305e-07,
+      "loss": 0.7398,
+      "step": 9037
+    },
+    {
+      "epoch": 0.9458922030350602,
+      "grad_norm": 2.181589596999383,
+      "learning_rate": 1.5316763088537135e-07,
+      "loss": 0.8476,
+      "step": 9038
+    },
+    {
+      "epoch": 0.9459968602825746,
+      "grad_norm": 2.1307840848771447,
+      "learning_rate": 1.52577189170271e-07,
+      "loss": 0.8192,
+      "step": 9039
+    },
+    {
+      "epoch": 0.946101517530089,
+      "grad_norm": 2.360122477075766,
+      "learning_rate": 1.5198787894401433e-07,
+      "loss": 0.9385,
+      "step": 9040
+    },
+    {
+      "epoch": 0.9462061747776034,
+      "grad_norm": 2.456600162959172,
+      "learning_rate": 1.5139970027431505e-07,
+      "loss": 0.87,
+      "step": 9041
+    },
+    {
+      "epoch": 0.9463108320251178,
+      "grad_norm": 1.9169412412415379,
+      "learning_rate": 1.508126532287557e-07,
+      "loss": 0.8699,
+      "step": 9042
+    },
+    {
+      "epoch": 0.9464154892726321,
+      "grad_norm": 2.2877674182580297,
+      "learning_rate": 1.5022673787478793e-07,
+      "loss": 0.8462,
+      "step": 9043
+    },
+    {
+      "epoch": 0.9465201465201465,
+      "grad_norm": 2.039640660135877,
+      "learning_rate": 1.496419542797356e-07,
+      "loss": 0.7836,
+      "step": 9044
+    },
+    {
+      "epoch": 0.9466248037676609,
+      "grad_norm": 1.9739212006187399,
+      "learning_rate": 1.4905830251079167e-07,
+      "loss": 0.9003,
+      "step": 9045
+    },
+    {
+      "epoch": 0.9467294610151753,
+      "grad_norm": 1.9770702491808176,
+      "learning_rate": 1.4847578263501916e-07,
+      "loss": 0.8087,
+      "step": 9046
+    },
+    {
+      "epoch": 0.9468341182626897,
+      "grad_norm": 2.25033314641664,
+      "learning_rate": 1.4789439471935008e-07,
+      "loss": 0.9281,
+      "step": 9047
+    },
+    {
+      "epoch": 0.9469387755102041,
+      "grad_norm": 2.2873109923661126,
+      "learning_rate": 1.4731413883058655e-07,
+      "loss": 0.9335,
+      "step": 9048
+    },
+    {
+      "epoch": 0.9470434327577185,
+      "grad_norm": 2.151440014464125,
+      "learning_rate": 1.4673501503540187e-07,
+      "loss": 0.8403,
+      "step": 9049
+    },
+    {
+      "epoch": 0.9471480900052328,
+      "grad_norm": 1.8386378234588971,
+      "learning_rate": 1.4615702340033954e-07,
+      "loss": 0.8304,
+      "step": 9050
+    },
+    {
+      "epoch": 0.9472527472527472,
+      "grad_norm": 2.11039769161247,
+      "learning_rate": 1.4558016399181086e-07,
+      "loss": 0.8667,
+      "step": 9051
+    },
+    {
+      "epoch": 0.9473574045002616,
+      "grad_norm": 2.2668021084417527,
+      "learning_rate": 1.4500443687609945e-07,
+      "loss": 0.9028,
+      "step": 9052
+    },
+    {
+      "epoch": 0.947462061747776,
+      "grad_norm": 2.5925480089352844,
+      "learning_rate": 1.4442984211935463e-07,
+      "loss": 0.9285,
+      "step": 9053
+    },
+    {
+      "epoch": 0.9475667189952904,
+      "grad_norm": 1.9257446949630599,
+      "learning_rate": 1.4385637978760026e-07,
+      "loss": 0.8561,
+      "step": 9054
+    },
+    {
+      "epoch": 0.9476713762428048,
+      "grad_norm": 2.2283800756599264,
+      "learning_rate": 1.4328404994672917e-07,
+      "loss": 0.8976,
+      "step": 9055
+    },
+    {
+      "epoch": 0.9477760334903192,
+      "grad_norm": 1.8811361955772714,
+      "learning_rate": 1.4271285266250323e-07,
+      "loss": 0.8489,
+      "step": 9056
+    },
+    {
+      "epoch": 0.9478806907378335,
+      "grad_norm": 2.2540466745301493,
+      "learning_rate": 1.4214278800055102e-07,
+      "loss": 0.7986,
+      "step": 9057
+    },
+    {
+      "epoch": 0.947985347985348,
+      "grad_norm": 1.7605020970876353,
+      "learning_rate": 1.4157385602637685e-07,
+      "loss": 0.7758,
+      "step": 9058
+    },
+    {
+      "epoch": 0.9480900052328624,
+      "grad_norm": 2.0483519247325197,
+      "learning_rate": 1.4100605680535284e-07,
+      "loss": 0.8848,
+      "step": 9059
+    },
+    {
+      "epoch": 0.9481946624803768,
+      "grad_norm": 1.814643560597211,
+      "learning_rate": 1.4043939040272014e-07,
+      "loss": 0.8009,
+      "step": 9060
+    },
+    {
+      "epoch": 0.9482993197278912,
+      "grad_norm": 2.1844789581596333,
+      "learning_rate": 1.398738568835878e-07,
+      "loss": 0.848,
+      "step": 9061
+    },
+    {
+      "epoch": 0.9484039769754056,
+      "grad_norm": 2.08957785838643,
+      "learning_rate": 1.3930945631293712e-07,
+      "loss": 0.9124,
+      "step": 9062
+    },
+    {
+      "epoch": 0.94850863422292,
+      "grad_norm": 2.1549858224284826,
+      "learning_rate": 1.3874618875561963e-07,
+      "loss": 0.9237,
+      "step": 9063
+    },
+    {
+      "epoch": 0.9486132914704343,
+      "grad_norm": 2.029197205102344,
+      "learning_rate": 1.3818405427635683e-07,
+      "loss": 0.8428,
+      "step": 9064
+    },
+    {
+      "epoch": 0.9487179487179487,
+      "grad_norm": 2.162465083314952,
+      "learning_rate": 1.376230529397371e-07,
+      "loss": 0.953,
+      "step": 9065
+    },
+    {
+      "epoch": 0.9488226059654631,
+      "grad_norm": 2.2719870379626967,
+      "learning_rate": 1.370631848102233e-07,
+      "loss": 0.9192,
+      "step": 9066
+    },
+    {
+      "epoch": 0.9489272632129775,
+      "grad_norm": 2.3224845795701117,
+      "learning_rate": 1.3650444995214286e-07,
+      "loss": 0.8813,
+      "step": 9067
+    },
+    {
+      "epoch": 0.9490319204604919,
+      "grad_norm": 2.1462838164243014,
+      "learning_rate": 1.3594684842969664e-07,
+      "loss": 0.8676,
+      "step": 9068
+    },
+    {
+      "epoch": 0.9491365777080063,
+      "grad_norm": 1.815945744671672,
+      "learning_rate": 1.3539038030695338e-07,
+      "loss": 0.8986,
+      "step": 9069
+    },
+    {
+      "epoch": 0.9492412349555207,
+      "grad_norm": 1.906488906851326,
+      "learning_rate": 1.3483504564785422e-07,
+      "loss": 0.8174,
+      "step": 9070
+    },
+    {
+      "epoch": 0.9493458922030351,
+      "grad_norm": 2.236795411684707,
+      "learning_rate": 1.3428084451620694e-07,
+      "loss": 0.8814,
+      "step": 9071
+    },
+    {
+      "epoch": 0.9494505494505494,
+      "grad_norm": 2.0557267418605902,
+      "learning_rate": 1.337277769756895e-07,
+      "loss": 0.8568,
+      "step": 9072
+    },
+    {
+      "epoch": 0.9495552066980638,
+      "grad_norm": 2.6957377041769504,
+      "learning_rate": 1.331758430898522e-07,
+      "loss": 0.9719,
+      "step": 9073
+    },
+    {
+      "epoch": 0.9496598639455782,
+      "grad_norm": 1.8970879310242872,
+      "learning_rate": 1.326250429221121e-07,
+      "loss": 0.8987,
+      "step": 9074
+    },
+    {
+      "epoch": 0.9497645211930926,
+      "grad_norm": 1.934705112665449,
+      "learning_rate": 1.3207537653575851e-07,
+      "loss": 0.8891,
+      "step": 9075
+    },
+    {
+      "epoch": 0.949869178440607,
+      "grad_norm": 2.2063646405916124,
+      "learning_rate": 1.3152684399394877e-07,
+      "loss": 0.8031,
+      "step": 9076
+    },
+    {
+      "epoch": 0.9499738356881214,
+      "grad_norm": 1.979196270712241,
+      "learning_rate": 1.3097944535970908e-07,
+      "loss": 0.8487,
+      "step": 9077
+    },
+    {
+      "epoch": 0.9500784929356358,
+      "grad_norm": 2.3747448704287173,
+      "learning_rate": 1.3043318069593914e-07,
+      "loss": 0.9484,
+      "step": 9078
+    },
+    {
+      "epoch": 0.9501831501831501,
+      "grad_norm": 1.9720183796039061,
+      "learning_rate": 1.2988805006540317e-07,
+      "loss": 0.7593,
+      "step": 9079
+    },
+    {
+      "epoch": 0.9502878074306645,
+      "grad_norm": 1.66623523158947,
+      "learning_rate": 1.2934405353073997e-07,
+      "loss": 0.7237,
+      "step": 9080
+    },
+    {
+      "epoch": 0.950392464678179,
+      "grad_norm": 2.010121946011193,
+      "learning_rate": 1.288011911544551e-07,
+      "loss": 0.781,
+      "step": 9081
+    },
+    {
+      "epoch": 0.9504971219256934,
+      "grad_norm": 2.338519331926691,
+      "learning_rate": 1.282594629989231e-07,
+      "loss": 0.8435,
+      "step": 9082
+    },
+    {
+      "epoch": 0.9506017791732078,
+      "grad_norm": 2.1718237588653144,
+      "learning_rate": 1.2771886912639088e-07,
+      "loss": 0.8509,
+      "step": 9083
+    },
+    {
+      "epoch": 0.9507064364207222,
+      "grad_norm": 2.2698280951956,
+      "learning_rate": 1.2717940959897424e-07,
+      "loss": 0.857,
+      "step": 9084
+    },
+    {
+      "epoch": 0.9508110936682366,
+      "grad_norm": 2.1107172447457065,
+      "learning_rate": 1.2664108447865918e-07,
+      "loss": 0.7931,
+      "step": 9085
+    },
+    {
+      "epoch": 0.9509157509157509,
+      "grad_norm": 1.9797524667575515,
+      "learning_rate": 1.2610389382729738e-07,
+      "loss": 0.7899,
+      "step": 9086
+    },
+    {
+      "epoch": 0.9510204081632653,
+      "grad_norm": 2.057673027423047,
+      "learning_rate": 1.2556783770661497e-07,
+      "loss": 0.9024,
+      "step": 9087
+    },
+    {
+      "epoch": 0.9511250654107797,
+      "grad_norm": 2.1671668650036535,
+      "learning_rate": 1.250329161782049e-07,
+      "loss": 0.8494,
+      "step": 9088
+    },
+    {
+      "epoch": 0.9512297226582941,
+      "grad_norm": 2.0877723433934086,
+      "learning_rate": 1.2449912930353248e-07,
+      "loss": 0.885,
+      "step": 9089
+    },
+    {
+      "epoch": 0.9513343799058085,
+      "grad_norm": 2.177290420704625,
+      "learning_rate": 1.2396647714392974e-07,
+      "loss": 0.8511,
+      "step": 9090
+    },
+    {
+      "epoch": 0.9514390371533229,
+      "grad_norm": 2.051463049714876,
+      "learning_rate": 1.2343495976059882e-07,
+      "loss": 0.9175,
+      "step": 9091
+    },
+    {
+      "epoch": 0.9515436944008373,
+      "grad_norm": 1.9910691736016368,
+      "learning_rate": 1.2290457721461314e-07,
+      "loss": 0.8602,
+      "step": 9092
+    },
+    {
+      "epoch": 0.9516483516483516,
+      "grad_norm": 1.9342141459380122,
+      "learning_rate": 1.22375329566915e-07,
+      "loss": 0.8365,
+      "step": 9093
+    },
+    {
+      "epoch": 0.951753008895866,
+      "grad_norm": 1.800982384325224,
+      "learning_rate": 1.2184721687831468e-07,
+      "loss": 0.847,
+      "step": 9094
+    },
+    {
+      "epoch": 0.9518576661433804,
+      "grad_norm": 2.176163511782998,
+      "learning_rate": 1.2132023920949586e-07,
+      "loss": 0.8672,
+      "step": 9095
+    },
+    {
+      "epoch": 0.9519623233908948,
+      "grad_norm": 2.0098442785162343,
+      "learning_rate": 1.2079439662100567e-07,
+      "loss": 0.7643,
+      "step": 9096
+    },
+    {
+      "epoch": 0.9520669806384092,
+      "grad_norm": 2.1686254394715867,
+      "learning_rate": 1.20269689173268e-07,
+      "loss": 0.8923,
+      "step": 9097
+    },
+    {
+      "epoch": 0.9521716378859236,
+      "grad_norm": 2.4479350828010182,
+      "learning_rate": 1.1974611692657124e-07,
+      "loss": 0.9458,
+      "step": 9098
+    },
+    {
+      "epoch": 0.952276295133438,
+      "grad_norm": 1.7758983963507677,
+      "learning_rate": 1.1922367994107397e-07,
+      "loss": 0.798,
+      "step": 9099
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 1.9312505481701443,
+      "learning_rate": 1.1870237827680708e-07,
+      "loss": 0.8766,
+      "step": 9100
+    },
+    {
+      "epoch": 0.9524856096284667,
+      "grad_norm": 2.1489020046415384,
+      "learning_rate": 1.181822119936682e-07,
+      "loss": 0.8693,
+      "step": 9101
+    },
+    {
+      "epoch": 0.9525902668759811,
+      "grad_norm": 1.9604487123502172,
+      "learning_rate": 1.1766318115142505e-07,
+      "loss": 0.9279,
