diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,26844 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 134060,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 8.0891,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.04e-06,
+      "loss": 7.3717,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.24e-06,
+      "loss": 5.2176,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.44e-06,
+      "loss": 4.35,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 3.4418,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.84e-06,
+      "loss": 2.8796,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.04e-06,
+      "loss": 2.5285,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.24e-06,
+      "loss": 2.2902,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 2.0726,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0640000000000001e-05,
+      "loss": 1.9922,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.184e-05,
+      "loss": 1.8281,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3040000000000002e-05,
+      "loss": 1.8522,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 1.694,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.544e-05,
+      "loss": 1.5824,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.664e-05,
+      "loss": 1.52,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7840000000000002e-05,
+      "loss": 1.5005,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.904e-05,
+      "loss": 1.474,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9881069312188603e-05,
+      "loss": 1.4102,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9316685232156396e-05,
+      "loss": 1.3253,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.879778950992281e-05,
+      "loss": 1.3384,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8318582636182796e-05,
+      "loss": 1.1948,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7874250135079064e-05,
+      "loss": 1.1314,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7460757394239457e-05,
+      "loss": 1.062,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.707469441906277e-05,
+      "loss": 1.0079,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6713156761621893e-05,
+      "loss": 1.0524,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6373653066597823e-05,
+      "loss": 1.0395,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.605403247669839e-05,
+      "loss": 0.9386,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5752427045264396e-05,
+      "loss": 0.9778,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.546720562224365e-05,
+      "loss": 0.9788,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5196936606339123e-05,
+      "loss": 0.9897,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.494035761667992e-05,
+      "loss": 0.9103,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4696350614431104e-05,
+      "loss": 0.916,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4463921353509297e-05,
+      "loss": 0.8597,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4242182297397127e-05,
+      "loss": 0.7784,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4030338331657845e-05,
+      "loss": 0.8172,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3827674747047456e-05,
+      "loss": 0.8181,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3633547078730296e-05,
+      "loss": 0.7576,
+      "step": 1080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3447372472027636e-05,
+      "loss": 0.7949,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3268622310856882e-05,
+      "loss": 0.7204,
+      "step": 1140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3096815896275181e-05,
+      "loss": 0.7064,
+      "step": 1170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2931515002796793e-05,
+      "loss": 0.7781,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2772319171982632e-05,
+      "loss": 0.6695,
+      "step": 1230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.261886162812672e-05,
+      "loss": 0.7092,
+      "step": 1260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2470805721138076e-05,
+      "loss": 0.6964,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2327841818038448e-05,
+      "loss": 0.6547,
+      "step": 1320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2189684577707986e-05,
+      "loss": 0.6966,
+      "step": 1350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2056070554260304e-05,
+      "loss": 0.707,
+      "step": 1380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1926756083217403e-05,
+      "loss": 0.7237,
+      "step": 1410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1801515411874576e-05,
+      "loss": 0.6281,
+      "step": 1440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1680139041202725e-05,
+      "loss": 0.6783,
+      "step": 1470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.156243225157201e-05,
+      "loss": 0.628,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1448213788688051e-05,
+      "loss": 0.6361,
+      "step": 1530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1337314689562955e-05,
+      "loss": 0.5683,
+      "step": 1560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1229577231219766e-05,
+      "loss": 0.5641,
+      "step": 1590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.112485398724962e-05,
+      "loss": 0.6237,
+      "step": 1620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.1023006979384424e-05,
+      "loss": 0.5927,
+      "step": 1650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0923906912979292e-05,
+      "loss": 0.6144,
+      "step": 1680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0827432486770289e-05,
+      "loss": 0.5461,
+      "step": 1710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0733469768527298e-05,
+      "loss": 0.5728,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0641911629294233e-05,
+      "loss": 0.5896,
+      "step": 1770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0552657229828183e-05,
+      "loss": 0.5095,
