diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4816 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.001098705829648,
+  "global_step": 400000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014993743133088564,
+      "loss": 9.1175,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001498748626617713,
+      "loss": 8.3544,
+      "step": 1000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014981229399265691,
+      "loss": 8.0502,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014974972532354257,
+      "loss": 7.8515,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014968715665442822,
+      "loss": 7.681,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014962458798531387,
+      "loss": 7.5517,
+      "step": 3000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014956201931619952,
+      "loss": 7.4159,
+      "step": 3500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014949945064708517,
+      "loss": 7.2876,
+      "step": 4000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014943688197797082,
+      "loss": 7.1397,
+      "step": 4500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00014937431330885645,
+      "loss": 6.9971,
+      "step": 5000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001493117446397421,
+      "loss": 6.8653,
+      "step": 5500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014924917597062775,
+      "loss": 6.7289,
+      "step": 6000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001491866073015134,
+      "loss": 6.6215,
+      "step": 6500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014912403863239903,
+      "loss": 6.5087,
+      "step": 7000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014906146996328468,
+      "loss": 6.4134,
+      "step": 7500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014899890129417033,
+      "loss": 6.3202,
+      "step": 8000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014893633262505598,
+      "loss": 6.2245,
+      "step": 8500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014887376395594163,
+      "loss": 6.1525,
+      "step": 9000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014881119528682729,
+      "loss": 6.0713,
+      "step": 9500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014874862661771294,
+      "loss": 5.9881,
+      "step": 10000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014868605794859856,
+      "loss": 5.9234,
+      "step": 10500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014862348927948421,
+      "loss": 5.8453,
+      "step": 11000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014856092061036987,
+      "loss": 5.7843,
+      "step": 11500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014849835194125552,
+      "loss": 5.7166,
+      "step": 12000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014843578327214117,
+      "loss": 5.6656,
+      "step": 12500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001483732146030268,
+      "loss": 5.6084,
+      "step": 13000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00014831064593391247,
+      "loss": 5.5446,
+      "step": 13500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001482480772647981,
+      "loss": 5.5103,
+      "step": 14000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014818550859568375,
+      "loss": 5.4623,
+      "step": 14500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001481229399265694,
+      "loss": 5.4276,
+      "step": 15000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014806037125745505,
+      "loss": 5.379,
+      "step": 15500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001479978025883407,
+      "loss": 5.3413,
+      "step": 16000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014793523391922633,
+      "loss": 5.3178,
+      "step": 16500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014787266525011198,
+      "loss": 5.2804,
+      "step": 17000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00014781009658099763,
+      "loss": 5.2634,
+      "step": 17500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014774752791188328,
+      "loss": 5.2208,
+      "step": 18000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001476849592427689,
+      "loss": 5.1902,
+      "step": 18500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014762239057365459,
+      "loss": 5.1797,
+      "step": 19000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014755982190454024,
+      "loss": 5.1425,
+      "step": 19500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014749725323542586,
+      "loss": 5.1115,
+      "step": 20000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001474346845663115,
+      "loss": 5.0887,
+      "step": 20500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014737211589719716,
+      "loss": 5.0949,
+      "step": 21000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00014730954722808282,
+      "loss": 5.0657,
+      "step": 21500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014724697855896844,
+      "loss": 5.0448,
+      "step": 22000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001471844098898541,
+      "loss": 5.0233,
+      "step": 22500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014712184122073974,
+      "loss": 4.9993,
+      "step": 23000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001470592725516254,
+      "loss": 4.9959,
+      "step": 23500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014699670388251105,
+      "loss": 4.9717,
+      "step": 24000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001469341352133967,
+      "loss": 4.9487,
+      "step": 24500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014687156654428235,
+      "loss": 4.944,
+      "step": 25000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00014680899787516798,
+      "loss": 4.926,
+      "step": 25500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014674642920605363,
+      "loss": 4.9112,
+      "step": 26000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014668386053693928,
+      "loss": 4.9049,
+      "step": 26500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014662129186782493,
+      "loss": 4.8885,
+      "step": 27000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014655872319871058,
+      "loss": 4.871,
+      "step": 27500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001464961545295962,
+      "loss": 4.8603,
+      "step": 28000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014643358586048186,
+      "loss": 4.8336,
+      "step": 28500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001463710171913675,
+      "loss": 4.8393,
+      "step": 29000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00014630844852225316,
+      "loss": 4.8184,
+      "step": 29500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001462458798531388,
+      "loss": 4.8204,
+      "step": 30000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00014618331118402446,
+      "loss": 4.8008,
+      "step": 30500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00014612074251491012,
+      "loss": 4.7858,
+      "step": 31000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00014605817384579574,
+      "loss": 4.7695,
+      "step": 31500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001459956051766814,
+      "loss": 4.7705,
+      "step": 32000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00014593303650756704,
+      "loss": 4.7617,
+      "step": 32500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001458704678384527,
+      "loss": 4.7498,
+      "step": 33000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00014580789916933832,
+      "loss": 4.7427,
+      "step": 33500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00014574533050022397,
+      "loss": 4.7235,
+      "step": 34000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00014568276183110962,
+      "loss": 4.7262,
+      "step": 34500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00014562019316199527,
+      "loss": 4.7172,
+      "step": 35000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00014555762449288093,
+      "loss": 4.7051,
+      "step": 35500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00014549505582376658,
+      "loss": 4.6796,
+      "step": 36000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00014543248715465223,
+      "loss": 4.665,
+      "step": 36500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00014536991848553788,
+      "loss": 4.6904,
+      "step": 37000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001453073498164235,
+      "loss": 4.6701,
+      "step": 37500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00014524478114730916,
+      "loss": 4.6592,
+      "step": 38000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001451822124781948,
+      "loss": 4.6409,
+      "step": 38500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00014511964380908046,
+      "loss": 4.6627,
+      "step": 39000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00014505707513996608,
+      "loss": 4.6439,
+      "step": 39500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00014499450647085174,
+      "loss": 4.6214,
+      "step": 40000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00014493193780173742,
+      "loss": 4.6419,
+      "step": 40500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00014486936913262304,
+      "loss": 4.6312,
+      "step": 41000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001448068004635087,
+      "loss": 4.6204,
+      "step": 41500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014474423179439434,
+      "loss": 4.6197,
+      "step": 42000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014468166312528,
+      "loss": 4.6011,
+      "step": 42500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014461909445616562,
+      "loss": 4.5853,
+      "step": 43000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014455652578705127,
+      "loss": 4.5793,
+      "step": 43500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014449395711793692,
+      "loss": 4.576,
+      "step": 44000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014443138844882257,
+      "loss": 4.5698,
+      "step": 44500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001443688197797082,
+      "loss": 4.5658,
+      "step": 45000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00014430625111059385,
+      "loss": 4.5533,
+      "step": 45500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014424368244147953,
+      "loss": 4.551,
+      "step": 46000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014418111377236515,
+      "loss": 4.5576,
+      "step": 46500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001441185451032508,
+      "loss": 4.5412,
+      "step": 47000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014405597643413646,
+      "loss": 4.5437,
+      "step": 47500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001439934077650221,
+      "loss": 4.5266,
+      "step": 48000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014393083909590776,
+      "loss": 4.5125,
+      "step": 48500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014386827042679338,
+      "loss": 4.5353,
+      "step": 49000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00014380570175767904,
+      "loss": 4.5159,
+      "step": 49500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001437431330885647,
+      "loss": 4.5024,
+      "step": 50000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014368056441945034,