+      "step": 9102
+    },
+    {
+      "epoch": 0.9526949241234955,
+      "grad_norm": 1.979028252970426,
+      "learning_rate": 1.1714528580971774e-07,
+      "loss": 0.9074,
+      "step": 9103
+    },
+    {
+      "epoch": 0.95279958137101,
+      "grad_norm": 2.0147250746415244,
+      "learning_rate": 1.1662852602805087e-07,
+      "loss": 0.9098,
+      "step": 9104
+    },
+    {
+      "epoch": 0.9529042386185244,
+      "grad_norm": 1.941672354841768,
+      "learning_rate": 1.1611290186580137e-07,
+      "loss": 0.8705,
+      "step": 9105
+    },
+    {
+      "epoch": 0.9530088958660388,
+      "grad_norm": 2.081202528742428,
+      "learning_rate": 1.1559841338221633e-07,
+      "loss": 0.8618,
+      "step": 9106
+    },
+    {
+      "epoch": 0.9531135531135531,
+      "grad_norm": 1.900573593944078,
+      "learning_rate": 1.1508506063641178e-07,
+      "loss": 0.7274,
+      "step": 9107
+    },
+    {
+      "epoch": 0.9532182103610675,
+      "grad_norm": 2.2879689467033897,
+      "learning_rate": 1.1457284368737276e-07,
+      "loss": 0.8774,
+      "step": 9108
+    },
+    {
+      "epoch": 0.9533228676085819,
+      "grad_norm": 2.0021870264860553,
+      "learning_rate": 1.140617625939533e-07,
+      "loss": 0.8242,
+      "step": 9109
+    },
+    {
+      "epoch": 0.9534275248560963,
+      "grad_norm": 2.189380584746929,
+      "learning_rate": 1.1355181741487864e-07,
+      "loss": 0.9973,
+      "step": 9110
+    },
+    {
+      "epoch": 0.9535321821036107,
+      "grad_norm": 2.046145341571205,
+      "learning_rate": 1.1304300820874192e-07,
+      "loss": 0.9143,
+      "step": 9111
+    },
+    {
+      "epoch": 0.9536368393511251,
+      "grad_norm": 1.9336419044364952,
+      "learning_rate": 1.1253533503400638e-07,
+      "loss": 0.8793,
+      "step": 9112
+    },
+    {
+      "epoch": 0.9537414965986395,
+      "grad_norm": 2.1511547350474616,
+      "learning_rate": 1.1202879794900535e-07,
+      "loss": 0.8096,
+      "step": 9113
+    },
+    {
+      "epoch": 0.9538461538461539,
+      "grad_norm": 1.9734945556556622,
+      "learning_rate": 1.1152339701194004e-07,
+      "loss": 0.808,
+      "step": 9114
+    },
+    {
+      "epoch": 0.9539508110936682,
+      "grad_norm": 2.0081235379107203,
+      "learning_rate": 1.1101913228088179e-07,
+      "loss": 0.8715,
+      "step": 9115
+    },
+    {
+      "epoch": 0.9540554683411826,
+      "grad_norm": 2.117755926561859,
+      "learning_rate": 1.1051600381377315e-07,
+      "loss": 0.8589,
+      "step": 9116
+    },
+    {
+      "epoch": 0.954160125588697,
+      "grad_norm": 2.0526210136326695,
+      "learning_rate": 1.100140116684234e-07,
+      "loss": 0.8793,
+      "step": 9117
+    },
+    {
+      "epoch": 0.9542647828362114,
+      "grad_norm": 2.112405339433251,
+      "learning_rate": 1.095131559025131e-07,
+      "loss": 0.8667,
+      "step": 9118
+    },
+    {
+      "epoch": 0.9543694400837258,
+      "grad_norm": 2.533576036279591,
+      "learning_rate": 1.0901343657359175e-07,
+      "loss": 0.7516,
+      "step": 9119
+    },
+    {
+      "epoch": 0.9544740973312402,
+      "grad_norm": 2.154561014085007,
+      "learning_rate": 1.0851485373907677e-07,
+      "loss": 0.9504,
+      "step": 9120
+    },
+    {
+      "epoch": 0.9545787545787546,
+      "grad_norm": 2.4408447371630464,
+      "learning_rate": 1.0801740745625677e-07,
+      "loss": 0.9668,
+      "step": 9121
+    },
+    {
+      "epoch": 0.9546834118262689,
+      "grad_norm": 1.9109247782495005,
+      "learning_rate": 1.0752109778229158e-07,
+      "loss": 0.8883,
+      "step": 9122
+    },
+    {
+      "epoch": 0.9547880690737833,
+      "grad_norm": 2.013558136876636,
+      "learning_rate": 1.070259247742056e-07,
+      "loss": 0.885,
+      "step": 9123
+    },
+    {
+      "epoch": 0.9548927263212977,
+      "grad_norm": 2.136904562312622,
+      "learning_rate": 1.0653188848889551e-07,
+      "loss": 0.8507,
+      "step": 9124
+    },
+    {
+      "epoch": 0.9549973835688121,
+      "grad_norm": 1.978016541353706,
+      "learning_rate": 1.0603898898312704e-07,
+      "loss": 0.8988,
+      "step": 9125
+    },
+    {
+      "epoch": 0.9551020408163265,
+      "grad_norm": 2.292835501413161,
+      "learning_rate": 1.0554722631353709e-07,
+      "loss": 0.9214,
+      "step": 9126
+    },
+    {
+      "epoch": 0.955206698063841,
+      "grad_norm": 1.9578723859363283,
+      "learning_rate": 1.0505660053662825e-07,
+      "loss": 0.807,
+      "step": 9127
+    },
+    {
+      "epoch": 0.9553113553113554,
+      "grad_norm": 2.074313232154308,
+      "learning_rate": 1.0456711170877543e-07,
+      "loss": 0.865,
+      "step": 9128
+    },
+    {
+      "epoch": 0.9554160125588697,
+      "grad_norm": 2.4188323428193166,
+      "learning_rate": 1.0407875988622252e-07,
+      "loss": 0.8573,
+      "step": 9129
+    },
+    {
+      "epoch": 0.9555206698063841,
+      "grad_norm": 1.9965673098227608,
+      "learning_rate": 1.0359154512508019e-07,
+      "loss": 0.8046,
+      "step": 9130
+    },
+    {
+      "epoch": 0.9556253270538985,
+      "grad_norm": 2.424445738477294,
+      "learning_rate": 1.0310546748133143e-07,
+      "loss": 1.0085,
+      "step": 9131
+    },
+    {
+      "epoch": 0.9557299843014129,
+      "grad_norm": 1.9689504274155718,
+      "learning_rate": 1.0262052701082825e-07,
+      "loss": 0.8351,
+      "step": 9132
+    },
+    {
+      "epoch": 0.9558346415489273,
+      "grad_norm": 2.108713019030409,
+      "learning_rate": 1.021367237692894e-07,
+      "loss": 0.9932,
+      "step": 9133
+    },
+    {
+      "epoch": 0.9559392987964417,
+      "grad_norm": 1.7814011137810855,
+      "learning_rate": 1.0165405781230708e-07,
+      "loss": 0.8543,
+      "step": 9134
+    },
+    {
+      "epoch": 0.9560439560439561,
+      "grad_norm": 2.458888218864724,
+      "learning_rate": 1.0117252919533804e-07,
+      "loss": 0.9106,
+      "step": 9135
+    },
+    {
+      "epoch": 0.9561486132914704,
+      "grad_norm": 2.454880605782708,
+      "learning_rate": 1.0069213797371357e-07,
+      "loss": 0.8216,
+      "step": 9136
+    },
+    {
+      "epoch": 0.9562532705389848,
+      "grad_norm": 2.3292738121246934,
+      "learning_rate": 1.0021288420263065e-07,
+      "loss": 0.8631,
+      "step": 9137
+    },
+    {
+      "epoch": 0.9563579277864992,
+      "grad_norm": 1.9940769879892457,
+      "learning_rate": 9.97347679371563e-08,
+      "loss": 0.9346,
+      "step": 9138
+    },
+    {
+      "epoch": 0.9564625850340136,
+      "grad_norm": 2.168224739156848,
+      "learning_rate": 9.925778923222773e-08,
+      "loss": 0.9215,
+      "step": 9139
+    },
+    {
+      "epoch": 0.956567242281528,
+      "grad_norm": 2.0395622501364605,
+      "learning_rate": 9.878194814264886e-08,
+      "loss": 0.9338,
+      "step": 9140
+    },
+    {
+      "epoch": 0.9566718995290424,
+      "grad_norm": 2.2995306333544696,
+      "learning_rate": 9.830724472309594e-08,
+      "loss": 0.9621,
+      "step": 9141
+    },
+    {
+      "epoch": 0.9567765567765568,
+      "grad_norm": 1.9195393542003314,
+      "learning_rate": 9.783367902811424e-08,
+      "loss": 0.9511,
+      "step": 9142
+    },
+    {
+      "epoch": 0.9568812140240711,
+      "grad_norm": 2.001176501644658,
+      "learning_rate": 9.73612511121158e-08,
+      "loss": 0.9361,
+      "step": 9143
+    },
+    {
+      "epoch": 0.9569858712715855,
+      "grad_norm": 2.1602312695732224,
+      "learning_rate": 9.688996102938497e-08,
+      "loss": 0.7655,
+      "step": 9144
+    },
+    {
+      "epoch": 0.9570905285190999,
+      "grad_norm": 2.3661293111777515,
+      "learning_rate": 9.641980883407287e-08,
+      "loss": 0.9697,
+      "step": 9145
+    },
+    {
+      "epoch": 0.9571951857666143,
+      "grad_norm": 1.9875384007327113,
+      "learning_rate": 9.595079458020185e-08,
+      "loss": 0.7232,
+      "step": 9146
+    },
+    {
+      "epoch": 0.9572998430141287,
+      "grad_norm": 1.5923915548495342,
+      "learning_rate": 9.548291832166213e-08,
+      "loss": 0.8366,
+      "step": 9147
+    },
+    {
+      "epoch": 0.9574045002616431,
+      "grad_norm": 2.2856279399112225,
+      "learning_rate": 9.501618011221291e-08,
+      "loss": 0.9121,
+      "step": 9148
+    },
+    {
+      "epoch": 0.9575091575091575,
+      "grad_norm": 2.1702328491811045,
+      "learning_rate": 9.455058000548467e-08,
+      "loss": 0.8221,
+      "step": 9149
+    },
+    {
+      "epoch": 0.957613814756672,
+      "grad_norm": 2.169354118160728,
+      "learning_rate": 9.408611805497459e-08,
+      "loss": 0.8867,
+      "step": 9150
+    },
+    {
+      "epoch": 0.9577184720041862,
+      "grad_norm": 1.9949827280347652,
+      "learning_rate": 9.36227943140522e-08,
+      "loss": 0.8586,
+      "step": 9151
+    },
+    {
+      "epoch": 0.9578231292517007,
+      "grad_norm": 2.6143300269638288,
+      "learning_rate": 9.316060883595158e-08,
+      "loss": 0.8846,
+      "step": 9152
+    },
+    {
+      "epoch": 0.9579277864992151,
+      "grad_norm": 2.0238769325890096,
+      "learning_rate": 9.269956167378136e-08,
+      "loss": 0.863,
+      "step": 9153
+    },
+    {
+      "epoch": 0.9580324437467295,
+      "grad_norm": 1.7312936279188003,
+      "learning_rate": 9.223965288051474e-08,
+      "loss": 0.775,
+      "step": 9154
+    },
+    {
+      "epoch": 0.9581371009942439,
+      "grad_norm": 1.679219951168678,
+      "learning_rate": 9.17808825089983e-08,
+      "loss": 0.7436,
+      "step": 9155
+    },
+    {
+      "epoch": 0.9582417582417583,
+      "grad_norm": 2.0696046017408114,
+      "learning_rate": 9.132325061194325e-08,
+      "loss": 0.8891,
+      "step": 9156
+    },
+    {
+      "epoch": 0.9583464154892727,
+      "grad_norm": 1.9650852136731827,
+      "learning_rate": 9.086675724193528e-08,
+      "loss": 0.8477,
+      "step": 9157
+    },
+    {
+      "epoch": 0.958451072736787,
+      "grad_norm": 1.9933766746277393,
+      "learning_rate": 9.041140245142355e-08,
+      "loss": 0.9044,
+      "step": 9158
+    },
+    {
+      "epoch": 0.9585557299843014,
+      "grad_norm": 2.43537039286737,
+      "learning_rate": 8.995718629273065e-08,
+      "loss": 0.9137,
+      "step": 9159
+    },
+    {
+      "epoch": 0.9586603872318158,
+      "grad_norm": 2.0921310588419493,
+      "learning_rate": 8.950410881804706e-08,
+      "loss": 0.9069,
+      "step": 9160
+    },
+    {
+      "epoch": 0.9587650444793302,
+      "grad_norm": 2.1188665014531147,
+      "learning_rate": 8.905217007943223e-08,
+      "loss": 0.8711,
+      "step": 9161
+    },
+    {
+      "epoch": 0.9588697017268446,
+      "grad_norm": 1.8922155563080787,
+      "learning_rate": 8.860137012881465e-08,
+      "loss": 0.9361,
+      "step": 9162
+    },
+    {
+      "epoch": 0.958974358974359,
+      "grad_norm": 1.711472138483851,
+      "learning_rate": 8.815170901799174e-08,
+      "loss": 0.7832,
+      "step": 9163
+    },
+    {
+      "epoch": 0.9590790162218734,
+      "grad_norm": 2.05901764490939,
+      "learning_rate": 8.770318679862999e-08,
+      "loss": 0.791,
+      "step": 9164
+    },
+    {
+      "epoch": 0.9591836734693877,
+      "grad_norm": 2.0760777814710285,
+      "learning_rate": 8.725580352226815e-08,
+      "loss": 0.8347,
+      "step": 9165
+    },
+    {
+      "epoch": 0.9592883307169021,
+      "grad_norm": 1.904314569738717,
+      "learning_rate": 8.680955924030954e-08,
+      "loss": 0.8394,
+      "step": 9166
+    },
+    {
+      "epoch": 0.9593929879644165,
+      "grad_norm": 1.5764956424130905,
+      "learning_rate": 8.63644540040276e-08,
+      "loss": 0.6921,
+      "step": 9167