+      "step": 1800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0465611553639654e-05,
+      "loss": 0.58,
+      "step": 1830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0380684981717496e-05,
+      "loss": 0.5911,
+      "step": 1860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0297792904611e-05,
+      "loss": 0.5328,
+      "step": 1890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0216855368051905e-05,
+      "loss": 0.5636,
+      "step": 1920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0137796748742198e-05,
+      "loss": 0.5733,
+      "step": 1950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0060545457319173e-05,
+      "loss": 0.5326,
+      "step": 1980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98503366584589e-06,
+      "loss": 0.5016,
+      "step": 2010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.911197057469108e-06,
+      "loss": 0.5416,
+      "step": 2040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.838974596144754e-06,
+      "loss": 0.6255,
+      "step": 2070
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.768308314557045e-06,
+      "loss": 0.4626,
+      "step": 2100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.699143118556543e-06,
+      "loss": 0.496,
+      "step": 2130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.631426606617744e-06,
+      "loss": 0.5069,
+      "step": 2160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.565108902971731e-06,
+      "loss": 0.4882,
+      "step": 2190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.50014250320633e-06,
+      "loss": 0.5291,
+      "step": 2220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.436482131247697e-06,
+      "loss": 0.541,
+      "step": 2250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.374084606744877e-06,
+      "loss": 0.5595,
+      "step": 2280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.312908721974755e-06,
+      "loss": 0.5041,
+      "step": 2310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.252915127470068e-06,
+      "loss": 0.568,
+      "step": 2340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.194066225649182e-06,
+      "loss": 0.4467,
+      "step": 2370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.13632607179441e-06,
+      "loss": 0.471,
+      "step": 2400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.079660281786226e-06,
+      "loss": 0.5507,
+      "step": 2430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.02403594605542e-06,
+      "loss": 0.4895,
+      "step": 2460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.969421549263868e-06,
+      "loss": 0.5492,
+      "step": 2490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.915786895268652e-06,
+      "loss": 0.4661,
+      "step": 2520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.863103036963604e-06,
+      "loss": 0.4957,
+      "step": 2550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.811342210628018e-06,
+      "loss": 0.4698,
+      "step": 2580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.760477774444217e-06,
+      "loss": 0.4813,
+      "step": 2610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.710484150874759e-06,
+      "loss": 0.4382,
+      "step": 2640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.661336772616119e-06,
+      "loss": 0.5073,
+      "step": 2670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.613012031869432e-06,
+      "loss": 0.4712,
+      "step": 2700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.5654872326904e-06,
+      "loss": 0.5269,
+      "step": 2730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.51874054619982e-06,
+      "loss": 0.4369,
+      "step": 2760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 8.472750968454027e-06,
+      "loss": 0.4954,
+      "step": 2790
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.427498280790527e-06,
+      "loss": 0.4503,
+      "step": 2820
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.382963012478792e-06,
+      "loss": 0.4511,
+      "step": 2850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.339126405519484e-06,
+      "loss": 0.4604,
+      "step": 2880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.295970381447577e-06,
+      "loss": 0.4256,
+      "step": 2910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.253477510005973e-06,
+      "loss": 0.4544,
+      "step": 2940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.211630979566229e-06,
+      "loss": 0.4404,
+      "step": 2970
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.170414569182507e-06,
+      "loss": 0.4183,
+      "step": 3000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.129812622173143e-06,
+      "loss": 0.4439,
+      "step": 3030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.08981002113217e-06,
+      "loss": 0.4319,
+      "step": 3060
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.050392164280187e-06,
+      "loss": 0.4402,
+      "step": 3090
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.011544943070564e-06,
+      "loss": 0.4871,
+      "step": 3120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.973254720972994e-06,
+      "loss": 0.4748,
+      "step": 3150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.935508313361898e-06,
+      "loss": 0.5215,
+      "step": 3180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.898292968442374e-06,
+      "loss": 0.4496,
+      "step": 3210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.861596349150974e-06,
+      "loss": 0.4912,
+      "step": 3240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.825406515973003e-06,
+      "loss": 0.4403,
+      "step": 3270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.78971191062195e-06,
+      "loss": 0.4216,
+      "step": 3300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.754501340530456e-06,
+      "loss": 0.4121,
+      "step": 3330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.719763964105497e-06,
+      "loss": 0.4315,
+      "step": 3360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.68548927670368e-06,