+      "loss": 4.4975,
+      "step": 50500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014361799575033596,
+      "loss": 4.5025,
+      "step": 51000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014355542708122164,
+      "loss": 4.5117,
+      "step": 51500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001434928584121073,
+      "loss": 4.5049,
+      "step": 52000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014343028974299292,
+      "loss": 4.5069,
+      "step": 52500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014336772107387857,
+      "loss": 4.4901,
+      "step": 53000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00014330515240476422,
+      "loss": 4.4853,
+      "step": 53500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014324258373564987,
+      "loss": 4.4722,
+      "step": 54000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001431800150665355,
+      "loss": 4.4653,
+      "step": 54500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014311744639742115,
+      "loss": 4.4651,
+      "step": 55000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001430548777283068,
+      "loss": 4.4379,
+      "step": 55500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014299230905919245,
+      "loss": 4.4491,
+      "step": 56000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014292974039007808,
+      "loss": 4.4594,
+      "step": 56500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00014286717172096376,
+      "loss": 4.4491,
+      "step": 57000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001428046030518494,
+      "loss": 4.4344,
+      "step": 57500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014274203438273503,
+      "loss": 4.4358,
+      "step": 58000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014267946571362068,
+      "loss": 4.4493,
+      "step": 58500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014261689704450633,
+      "loss": 4.4361,
+      "step": 59000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000142554328375392,
+      "loss": 4.4308,
+      "step": 59500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014249175970627764,
+      "loss": 4.4219,
+      "step": 60000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014242919103716326,
+      "loss": 4.4086,
+      "step": 60500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014236662236804891,
+      "loss": 4.4285,
+      "step": 61000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00014230405369893457,
+      "loss": 4.4069,
+      "step": 61500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014224148502982022,
+      "loss": 4.4121,
+      "step": 62000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014217891636070587,
+      "loss": 4.421,
+      "step": 62500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014211634769159152,
+      "loss": 4.3855,
+      "step": 63000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014205377902247717,
+      "loss": 4.397,
+      "step": 63500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001419912103533628,
+      "loss": 4.3656,
+      "step": 64000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014192864168424845,
+      "loss": 4.3959,
+      "step": 64500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001418660730151341,
+      "loss": 4.3878,
+      "step": 65000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00014180350434601975,
+      "loss": 4.3796,
+      "step": 65500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014174093567690538,
+      "loss": 4.3698,
+      "step": 66000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014167836700779103,
+      "loss": 4.3745,
+      "step": 66500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014161579833867668,
+      "loss": 4.3698,
+      "step": 67000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014155322966956233,
+      "loss": 4.3795,
+      "step": 67500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014149066100044798,
+      "loss": 4.3687,
+      "step": 68000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014142809233133363,
+      "loss": 4.3577,
+      "step": 68500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00014136552366221929,
+      "loss": 4.3724,
+      "step": 69000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001413029549931049,
+      "loss": 4.3588,
+      "step": 69500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00014124038632399056,
+      "loss": 4.3452,
+      "step": 70000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001411778176548762,
+      "loss": 4.3588,
+      "step": 70500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00014111524898576187,
+      "loss": 4.3468,
+      "step": 71000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00014105268031664752,
+      "loss": 4.3428,
+      "step": 71500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00014099011164753314,
+      "loss": 4.3454,
+      "step": 72000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001409275429784188,
+      "loss": 4.3414,
+      "step": 72500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00014086497430930444,
+      "loss": 4.3335,
+      "step": 73000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001408024056401901,
+      "loss": 4.3403,
+      "step": 73500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014073983697107575,
+      "loss": 4.3481,
+      "step": 74000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001406772683019614,
+      "loss": 4.3331,
+      "step": 74500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014061469963284705,
+      "loss": 4.3437,
+      "step": 75000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014055213096373268,
+      "loss": 4.3193,
+      "step": 75500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014048956229461833,
+      "loss": 4.2995,
+      "step": 76000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014042699362550398,
+      "loss": 4.3271,
+      "step": 76500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014036442495638963,
+      "loss": 4.3122,
+      "step": 77000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00014030185628727525,
+      "loss": 4.3048,
+      "step": 77500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001402392876181609,
+      "loss": 4.3183,
+      "step": 78000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00014017671894904659,
+      "loss": 4.3096,
+      "step": 78500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001401141502799322,
+      "loss": 4.2995,
+      "step": 79000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00014005158161081786,
+      "loss": 4.3082,
+      "step": 79500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001399890129417035,
+      "loss": 4.297,
+      "step": 80000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00013992644427258916,
+      "loss": 4.2907,
+      "step": 80500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001398638756034748,
+      "loss": 4.2937,
+      "step": 81000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00013980130693436044,
+      "loss": 4.289,
+      "step": 81500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001397387382652461,
+      "loss": 4.2851,
+      "step": 82000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00013967616959613174,
+      "loss": 4.298,
+      "step": 82500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001396136009270174,
+      "loss": 4.286,
+      "step": 83000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00013955103225790302,
+      "loss": 4.2706,
+      "step": 83500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001394884635887887,
+      "loss": 4.2687,
+      "step": 84000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00013942589491967432,
+      "loss": 4.2638,
+      "step": 84500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00013936332625055997,
+      "loss": 4.2636,
+      "step": 85000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00013930075758144563,
+      "loss": 4.2626,
+      "step": 85500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00013923818891233128,
+      "loss": 4.2701,
+      "step": 86000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00013917562024321693,
+      "loss": 4.2779,
+      "step": 86500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00013911305157410255,
+      "loss": 4.2702,
+      "step": 87000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001390504829049882,
+      "loss": 4.2689,
+      "step": 87500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00013898791423587386,
+      "loss": 4.2625,
+      "step": 88000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001389253455667595,
+      "loss": 4.2523,
+      "step": 88500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00013886277689764516,
+      "loss": 4.2561,
+      "step": 89000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001388002082285308,
+      "loss": 4.2425,
+      "step": 89500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00013873763955941646,
+      "loss": 4.253,
+      "step": 90000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001386750708903021,
+      "loss": 4.2466,
+      "step": 90500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00013861250222118774,
+      "loss": 4.2531,
+      "step": 91000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001385499335520734,
+      "loss": 4.2428,
+      "step": 91500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00013848736488295904,
+      "loss": 4.2498,
+      "step": 92000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00013842479621384467,
+      "loss": 4.2349,
+      "step": 92500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00013836222754473032,
+      "loss": 4.2522,
+      "step": 93000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00013829965887561597,
+      "loss": 4.2362,
+      "step": 93500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00013823709020650162,
+      "loss": 4.2368,
+      "step": 94000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00013817452153738727,
+      "loss": 4.2235,
+      "step": 94500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00013811195286827293,
+      "loss": 4.2134,
+      "step": 95000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00013804938419915858,
+      "loss": 4.2103,
+      "step": 95500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001379868155300442,
+      "loss": 4.2198,
+      "step": 96000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00013792424686092985,
+      "loss": 4.2141,
+      "step": 96500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001378616781918155,
+      "loss": 4.2192,
+      "step": 97000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00013779910952270116,
+      "loss": 4.2225,
+      "step": 97500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001377365408535868,
+      "loss": 4.2167,
+      "step": 98000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00013767397218447243,
+      "loss": 4.1967,
+      "step": 98500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00013761140351535808,
+      "loss": 4.2036,
+      "step": 99000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00013754883484624374,
+      "loss": 4.23,
+      "step": 99500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001374862661771294,
+      "loss": 4.2076,
+      "step": 100000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00013742369750801504,
+      "loss": 4.198,
+      "step": 100500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001373611288389007,