+    },
+    {
+      "epoch": 0.9594976452119309,
+      "grad_norm": 1.9033111511642289,
+      "learning_rate": 8.592048786456698e-08,
+      "loss": 0.7482,
+      "step": 9168
+    },
+    {
+      "epoch": 0.9596023024594453,
+      "grad_norm": 2.07972888411746,
+      "learning_rate": 8.547766087294018e-08,
+      "loss": 0.8112,
+      "step": 9169
+    },
+    {
+      "epoch": 0.9597069597069597,
+      "grad_norm": 2.1173563174976393,
+      "learning_rate": 8.503597308002986e-08,
+      "loss": 0.7684,
+      "step": 9170
+    },
+    {
+      "epoch": 0.9598116169544741,
+      "grad_norm": 2.3978904207022067,
+      "learning_rate": 8.459542453658653e-08,
+      "loss": 0.9662,
+      "step": 9171
+    },
+    {
+      "epoch": 0.9599162742019884,
+      "grad_norm": 2.015486093729567,
+      "learning_rate": 8.415601529322858e-08,
+      "loss": 0.9097,
+      "step": 9172
+    },
+    {
+      "epoch": 0.9600209314495028,
+      "grad_norm": 2.100667884440473,
+      "learning_rate": 8.371774540044564e-08,
+      "loss": 0.8485,
+      "step": 9173
+    },
+    {
+      "epoch": 0.9601255886970173,
+      "grad_norm": 2.2677052547515455,
+      "learning_rate": 8.328061490859629e-08,
+      "loss": 0.805,
+      "step": 9174
+    },
+    {
+      "epoch": 0.9602302459445317,
+      "grad_norm": 2.502950900699366,
+      "learning_rate": 8.284462386790815e-08,
+      "loss": 0.8131,
+      "step": 9175
+    },
+    {
+      "epoch": 0.9603349031920461,
+      "grad_norm": 1.845406579180182,
+      "learning_rate": 8.240977232847669e-08,
+      "loss": 0.8898,
+      "step": 9176
+    },
+    {
+      "epoch": 0.9604395604395605,
+      "grad_norm": 2.644639245831181,
+      "learning_rate": 8.19760603402675e-08,
+      "loss": 0.9884,
+      "step": 9177
+    },
+    {
+      "epoch": 0.9605442176870749,
+      "grad_norm": 2.0136239922533767,
+      "learning_rate": 8.154348795311518e-08,
+      "loss": 0.9439,
+      "step": 9178
+    },
+    {
+      "epoch": 0.9606488749345892,
+      "grad_norm": 2.508687484109057,
+      "learning_rate": 8.111205521672105e-08,
+      "loss": 0.842,
+      "step": 9179
+    },
+    {
+      "epoch": 0.9607535321821036,
+      "grad_norm": 2.247230140286789,
+      "learning_rate": 8.068176218066104e-08,
+      "loss": 0.7672,
+      "step": 9180
+    },
+    {
+      "epoch": 0.960858189429618,
+      "grad_norm": 2.1501454719365465,
+      "learning_rate": 8.025260889437558e-08,
+      "loss": 0.906,
+      "step": 9181
+    },
+    {
+      "epoch": 0.9609628466771324,
+      "grad_norm": 2.2857097799522395,
+      "learning_rate": 7.982459540717413e-08,
+      "loss": 0.8782,
+      "step": 9182
+    },
+    {
+      "epoch": 0.9610675039246468,
+      "grad_norm": 2.418986186789819,
+      "learning_rate": 7.939772176823735e-08,
+      "loss": 0.8045,
+      "step": 9183
+    },
+    {
+      "epoch": 0.9611721611721612,
+      "grad_norm": 2.1459549533496594,
+      "learning_rate": 7.897198802661266e-08,
+      "loss": 0.9515,
+      "step": 9184
+    },
+    {
+      "epoch": 0.9612768184196756,
+      "grad_norm": 2.109672695130541,
+      "learning_rate": 7.854739423121982e-08,
+      "loss": 0.8162,
+      "step": 9185
+    },
+    {
+      "epoch": 0.9613814756671899,
+      "grad_norm": 2.0736403186020205,
+      "learning_rate": 7.812394043084315e-08,
+      "loss": 0.9333,
+      "step": 9186
+    },
+    {
+      "epoch": 0.9614861329147043,
+      "grad_norm": 1.9672228441543578,
+      "learning_rate": 7.770162667413928e-08,
+      "loss": 0.878,
+      "step": 9187
+    },
+    {
+      "epoch": 0.9615907901622187,
+      "grad_norm": 2.152750014034463,
+      "learning_rate": 7.728045300963383e-08,
+      "loss": 0.8695,
+      "step": 9188
+    },
+    {
+      "epoch": 0.9616954474097331,
+      "grad_norm": 2.1946243167100854,
+      "learning_rate": 7.686041948571921e-08,
+      "loss": 0.8581,
+      "step": 9189
+    },
+    {
+      "epoch": 0.9618001046572475,
+      "grad_norm": 1.7958639695003695,
+      "learning_rate": 7.644152615065903e-08,
+      "loss": 0.9317,
+      "step": 9190
+    },
+    {
+      "epoch": 0.9619047619047619,
+      "grad_norm": 2.103440015443088,
+      "learning_rate": 7.602377305258479e-08,
+      "loss": 0.8852,
+      "step": 9191
+    },
+    {
+      "epoch": 0.9620094191522763,
+      "grad_norm": 1.9632826033869528,
+      "learning_rate": 7.560716023949699e-08,
+      "loss": 0.9123,
+      "step": 9192
+    },
+    {
+      "epoch": 0.9621140763997907,
+      "grad_norm": 2.582600940053974,
+      "learning_rate": 7.51916877592651e-08,
+      "loss": 0.7138,
+      "step": 9193
+    },
+    {
+      "epoch": 0.962218733647305,
+      "grad_norm": 2.426593772598077,
+      "learning_rate": 7.477735565962873e-08,
+      "loss": 0.9125,
+      "step": 9194
+    },
+    {
+      "epoch": 0.9623233908948194,
+      "grad_norm": 2.0534735800973913,
+      "learning_rate": 7.436416398819313e-08,
+      "loss": 0.9127,
+      "step": 9195
+    },
+    {
+      "epoch": 0.9624280481423338,
+      "grad_norm": 1.8146955476652327,
+      "learning_rate": 7.395211279243808e-08,
+      "loss": 0.7993,
+      "step": 9196
+    },
+    {
+      "epoch": 0.9625327053898483,
+      "grad_norm": 2.2323319981904874,
+      "learning_rate": 7.354120211970684e-08,
+      "loss": 0.9055,
+      "step": 9197
+    },
+    {
+      "epoch": 0.9626373626373627,
+      "grad_norm": 1.846938976318557,
+      "learning_rate": 7.313143201721384e-08,
+      "loss": 0.7602,
+      "step": 9198
+    },
+    {
+      "epoch": 0.9627420198848771,
+      "grad_norm": 1.9136272699219579,
+      "learning_rate": 7.272280253204367e-08,
+      "loss": 0.7391,
+      "step": 9199
+    },
+    {
+      "epoch": 0.9628466771323915,
+      "grad_norm": 2.430133835674404,
+      "learning_rate": 7.231531371114875e-08,
+      "loss": 0.8995,
+      "step": 9200
+    },
+    {
+      "epoch": 0.9629513343799058,
+      "grad_norm": 2.131458703559915,
+      "learning_rate": 7.190896560134942e-08,
+      "loss": 0.8002,
+      "step": 9201
+    },
+    {
+      "epoch": 0.9630559916274202,
+      "grad_norm": 1.9827611709558692,
+      "learning_rate": 7.150375824933608e-08,
+      "loss": 0.8915,
+      "step": 9202
+    },
+    {
+      "epoch": 0.9631606488749346,
+      "grad_norm": 2.1954279263735317,
+      "learning_rate": 7.10996917016682e-08,
+      "loss": 0.9135,
+      "step": 9203
+    },
+    {
+      "epoch": 0.963265306122449,
+      "grad_norm": 2.150004140942097,
+      "learning_rate": 7.069676600477304e-08,
+      "loss": 0.8268,
+      "step": 9204
+    },
+    {
+      "epoch": 0.9633699633699634,
+      "grad_norm": 2.2008699779019256,
+      "learning_rate": 7.029498120494916e-08,
+      "loss": 0.8064,
+      "step": 9205
+    },
+    {
+      "epoch": 0.9634746206174778,
+      "grad_norm": 1.9973249721182413,
+      "learning_rate": 6.989433734836182e-08,
+      "loss": 0.8953,
+      "step": 9206
+    },
+    {
+      "epoch": 0.9635792778649922,
+      "grad_norm": 2.1012015027082023,
+      "learning_rate": 6.949483448104422e-08,
+      "loss": 0.8212,
+      "step": 9207
+    },
+    {
+      "epoch": 0.9636839351125065,
+      "grad_norm": 2.252680211089396,
+      "learning_rate": 6.909647264890296e-08,
+      "loss": 0.9004,
+      "step": 9208
+    },
+    {
+      "epoch": 0.9637885923600209,
+      "grad_norm": 2.0623106938542444,
+      "learning_rate": 6.869925189770809e-08,
+      "loss": 0.8154,
+      "step": 9209
+    },
+    {
+      "epoch": 0.9638932496075353,
+      "grad_norm": 2.4261216634526113,
+      "learning_rate": 6.830317227310202e-08,
+      "loss": 0.9477,
+      "step": 9210
+    },
+    {
+      "epoch": 0.9639979068550497,
+      "grad_norm": 2.2540675121561056,
+      "learning_rate": 6.790823382059497e-08,
+      "loss": 0.947,
+      "step": 9211
+    },
+    {
+      "epoch": 0.9641025641025641,
+      "grad_norm": 2.1048027694750844,
+      "learning_rate": 6.751443658556733e-08,
+      "loss": 0.9247,
+      "step": 9212
+    },
+    {
+      "epoch": 0.9642072213500785,
+      "grad_norm": 2.0158913138500045,
+      "learning_rate": 6.712178061326513e-08,
+      "loss": 0.9507,
+      "step": 9213
+    },
+    {
+      "epoch": 0.9643118785975929,
+      "grad_norm": 2.003911985362856,
+      "learning_rate": 6.673026594880672e-08,
+      "loss": 0.8659,
+      "step": 9214
+    },
+    {
+      "epoch": 0.9644165358451072,
+      "grad_norm": 1.9811498476133749,
+      "learning_rate": 6.633989263717722e-08,
+      "loss": 0.8903,
+      "step": 9215
+    },
+    {
+      "epoch": 0.9645211930926216,
+      "grad_norm": 2.232106088676546,
+      "learning_rate": 6.595066072323297e-08,
+      "loss": 0.8571,
+      "step": 9216
+    },
+    {
+      "epoch": 0.964625850340136,
+      "grad_norm": 2.5558019802236105,
+      "learning_rate": 6.556257025169488e-08,
+      "loss": 0.787,
+      "step": 9217
+    },
+    {
+      "epoch": 0.9647305075876504,
+      "grad_norm": 1.8740286415803826,
+      "learning_rate": 6.517562126715726e-08,
+      "loss": 0.7737,
+      "step": 9218
+    },
+    {
+      "epoch": 0.9648351648351648,
+      "grad_norm": 2.2998119940125403,
+      "learning_rate": 6.478981381408012e-08,
+      "loss": 0.8484,
+      "step": 9219
+    },
+    {
+      "epoch": 0.9649398220826793,
+      "grad_norm": 2.1309929398704592,
+      "learning_rate": 6.440514793679465e-08,
+      "loss": 0.9934,
+      "step": 9220
+    },
+    {
+      "epoch": 0.9650444793301937,
+      "grad_norm": 1.6771005831362835,
+      "learning_rate": 6.402162367949883e-08,
+      "loss": 0.7815,
+      "step": 9221
+    },
+    {
+      "epoch": 0.965149136577708,
+      "grad_norm": 2.1014499143649603,
+      "learning_rate": 6.363924108626074e-08,
+      "loss": 0.8083,
+      "step": 9222
+    },
+    {
+      "epoch": 0.9652537938252224,
+      "grad_norm": 1.896185187952413,
+      "learning_rate": 6.325800020101747e-08,
+      "loss": 0.7202,
+      "step": 9223
+    },
+    {
+      "epoch": 0.9653584510727368,
+      "grad_norm": 1.9377624373854216,
+      "learning_rate": 6.287790106757396e-08,
+      "loss": 0.8109,
+      "step": 9224
+    },
+    {
+      "epoch": 0.9654631083202512,
+      "grad_norm": 2.079814185022681,
+      "learning_rate": 6.249894372960419e-08,
+      "loss": 0.8617,
+      "step": 9225
+    },
+    {
+      "epoch": 0.9655677655677656,
+      "grad_norm": 1.9888426701736415,
+      "learning_rate": 6.212112823065109e-08,
+      "loss": 0.7941,
+      "step": 9226
+    },
+    {
+      "epoch": 0.96567242281528,
+      "grad_norm": 2.014484823659193,
+      "learning_rate": 6.174445461412659e-08,
+      "loss": 0.9727,
+      "step": 9227
+    },
+    {
+      "epoch": 0.9657770800627944,
+      "grad_norm": 2.114946519857328,
+      "learning_rate": 6.136892292331165e-08,
+      "loss": 0.9009,
+      "step": 9228
+    },
+    {
+      "epoch": 0.9658817373103087,
+      "grad_norm": 2.464678652778644,
+      "learning_rate": 6.099453320135396e-08,
+      "loss": 0.8523,
+      "step": 9229
+    },
+    {
+      "epoch": 0.9659863945578231,
+      "grad_norm": 2.064257270218177,
+      "learning_rate": 6.062128549127355e-08,
+      "loss": 0.7611,
+      "step": 9230
+    },
+    {
+      "epoch": 0.9660910518053375,
+      "grad_norm": 2.434856201533525,
+      "learning_rate": 6.024917983595613e-08,
+      "loss": 0.8406,
+      "step": 9231
+    },
+    {
+      "epoch": 0.9661957090528519,
+      "grad_norm": 2.3887496596256113,
+      "learning_rate": 5.98782162781586e-08,
+      "loss": 0.8096,
+      "step": 9232
+    },
+    {
+      "epoch": 0.9663003663003663,