+      "loss": 0.4272,
+      "step": 3390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.651667097285501e-06,
+      "loss": 0.4234,
+      "step": 3420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.618287555709996e-06,
+      "loss": 0.5091,
+      "step": 3450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.585341080633831e-06,
+      "loss": 0.4435,
+      "step": 3480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.552818387981102e-06,
+      "loss": 0.3596,
+      "step": 3510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.520710469952336e-06,
+      "loss": 0.4096,
+      "step": 3540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.489008584543126e-06,
+      "loss": 0.3699,
+      "step": 3570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.457704245544709e-06,
+      "loss": 0.4257,
+      "step": 3600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.4267892130005204e-06,
+      "loss": 0.3542,
+      "step": 3630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.396255484094341e-06,
+      "loss": 0.4619,
+      "step": 3660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.366095284447187e-06,
+      "loss": 0.442,
+      "step": 3690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.336301059801395e-06,
+      "loss": 0.3701,
+      "step": 3720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.306865468071773e-06,
+      "loss": 0.4379,
+      "step": 3750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.277781371744775e-06,
+      "loss": 0.4511,
+      "step": 3780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.24904183060784e-06,
+      "loss": 0.3869,
+      "step": 3810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.220640094792103e-06,
+      "loss": 0.3576,
+      "step": 3840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.1925695981126075e-06,
+      "loss": 0.344,
+      "step": 3870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.164823951691131e-06,
+      "loss": 0.3708,
+      "step": 3900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.137396937847532e-06,
+      "loss": 0.414,
+      "step": 3930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.110282504246376e-06,
+      "loss": 0.4553,
+      "step": 3960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.083474758286323e-06,
+      "loss": 0.4011,
+      "step": 3990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.056967961720459e-06,
+      "loss": 0.3563,
+      "step": 4020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.0307565254964336e-06,
+      "loss": 0.3607,
+      "step": 4050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 7.004835004805859e-06,
+      "loss": 0.3634,
+      "step": 4080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.979198094333013e-06,
+      "loss": 0.3656,
+      "step": 4110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.953840623693443e-06,
+      "loss": 0.4157,
+      "step": 4140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.928757553053551e-06,
+      "loss": 0.3799,
+      "step": 4170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.9039439689227496e-06,
+      "loss": 0.3595,
+      "step": 4200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.879395080110185e-06,
+      "loss": 0.3912,
+      "step": 4230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.855106213838523e-06,
+      "loss": 0.3785,
+      "step": 4260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.83107281200758e-06,
+      "loss": 0.4702,
+      "step": 4290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.807290427601057e-06,
+      "loss": 0.3987,
+      "step": 4320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.7837547212299395e-06,
+      "loss": 0.3743,
+      "step": 4350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.760461457806433e-06,
+      "loss": 0.3816,
+      "step": 4380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.737406503342708e-06,
+      "loss": 0.3293,
+      "step": 4410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.714585821868879e-06,
+      "loss": 0.4261,
+      "step": 4440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.691995472465075e-06,
+      "loss": 0.3562,
+      "step": 4470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.6696316064026025e-06,
+      "loss": 0.3437,
+      "step": 4500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.6474904643895035e-06,
+      "loss": 0.3696,
+      "step": 4530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.625568373916034e-06,
+      "loss": 0.4247,
+      "step": 4560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.603861746695817e-06,
+      "loss": 0.3945,
+      "step": 4590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.582367076198595e-06,
+      "loss": 0.3798,
+      "step": 4620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.561080935270765e-06,
+      "loss": 0.3464,
+      "step": 4650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.53999997384e-06,
+      "loss": 0.3523,
+      "step": 4680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.5191209167004895e-06,
+      "loss": 0.363,
+      "step": 4710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.4984405613754545e-06,
+      "loss": 0.3986,
+      "step": 4740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.477955776053786e-06,
+      "loss": 0.3417,
+      "step": 4770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.457663497597783e-06,
+      "loss": 0.366,
+      "step": 4800
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.437560729619116e-06,
+      "loss": 0.3212,
+      "step": 4830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.41764454062025e-06,
+      "loss": 0.42,
+      "step": 4860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.3979120621987555e-06,
+      "loss": 0.3657,
+      "step": 4890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.378360487311966e-06,
+      "loss": 0.3274,
+      "step": 4920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.358987068599611e-06,
+      "loss": 0.3556,
+      "step": 4950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.339789116762173e-06,