+      "loss": 4.2227,
+      "step": 101000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00013729856016978634,
+      "loss": 4.1795,
+      "step": 101500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00013723599150067197,
+      "loss": 4.2043,
+      "step": 102000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00013717342283155762,
+      "loss": 4.1902,
+      "step": 102500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00013711085416244327,
+      "loss": 4.1818,
+      "step": 103000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00013704828549332892,
+      "loss": 4.1775,
+      "step": 103500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00013698571682421455,
+      "loss": 4.2037,
+      "step": 104000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001369231481551002,
+      "loss": 4.1918,
+      "step": 104500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00013686057948598585,
+      "loss": 4.2047,
+      "step": 105000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001367980108168715,
+      "loss": 4.182,
+      "step": 105500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00013673544214775715,
+      "loss": 4.1929,
+      "step": 106000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001366728734786428,
+      "loss": 4.2035,
+      "step": 106500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00013661030480952846,
+      "loss": 4.1702,
+      "step": 107000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00013654773614041408,
+      "loss": 4.1796,
+      "step": 107500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00013648516747129973,
+      "loss": 4.1841,
+      "step": 108000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00013642259880218538,
+      "loss": 4.1764,
+      "step": 108500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00013636003013307104,
+      "loss": 4.1943,
+      "step": 109000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001362974614639567,
+      "loss": 4.162,
+      "step": 109500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001362348927948423,
+      "loss": 4.1719,
+      "step": 110000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00013617232412572796,
+      "loss": 4.173,
+      "step": 110500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00013610975545661364,
+      "loss": 4.1535,
+      "step": 111000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00013604718678749927,
+      "loss": 4.1674,
+      "step": 111500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00013598461811838492,
+      "loss": 4.181,
+      "step": 112000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00013592204944927057,
+      "loss": 4.1721,
+      "step": 112500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00013585948078015622,
+      "loss": 4.2129,
+      "step": 113000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00013579691211104185,
+      "loss": 4.1508,
+      "step": 113500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001357343434419275,
+      "loss": 4.1705,
+      "step": 114000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00013567177477281315,
+      "loss": 4.1426,
+      "step": 114500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001356092061036988,
+      "loss": 4.1467,
+      "step": 115000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00013554663743458442,
+      "loss": 4.1589,
+      "step": 115500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001354840687654701,
+      "loss": 4.1458,
+      "step": 116000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00013542150009635576,
+      "loss": 4.1691,
+      "step": 116500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00013535893142724138,
+      "loss": 4.1407,
+      "step": 117000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00013529636275812703,
+      "loss": 4.1473,
+      "step": 117500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00013523379408901268,
+      "loss": 4.1471,
+      "step": 118000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00013517122541989833,
+      "loss": 4.149,
+      "step": 118500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00013510865675078396,
+      "loss": 4.1389,
+      "step": 119000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001350460880816696,
+      "loss": 4.1607,
+      "step": 119500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00013498351941255526,
+      "loss": 4.1431,
+      "step": 120000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00013492095074344091,
+      "loss": 4.136,
+      "step": 120500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00013485838207432657,
+      "loss": 4.142,
+      "step": 121000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00013479581340521222,
+      "loss": 4.1344,
+      "step": 121500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00013473324473609787,
+      "loss": 4.1253,
+      "step": 122000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00013467067606698352,
+      "loss": 4.1441,
+      "step": 122500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00013460810739786914,
+      "loss": 4.1341,
+      "step": 123000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001345455387287548,
+      "loss": 4.1466,
+      "step": 123500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00013448297005964045,
+      "loss": 4.1522,
+      "step": 124000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001344204013905261,
+      "loss": 4.1288,
+      "step": 124500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00013435783272141172,
+      "loss": 4.1256,
+      "step": 125000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00013429526405229738,
+      "loss": 4.1421,
+      "step": 125500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00013423269538318303,
+      "loss": 4.1247,
+      "step": 126000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00013417012671406868,
+      "loss": 4.129,
+      "step": 126500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00013410755804495433,
+      "loss": 4.1164,
+      "step": 127000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00013404498937583998,
+      "loss": 4.1279,
+      "step": 127500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00013398242070672563,
+      "loss": 4.1218,
+      "step": 128000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00013391985203761126,
+      "loss": 4.1262,
+      "step": 128500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001338572833684969,
+      "loss": 4.1298,
+      "step": 129000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00013379471469938256,
+      "loss": 4.1108,
+      "step": 129500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001337321460302682,
+      "loss": 4.1112,
+      "step": 130000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00013366957736115384,
+      "loss": 4.1169,
+      "step": 130500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001336070086920395,
+      "loss": 4.1235,
+      "step": 131000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00013354444002292514,
+      "loss": 4.1291,
+      "step": 131500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001334818713538108,
+      "loss": 4.1244,
+      "step": 132000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00013341930268469644,
+      "loss": 4.1072,
+      "step": 132500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001333567340155821,
+      "loss": 4.109,
+      "step": 133000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00013329416534646775,
+      "loss": 4.1071,
+      "step": 133500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001332315966773534,
+      "loss": 4.1076,
+      "step": 134000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013316902800823902,
+      "loss": 4.1154,
+      "step": 134500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013310645933912468,
+      "loss": 4.1136,
+      "step": 135000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013304389067001033,
+      "loss": 4.1053,
+      "step": 135500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013298132200089598,
+      "loss": 4.1116,
+      "step": 136000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001329187533317816,
+      "loss": 4.0951,
+      "step": 136500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013285618466266725,
+      "loss": 4.089,
+      "step": 137000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00013279361599355293,
+      "loss": 4.1044,
+      "step": 137500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013273104732443856,
+      "loss": 4.1043,
+      "step": 138000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001326684786553242,
+      "loss": 4.0835,
+      "step": 138500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013260590998620986,
+      "loss": 4.0865,
+      "step": 139000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001325433413170955,
+      "loss": 4.0892,
+      "step": 139500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013248077264798114,
+      "loss": 4.1022,
+      "step": 140000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001324182039788668,
+      "loss": 4.088,
+      "step": 140500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00013235563530975244,
+      "loss": 4.093,
+      "step": 141000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001322930666406381,
+      "loss": 4.099,
+      "step": 141500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013223049797152374,
+      "loss": 4.079,
+      "step": 142000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013216792930240937,
+      "loss": 4.0908,
+      "step": 142500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013210536063329505,
+      "loss": 4.1,
+      "step": 143000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013204279196418067,
+      "loss": 4.0889,
+      "step": 143500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013198022329506632,
+      "loss": 4.0923,
+      "step": 144000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013191765462595197,
+      "loss": 4.0708,
+      "step": 144500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013185508595683763,
+      "loss": 4.0838,
+      "step": 145000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00013179251728772328,
+      "loss": 4.0742,
+      "step": 145500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001317299486186089,
+      "loss": 4.0786,
+      "step": 146000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013166737994949455,
+      "loss": 4.0673,
+      "step": 146500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001316048112803802,
+      "loss": 4.0746,
+      "step": 147000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013154224261126586,
+      "loss": 4.0839,
+      "step": 147500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013147967394215148,
+      "loss": 4.0709,
+      "step": 148000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013141710527303716,
+      "loss": 4.0491,
+      "step": 148500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001313545366039228,
+      "loss": 4.0832,
+      "step": 149000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00013129196793480844,
+      "loss": 4.064,
+      "step": 149500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001312293992656941,
+      "loss": 4.075,
+      "step": 150000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013116683059657974,
+      "loss": 4.0708,