+      "grad_norm": 2.365481278236734,
+      "learning_rate": 5.9508394860504635e-08,
+      "loss": 0.8159,
+      "step": 9233
+    },
+    {
+      "epoch": 0.9664050235478807,
+      "grad_norm": 2.3102168301713832,
+      "learning_rate": 5.9139715625488036e-08,
+      "loss": 0.9929,
+      "step": 9234
+    },
+    {
+      "epoch": 0.9665096807953951,
+      "grad_norm": 2.0516025566151264,
+      "learning_rate": 5.8772178615469355e-08,
+      "loss": 0.8985,
+      "step": 9235
+    },
+    {
+      "epoch": 0.9666143380429095,
+      "grad_norm": 1.9481670124245183,
+      "learning_rate": 5.8405783872680364e-08,
+      "loss": 0.9224,
+      "step": 9236
+    },
+    {
+      "epoch": 0.9667189952904238,
+      "grad_norm": 1.936915127599676,
+      "learning_rate": 5.804053143922184e-08,
+      "loss": 0.9193,
+      "step": 9237
+    },
+    {
+      "epoch": 0.9668236525379382,
+      "grad_norm": 2.159472821731071,
+      "learning_rate": 5.76764213570602e-08,
+      "loss": 0.933,
+      "step": 9238
+    },
+    {
+      "epoch": 0.9669283097854526,
+      "grad_norm": 2.1432057819082044,
+      "learning_rate": 5.731345366803198e-08,
+      "loss": 0.7927,
+      "step": 9239
+    },
+    {
+      "epoch": 0.967032967032967,
+      "grad_norm": 2.1392093036071356,
+      "learning_rate": 5.695162841384383e-08,
+      "loss": 0.8698,
+      "step": 9240
+    },
+    {
+      "epoch": 0.9671376242804814,
+      "grad_norm": 1.6287695667451592,
+      "learning_rate": 5.659094563607137e-08,
+      "loss": 0.6806,
+      "step": 9241
+    },
+    {
+      "epoch": 0.9672422815279959,
+      "grad_norm": 1.9649345198964663,
+      "learning_rate": 5.62314053761559e-08,
+      "loss": 1.017,
+      "step": 9242
+    },
+    {
+      "epoch": 0.9673469387755103,
+      "grad_norm": 1.9674398387069323,
+      "learning_rate": 5.587300767540993e-08,
+      "loss": 0.8725,
+      "step": 9243
+    },
+    {
+      "epoch": 0.9674515960230246,
+      "grad_norm": 2.1414946410927294,
+      "learning_rate": 5.551575257501496e-08,
+      "loss": 1.0452,
+      "step": 9244
+    },
+    {
+      "epoch": 0.967556253270539,
+      "grad_norm": 1.9468821502685336,
+      "learning_rate": 5.5159640116019264e-08,
+      "loss": 0.8087,
+      "step": 9245
+    },
+    {
+      "epoch": 0.9676609105180534,
+      "grad_norm": 1.6540577154311573,
+      "learning_rate": 5.4804670339341225e-08,
+      "loss": 0.7482,
+      "step": 9246
+    },
+    {
+      "epoch": 0.9677655677655678,
+      "grad_norm": 1.7009899889159676,
+      "learning_rate": 5.445084328576711e-08,
+      "loss": 0.7608,
+      "step": 9247
+    },
+    {
+      "epoch": 0.9678702250130822,
+      "grad_norm": 2.2598483825041096,
+      "learning_rate": 5.4098158995953274e-08,
+      "loss": 0.9915,
+      "step": 9248
+    },
+    {
+      "epoch": 0.9679748822605966,
+      "grad_norm": 2.3568793617507335,
+      "learning_rate": 5.374661751042287e-08,
+      "loss": 0.8288,
+      "step": 9249
+    },
+    {
+      "epoch": 0.968079539508111,
+      "grad_norm": 2.1736820945639828,
+      "learning_rate": 5.339621886957025e-08,
+      "loss": 0.9018,
+      "step": 9250
+    },
+    {
+      "epoch": 0.9681841967556253,
+      "grad_norm": 2.101174708101391,
+      "learning_rate": 5.304696311365543e-08,
+      "loss": 0.8313,
+      "step": 9251
+    },
+    {
+      "epoch": 0.9682888540031397,
+      "grad_norm": 2.3344750804334122,
+      "learning_rate": 5.269885028280963e-08,
+      "loss": 0.8387,
+      "step": 9252
+    },
+    {
+      "epoch": 0.9683935112506541,
+      "grad_norm": 2.2550389618502407,
+      "learning_rate": 5.2351880417030874e-08,
+      "loss": 0.8327,
+      "step": 9253
+    },
+    {
+      "epoch": 0.9684981684981685,
+      "grad_norm": 1.8950935631200951,
+      "learning_rate": 5.2006053556186155e-08,
+      "loss": 0.9351,
+      "step": 9254
+    },
+    {
+      "epoch": 0.9686028257456829,
+      "grad_norm": 2.3653393297563,
+      "learning_rate": 5.166136974001368e-08,
+      "loss": 0.8625,
+      "step": 9255
+    },
+    {
+      "epoch": 0.9687074829931973,
+      "grad_norm": 2.098650984633713,
+      "learning_rate": 5.1317829008116215e-08,
+      "loss": 0.9453,
+      "step": 9256
+    },
+    {
+      "epoch": 0.9688121402407117,
+      "grad_norm": 2.0350999152809006,
+      "learning_rate": 5.097543139996886e-08,
+      "loss": 0.8874,
+      "step": 9257
+    },
+    {
+      "epoch": 0.968916797488226,
+      "grad_norm": 1.7287186179945104,
+      "learning_rate": 5.0634176954913464e-08,
+      "loss": 0.8007,
+      "step": 9258
+    },
+    {
+      "epoch": 0.9690214547357404,
+      "grad_norm": 2.243175615621909,
+      "learning_rate": 5.029406571216089e-08,
+      "loss": 0.8416,
+      "step": 9259
+    },
+    {
+      "epoch": 0.9691261119832548,
+      "grad_norm": 2.089301015163357,
+      "learning_rate": 4.995509771078877e-08,
+      "loss": 0.8799,
+      "step": 9260
+    },
+    {
+      "epoch": 0.9692307692307692,
+      "grad_norm": 1.8265334250247782,
+      "learning_rate": 4.9617272989748166e-08,
+      "loss": 0.7664,
+      "step": 9261
+    },
+    {
+      "epoch": 0.9693354264782836,
+      "grad_norm": 2.1422139711001673,
+      "learning_rate": 4.92805915878547e-08,
+      "loss": 0.8051,
+      "step": 9262
+    },
+    {
+      "epoch": 0.969440083725798,
+      "grad_norm": 1.7236231384803589,
+      "learning_rate": 4.8945053543792975e-08,
+      "loss": 0.7868,
+      "step": 9263
+    },
+    {
+      "epoch": 0.9695447409733124,
+      "grad_norm": 1.9977730114261136,
+      "learning_rate": 4.86106588961166e-08,
+      "loss": 0.8113,
+      "step": 9264
+    },
+    {
+      "epoch": 0.9696493982208267,
+      "grad_norm": 2.196435687000056,
+      "learning_rate": 4.827740768324929e-08,
+      "loss": 0.88,
+      "step": 9265
+    },
+    {
+      "epoch": 0.9697540554683411,
+      "grad_norm": 2.154480773213278,
+      "learning_rate": 4.794529994348374e-08,
+      "loss": 0.8232,
+      "step": 9266
+    },
+    {
+      "epoch": 0.9698587127158556,
+      "grad_norm": 2.054282523683849,
+      "learning_rate": 4.76143357149772e-08,
+      "loss": 0.9648,
+      "step": 9267
+    },
+    {
+      "epoch": 0.96996336996337,
+      "grad_norm": 2.1057006673654297,
+      "learning_rate": 4.728451503575815e-08,
+      "loss": 0.8939,
+      "step": 9268
+    },
+    {
+      "epoch": 0.9700680272108844,
+      "grad_norm": 2.1885036662303445,
+      "learning_rate": 4.695583794372516e-08,
+      "loss": 0.9587,
+      "step": 9269
+    },
+    {
+      "epoch": 0.9701726844583988,
+      "grad_norm": 2.15267308775205,
+      "learning_rate": 4.662830447664357e-08,
+      "loss": 0.9066,
+      "step": 9270
+    },
+    {
+      "epoch": 0.9702773417059132,
+      "grad_norm": 2.4891197615275202,
+      "learning_rate": 4.630191467214773e-08,
+      "loss": 0.9047,
+      "step": 9271
+    },
+    {
+      "epoch": 0.9703819989534275,
+      "grad_norm": 2.152695134921288,
+      "learning_rate": 4.5976668567739856e-08,
+      "loss": 0.8575,
+      "step": 9272
+    },
+    {
+      "epoch": 0.9704866562009419,
+      "grad_norm": 2.086583458860564,
+      "learning_rate": 4.565256620079117e-08,
+      "loss": 0.9318,
+      "step": 9273
+    },
+    {
+      "epoch": 0.9705913134484563,
+      "grad_norm": 1.9990374714123995,
+      "learning_rate": 4.532960760854299e-08,
+      "loss": 0.8295,
+      "step": 9274
+    },
+    {
+      "epoch": 0.9706959706959707,
+      "grad_norm": 2.045660681387157,
+      "learning_rate": 4.5007792828102305e-08,
+      "loss": 0.7581,
+      "step": 9275
+    },
+    {
+      "epoch": 0.9708006279434851,
+      "grad_norm": 2.1535992375534088,
+      "learning_rate": 4.4687121896447305e-08,
+      "loss": 0.8037,
+      "step": 9276
+    },
+    {
+      "epoch": 0.9709052851909995,
+      "grad_norm": 1.933861715496548,
+      "learning_rate": 4.436759485042408e-08,
+      "loss": 0.9047,
+      "step": 9277
+    },
+    {
+      "epoch": 0.9710099424385139,
+      "grad_norm": 2.460250636187858,
+      "learning_rate": 4.404921172674659e-08,
+      "loss": 0.8763,
+      "step": 9278
+    },
+    {
+      "epoch": 0.9711145996860283,
+      "grad_norm": 2.05066277066522,
+      "learning_rate": 4.373197256199668e-08,
+      "loss": 0.8936,
+      "step": 9279
+    },
+    {
+      "epoch": 0.9712192569335426,
+      "grad_norm": 2.0608607605365914,
+      "learning_rate": 4.3415877392626314e-08,
+      "loss": 0.8599,
+      "step": 9280
+    },
+    {
+      "epoch": 0.971323914181057,
+      "grad_norm": 1.9005786956775899,
+      "learning_rate": 4.3100926254956434e-08,
+      "loss": 0.9429,
+      "step": 9281
+    },
+    {
+      "epoch": 0.9714285714285714,
+      "grad_norm": 2.3320969105885028,
+      "learning_rate": 4.2787119185174756e-08,
+      "loss": 0.9442,
+      "step": 9282
+    },
+    {
+      "epoch": 0.9715332286760858,
+      "grad_norm": 1.9467790462210524,
+      "learning_rate": 4.247445621933799e-08,
+      "loss": 0.9022,
+      "step": 9283
+    },
+    {
+      "epoch": 0.9716378859236002,
+      "grad_norm": 2.1353572789569237,
+      "learning_rate": 4.216293739337296e-08,
+      "loss": 0.8758,
+      "step": 9284
+    },
+    {
+      "epoch": 0.9717425431711146,
+      "grad_norm": 2.0519015338547826,
+      "learning_rate": 4.1852562743072144e-08,
+      "loss": 0.8413,
+      "step": 9285
+    },
+    {
+      "epoch": 0.971847200418629,
+      "grad_norm": 2.07726966403895,
+      "learning_rate": 4.154333230410035e-08,
+      "loss": 0.9582,
+      "step": 9286
+    },
+    {
+      "epoch": 0.9719518576661433,
+      "grad_norm": 1.734959100833163,
+      "learning_rate": 4.1235246111986925e-08,
+      "loss": 0.715,
+      "step": 9287
+    },
+    {
+      "epoch": 0.9720565149136577,
+      "grad_norm": 2.3376164950199807,
+      "learning_rate": 4.092830420213134e-08,
+      "loss": 0.7634,
+      "step": 9288
+    },
+    {
+      "epoch": 0.9721611721611721,
+      "grad_norm": 2.0354642090531447,
+      "learning_rate": 4.062250660980427e-08,
+      "loss": 0.9849,
+      "step": 9289
+    },
+    {
+      "epoch": 0.9722658294086866,
+      "grad_norm": 2.4203466382212264,
+      "learning_rate": 4.0317853370139826e-08,
+      "loss": 0.9165,
+      "step": 9290
+    },
+    {
+      "epoch": 0.972370486656201,
+      "grad_norm": 2.2074429056634024,
+      "learning_rate": 4.0014344518145566e-08,
+      "loss": 0.8997,
+      "step": 9291
+    },
+    {
+      "epoch": 0.9724751439037154,
+      "grad_norm": 1.9811031986210714,
+      "learning_rate": 3.971198008869359e-08,
+      "loss": 0.8448,
+      "step": 9292
+    },
+    {
+      "epoch": 0.9725798011512298,
+      "grad_norm": 2.140986208627522,
+      "learning_rate": 3.9410760116526115e-08,
+      "loss": 0.9095,
+      "step": 9293
+    },
+    {
+      "epoch": 0.9726844583987441,
+      "grad_norm": 2.1998429755193047,
+      "learning_rate": 3.9110684636254334e-08,
+      "loss": 0.8572,
+      "step": 9294
+    },
+    {
+      "epoch": 0.9727891156462585,
+      "grad_norm": 1.791680128166834,
+      "learning_rate": 3.8811753682358454e-08,
+      "loss": 0.8926,
+      "step": 9295
+    },
+    {
+      "epoch": 0.9728937728937729,
+      "grad_norm": 2.0214903816242873,
+      "learning_rate": 3.851396728918544e-08,
+      "loss": 0.9469,
+      "step": 9296
+    },
+    {
+      "epoch": 0.9729984301412873,
+      "grad_norm": 2.086981122937348,
+      "learning_rate": 3.8217325490952365e-08,
+      "loss": 0.8404,
+      "step": 9297
+    },
+    {
+      "epoch": 0.9731030873888017,