+      "loss": 0.3901,
+      "step": 4980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.320763998992753e-06,
+      "loss": 0.3037,
+      "step": 5010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.301909137460409e-06,
+      "loss": 0.3604,
+      "step": 5040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.283222007842957e-06,
+      "loss": 0.3274,
+      "step": 5070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.264700137907351e-06,
+      "loss": 0.349,
+      "step": 5100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.246341106135824e-06,
+      "loss": 0.3592,
+      "step": 5130
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.22814254039606e-06,
+      "loss": 0.3222,
+      "step": 5160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.21010211665374e-06,
+      "loss": 0.3953,
+      "step": 5190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 6.192217557725875e-06,
+      "loss": 0.3456,
+      "step": 5220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.174486632073397e-06,
+      "loss": 0.2952,
+      "step": 5250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.156907152631576e-06,
+      "loss": 0.3385,
+      "step": 5280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.1394769756768524e-06,
+      "loss": 0.3676,
+      "step": 5310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.12219399972876e-06,
+      "loss": 0.3084,
+      "step": 5340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.105056164485665e-06,
+      "loss": 0.3262,
+      "step": 5370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.0880614497930825e-06,
+      "loss": 0.3254,
+      "step": 5400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.071207874643425e-06,
+      "loss": 0.3312,
+      "step": 5430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.054493496206035e-06,
+      "loss": 0.2912,
+      "step": 5460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 6.037916408886427e-06,
+      "loss": 0.3067,
+      "step": 5490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.021474743413714e-06,
+      "loss": 0.2967,
+      "step": 5520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 6.005166665955221e-06,
+      "loss": 0.3792,
+      "step": 5550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.988990377257332e-06,
+      "loss": 0.3027,
+      "step": 5580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.972944111811658e-06,
+      "loss": 0.3165,
+      "step": 5610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9570261370456484e-06,
+      "loss": 0.331,
+      "step": 5640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.941234752536812e-06,
+      "loss": 0.3406,
+      "step": 5670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.925568289249716e-06,
+      "loss": 0.3086,
+      "step": 5700
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.910025108795013e-06,
+      "loss": 0.335,
+      "step": 5730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.894603602709729e-06,
+      "loss": 0.3411,
+      "step": 5760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.879302191758114e-06,
+      "loss": 0.3122,
+      "step": 5790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.864119325252328e-06,
+      "loss": 0.3138,
+      "step": 5820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8490534803923536e-06,
+      "loss": 0.2888,
+      "step": 5850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8341031616244595e-06,
+      "loss": 0.3032,
+      "step": 5880
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.819266900017594e-06,
+      "loss": 0.3255,
+      "step": 5910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.804543252657151e-06,
+      "loss": 0.3174,
+      "step": 5940
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.789930802055526e-06,
+      "loss": 0.3083,
+      "step": 5970
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.775428155578901e-06,
+      "loss": 0.292,
+      "step": 6000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.761033944889748e-06,
+      "loss": 0.3553,
+      "step": 6030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.746746825404546e-06,
+      "loss": 0.316,
+      "step": 6060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.7325654757662186e-06,
+      "loss": 0.3385,
+      "step": 6090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.718956157005957e-06,
+      "loss": 0.3383,
+      "step": 6120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.704979053851296e-06,
+      "loss": 0.3152,
+      "step": 6150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.691103932244723e-06,
+      "loss": 0.2978,
+      "step": 6180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.677329558036595e-06,
+      "loss": 0.3089,
+      "step": 6210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.66365471788599e-06,
+      "loss": 0.3295,
+      "step": 6240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.650078218811767e-06,
+      "loss": 0.2739,
+      "step": 6270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.636598887755416e-06,
+      "loss": 0.3879,
+      "step": 6300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.623215571155333e-06,
+      "loss": 0.2913,
+      "step": 6330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.609927134532144e-06,
+      "loss": 0.3459,
+      "step": 6360
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.596732462084774e-06,
+      "loss": 0.3362,
+      "step": 6390
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.583630456296903e-06,
+      "loss": 0.3566,
+      "step": 6420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.570620037553528e-06,
+      "loss": 0.2793,
+      "step": 6450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.557700143767285e-06,
+      "loss": 0.3272,
+      "step": 6480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.544869730014276e-06,
+      "loss": 0.2916,
+      "step": 6510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.532127768179103e-06,
+      "loss": 0.327,