+      "step": 150500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001311042619274654,
+      "loss": 4.0776,
+      "step": 151000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013104169325835102,
+      "loss": 4.0503,
+      "step": 151500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013097912458923667,
+      "loss": 4.059,
+      "step": 152000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013091655592012232,
+      "loss": 4.0565,
+      "step": 152500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013085398725100797,
+      "loss": 4.0663,
+      "step": 153000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00013079141858189362,
+      "loss": 4.0696,
+      "step": 153500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013072884991277927,
+      "loss": 4.0619,
+      "step": 154000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013066628124366493,
+      "loss": 4.0653,
+      "step": 154500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013060371257455055,
+      "loss": 4.0456,
+      "step": 155000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001305411439054362,
+      "loss": 4.076,
+      "step": 155500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013047857523632185,
+      "loss": 4.0597,
+      "step": 156000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001304160065672075,
+      "loss": 4.0575,
+      "step": 156500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013035343789809316,
+      "loss": 4.0713,
+      "step": 157000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013029086922897878,
+      "loss": 4.0654,
+      "step": 157500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013022830055986443,
+      "loss": 4.0516,
+      "step": 158000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013016573189075008,
+      "loss": 4.0511,
+      "step": 158500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013010316322163574,
+      "loss": 4.0686,
+      "step": 159000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001300405945525214,
+      "loss": 4.0244,
+      "step": 159500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012997802588340704,
+      "loss": 4.0521,
+      "step": 160000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001299154572142927,
+      "loss": 4.0511,
+      "step": 160500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012985288854517831,
+      "loss": 4.0527,
+      "step": 161000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00012979031987606397,
+      "loss": 4.0404,
+      "step": 161500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012972775120694962,
+      "loss": 4.0419,
+      "step": 162000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012966518253783527,
+      "loss": 4.0668,
+      "step": 162500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001296026138687209,
+      "loss": 4.0317,
+      "step": 163000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012954004519960655,
+      "loss": 4.0394,
+      "step": 163500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001294774765304922,
+      "loss": 4.0458,
+      "step": 164000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012941490786137785,
+      "loss": 4.0349,
+      "step": 164500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001293523391922635,
+      "loss": 4.2634,
+      "step": 165000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00012928977052314915,
+      "loss": 4.0683,
+      "step": 165500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001292272018540348,
+      "loss": 4.0383,
+      "step": 166000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012916463318492043,
+      "loss": 4.0375,
+      "step": 166500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012910206451580608,
+      "loss": 4.0459,
+      "step": 167000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012903949584669173,
+      "loss": 4.0305,
+      "step": 167500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012897692717757738,
+      "loss": 4.0304,
+      "step": 168000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012891435850846303,
+      "loss": 4.0349,
+      "step": 168500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012885178983934866,
+      "loss": 4.0283,
+      "step": 169000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001287892211702343,
+      "loss": 4.0258,
+      "step": 169500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012872665250111996,
+      "loss": 4.0319,
+      "step": 170000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012866408383200561,
+      "loss": 4.0239,
+      "step": 170500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012860151516289127,
+      "loss": 4.0318,
+      "step": 171000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012853894649377692,
+      "loss": 4.0078,
+      "step": 171500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012847637782466257,
+      "loss": 4.0309,
+      "step": 172000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001284138091555482,
+      "loss": 4.0206,
+      "step": 172500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012835124048643385,
+      "loss": 4.0118,
+      "step": 173000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001282886718173195,
+      "loss": 4.0408,
+      "step": 173500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012822610314820515,
+      "loss": 4.0364,
+      "step": 174000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012816353447909077,
+      "loss": 4.0211,
+      "step": 174500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012810096580997642,
+      "loss": 4.0168,
+      "step": 175000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001280383971408621,
+      "loss": 4.0254,
+      "step": 175500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012797582847174773,
+      "loss": 4.0127,
+      "step": 176000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012791325980263338,
+      "loss": 3.9996,
+      "step": 176500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012785069113351903,
+      "loss": 4.0224,
+      "step": 177000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012778812246440468,
+      "loss": 4.0247,
+      "step": 177500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001277255537952903,
+      "loss": 4.0129,
+      "step": 178000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012766298512617596,
+      "loss": 4.0236,
+      "step": 178500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001276004164570616,
+      "loss": 4.0203,
+      "step": 179000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012753784778794726,
+      "loss": 4.0008,
+      "step": 179500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001274752791188329,
+      "loss": 4.0208,
+      "step": 180000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012741271044971854,
+      "loss": 4.0087,
+      "step": 180500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012735014178060422,
+      "loss": 4.0199,
+      "step": 181000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012728757311148984,
+      "loss": 4.0143,
+      "step": 181500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001272250044423755,
+      "loss": 4.0345,
+      "step": 182000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012716243577326114,
+      "loss": 4.0067,
+      "step": 182500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001270998671041468,
+      "loss": 4.0124,
+      "step": 183000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012703729843503245,
+      "loss": 4.0067,
+      "step": 183500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012697472976591807,
+      "loss": 3.9995,
+      "step": 184000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012691216109680372,
+      "loss": 4.007,
+      "step": 184500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012684959242768938,
+      "loss": 4.0085,
+      "step": 185000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00012678702375857503,
+      "loss": 4.0163,
+      "step": 185500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012672445508946068,
+      "loss": 3.9997,
+      "step": 186000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012666188642034633,
+      "loss": 3.9908,
+      "step": 186500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012659931775123198,
+      "loss": 3.9906,
+      "step": 187000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001265367490821176,
+      "loss": 4.0107,
+      "step": 187500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012647418041300326,
+      "loss": 4.0004,
+      "step": 188000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001264116117438889,
+      "loss": 4.0065,
+      "step": 188500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012634904307477456,
+      "loss": 4.002,
+      "step": 189000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00012628647440566019,
+      "loss": 4.0029,
+      "step": 189500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012622390573654584,
+      "loss": 3.9801,
+      "step": 190000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001261613370674315,
+      "loss": 3.9934,
+      "step": 190500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012609876839831714,
+      "loss": 4.0027,
+      "step": 191000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001260361997292028,
+      "loss": 4.0057,
+      "step": 191500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012597363106008844,
+      "loss": 4.0006,
+      "step": 192000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001259110623909741,
+      "loss": 3.9971,
+      "step": 192500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012584849372185972,
+      "loss": 3.989,
+      "step": 193000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00012578592505274537,
+      "loss": 3.9951,
+      "step": 193500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012572335638363102,
+      "loss": 3.9776,
+      "step": 194000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012566078771451667,
+      "loss": 3.9995,
+      "step": 194500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012559821904540233,
+      "loss": 3.9752,
+      "step": 195000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012553565037628795,
+      "loss": 3.9873,
+      "step": 195500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001254730817071736,
+      "loss": 3.985,
+      "step": 196000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012541051303805925,
+      "loss": 3.9924,
+      "step": 196500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001253479443689449,
+      "loss": 3.9778,
+      "step": 197000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00012528537569983056,
+      "loss": 3.9873,
+      "step": 197500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001252228070307162,
+      "loss": 3.9739,
+      "step": 198000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00012516023836160186,
+      "loss": 3.9946,
+      "step": 198500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00012509766969248748,
+      "loss": 4.008,
+      "step": 199000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00012503510102337314,
+      "loss": 3.9697,
+      "step": 199500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001249725323542588,
+      "loss": 3.9788,
+      "step": 200000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00012490996368514444,
+      "loss": 3.9803,
+      "step": 200500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00012484739501603006,