+      "grad_norm": 2.175421733396647,
+      "learning_rate": 3.792182832174196e-08,
+      "loss": 0.9427,
+      "step": 9298
+    },
+    {
+      "epoch": 0.9732077446363161,
+      "grad_norm": 1.9291300718410933,
+      "learning_rate": 3.76274758155093e-08,
+      "loss": 0.8368,
+      "step": 9299
+    },
+    {
+      "epoch": 0.9733124018838305,
+      "grad_norm": 2.117352303798171,
+      "learning_rate": 3.7334268006075094e-08,
+      "loss": 0.9744,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9734170591313448,
+      "grad_norm": 1.675142403551649,
+      "learning_rate": 3.7042204927130177e-08,
+      "loss": 0.7633,
+      "step": 9301
+    },
+    {
+      "epoch": 0.9735217163788592,
+      "grad_norm": 2.036570056386732,
+      "learning_rate": 3.675128661223326e-08,
+      "loss": 0.6977,
+      "step": 9302
+    },
+    {
+      "epoch": 0.9736263736263736,
+      "grad_norm": 1.8889553909422538,
+      "learning_rate": 3.646151309481094e-08,
+      "loss": 0.792,
+      "step": 9303
+    },
+    {
+      "epoch": 0.973731030873888,
+      "grad_norm": 2.209590888127619,
+      "learning_rate": 3.6172884408158805e-08,
+      "loss": 0.874,
+      "step": 9304
+    },
+    {
+      "epoch": 0.9738356881214024,
+      "grad_norm": 1.9542284512171744,
+      "learning_rate": 3.5885400585441435e-08,
+      "loss": 0.8408,
+      "step": 9305
+    },
+    {
+      "epoch": 0.9739403453689168,
+      "grad_norm": 2.079751071735509,
+      "learning_rate": 3.559906165969018e-08,
+      "loss": 0.8197,
+      "step": 9306
+    },
+    {
+      "epoch": 0.9740450026164312,
+      "grad_norm": 1.7986235225721088,
+      "learning_rate": 3.531386766380762e-08,
+      "loss": 0.8663,
+      "step": 9307
+    },
+    {
+      "epoch": 0.9741496598639455,
+      "grad_norm": 2.120431133086565,
+      "learning_rate": 3.502981863056088e-08,
+      "loss": 1.016,
+      "step": 9308
+    },
+    {
+      "epoch": 0.9742543171114599,
+      "grad_norm": 2.27161995008741,
+      "learning_rate": 3.4746914592590496e-08,
+      "loss": 0.8601,
+      "step": 9309
+    },
+    {
+      "epoch": 0.9743589743589743,
+      "grad_norm": 1.8602129108469572,
+      "learning_rate": 3.446515558240049e-08,
+      "loss": 0.8764,
+      "step": 9310
+    },
+    {
+      "epoch": 0.9744636316064887,
+      "grad_norm": 2.5728016050016245,
+      "learning_rate": 3.4184541632366066e-08,
+      "loss": 1.0023,
+      "step": 9311
+    },
+    {
+      "epoch": 0.9745682888540032,
+      "grad_norm": 2.3187035734822334,
+      "learning_rate": 3.390507277473143e-08,
+      "loss": 0.9637,
+      "step": 9312
+    },
+    {
+      "epoch": 0.9746729461015176,
+      "grad_norm": 1.8523908673682685,
+      "learning_rate": 3.3626749041606454e-08,
+      "loss": 0.9465,
+      "step": 9313
+    },
+    {
+      "epoch": 0.974777603349032,
+      "grad_norm": 1.9384599924677752,
+      "learning_rate": 3.3349570464973336e-08,
+      "loss": 0.7849,
+      "step": 9314
+    },
+    {
+      "epoch": 0.9748822605965463,
+      "grad_norm": 1.7474977395994695,
+      "learning_rate": 3.307353707667771e-08,
+      "loss": 0.7028,
+      "step": 9315
+    },
+    {
+      "epoch": 0.9749869178440607,
+      "grad_norm": 1.7458067005238376,
+      "learning_rate": 3.279864890843865e-08,
+      "loss": 0.9057,
+      "step": 9316
+    },
+    {
+      "epoch": 0.9750915750915751,
+      "grad_norm": 1.93802753381986,
+      "learning_rate": 3.2524905991839775e-08,
+      "loss": 0.7735,
+      "step": 9317
+    },
+    {
+      "epoch": 0.9751962323390895,
+      "grad_norm": 2.0070696055706616,
+      "learning_rate": 3.225230835833593e-08,
+      "loss": 0.8985,
+      "step": 9318
+    },
+    {
+      "epoch": 0.9753008895866039,
+      "grad_norm": 2.290676066932009,
+      "learning_rate": 3.198085603924761e-08,
+      "loss": 0.8573,
+      "step": 9319
+    },
+    {
+      "epoch": 0.9754055468341183,
+      "grad_norm": 1.9242114892590325,
+      "learning_rate": 3.1710549065767645e-08,
+      "loss": 0.8236,
+      "step": 9320
+    },
+    {
+      "epoch": 0.9755102040816327,
+      "grad_norm": 2.0179022005643312,
+      "learning_rate": 3.144138746895231e-08,
+      "loss": 0.9594,
+      "step": 9321
+    },
+    {
+      "epoch": 0.9756148613291471,
+      "grad_norm": 1.688029558610551,
+      "learning_rate": 3.117337127973019e-08,
+      "loss": 0.6891,
+      "step": 9322
+    },
+    {
+      "epoch": 0.9757195185766614,
+      "grad_norm": 2.123029102476217,
+      "learning_rate": 3.0906500528896656e-08,
+      "loss": 0.9667,
+      "step": 9323
+    },
+    {
+      "epoch": 0.9758241758241758,
+      "grad_norm": 2.114035431835684,
+      "learning_rate": 3.064077524711606e-08,
+      "loss": 0.8246,
+      "step": 9324
+    },
+    {
+      "epoch": 0.9759288330716902,
+      "grad_norm": 1.9867461075805437,
+      "learning_rate": 3.0376195464920655e-08,
+      "loss": 0.9916,
+      "step": 9325
+    },
+    {
+      "epoch": 0.9760334903192046,
+      "grad_norm": 2.2706805982649962,
+      "learning_rate": 3.0112761212709454e-08,
+      "loss": 0.716,
+      "step": 9326
+    },
+    {
+      "epoch": 0.976138147566719,
+      "grad_norm": 2.2268234648526413,
+      "learning_rate": 2.9850472520753795e-08,
+      "loss": 0.84,
+      "step": 9327
+    },
+    {
+      "epoch": 0.9762428048142334,
+      "grad_norm": 2.217677339428372,
+      "learning_rate": 2.9589329419190683e-08,
+      "loss": 0.8897,
+      "step": 9328
+    },
+    {
+      "epoch": 0.9763474620617478,
+      "grad_norm": 2.093519971186103,
+      "learning_rate": 2.9329331938026117e-08,
+      "loss": 0.8808,
+      "step": 9329
+    },
+    {
+      "epoch": 0.9764521193092621,
+      "grad_norm": 1.9406821046190856,
+      "learning_rate": 2.907048010713398e-08,
+      "loss": 0.8893,
+      "step": 9330
+    },
+    {
+      "epoch": 0.9765567765567765,
+      "grad_norm": 1.9257719073652975,
+      "learning_rate": 2.8812773956256034e-08,
+      "loss": 0.8714,
+      "step": 9331
+    },
+    {
+      "epoch": 0.9766614338042909,
+      "grad_norm": 2.330309710581296,
+      "learning_rate": 2.855621351500415e-08,
+      "loss": 0.9156,
+      "step": 9332
+    },
+    {
+      "epoch": 0.9767660910518053,
+      "grad_norm": 1.742461766113821,
+      "learning_rate": 2.8300798812858078e-08,
+      "loss": 0.7883,
+      "step": 9333
+    },
+    {
+      "epoch": 0.9768707482993197,
+      "grad_norm": 2.150066157396379,
+      "learning_rate": 2.8046529879164343e-08,
+      "loss": 0.7517,
+      "step": 9334
+    },
+    {
+      "epoch": 0.9769754055468342,
+      "grad_norm": 2.162913910208753,
+      "learning_rate": 2.7793406743139572e-08,
+      "loss": 0.8699,
+      "step": 9335
+    },
+    {
+      "epoch": 0.9770800627943486,
+      "grad_norm": 3.0655887965284343,
+      "learning_rate": 2.7541429433869393e-08,
+      "loss": 0.8514,
+      "step": 9336
+    },
+    {
+      "epoch": 0.9771847200418629,
+      "grad_norm": 1.9873130806621349,
+      "learning_rate": 2.729059798030398e-08,
+      "loss": 0.7885,
+      "step": 9337
+    },
+    {
+      "epoch": 0.9772893772893773,
+      "grad_norm": 2.2769678998356295,
+      "learning_rate": 2.7040912411265828e-08,
+      "loss": 0.9203,
+      "step": 9338
+    },
+    {
+      "epoch": 0.9773940345368917,
+      "grad_norm": 1.9774891685632308,
+      "learning_rate": 2.679237275544422e-08,
+      "loss": 0.8983,
+      "step": 9339
+    },
+    {
+      "epoch": 0.9774986917844061,
+      "grad_norm": 2.2420458226870985,
+      "learning_rate": 2.6544979041397412e-08,
+      "loss": 0.7687,
+      "step": 9340
+    },
+    {
+      "epoch": 0.9776033490319205,
+      "grad_norm": 2.119669776522172,
+      "learning_rate": 2.6298731297550452e-08,
+      "loss": 0.8316,
+      "step": 9341
+    },
+    {
+      "epoch": 0.9777080062794349,
+      "grad_norm": 2.4118682693637385,
+      "learning_rate": 2.605362955219737e-08,
+      "loss": 0.7018,
+      "step": 9342
+    },
+    {
+      "epoch": 0.9778126635269493,
+      "grad_norm": 2.0233107477477787,
+      "learning_rate": 2.5809673833502303e-08,
+      "loss": 0.8836,
+      "step": 9343
+    },
+    {
+      "epoch": 0.9779173207744636,
+      "grad_norm": 1.9871161511437192,
+      "learning_rate": 2.5566864169495054e-08,
+      "loss": 0.735,
+      "step": 9344
+    },
+    {
+      "epoch": 0.978021978021978,
+      "grad_norm": 2.292335147286737,
+      "learning_rate": 2.5325200588076636e-08,
+      "loss": 0.839,
+      "step": 9345
+    },
+    {
+      "epoch": 0.9781266352694924,
+      "grad_norm": 2.2868212579297644,
+      "learning_rate": 2.508468311701151e-08,
+      "loss": 0.8197,
+      "step": 9346
+    },
+    {
+      "epoch": 0.9782312925170068,
+      "grad_norm": 2.1010720301980004,
+      "learning_rate": 2.4845311783938676e-08,
+      "loss": 0.9124,
+      "step": 9347
+    },
+    {
+      "epoch": 0.9783359497645212,
+      "grad_norm": 1.8598221177836944,
+      "learning_rate": 2.4607086616361687e-08,
+      "loss": 0.8497,
+      "step": 9348
+    },
+    {
+      "epoch": 0.9784406070120356,
+      "grad_norm": 2.237038039840748,
+      "learning_rate": 2.4370007641651983e-08,
+      "loss": 0.8217,
+      "step": 9349
+    },
+    {
+      "epoch": 0.97854526425955,
+      "grad_norm": 2.085673606140328,
+      "learning_rate": 2.4134074887051108e-08,
+      "loss": 0.8642,
+      "step": 9350
+    },
+    {
+      "epoch": 0.9786499215070643,
+      "grad_norm": 1.9367343428587138,
+      "learning_rate": 2.3899288379668483e-08,
+      "loss": 0.9598,
+      "step": 9351
+    },
+    {
+      "epoch": 0.9787545787545787,
+      "grad_norm": 1.968532478877287,
+      "learning_rate": 2.3665648146480315e-08,
+      "loss": 0.9346,
+      "step": 9352
+    },
+    {
+      "epoch": 0.9788592360020931,
+      "grad_norm": 2.2698739958097547,
+      "learning_rate": 2.343315421433401e-08,
+      "loss": 0.9245,
+      "step": 9353
+    },
+    {
+      "epoch": 0.9789638932496075,
+      "grad_norm": 2.1111746202775517,
+      "learning_rate": 2.3201806609943756e-08,
+      "loss": 0.8771,
+      "step": 9354
+    },
+    {
+      "epoch": 0.9790685504971219,
+      "grad_norm": 2.395697271895756,
+      "learning_rate": 2.2971605359890514e-08,
+      "loss": 0.7715,
+      "step": 9355
+    },
+    {
+      "epoch": 0.9791732077446363,
+      "grad_norm": 2.048806092369536,
+      "learning_rate": 2.2742550490624237e-08,
+      "loss": 0.8791,
+      "step": 9356
+    },
+    {
+      "epoch": 0.9792778649921507,
+      "grad_norm": 2.0576345151800197,
+      "learning_rate": 2.2514642028466092e-08,
+      "loss": 0.9454,
+      "step": 9357
+    },
+    {
+      "epoch": 0.979382522239665,
+      "grad_norm": 2.1060942188116294,
+      "learning_rate": 2.2287879999601804e-08,
+      "loss": 0.8494,
+      "step": 9358
+    },
+    {
+      "epoch": 0.9794871794871794,
+      "grad_norm": 2.0030711225076745,
+      "learning_rate": 2.2062264430087187e-08,
+      "loss": 0.7593,
+      "step": 9359
+    },
+    {
+      "epoch": 0.9795918367346939,
+      "grad_norm": 2.18284415979648,
+      "learning_rate": 2.1837795345844847e-08,
+      "loss": 0.8958,
+      "step": 9360
+    },
+    {
+      "epoch": 0.9796964939822083,
+      "grad_norm": 1.9127644215367114,
+      "learning_rate": 2.1614472772668592e-08,
+      "loss": 0.9716,
+      "step": 9361
+    },
+    {
+      "epoch": 0.9798011512297227,
+      "grad_norm": 1.8495943717092973,
+      "learning_rate": 2.13922967362179e-08,
+      "loss": 0.8342,
+      "step": 9362
+    },
+    {
+      "epoch": 0.9799058084772371,
+      "grad_norm": 2.1175372567636703,