+      "loss": 3.9948,
+      "step": 201000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00012478482634691572,
+      "loss": 3.978,
+      "step": 201500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00012472225767780137,
+      "loss": 3.968,
+      "step": 202000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00012465968900868702,
+      "loss": 3.9703,
+      "step": 202500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00012459712033957267,
+      "loss": 3.979,
+      "step": 203000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00012453455167045832,
+      "loss": 3.9788,
+      "step": 203500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00012447198300134397,
+      "loss": 3.9779,
+      "step": 204000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00012440941433222963,
+      "loss": 3.9707,
+      "step": 204500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00012434684566311525,
+      "loss": 3.958,
+      "step": 205000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001242842769940009,
+      "loss": 3.9778,
+      "step": 205500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00012422170832488655,
+      "loss": 3.9846,
+      "step": 206000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001241591396557722,
+      "loss": 3.9743,
+      "step": 206500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00012409657098665783,
+      "loss": 3.9612,
+      "step": 207000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00012403400231754348,
+      "loss": 3.9537,
+      "step": 207500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00012397143364842916,
+      "loss": 3.9754,
+      "step": 208000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00012390886497931478,
+      "loss": 3.9616,
+      "step": 208500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00012384629631020044,
+      "loss": 3.9575,
+      "step": 209000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001237837276410861,
+      "loss": 3.9535,
+      "step": 209500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00012372115897197174,
+      "loss": 3.9767,
+      "step": 210000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00012365859030285736,
+      "loss": 3.9582,
+      "step": 210500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00012359602163374302,
+      "loss": 3.9737,
+      "step": 211000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00012353345296462867,
+      "loss": 3.9652,
+      "step": 211500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00012347088429551432,
+      "loss": 3.9688,
+      "step": 212000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00012340831562639994,
+      "loss": 3.9344,
+      "step": 212500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00012334574695728562,
+      "loss": 3.9645,
+      "step": 213000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00012328317828817127,
+      "loss": 3.9539,
+      "step": 213500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001232206096190569,
+      "loss": 3.9613,
+      "step": 214000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00012315804094994255,
+      "loss": 3.9479,
+      "step": 214500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001230954722808282,
+      "loss": 3.9571,
+      "step": 215000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00012303290361171385,
+      "loss": 3.9622,
+      "step": 215500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001229703349425995,
+      "loss": 3.952,
+      "step": 216000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00012290776627348513,
+      "loss": 3.9619,
+      "step": 216500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00012284519760437078,
+      "loss": 3.9535,
+      "step": 217000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00012278262893525643,
+      "loss": 3.9575,
+      "step": 217500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00012272006026614208,
+      "loss": 3.9497,
+      "step": 218000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00012265749159702774,
+      "loss": 3.9577,
+      "step": 218500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001225949229279134,
+      "loss": 3.9517,
+      "step": 219000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00012253235425879904,
+      "loss": 3.9447,
+      "step": 219500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00012246978558968466,
+      "loss": 3.9566,
+      "step": 220000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00012240721692057031,
+      "loss": 3.9493,
+      "step": 220500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00012234464825145597,
+      "loss": 3.9701,
+      "step": 221000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00012228207958234162,
+      "loss": 3.9576,
+      "step": 221500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00012221951091322724,
+      "loss": 3.9513,
+      "step": 222000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001221569422441129,
+      "loss": 3.9434,
+      "step": 222500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00012209437357499855,
+      "loss": 3.9586,
+      "step": 223000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00012203180490588418,
+      "loss": 3.9503,
+      "step": 223500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00012196923623676985,
+      "loss": 3.9385,
+      "step": 224000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001219066675676555,
+      "loss": 3.9662,
+      "step": 224500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00012184409889854114,
+      "loss": 3.9432,
+      "step": 225000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00012178153022942679,
+      "loss": 3.9362,
+      "step": 225500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00012171896156031243,
+      "loss": 3.9571,
+      "step": 226000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00012165639289119808,
+      "loss": 3.9502,
+      "step": 226500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00012159382422208372,
+      "loss": 3.9388,
+      "step": 227000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00012153125555296937,
+      "loss": 3.9471,
+      "step": 227500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00012146868688385502,
+      "loss": 3.9667,
+      "step": 228000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00012140611821474066,
+      "loss": 3.9273,
+      "step": 228500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00012134354954562631,
+      "loss": 3.9378,
+      "step": 229000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00012128098087651196,
+      "loss": 3.9617,
+      "step": 229500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00012121841220739761,
+      "loss": 3.9304,
+      "step": 230000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00012115584353828327,
+      "loss": 3.9296,
+      "step": 230500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001210932748691689,
+      "loss": 3.9526,
+      "step": 231000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00012103070620005456,
+      "loss": 3.9438,
+      "step": 231500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00012096813753094019,
+      "loss": 3.9403,
+      "step": 232000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00012090556886182584,
+      "loss": 3.9372,
+      "step": 232500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00012084300019271148,
+      "loss": 3.9406,
+      "step": 233000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00012078043152359713,
+      "loss": 3.9351,
+      "step": 233500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00012071786285448277,
+      "loss": 3.942,
+      "step": 234000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00012065529418536844,
+      "loss": 3.9334,
+      "step": 234500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00012059272551625409,
+      "loss": 3.9346,
+      "step": 235000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00012053015684713973,
+      "loss": 3.9501,
+      "step": 235500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00012046758817802538,
+      "loss": 3.9416,
+      "step": 236000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00012040501950891102,
+      "loss": 3.9413,
+      "step": 236500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00012034245083979667,
+      "loss": 3.9312,
+      "step": 237000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001202798821706823,
+      "loss": 3.9418,
+      "step": 237500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00012021731350156796,
+      "loss": 3.9364,
+      "step": 238000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00012015474483245361,
+      "loss": 3.9398,
+      "step": 238500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00012009217616333925,
+      "loss": 3.9238,
+      "step": 239000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001200296074942249,
+      "loss": 4.5988,
+      "step": 239500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00011996703882511055,
+      "loss": 3.9397,
+      "step": 240000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001199044701559962,
+      "loss": 3.9328,
+      "step": 240500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00011984190148688184,
+      "loss": 3.931,
+      "step": 241000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00011977933281776749,
+      "loss": 3.9197,
+      "step": 241500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011971676414865314,
+      "loss": 3.931,
+      "step": 242000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011965419547953878,
+      "loss": 3.9401,
+      "step": 242500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011959162681042443,
+      "loss": 3.9206,
+      "step": 243000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011952905814131007,
+      "loss": 3.918,
+      "step": 243500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011946648947219572,
+      "loss": 3.9236,
+      "step": 244000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011940392080308136,
+      "loss": 3.9311,
+      "step": 244500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011934135213396701,
+      "loss": 3.9148,
+      "step": 245000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011927878346485268,
+      "loss": 3.9347,
+      "step": 245500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011921621479573832,
+      "loss": 3.9306,
+      "step": 246000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011915364612662397,
+      "loss": 3.9145,
+      "step": 246500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001190910774575096,
+      "loss": 3.9325,
+      "step": 247000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011902850878839526,
+      "loss": 3.9221,
+      "step": 247500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001189659401192809,
+      "loss": 3.9248,
+      "step": 248000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011890337145016655,
+      "loss": 3.9249,
+      "step": 248500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011884080278105219,
+      "loss": 3.9239,
+      "step": 249000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011877823411193784,
+      "loss": 3.9186,
+      "step": 249500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011871566544282349,
+      "loss": 3.9135,
+      "step": 250000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011865309677370913,
+      "loss": 3.9151,
+      "step": 250500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011859052810459479,
+      "loss": 3.9165,
+      "step": 251000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011852795943548043,
+      "loss": 3.9201,
+      "step": 251500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011846539076636608,
+      "loss": 3.9049,
+      "step": 252000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011840282209725172,
+      "loss": 3.9179,
+      "step": 252500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011834025342813737,
+      "loss": 3.9113,
+      "step": 253000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011827768475902302,
+      "loss": 3.9094,
+      "step": 253500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011821511608990866,
+      "loss": 3.9265,
+      "step": 254000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011815254742079431,
+      "loss": 3.8923,
+      "step": 254500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011808997875167995,
+      "loss": 3.9079,
+      "step": 255000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001180274100825656,
+      "loss": 3.9078,
+      "step": 255500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011796484141345124,
+      "loss": 3.9074,
+      "step": 256000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001179022727443369,
+      "loss": 3.9031,
+      "step": 256500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00011783970407522256,
+      "loss": 3.9024,
+      "step": 257000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001177771354061082,
+      "loss": 3.9096,
+      "step": 257500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011771456673699385,
+      "loss": 3.9072,
+      "step": 258000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011765199806787948,
+      "loss": 3.9173,
+      "step": 258500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011758942939876514,
+      "loss": 3.9193,
+      "step": 259000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011752686072965077,
+      "loss": 3.9022,
+      "step": 259500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011746429206053643,
+      "loss": 3.9117,
+      "step": 260000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011740172339142206,
+      "loss": 3.9061,
+      "step": 260500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011733915472230772,
+      "loss": 3.914,
+      "step": 261000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00011727658605319338,
+      "loss": 3.8996,
+      "step": 261500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011721401738407902,
+      "loss": 3.8946,
+      "step": 262000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011715144871496467,
+      "loss": 3.9097,
+      "step": 262500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011708888004585031,
+      "loss": 3.9106,
+      "step": 263000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011702631137673596,
+      "loss": 3.9022,
+      "step": 263500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011696374270762161,
+      "loss": 3.9096,
+      "step": 264000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011690117403850725,
+      "loss": 3.9113,
+      "step": 264500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001168386053693929,
+      "loss": 3.9066,
+      "step": 265000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00011677603670027854,
+      "loss": 3.8952,
+      "step": 265500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011671346803116419,
+      "loss": 3.9009,
+      "step": 266000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011665089936204983,
+      "loss": 3.8977,
+      "step": 266500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001165883306929355,
+      "loss": 3.91,
+      "step": 267000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011652576202382115,
+      "loss": 3.9027,
+      "step": 267500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011646319335470678,
+      "loss": 3.921,
+      "step": 268000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011640062468559244,
+      "loss": 3.8889,
+      "step": 268500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011633805601647807,
+      "loss": 3.8891,
+      "step": 269000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00011627548734736373,
+      "loss": 3.91,
+      "step": 269500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011621291867824936,
+      "loss": 3.8922,
+      "step": 270000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011615035000913501,
+      "loss": 3.8894,
+      "step": 270500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011608778134002065,
+      "loss": 3.8887,
+      "step": 271000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001160252126709063,
+      "loss": 3.9053,
+      "step": 271500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011596264400179194,
+      "loss": 3.9023,
+      "step": 272000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011590007533267761,
+      "loss": 3.89,
+      "step": 272500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00011583750666356326,
+      "loss": 3.9114,
+      "step": 273000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001157749379944489,
+      "loss": 3.8969,
+      "step": 273500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011571236932533455,
+      "loss": 3.8974,
+      "step": 274000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011564980065622019,
+      "loss": 3.8966,
+      "step": 274500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011558723198710584,
+      "loss": 3.8898,
+      "step": 275000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011552466331799149,
+      "loss": 3.8769,
+      "step": 275500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011546209464887713,
+      "loss": 3.8784,
+      "step": 276000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011539952597976278,
+      "loss": 3.9041,
+      "step": 276500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011533695731064842,
+      "loss": 3.8765,
+      "step": 277000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00011527438864153407,
+      "loss": 3.8758,
+      "step": 277500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00011521181997241972,
+      "loss": 3.8971,
+      "step": 278000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00011514925130330537,
+      "loss": 3.8909,
+      "step": 278500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00011508668263419102,
+      "loss": 3.8881,
+      "step": 279000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00011502411396507666,
+      "loss": 3.8766,
+      "step": 279500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00011496154529596231,
+      "loss": 3.8711,
+      "step": 280000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00011489897662684795,
+      "loss": 3.8868,
+      "step": 280500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001148364079577336,
+      "loss": 3.8791,
+      "step": 281000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00011477383928861924,
+      "loss": 3.8802,
+      "step": 281500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001147112706195049,
+      "loss": 3.8802,
+      "step": 282000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011464870195039053,
+      "loss": 3.885,
+      "step": 282500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011458613328127618,
+      "loss": 3.8846,
+      "step": 283000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011452356461216185,
+      "loss": 3.8984,
+      "step": 283500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011446099594304749,
+      "loss": 3.8829,
+      "step": 284000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011439842727393314,
+      "loss": 3.8881,
+      "step": 284500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011433585860481878,
+      "loss": 3.8964,
+      "step": 285000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00011427328993570443,
+      "loss": 3.8881,
+      "step": 285500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011421072126659007,
+      "loss": 3.8769,
+      "step": 286000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011414815259747572,
+      "loss": 3.878,
+      "step": 286500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011408558392836137,
+      "loss": 3.91,
+      "step": 287000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011402301525924701,
+      "loss": 3.8875,
+      "step": 287500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011396044659013266,
+      "loss": 3.8843,
+      "step": 288000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011389787792101831,
+      "loss": 3.8807,
+      "step": 288500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00011383530925190396,
+      "loss": 3.8784,
+      "step": 289000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001137727405827896,
+      "loss": 3.8835,
+      "step": 289500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011371017191367525,
+      "loss": 3.8723,
+      "step": 290000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001136476032445609,
+      "loss": 3.8745,
+      "step": 290500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011358503457544654,
+      "loss": 3.896,
+      "step": 291000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011352246590633219,
+      "loss": 3.8835,
+      "step": 291500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011345989723721783,
+      "loss": 3.8634,
+      "step": 292000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011339732856810348,
+      "loss": 3.8514,
+      "step": 292500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011333475989898912,
+      "loss": 3.867,
+      "step": 293000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00011327219122987477,
+      "loss": 3.8895,
+      "step": 293500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011320962256076044,
+      "loss": 3.8819,
+      "step": 294000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011314705389164608,
+      "loss": 3.8708,
+      "step": 294500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011308448522253173,
+      "loss": 3.8904,
+      "step": 295000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011302191655341737,
+      "loss": 3.8782,
+      "step": 295500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011295934788430302,
+      "loss": 3.8866,
+      "step": 296000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011289677921518865,
+      "loss": 3.8742,
+      "step": 296500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001128342105460743,
+      "loss": 3.8854,
+      "step": 297000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00011277164187695994,
+      "loss": 3.867,
+      "step": 297500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001127090732078456,
+      "loss": 3.877,
+      "step": 298000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00011264650453873125,
+      "loss": 3.8697,
+      "step": 298500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00011258393586961689,
+      "loss": 3.8574,
+      "step": 299000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00011252136720050255,
+      "loss": 3.8845,
+      "step": 299500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00011245879853138819,
+      "loss": 3.8626,
+      "step": 300000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00011239622986227384,
+      "loss": 3.8609,