+      "learning_rate": 2.1171267262020124e-08,
+      "loss": 0.9731,
+      "step": 9363
+    },
+    {
+      "epoch": 0.9800104657247515,
+      "grad_norm": 1.961813987025682,
+      "learning_rate": 2.0951384375473837e-08,
+      "loss": 0.8674,
+      "step": 9364
+    },
+    {
+      "epoch": 0.9801151229722659,
+      "grad_norm": 2.2096628727490555,
+      "learning_rate": 2.073264810184328e-08,
+      "loss": 0.8222,
+      "step": 9365
+    },
+    {
+      "epoch": 0.9802197802197802,
+      "grad_norm": 3.0174110994465653,
+      "learning_rate": 2.0515058466260563e-08,
+      "loss": 0.7058,
+      "step": 9366
+    },
+    {
+      "epoch": 0.9803244374672946,
+      "grad_norm": 1.9638967173681348,
+      "learning_rate": 2.0298615493729025e-08,
+      "loss": 0.8708,
+      "step": 9367
+    },
+    {
+      "epoch": 0.980429094714809,
+      "grad_norm": 1.8984784000934047,
+      "learning_rate": 2.0083319209116547e-08,
+      "loss": 0.8403,
+      "step": 9368
+    },
+    {
+      "epoch": 0.9805337519623234,
+      "grad_norm": 1.9998734455651994,
+      "learning_rate": 1.9869169637162232e-08,
+      "loss": 0.8549,
+      "step": 9369
+    },
+    {
+      "epoch": 0.9806384092098378,
+      "grad_norm": 2.7056047403555916,
+      "learning_rate": 1.9656166802473066e-08,
+      "loss": 0.7828,
+      "step": 9370
+    },
+    {
+      "epoch": 0.9807430664573522,
+      "grad_norm": 1.885856551772748,
+      "learning_rate": 1.9444310729521686e-08,
+      "loss": 0.8384,
+      "step": 9371
+    },
+    {
+      "epoch": 0.9808477237048666,
+      "grad_norm": 2.3439297429643835,
+      "learning_rate": 1.9233601442653073e-08,
+      "loss": 0.8945,
+      "step": 9372
+    },
+    {
+      "epoch": 0.9809523809523809,
+      "grad_norm": 2.273645673597301,
+      "learning_rate": 1.902403896607563e-08,
+      "loss": 0.8892,
+      "step": 9373
+    },
+    {
+      "epoch": 0.9810570381998953,
+      "grad_norm": 2.101477849817953,
+      "learning_rate": 1.8815623323870102e-08,
+      "loss": 0.7195,
+      "step": 9374
+    },
+    {
+      "epoch": 0.9811616954474097,
+      "grad_norm": 2.067407200022368,
+      "learning_rate": 1.8608354539982886e-08,
+      "loss": 0.931,
+      "step": 9375
+    },
+    {
+      "epoch": 0.9812663526949241,
+      "grad_norm": 2.0656438344157837,
+      "learning_rate": 1.8402232638230488e-08,
+      "loss": 0.9159,
+      "step": 9376
+    },
+    {
+      "epoch": 0.9813710099424385,
+      "grad_norm": 1.989366318840071,
+      "learning_rate": 1.8197257642296183e-08,
+      "loss": 0.9061,
+      "step": 9377
+    },
+    {
+      "epoch": 0.9814756671899529,
+      "grad_norm": 2.1104290095055007,
+      "learning_rate": 1.7993429575732247e-08,
+      "loss": 0.7356,
+      "step": 9378
+    },
+    {
+      "epoch": 0.9815803244374673,
+      "grad_norm": 2.3574441233228804,
+      "learning_rate": 1.779074846195883e-08,
+      "loss": 0.7541,
+      "step": 9379
+    },
+    {
+      "epoch": 0.9816849816849816,
+      "grad_norm": 2.053364465624434,
+      "learning_rate": 1.758921432426397e-08,
+      "loss": 0.929,
+      "step": 9380
+    },
+    {
+      "epoch": 0.981789638932496,
+      "grad_norm": 1.9733850210433772,
+      "learning_rate": 1.738882718580581e-08,
+      "loss": 0.8051,
+      "step": 9381
+    },
+    {
+      "epoch": 0.9818942961800105,
+      "grad_norm": 2.010472060114112,
+      "learning_rate": 1.718958706960816e-08,
+      "loss": 0.8714,
+      "step": 9382
+    },
+    {
+      "epoch": 0.9819989534275249,
+      "grad_norm": 1.9418541630139494,
+      "learning_rate": 1.6991493998563814e-08,
+      "loss": 0.9111,
+      "step": 9383
+    },
+    {
+      "epoch": 0.9821036106750393,
+      "grad_norm": 2.0017027127013303,
+      "learning_rate": 1.6794547995434563e-08,
+      "loss": 0.8167,
+      "step": 9384
+    },
+    {
+      "epoch": 0.9822082679225537,
+      "grad_norm": 2.0504485758602895,
+      "learning_rate": 1.6598749082850084e-08,
+      "loss": 0.8904,
+      "step": 9385
+    },
+    {
+      "epoch": 0.9823129251700681,
+      "grad_norm": 2.360758373402244,
+      "learning_rate": 1.640409728330794e-08,
+      "loss": 0.9911,
+      "step": 9386
+    },
+    {
+      "epoch": 0.9824175824175824,
+      "grad_norm": 1.99459443178208,
+      "learning_rate": 1.621059261917357e-08,
+      "loss": 0.844,
+      "step": 9387
+    },
+    {
+      "epoch": 0.9825222396650968,
+      "grad_norm": 1.8418136023305347,
+      "learning_rate": 1.6018235112681412e-08,
+      "loss": 0.8789,
+      "step": 9388
+    },
+    {
+      "epoch": 0.9826268969126112,
+      "grad_norm": 1.6553513965603035,
+      "learning_rate": 1.5827024785933785e-08,
+      "loss": 0.7062,
+      "step": 9389
+    },
+    {
+      "epoch": 0.9827315541601256,
+      "grad_norm": 1.9968661967922576,
+      "learning_rate": 1.5636961660899787e-08,
+      "loss": 0.7489,
+      "step": 9390
+    },
+    {
+      "epoch": 0.98283621140764,
+      "grad_norm": 2.3529232712311807,
+      "learning_rate": 1.5448045759420827e-08,
+      "loss": 0.9545,
+      "step": 9391
+    },
+    {
+      "epoch": 0.9829408686551544,
+      "grad_norm": 1.8819806519766973,
+      "learning_rate": 1.526027710320177e-08,
+      "loss": 0.8448,
+      "step": 9392
+    },
+    {
+      "epoch": 0.9830455259026688,
+      "grad_norm": 2.138462511740512,
+      "learning_rate": 1.5073655713818692e-08,
+      "loss": 0.9485,
+      "step": 9393
+    },
+    {
+      "epoch": 0.9831501831501831,
+      "grad_norm": 2.3151261223417703,
+      "learning_rate": 1.4888181612713327e-08,
+      "loss": 0.9372,
+      "step": 9394
+    },
+    {
+      "epoch": 0.9832548403976975,
+      "grad_norm": 2.1342293657223292,
+      "learning_rate": 1.4703854821197516e-08,
+      "loss": 0.7339,
+      "step": 9395
+    },
+    {
+      "epoch": 0.9833594976452119,
+      "grad_norm": 1.9715185392028405,
+      "learning_rate": 1.4520675360450987e-08,
+      "loss": 0.9157,
+      "step": 9396
+    },
+    {
+      "epoch": 0.9834641548927263,
+      "grad_norm": 2.6450142775406964,
+      "learning_rate": 1.4338643251522456e-08,
+      "loss": 0.9199,
+      "step": 9397
+    },
+    {
+      "epoch": 0.9835688121402407,
+      "grad_norm": 1.8046838157595604,
+      "learning_rate": 1.4157758515326303e-08,
+      "loss": 0.749,
+      "step": 9398
+    },
+    {
+      "epoch": 0.9836734693877551,
+      "grad_norm": 1.8899814023719925,
+      "learning_rate": 1.3978021172648126e-08,
+      "loss": 0.93,
+      "step": 9399
+    },
+    {
+      "epoch": 0.9837781266352695,
+      "grad_norm": 1.6809007606749307,
+      "learning_rate": 1.3799431244138073e-08,
+      "loss": 0.7007,
+      "step": 9400
+    },
+    {
+      "epoch": 0.9838827838827838,
+      "grad_norm": 1.6089297962761584,
+      "learning_rate": 1.3621988750317505e-08,
+      "loss": 0.7509,
+      "step": 9401
+    },
+    {
+      "epoch": 0.9839874411302982,
+      "grad_norm": 2.032535731745511,
+      "learning_rate": 1.344569371157567e-08,
+      "loss": 0.8048,
+      "step": 9402
+    },
+    {
+      "epoch": 0.9840920983778126,
+      "grad_norm": 1.9666083217252117,
+      "learning_rate": 1.3270546148167474e-08,
+      "loss": 0.8324,
+      "step": 9403
+    },
+    {
+      "epoch": 0.984196755625327,
+      "grad_norm": 2.0185643963937054,
+      "learning_rate": 1.3096546080219042e-08,
+      "loss": 0.8746,
+      "step": 9404
+    },
+    {
+      "epoch": 0.9843014128728415,
+      "grad_norm": 1.9656375458562914,
+      "learning_rate": 1.292369352772327e-08,
+      "loss": 0.8422,
+      "step": 9405
+    },
+    {
+      "epoch": 0.9844060701203559,
+      "grad_norm": 2.331480394760753,
+      "learning_rate": 1.2751988510540937e-08,
+      "loss": 0.9247,
+      "step": 9406
+    },
+    {
+      "epoch": 0.9845107273678703,
+      "grad_norm": 1.9029882738516415,
+      "learning_rate": 1.2581431048401815e-08,
+      "loss": 0.8051,
+      "step": 9407
+    },
+    {
+      "epoch": 0.9846153846153847,
+      "grad_norm": 1.9192845614133922,
+      "learning_rate": 1.2412021160902454e-08,
+      "loss": 0.851,
+      "step": 9408
+    },
+    {
+      "epoch": 0.984720041862899,
+      "grad_norm": 1.8685299347220463,
+      "learning_rate": 1.2243758867508393e-08,
+      "loss": 0.798,
+      "step": 9409
+    },
+    {
+      "epoch": 0.9848246991104134,
+      "grad_norm": 2.171112776891488,
+      "learning_rate": 1.2076644187554165e-08,
+      "loss": 0.9017,
+      "step": 9410
+    },
+    {
+      "epoch": 0.9849293563579278,
+      "grad_norm": 2.3532297197914427,
+      "learning_rate": 1.1910677140241077e-08,
+      "loss": 0.902,
+      "step": 9411
+    },
+    {
+      "epoch": 0.9850340136054422,
+      "grad_norm": 2.1352803267811944,
+      "learning_rate": 1.1745857744639433e-08,
+      "loss": 0.9434,
+      "step": 9412
+    },
+    {
+      "epoch": 0.9851386708529566,
+      "grad_norm": 2.0466470004257444,
+      "learning_rate": 1.1582186019686303e-08,
+      "loss": 0.854,
+      "step": 9413
+    },
+    {
+      "epoch": 0.985243328100471,
+      "grad_norm": 1.8753970699615763,
+      "learning_rate": 1.1419661984189978e-08,
+      "loss": 0.885,
+      "step": 9414
+    },
+    {
+      "epoch": 0.9853479853479854,
+      "grad_norm": 1.9062133085606772,
+      "learning_rate": 1.1258285656822187e-08,
+      "loss": 0.8093,
+      "step": 9415
+    },
+    {
+      "epoch": 0.9854526425954997,
+      "grad_norm": 2.2126700579173644,
+      "learning_rate": 1.1098057056126987e-08,
+      "loss": 0.9876,
+      "step": 9416
+    },
+    {
+      "epoch": 0.9855572998430141,
+      "grad_norm": 2.0255896960925255,
+      "learning_rate": 1.0938976200515206e-08,
+      "loss": 0.742,
+      "step": 9417
+    },
+    {
+      "epoch": 0.9856619570905285,
+      "grad_norm": 2.1729727012223328,
+      "learning_rate": 1.0781043108264444e-08,
+      "loss": 0.8281,
+      "step": 9418
+    },
+    {
+      "epoch": 0.9857666143380429,
+      "grad_norm": 2.5225886899280794,
+      "learning_rate": 1.0624257797522408e-08,
+      "loss": 0.8728,
+      "step": 9419
+    },
+    {
+      "epoch": 0.9858712715855573,
+      "grad_norm": 2.0350945295517993,
+      "learning_rate": 1.0468620286303577e-08,
+      "loss": 0.8491,
+      "step": 9420
+    },
+    {
+      "epoch": 0.9859759288330717,
+      "grad_norm": 2.2700948009138253,
+      "learning_rate": 1.0314130592490313e-08,
+      "loss": 0.9252,
+      "step": 9421
+    },
+    {
+      "epoch": 0.9860805860805861,
+      "grad_norm": 1.9846202839221025,
+      "learning_rate": 1.0160788733836191e-08,
+      "loss": 0.8137,
+      "step": 9422
+    },
+    {
+      "epoch": 0.9861852433281004,
+      "grad_norm": 2.3145335457286462,
+      "learning_rate": 1.0008594727958232e-08,
+      "loss": 0.6612,
+      "step": 9423
+    },
+    {
+      "epoch": 0.9862899005756148,
+      "grad_norm": 1.6193909422722288,
+      "learning_rate": 9.857548592343558e-09,
+      "loss": 0.7132,
+      "step": 9424
+    },
+    {
+      "epoch": 0.9863945578231292,
+      "grad_norm": 2.3517435794568353,
+      "learning_rate": 9.707650344350505e-09,
+      "loss": 0.8793,
+      "step": 9425
+    },
+    {
+      "epoch": 0.9864992150706436,
+      "grad_norm": 1.9920534058323922,
+      "learning_rate": 9.558900001199745e-09,
+      "loss": 0.8218,
+      "step": 9426
+    },
+    {
+      "epoch": 0.986603872318158,
+      "grad_norm": 2.427971815968549,
+      "learning_rate": 9.411297579984269e-09,
+      "loss": 0.9036,
+      "step": 9427
+    },
+    {
+      "epoch": 0.9867085295656725,