+      "step": 300500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00011233366119315949,
+      "loss": 3.8632,
+      "step": 301000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00011227109252404513,
+      "loss": 3.8588,
+      "step": 301500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00011220852385493078,
+      "loss": 3.8738,
+      "step": 302000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00011214595518581642,
+      "loss": 3.8531,
+      "step": 302500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00011208338651670207,
+      "loss": 3.8755,
+      "step": 303000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00011202081784758771,
+      "loss": 3.8675,
+      "step": 303500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00011195824917847336,
+      "loss": 3.8623,
+      "step": 304000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000111895680509359,
+      "loss": 3.8766,
+      "step": 304500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00011183311184024466,
+      "loss": 3.8763,
+      "step": 305000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00011177054317113032,
+      "loss": 3.8474,
+      "step": 305500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00011170797450201595,
+      "loss": 3.8749,
+      "step": 306000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001116454058329016,
+      "loss": 3.8519,
+      "step": 306500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00011158283716378724,
+      "loss": 3.8551,
+      "step": 307000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001115202684946729,
+      "loss": 3.8674,
+      "step": 307500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00011145769982555853,
+      "loss": 3.8618,
+      "step": 308000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00011139513115644418,
+      "loss": 3.8568,
+      "step": 308500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00011133256248732982,
+      "loss": 3.8617,
+      "step": 309000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00011126999381821547,
+      "loss": 3.8634,
+      "step": 309500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011120742514910114,
+      "loss": 3.8671,
+      "step": 310000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011114485647998678,
+      "loss": 3.8641,
+      "step": 310500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011108228781087243,
+      "loss": 3.859,
+      "step": 311000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011101971914175807,
+      "loss": 3.8586,
+      "step": 311500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011095715047264372,
+      "loss": 3.8569,
+      "step": 312000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011089458180352937,
+      "loss": 3.8563,
+      "step": 312500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011083201313441501,
+      "loss": 3.8736,
+      "step": 313000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011076944446530066,
+      "loss": 3.8589,
+      "step": 313500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001107068757961863,
+      "loss": 3.8563,
+      "step": 314000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011064430712707195,
+      "loss": 3.8621,
+      "step": 314500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011058173845795759,
+      "loss": 3.8506,
+      "step": 315000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011051916978884325,
+      "loss": 3.8598,
+      "step": 315500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001104566011197289,
+      "loss": 3.8423,
+      "step": 316000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011039403245061454,
+      "loss": 3.8426,
+      "step": 316500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001103314637815002,
+      "loss": 3.8441,
+      "step": 317000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011026889511238583,
+      "loss": 3.8521,
+      "step": 317500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00011020632644327148,
+      "loss": 3.8747,
+      "step": 318000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00011014375777415712,
+      "loss": 3.8593,
+      "step": 318500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00011008118910504277,
+      "loss": 3.8664,
+      "step": 319000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00011001862043592841,
+      "loss": 3.8536,
+      "step": 319500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010995605176681406,
+      "loss": 3.8628,
+      "step": 320000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001098934830976997,
+      "loss": 3.8362,
+      "step": 320500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010983091442858537,
+      "loss": 3.8512,
+      "step": 321000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010976834575947102,
+      "loss": 3.8431,
+      "step": 321500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010970577709035666,
+      "loss": 3.8705,
+      "step": 322000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010964320842124231,
+      "loss": 3.8634,
+      "step": 322500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010958063975212795,
+      "loss": 3.8493,
+      "step": 323000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001095180710830136,
+      "loss": 3.8565,
+      "step": 323500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010945550241389925,
+      "loss": 3.8613,
+      "step": 324000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010939293374478489,
+      "loss": 3.8443,
+      "step": 324500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010933036507567054,
+      "loss": 3.8538,
+      "step": 325000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010926779640655618,
+      "loss": 3.8485,
+      "step": 325500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010920522773744183,
+      "loss": 3.8492,
+      "step": 326000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010914265906832748,
+      "loss": 3.8408,
+      "step": 326500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010908009039921313,
+      "loss": 3.856,
+      "step": 327000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010901752173009878,
+      "loss": 3.8426,
+      "step": 327500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010895495306098442,
+      "loss": 3.853,
+      "step": 328000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010889238439187007,
+      "loss": 3.8405,
+      "step": 328500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010882981572275571,
+      "loss": 3.8473,
+      "step": 329000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010876724705364136,
+      "loss": 3.858,
+      "step": 329500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.000108704678384527,
+      "loss": 3.8421,
+      "step": 330000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00010864210971541265,
+      "loss": 3.8442,
+      "step": 330500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00010857954104629829,
+      "loss": 3.8451,
+      "step": 331000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00010851697237718394,
+      "loss": 3.8518,
+      "step": 331500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00010845440370806961,
+      "loss": 3.8548,
+      "step": 332000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00010839183503895525,
+      "loss": 3.8413,
+      "step": 332500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001083292663698409,
+      "loss": 3.8245,
+      "step": 333000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00010826669770072654,
+      "loss": 3.8256,
+      "step": 333500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010820412903161219,
+      "loss": 3.8333,
+      "step": 334000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010814156036249782,
+      "loss": 3.8235,
+      "step": 334500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010807899169338348,
+      "loss": 3.8425,
+      "step": 335000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010801642302426913,
+      "loss": 3.8384,
+      "step": 335500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010795385435515477,
+      "loss": 3.839,
+      "step": 336000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010789128568604042,
+      "loss": 3.8271,
+      "step": 336500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010782871701692607,
+      "loss": 3.8388,
+      "step": 337000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010776614834781172,
+      "loss": 3.8421,
+      "step": 337500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010770357967869737,
+      "loss": 3.8411,
+      "step": 338000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010764101100958301,
+      "loss": 3.8433,
+      "step": 338500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010757844234046866,
+      "loss": 3.8453,
+      "step": 339000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001075158736713543,
+      "loss": 3.8367,
+      "step": 339500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010745330500223995,
+      "loss": 3.8419,
+      "step": 340000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010739073633312559,
+      "loss": 3.8371,
+      "step": 340500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010732816766401124,
+      "loss": 3.8296,
+      "step": 341000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010726559899489688,
+      "loss": 3.8465,
+      "step": 341500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010720303032578253,
+      "loss": 3.8562,
+      "step": 342000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001071404616566682,
+      "loss": 3.8395,
+      "step": 342500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010707789298755383,
+      "loss": 3.8356,
+      "step": 343000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010701532431843949,
+      "loss": 3.8115,
+      "step": 343500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010695275564932512,
+      "loss": 3.8326,
+      "step": 344000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010689018698021078,
+      "loss": 3.8343,
+      "step": 344500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010682761831109641,
+      "loss": 3.8243,
+      "step": 345000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010676504964198207,
+      "loss": 3.8421,
+      "step": 345500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001067024809728677,
+      "loss": 3.834,
+      "step": 346000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010663991230375336,
+      "loss": 3.8354,
+      "step": 346500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.000106577343634639,
+      "loss": 3.8305,
+      "step": 347000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010651477496552464,
+      "loss": 3.8433,
+      "step": 347500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010645220629641031,
+      "loss": 3.8429,
+      "step": 348000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010638963762729595,
+      "loss": 3.8368,
+      "step": 348500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001063270689581816,
+      "loss": 3.8596,
+      "step": 349000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010626450028906725,
+      "loss": 3.8286,
+      "step": 349500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010620193161995289,
+      "loss": 3.8355,
+      "step": 350000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010613936295083854,
+      "loss": 3.8319,