+      "grad_norm": 1.8261977196224324,
+      "learning_rate": 9.264843097663845e-09,
+      "loss": 0.776,
+      "step": 9428
+    },
+    {
+      "epoch": 0.9868131868131869,
+      "grad_norm": 2.197882541506156,
+      "learning_rate": 9.119536571066123e-09,
+      "loss": 0.8071,
+      "step": 9429
+    },
+    {
+      "epoch": 0.9869178440607012,
+      "grad_norm": 1.9625634256614233,
+      "learning_rate": 8.975378016887748e-09,
+      "loss": 0.7967,
+      "step": 9430
+    },
+    {
+      "epoch": 0.9870225013082156,
+      "grad_norm": 1.6165607498949757,
+      "learning_rate": 8.832367451692137e-09,
+      "loss": 0.7942,
+      "step": 9431
+    },
+    {
+      "epoch": 0.98712715855573,
+      "grad_norm": 2.0636190975616793,
+      "learning_rate": 8.6905048919117e-09,
+      "loss": 0.9357,
+      "step": 9432
+    },
+    {
+      "epoch": 0.9872318158032444,
+      "grad_norm": 2.0496132764727255,
+      "learning_rate": 8.549790353846731e-09,
+      "loss": 0.9125,
+      "step": 9433
+    },
+    {
+      "epoch": 0.9873364730507588,
+      "grad_norm": 1.98179465824096,
+      "learning_rate": 8.410223853665411e-09,
+      "loss": 0.7822,
+      "step": 9434
+    },
+    {
+      "epoch": 0.9874411302982732,
+      "grad_norm": 2.321799912705264,
+      "learning_rate": 8.27180540740491e-09,
+      "loss": 0.8323,
+      "step": 9435
+    },
+    {
+      "epoch": 0.9875457875457876,
+      "grad_norm": 1.8047308460509242,
+      "learning_rate": 8.134535030968061e-09,
+      "loss": 0.839,
+      "step": 9436
+    },
+    {
+      "epoch": 0.9876504447933019,
+      "grad_norm": 2.7150759739338293,
+      "learning_rate": 7.998412740130023e-09,
+      "loss": 0.9566,
+      "step": 9437
+    },
+    {
+      "epoch": 0.9877551020408163,
+      "grad_norm": 1.8975694108774446,
+      "learning_rate": 7.863438550529401e-09,
+      "loss": 0.8174,
+      "step": 9438
+    },
+    {
+      "epoch": 0.9878597592883307,
+      "grad_norm": 2.239706974304119,
+      "learning_rate": 7.729612477677118e-09,
+      "loss": 0.9733,
+      "step": 9439
+    },
+    {
+      "epoch": 0.9879644165358451,
+      "grad_norm": 1.7630296843202333,
+      "learning_rate": 7.596934536947542e-09,
+      "loss": 0.8704,
+      "step": 9440
+    },
+    {
+      "epoch": 0.9880690737833595,
+      "grad_norm": 2.1363895002879407,
+      "learning_rate": 7.465404743587368e-09,
+      "loss": 0.7886,
+      "step": 9441
+    },
+    {
+      "epoch": 0.9881737310308739,
+      "grad_norm": 2.2348274430335837,
+      "learning_rate": 7.33502311271006e-09,
+      "loss": 0.9167,
+      "step": 9442
+    },
+    {
+      "epoch": 0.9882783882783883,
+      "grad_norm": 2.9112946474193615,
+      "learning_rate": 7.205789659294748e-09,
+      "loss": 0.9517,
+      "step": 9443
+    },
+    {
+      "epoch": 0.9883830455259026,
+      "grad_norm": 1.9351507967777781,
+      "learning_rate": 7.0777043981917756e-09,
+      "loss": 0.9959,
+      "step": 9444
+    },
+    {
+      "epoch": 0.988487702773417,
+      "grad_norm": 2.205838160688878,
+      "learning_rate": 6.950767344118259e-09,
+      "loss": 1.0637,
+      "step": 9445
+    },
+    {
+      "epoch": 0.9885923600209314,
+      "grad_norm": 1.7747166910206333,
+      "learning_rate": 6.824978511660307e-09,
+      "loss": 0.7722,
+      "step": 9446
+    },
+    {
+      "epoch": 0.9886970172684458,
+      "grad_norm": 1.7660762190615527,
+      "learning_rate": 6.700337915269694e-09,
+      "loss": 0.8423,
+      "step": 9447
+    },
+    {
+      "epoch": 0.9888016745159602,
+      "grad_norm": 2.1500153429277287,
+      "learning_rate": 6.576845569269408e-09,
+      "loss": 0.9119,
+      "step": 9448
+    },
+    {
+      "epoch": 0.9889063317634746,
+      "grad_norm": 2.1119803885084334,
+      "learning_rate": 6.454501487848097e-09,
+      "loss": 0.8946,
+      "step": 9449
+    },
+    {
+      "epoch": 0.989010989010989,
+      "grad_norm": 2.045363078081982,
+      "learning_rate": 6.333305685064517e-09,
+      "loss": 0.814,
+      "step": 9450
+    },
+    {
+      "epoch": 0.9891156462585035,
+      "grad_norm": 2.143018279644762,
+      "learning_rate": 6.213258174841974e-09,
+      "loss": 0.8763,
+      "step": 9451
+    },
+    {
+      "epoch": 0.9892203035060178,
+      "grad_norm": 1.8414553993367668,
+      "learning_rate": 6.094358970976099e-09,
+      "loss": 0.8197,
+      "step": 9452
+    },
+    {
+      "epoch": 0.9893249607535322,
+      "grad_norm": 2.2429501404967374,
+      "learning_rate": 5.976608087128188e-09,
+      "loss": 0.9149,
+      "step": 9453
+    },
+    {
+      "epoch": 0.9894296180010466,
+      "grad_norm": 1.7802095129658109,
+      "learning_rate": 5.860005536828528e-09,
+      "loss": 0.829,
+      "step": 9454
+    },
+    {
+      "epoch": 0.989534275248561,
+      "grad_norm": 1.958887466082894,
+      "learning_rate": 5.74455133347418e-09,
+      "loss": 0.9254,
+      "step": 9455
+    },
+    {
+      "epoch": 0.9896389324960754,
+      "grad_norm": 2.0406723361448926,
+      "learning_rate": 5.630245490331199e-09,
+      "loss": 0.8935,
+      "step": 9456
+    },
+    {
+      "epoch": 0.9897435897435898,
+      "grad_norm": 1.985359908345336,
+      "learning_rate": 5.517088020533523e-09,
+      "loss": 0.7691,
+      "step": 9457
+    },
+    {
+      "epoch": 0.9898482469911042,
+      "grad_norm": 1.9928529617111481,
+      "learning_rate": 5.405078937082975e-09,
+      "loss": 0.8635,
+      "step": 9458
+    },
+    {
+      "epoch": 0.9899529042386185,
+      "grad_norm": 2.0172391384978616,
+      "learning_rate": 5.2942182528503695e-09,
+      "loss": 0.8513,
+      "step": 9459
+    },
+    {
+      "epoch": 0.9900575614861329,
+      "grad_norm": 2.3808186332110246,
+      "learning_rate": 5.184505980574406e-09,
+      "loss": 0.9489,
+      "step": 9460
+    },
+    {
+      "epoch": 0.9901622187336473,
+      "grad_norm": 2.318056374421155,
+      "learning_rate": 5.075942132859446e-09,
+      "loss": 0.9651,
+      "step": 9461
+    },
+    {
+      "epoch": 0.9902668759811617,
+      "grad_norm": 2.342430733838004,
+      "learning_rate": 4.968526722179956e-09,
+      "loss": 0.9706,
+      "step": 9462
+    },
+    {
+      "epoch": 0.9903715332286761,
+      "grad_norm": 2.210156685115261,
+      "learning_rate": 4.8622597608793954e-09,
+      "loss": 0.8883,
+      "step": 9463
+    },
+    {
+      "epoch": 0.9904761904761905,
+      "grad_norm": 2.100418573269556,
+      "learning_rate": 4.757141261167997e-09,
+      "loss": 0.9061,
+      "step": 9464
+    },
+    {
+      "epoch": 0.9905808477237049,
+      "grad_norm": 2.092025490922726,
+      "learning_rate": 4.653171235122767e-09,
+      "loss": 0.9194,
+      "step": 9465
+    },
+    {
+      "epoch": 0.9906855049712192,
+      "grad_norm": 2.2321294289010294,
+      "learning_rate": 4.550349694690814e-09,
+      "loss": 0.8867,
+      "step": 9466
+    },
+    {
+      "epoch": 0.9907901622187336,
+      "grad_norm": 2.032473032864792,
+      "learning_rate": 4.4486766516871335e-09,
+      "loss": 0.8543,
+      "step": 9467
+    },
+    {
+      "epoch": 0.990894819466248,
+      "grad_norm": 2.117448656836186,
+      "learning_rate": 4.348152117793491e-09,
+      "loss": 0.8434,
+      "step": 9468
+    },
+    {
+      "epoch": 0.9909994767137624,
+      "grad_norm": 2.2695523367962305,
+      "learning_rate": 4.248776104560648e-09,
+      "loss": 0.9899,
+      "step": 9469
+    },
+    {
+      "epoch": 0.9911041339612768,
+      "grad_norm": 2.053174387036949,
+      "learning_rate": 4.150548623406136e-09,
+      "loss": 0.8451,
+      "step": 9470
+    },
+    {
+      "epoch": 0.9912087912087912,
+      "grad_norm": 2.1987039378958233,
+      "learning_rate": 4.053469685617595e-09,
+      "loss": 0.7486,
+      "step": 9471
+    },
+    {
+      "epoch": 0.9913134484563056,
+      "grad_norm": 2.0752962011666423,
+      "learning_rate": 3.957539302349434e-09,
+      "loss": 0.9223,
+      "step": 9472
+    },
+    {
+      "epoch": 0.9914181057038199,
+      "grad_norm": 2.318495473294372,
+      "learning_rate": 3.862757484623947e-09,
+      "loss": 0.8358,
+      "step": 9473
+    },
+    {
+      "epoch": 0.9915227629513343,
+      "grad_norm": 2.0490582959624812,
+      "learning_rate": 3.769124243332423e-09,
+      "loss": 0.9201,
+      "step": 9474
+    },
+    {
+      "epoch": 0.9916274201988488,
+      "grad_norm": 2.1895429135365436,
+      "learning_rate": 3.676639589232922e-09,
+      "loss": 0.8669,
+      "step": 9475
+    },
+    {
+      "epoch": 0.9917320774463632,
+      "grad_norm": 2.081764463908597,
+      "learning_rate": 3.5853035329513898e-09,
+      "loss": 0.7138,
+      "step": 9476
+    },
+    {
+      "epoch": 0.9918367346938776,
+      "grad_norm": 1.9405197361065825,
+      "learning_rate": 3.4951160849838738e-09,
+      "loss": 0.9535,
+      "step": 9477
+    },
+    {
+      "epoch": 0.991941391941392,
+      "grad_norm": 1.962384250829884,
+      "learning_rate": 3.4060772556931965e-09,
+      "loss": 0.8824,
+      "step": 9478
+    },
+    {
+      "epoch": 0.9920460491889064,
+      "grad_norm": 1.9074856383254064,
+      "learning_rate": 3.3181870553089523e-09,
+      "loss": 0.889,
+      "step": 9479
+    },
+    {
+      "epoch": 0.9921507064364207,
+      "grad_norm": 2.044982555109191,
+      "learning_rate": 3.2314454939297304e-09,
+      "loss": 0.8188,
+      "step": 9480
+    },
+    {
+      "epoch": 0.9922553636839351,
+      "grad_norm": 1.9272346116373464,
+      "learning_rate": 3.1458525815231122e-09,
+      "loss": 0.7799,
+      "step": 9481
+    },
+    {
+      "epoch": 0.9923600209314495,
+      "grad_norm": 1.8056929489369389,
+      "learning_rate": 3.0614083279245644e-09,
+      "loss": 0.7488,
+      "step": 9482
+    },
+    {
+      "epoch": 0.9924646781789639,
+      "grad_norm": 2.111478786571531,
+      "learning_rate": 2.9781127428352153e-09,
+      "loss": 0.9358,
+      "step": 9483
+    },
+    {
+      "epoch": 0.9925693354264783,
+      "grad_norm": 2.2187962376027532,
+      "learning_rate": 2.8959658358274077e-09,
+      "loss": 0.8856,
+      "step": 9484
+    },
+    {
+      "epoch": 0.9926739926739927,
+      "grad_norm": 2.250021426198808,
+      "learning_rate": 2.814967616339148e-09,
+      "loss": 0.9267,
+      "step": 9485
+    },
+    {
+      "epoch": 0.9927786499215071,
+      "grad_norm": 1.9471155417501587,
+      "learning_rate": 2.735118093678546e-09,
+      "loss": 0.8243,
+      "step": 9486
+    },
+    {
+      "epoch": 0.9928833071690214,
+      "grad_norm": 2.2292413033155447,
+      "learning_rate": 2.656417277018264e-09,
+      "loss": 0.9037,
+      "step": 9487
+    },
+    {
+      "epoch": 0.9929879644165358,
+      "grad_norm": 2.1438692135997863,
+      "learning_rate": 2.578865175402179e-09,
+      "loss": 0.92,
+      "step": 9488
+    },
+    {
+      "epoch": 0.9930926216640502,
+      "grad_norm": 2.0987549084951564,
+      "learning_rate": 2.5024617977420507e-09,
+      "loss": 0.866,
+      "step": 9489
+    },
+    {
+      "epoch": 0.9931972789115646,
+      "grad_norm": 2.1710980962999833,
+      "learning_rate": 2.427207152815303e-09,
+      "loss": 0.8711,
+      "step": 9490
+    },
+    {
+      "epoch": 0.993301936159079,
+      "grad_norm": 1.9976864946771278,
+      "learning_rate": 2.3531012492705728e-09,
+      "loss": 0.8194,
+      "step": 9491
+    },
+    {
+      "epoch": 0.9934065934065934,
+      "grad_norm": 2.0151148991749315,
+      "learning_rate": 2.2801440956210506e-09,
+      "loss": 0.8259,
+      "step": 9492
+    },
+    {
+      "epoch": 0.9935112506541078,
+      "grad_norm": 1.7788086796827887,