+      "step": 350500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010607679428172418,
+      "loss": 3.8131,
+      "step": 351000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010601422561260983,
+      "loss": 3.8343,
+      "step": 351500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010595165694349547,
+      "loss": 3.8346,
+      "step": 352000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010588908827438112,
+      "loss": 3.827,
+      "step": 352500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010582651960526676,
+      "loss": 3.819,
+      "step": 353000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00010576395093615242,
+      "loss": 3.8327,
+      "step": 353500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010570138226703808,
+      "loss": 3.8175,
+      "step": 354000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010563881359792371,
+      "loss": 3.827,
+      "step": 354500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010557624492880936,
+      "loss": 3.8061,
+      "step": 355000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000105513676259695,
+      "loss": 3.8093,
+      "step": 355500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010545110759058065,
+      "loss": 3.8266,
+      "step": 356000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010538853892146629,
+      "loss": 3.8325,
+      "step": 356500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010532597025235194,
+      "loss": 3.84,
+      "step": 357000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00010526340158323758,
+      "loss": 3.8243,
+      "step": 357500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010520083291412323,
+      "loss": 3.8262,
+      "step": 358000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001051382642450089,
+      "loss": 3.8159,
+      "step": 358500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010507569557589454,
+      "loss": 3.8309,
+      "step": 359000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010501312690678019,
+      "loss": 3.8159,
+      "step": 359500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010495055823766583,
+      "loss": 3.8193,
+      "step": 360000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010488798956855148,
+      "loss": 3.8276,
+      "step": 360500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010482542089943713,
+      "loss": 3.8212,
+      "step": 361000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00010476285223032277,
+      "loss": 3.8248,
+      "step": 361500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010470028356120842,
+      "loss": 3.8317,
+      "step": 362000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010463771489209406,
+      "loss": 3.816,
+      "step": 362500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010457514622297971,
+      "loss": 3.8148,
+      "step": 363000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010451257755386535,
+      "loss": 3.8428,
+      "step": 363500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010445000888475101,
+      "loss": 3.8101,
+      "step": 364000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010438744021563666,
+      "loss": 3.8253,
+      "step": 364500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001043248715465223,
+      "loss": 3.8362,
+      "step": 365000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00010426230287740795,
+      "loss": 3.8227,
+      "step": 365500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010419973420829359,
+      "loss": 3.8267,
+      "step": 366000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010413716553917924,
+      "loss": 3.8239,
+      "step": 366500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010407459687006488,
+      "loss": 7.9568,
+      "step": 367000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010401202820095053,
+      "loss": 3.8325,
+      "step": 367500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010394945953183617,
+      "loss": 3.8136,
+      "step": 368000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010388689086272182,
+      "loss": 3.8197,
+      "step": 368500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010382432219360747,
+      "loss": 3.8178,
+      "step": 369000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00010376175352449313,
+      "loss": 3.8233,
+      "step": 369500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00010369918485537878,
+      "loss": 3.8212,
+      "step": 370000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00010363661618626442,
+      "loss": 3.8264,
+      "step": 370500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00010357404751715007,
+      "loss": 3.8236,
+      "step": 371000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001035114788480357,
+      "loss": 3.8209,
+      "step": 371500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00010344891017892136,
+      "loss": 3.8263,
+      "step": 372000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00010338634150980701,
+      "loss": 3.8064,
+      "step": 372500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00010332377284069265,
+      "loss": 3.7992,
+      "step": 373000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.0001032612041715783,
+      "loss": 3.8319,
+      "step": 373500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010319863550246394,
+      "loss": 3.815,
+      "step": 374000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010313606683334959,
+      "loss": 3.8187,
+      "step": 374500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010307349816423525,
+      "loss": 3.8098,
+      "step": 375000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010301092949512089,
+      "loss": 3.822,
+      "step": 375500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010294836082600654,
+      "loss": 3.8102,
+      "step": 376000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010288579215689218,
+      "loss": 3.8287,
+      "step": 376500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010282322348777783,
+      "loss": 3.8026,
+      "step": 377000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00010276065481866347,
+      "loss": 3.8152,
+      "step": 377500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010269808614954912,
+      "loss": 3.8109,
+      "step": 378000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010263551748043476,
+      "loss": 3.8145,
+      "step": 378500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010257294881132041,
+      "loss": 3.8125,
+      "step": 379000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010251038014220605,
+      "loss": 3.8163,
+      "step": 379500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001024478114730917,
+      "loss": 3.8342,
+      "step": 380000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010238524280397737,
+      "loss": 3.8269,
+      "step": 380500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.000102322674134863,
+      "loss": 3.8059,
+      "step": 381000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00010226010546574866,
+      "loss": 3.8022,
+      "step": 381500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001021975367966343,
+      "loss": 3.8129,
+      "step": 382000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010213496812751995,
+      "loss": 3.8221,
+      "step": 382500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010207239945840558,
+      "loss": 3.8143,
+      "step": 383000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010200983078929124,
+      "loss": 3.8103,
+      "step": 383500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010194726212017689,
+      "loss": 3.8045,
+      "step": 384000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010188469345106253,
+      "loss": 3.7872,
+      "step": 384500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010182212478194818,
+      "loss": 3.8087,
+      "step": 385000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00010175955611283383,
+      "loss": 3.7927,
+      "step": 385500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010169698744371948,
+      "loss": 3.8145,
+      "step": 386000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010163441877460513,
+      "loss": 3.8,
+      "step": 386500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010157185010549077,
+      "loss": 3.8151,
+      "step": 387000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010150928143637642,
+      "loss": 3.8143,
+      "step": 387500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010144671276726206,
+      "loss": 3.8224,
+      "step": 388000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010138414409814771,
+      "loss": 3.81,
+      "step": 388500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00010132157542903335,
+      "loss": 3.7963,
+      "step": 389000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.000101259006759919,
+      "loss": 3.8033,
+      "step": 389500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010119643809080464,
+      "loss": 3.8054,
+      "step": 390000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010113386942169029,
+      "loss": 3.8188,
+      "step": 390500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010107130075257596,
+      "loss": 3.8104,
+      "step": 391000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001010087320834616,
+      "loss": 3.8159,
+      "step": 391500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010094616341434725,
+      "loss": 3.8035,
+      "step": 392000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010088359474523288,
+      "loss": 3.8063,
+      "step": 392500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010082102607611853,
+      "loss": 3.7984,
+      "step": 393000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00010075845740700417,
+      "loss": 3.8028,
+      "step": 393500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010069588873788982,
+      "loss": 3.8001,
+      "step": 394000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010063332006877546,
+      "loss": 3.8099,
+      "step": 394500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010057075139966111,
+      "loss": 3.7928,
+      "step": 395000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010050818273054677,
+      "loss": 3.7977,
+      "step": 395500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001004456140614324,
+      "loss": 3.7979,
+      "step": 396000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010038304539231807,
+      "loss": 3.8007,
+      "step": 396500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010032047672320371,
+      "loss": 3.8017,
+      "step": 397000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00010025790805408936,
+      "loss": 3.7762,
+      "step": 397500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010019533938497501,
+      "loss": 3.8018,
+      "step": 398000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010013277071586065,
+      "loss": 3.8004,
+      "step": 398500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001000702020467463,
+      "loss": 3.7908,
+      "step": 399000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00010000763337763194,
+      "loss": 3.7934,
+      "step": 399500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.994506470851759e-05,
+      "loss": 3.781,
+      "step": 400000
+    }
+  ],
+  "max_steps": 1198683,
+  "num_train_epochs": 3,
+  "total_flos": 1.3368536226947138e+19,
+  "trial_name": null,
+  "trial_params": null
+}