+      "learning_rate": 2.208335700251141e-09,
+      "loss": 0.6933,
+      "step": 9493
+    },
+    {
+      "epoch": 0.9936159079016222,
+      "grad_norm": 2.243955438948143,
+      "learning_rate": 2.13767607141091e-09,
+      "loss": 1.0067,
+      "step": 9494
+    },
+    {
+      "epoch": 0.9937205651491365,
+      "grad_norm": 2.138700100591113,
+      "learning_rate": 2.06816521721942e-09,
+      "loss": 0.9177,
+      "step": 9495
+    },
+    {
+      "epoch": 0.9938252223966509,
+      "grad_norm": 2.0913847556873537,
+      "learning_rate": 1.9998031456636147e-09,
+      "loss": 0.9114,
+      "step": 9496
+    },
+    {
+      "epoch": 0.9939298796441653,
+      "grad_norm": 1.658736570996112,
+      "learning_rate": 1.9325898645983223e-09,
+      "loss": 0.8015,
+      "step": 9497
+    },
+    {
+      "epoch": 0.9940345368916798,
+      "grad_norm": 2.1338090873686615,
+      "learning_rate": 1.866525381747364e-09,
+      "loss": 0.9249,
+      "step": 9498
+    },
+    {
+      "epoch": 0.9941391941391942,
+      "grad_norm": 2.469627726192965,
+      "learning_rate": 1.8016097047002246e-09,
+      "loss": 0.8704,
+      "step": 9499
+    },
+    {
+      "epoch": 0.9942438513867086,
+      "grad_norm": 2.1414411950046026,
+      "learning_rate": 1.7378428409164927e-09,
+      "loss": 0.8624,
+      "step": 9500
+    },
+    {
+      "epoch": 0.994348508634223,
+      "grad_norm": 2.042150186442989,
+      "learning_rate": 1.6752247977236401e-09,
+      "loss": 0.8937,
+      "step": 9501
+    },
+    {
+      "epoch": 0.9944531658817373,
+      "grad_norm": 1.9900829634578683,
+      "learning_rate": 1.6137555823159123e-09,
+      "loss": 0.9209,
+      "step": 9502
+    },
+    {
+      "epoch": 0.9945578231292517,
+      "grad_norm": 2.410591588659901,
+      "learning_rate": 1.5534352017565479e-09,
+      "loss": 0.8172,
+      "step": 9503
+    },
+    {
+      "epoch": 0.9946624803767661,
+      "grad_norm": 2.236434312692957,
+      "learning_rate": 1.4942636629766693e-09,
+      "loss": 0.8395,
+      "step": 9504
+    },
+    {
+      "epoch": 0.9947671376242805,
+      "grad_norm": 2.1401814250288265,
+      "learning_rate": 1.4362409727752823e-09,
+      "loss": 0.8297,
+      "step": 9505
+    },
+    {
+      "epoch": 0.9948717948717949,
+      "grad_norm": 1.648097635917843,
+      "learning_rate": 1.3793671378181662e-09,
+      "loss": 0.7285,
+      "step": 9506
+    },
+    {
+      "epoch": 0.9949764521193093,
+      "grad_norm": 2.3796168135285676,
+      "learning_rate": 1.3236421646412034e-09,
+      "loss": 0.8813,
+      "step": 9507
+    },
+    {
+      "epoch": 0.9950811093668237,
+      "grad_norm": 1.9707127190807336,
+      "learning_rate": 1.2690660596481609e-09,
+      "loss": 0.7346,
+      "step": 9508
+    },
+    {
+      "epoch": 0.995185766614338,
+      "grad_norm": 2.161094149096642,
+      "learning_rate": 1.2156388291084675e-09,
+      "loss": 0.8541,
+      "step": 9509
+    },
+    {
+      "epoch": 0.9952904238618524,
+      "grad_norm": 2.010479082385231,
+      "learning_rate": 1.1633604791605468e-09,
+      "loss": 0.8944,
+      "step": 9510
+    },
+    {
+      "epoch": 0.9953950811093668,
+      "grad_norm": 2.4630434927929925,
+      "learning_rate": 1.1122310158129257e-09,
+      "loss": 1.0294,
+      "step": 9511
+    },
+    {
+      "epoch": 0.9954997383568812,
+      "grad_norm": 2.254975083731318,
+      "learning_rate": 1.0622504449409043e-09,
+      "loss": 0.9957,
+      "step": 9512
+    },
+    {
+      "epoch": 0.9956043956043956,
+      "grad_norm": 2.4241701416533905,
+      "learning_rate": 1.013418772285446e-09,
+      "loss": 0.8659,
+      "step": 9513
+    },
+    {
+      "epoch": 0.99570905285191,
+      "grad_norm": 2.3939669658872362,
+      "learning_rate": 9.65736003457618e-10,
+      "loss": 0.7979,
+      "step": 9514
+    },
+    {
+      "epoch": 0.9958137100994244,
+      "grad_norm": 2.2441183666873323,
+      "learning_rate": 9.19202143937481e-10,
+      "loss": 0.7062,
+      "step": 9515
+    },
+    {
+      "epoch": 0.9959183673469387,
+      "grad_norm": 1.9769218010919176,
+      "learning_rate": 8.738171990707589e-10,
+      "loss": 0.8455,
+      "step": 9516
+    },
+    {
+      "epoch": 0.9960230245944531,
+      "grad_norm": 1.9141948632707264,
+      "learning_rate": 8.295811740732796e-10,
+      "loss": 0.7915,
+      "step": 9517
+    },
+    {
+      "epoch": 0.9961276818419675,
+      "grad_norm": 2.3415485768220994,
+      "learning_rate": 7.864940740276439e-10,
+      "loss": 0.9133,
+      "step": 9518
+    },
+    {
+      "epoch": 0.9962323390894819,
+      "grad_norm": 2.1401706902733837,
+      "learning_rate": 7.445559038832262e-10,
+      "loss": 0.8676,
+      "step": 9519
+    },
+    {
+      "epoch": 0.9963369963369964,
+      "grad_norm": 1.8282468400440854,
+      "learning_rate": 7.037666684606148e-10,
+      "loss": 0.9874,
+      "step": 9520
+    },
+    {
+      "epoch": 0.9964416535845108,
+      "grad_norm": 2.2683960849665543,
+      "learning_rate": 6.641263724460612e-10,
+      "loss": 0.7478,
+      "step": 9521
+    },
+    {
+      "epoch": 0.9965463108320252,
+      "grad_norm": 2.1668339291877157,
+      "learning_rate": 6.256350203948103e-10,
+      "loss": 0.8291,
+      "step": 9522
+    },
+    {
+      "epoch": 0.9966509680795395,
+      "grad_norm": 2.4253965330798546,
+      "learning_rate": 5.882926167277703e-10,
+      "loss": 0.8643,
+      "step": 9523
+    },
+    {
+      "epoch": 0.9967556253270539,
+      "grad_norm": 2.1583972769715265,
+      "learning_rate": 5.520991657370633e-10,
+      "loss": 0.8868,
+      "step": 9524
+    },
+    {
+      "epoch": 0.9968602825745683,
+      "grad_norm": 2.3240808973920046,
+      "learning_rate": 5.17054671581585e-10,
+      "loss": 0.8017,
+      "step": 9525
+    },
+    {
+      "epoch": 0.9969649398220827,
+      "grad_norm": 2.194269062168786,
+      "learning_rate": 4.831591382870038e-10,
+      "loss": 0.9831,
+      "step": 9526
+    },
+    {
+      "epoch": 0.9970695970695971,
+      "grad_norm": 2.0354807231276872,
+      "learning_rate": 4.504125697490924e-10,
+      "loss": 0.9727,
+      "step": 9527
+    },
+    {
+      "epoch": 0.9971742543171115,
+      "grad_norm": 2.063899720456543,
+      "learning_rate": 4.188149697303967e-10,
+      "loss": 0.9148,
+      "step": 9528
+    },
+    {
+      "epoch": 0.9972789115646259,
+      "grad_norm": 2.2768683287427196,
+      "learning_rate": 3.883663418602357e-10,
+      "loss": 0.8469,
+      "step": 9529
+    },
+    {
+      "epoch": 0.9973835688121402,
+      "grad_norm": 2.0468757802028033,
+      "learning_rate": 3.5906668963803235e-10,
+      "loss": 0.8325,
+      "step": 9530
+    },
+    {
+      "epoch": 0.9974882260596546,
+      "grad_norm": 2.051294044317142,
+      "learning_rate": 3.3091601643109314e-10,
+      "loss": 0.8025,
+      "step": 9531
+    },
+    {
+      "epoch": 0.997592883307169,
+      "grad_norm": 2.105439857515838,
+      "learning_rate": 3.039143254723875e-10,
+      "loss": 0.9892,
+      "step": 9532
+    },
+    {
+      "epoch": 0.9976975405546834,
+      "grad_norm": 2.220984101222805,
+      "learning_rate": 2.7806161986609905e-10,
+      "loss": 0.7738,
+      "step": 9533
+    },
+    {
+      "epoch": 0.9978021978021978,
+      "grad_norm": 2.115650631099441,
+      "learning_rate": 2.5335790258207426e-10,
+      "loss": 0.8181,
+      "step": 9534
+    },
+    {
+      "epoch": 0.9979068550497122,
+      "grad_norm": 2.1913243479989855,
+      "learning_rate": 2.298031764591535e-10,
+      "loss": 0.8192,
+      "step": 9535
+    },
+    {
+      "epoch": 0.9980115122972266,
+      "grad_norm": 1.9652396497864315,
+      "learning_rate": 2.0739744420295026e-10,
+      "loss": 0.8074,
+      "step": 9536
+    },
+    {
+      "epoch": 0.998116169544741,
+      "grad_norm": 2.4031821017278814,
+      "learning_rate": 1.8614070838918198e-10,
+      "loss": 0.9575,
+      "step": 9537
+    },
+    {
+      "epoch": 0.9982208267922553,
+      "grad_norm": 2.037609844489397,
+      "learning_rate": 1.6603297145922902e-10,
+      "loss": 0.8618,
+      "step": 9538
+    },
+    {
+      "epoch": 0.9983254840397697,
+      "grad_norm": 1.6984142714399961,
+      "learning_rate": 1.4707423572346556e-10,
+      "loss": 0.754,
+      "step": 9539
+    },
+    {
+      "epoch": 0.9984301412872841,
+      "grad_norm": 1.627184498513617,
+      "learning_rate": 1.292645033612594e-10,
+      "loss": 0.7935,
+      "step": 9540
+    },
+    {
+      "epoch": 0.9985347985347985,
+      "grad_norm": 1.6578736068172026,
+      "learning_rate": 1.1260377641764131e-10,
+      "loss": 0.7769,
+      "step": 9541
+    },
+    {
+      "epoch": 0.998639455782313,
+      "grad_norm": 2.0902509066149784,
+      "learning_rate": 9.709205680885624e-11,
+      "loss": 0.7798,
+      "step": 9542
+    },
+    {
+      "epoch": 0.9987441130298274,
+      "grad_norm": 2.1775821639631023,
+      "learning_rate": 8.272934631459173e-11,
+      "loss": 0.8777,
+      "step": 9543
+    },
+    {
+      "epoch": 0.9988487702773418,
+      "grad_norm": 2.2443525221661678,
+      "learning_rate": 6.951564658796983e-11,
+      "loss": 0.85,
+      "step": 9544
+    },
+    {
+      "epoch": 0.998953427524856,
+      "grad_norm": 1.8463156292838934,
+      "learning_rate": 5.745095914555521e-11,
+      "loss": 0.8024,
+      "step": 9545
+    },
+    {
+      "epoch": 0.9990580847723705,
+      "grad_norm": 2.2214610429811676,
+      "learning_rate": 4.6535285374016416e-11,
+      "loss": 0.843,
+      "step": 9546
+    },
+    {
+      "epoch": 0.9991627420198849,
+      "grad_norm": 2.1287397785084075,
+      "learning_rate": 3.676862652790547e-11,
+      "loss": 0.8161,
+      "step": 9547
+    },
+    {
+      "epoch": 0.9992673992673993,
+      "grad_norm": 2.1670560388230014,
+      "learning_rate": 2.8150983728547632e-11,
+      "loss": 0.9613,
+      "step": 9548
+    },
+    {
+      "epoch": 0.9993720565149137,
+      "grad_norm": 1.8271825583030723,
+      "learning_rate": 2.0682357966261834e-11,
+      "loss": 0.8962,
+      "step": 9549
+    },
+    {
+      "epoch": 0.9994767137624281,
+      "grad_norm": 1.847917867704652,
+      "learning_rate": 1.4362750100360701e-11,
+      "loss": 0.8912,
+      "step": 9550
+    },
+    {
+      "epoch": 0.9995813710099425,
+      "grad_norm": 1.540095392303899,
+      "learning_rate": 9.192160856930088e-12,
+      "loss": 0.6913,
+      "step": 9551
+    },
+    {
+      "epoch": 0.9996860282574568,
+      "grad_norm": 1.8041355621927853,
+      "learning_rate": 5.170590828829092e-12,
+      "loss": 0.7122,
+      "step": 9552
+    },
+    {
+      "epoch": 0.9997906855049712,
+      "grad_norm": 2.0773412293224243,
+      "learning_rate": 2.2980404779104904e-12,
+      "loss": 0.8123,
+      "step": 9553
+    },
+    {
+      "epoch": 0.9998953427524856,
+      "grad_norm": 2.2586840266436203,
+      "learning_rate": 5.745101361309679e-13,
+      "loss": 0.8988,
+      "step": 9554
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.738276172232474,
+      "learning_rate": 0.0,
+      "loss": 0.7992,
+      "step": 9555
+    },
+    {
+      "epoch": 1.0,
+      "step": 9555,
+      "total_flos": 1276904626028544.0,
+      "train_loss": 0.9531744299042231,
+      "train_runtime": 26831.9909,
+      "train_samples_per_second": 22.79,
+      "train_steps_per_second": 0.356
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 9555,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": false,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1276904626028544.0,
+  "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null
 }