+      "step": 6540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.519473246608816e-06,
+      "loss": 0.33,
+      "step": 6570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.50690516977555e-06,
+      "loss": 0.2756,
+      "step": 6600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.494422557947562e-06,
+      "loss": 0.3566,
+      "step": 6630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.482024446868429e-06,
+      "loss": 0.3124,
+      "step": 6660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.469709887444195e-06,
+      "loss": 0.3601,
+      "step": 6690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.457477945438182e-06,
+      "loss": 0.3182,
+      "step": 6720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.445327701173302e-06,
+      "loss": 0.3418,
+      "step": 6750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.433258249241613e-06,
+      "loss": 0.2957,
+      "step": 6780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.4212686982209314e-06,
+      "loss": 0.3465,
+      "step": 6810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.409358170398289e-06,
+      "loss": 0.3435,
+      "step": 6840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.397525801500045e-06,
+      "loss": 0.3535,
+      "step": 6870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.385770740428468e-06,
+      "loss": 0.313,
+      "step": 6900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.37409214900459e-06,
+      "loss": 0.3222,
+      "step": 6930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.362489201717187e-06,
+      "loss": 0.2975,
+      "step": 6960
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.350961085477669e-06,
+      "loss": 0.3027,
+      "step": 6990
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.339506999380768e-06,
+      "loss": 0.2935,
+      "step": 7020
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.3281261544708195e-06,
+      "loss": 0.2793,
+      "step": 7050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.316817773513507e-06,
+      "loss": 0.2699,
+      "step": 7080
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.30558109077291e-06,
+      "loss": 0.3002,
+      "step": 7110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.2947864079915216e-06,
+      "loss": 0.3071,
+      "step": 7140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.283688541235345e-06,
+      "loss": 0.2798,
+      "step": 7170
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.272660166570372e-06,
+      "loss": 0.3221,
+      "step": 7200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.261700561773909e-06,
+      "loss": 0.2804,
+      "step": 7230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.250809015088132e-06,
+      "loss": 0.3268,
+      "step": 7260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.239984825025921e-06,
+      "loss": 0.3198,
+      "step": 7290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.229227300181094e-06,
+      "loss": 0.308,
+      "step": 7320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.218535759042913e-06,
+      "loss": 0.3606,
+      "step": 7350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.2079095298147244e-06,
+      "loss": 0.295,
+      "step": 7380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.197347950236668e-06,
+      "loss": 0.3313,
+      "step": 7410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.18685036741231e-06,
+      "loss": 0.3253,
+      "step": 7440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.176416137639122e-06,
+      "loss": 0.2724,
+      "step": 7470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.166044626242685e-06,
+      "loss": 0.3162,
+      "step": 7500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.155735207414551e-06,
+      "loss": 0.3187,
+      "step": 7530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.145487264053619e-06,
+      "loss": 0.2858,
+      "step": 7560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.135300187611006e-06,
+      "loss": 0.3,
+      "step": 7590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.125173377938251e-06,
+      "loss": 0.3222,
+      "step": 7620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.115106243138807e-06,
+      "loss": 0.2799,
+      "step": 7650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.105098199422739e-06,
+      "loss": 0.2617,
+      "step": 7680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.095148670964526e-06,
+      "loss": 0.3158,
+      "step": 7710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.085257089763904e-06,
+      "loss": 0.2533,
+      "step": 7740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.075422895509673e-06,
+      "loss": 0.2548,
+      "step": 7770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.065645535446375e-06,
+      "loss": 0.2719,
+      "step": 7800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.055924464243811e-06,
+      "loss": 0.2877,
+      "step": 7830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.046259143869278e-06,
+      "loss": 0.3081,
+      "step": 7860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.0366490434624956e-06,
+      "loss": 0.3128,
+      "step": 7890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.027093639213143e-06,
+      "loss": 0.323,
+      "step": 7920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.017592414240932e-06,
+      "loss": 0.2887,
+      "step": 7950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.008144858478176e-06,
+      "loss": 0.3268,
+      "step": 7980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9987504685547725e-06,
+      "loss": 0.3533,
+      "step": 8010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.989408747685563e-06,
+      "loss": 0.2842,
+      "step": 8040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.980119205559974e-06,
+      "loss": 0.254,
+      "step": 8070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.970881358233946e-06,
+      "loss": 0.2835,
+      "step": 8100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.961694728024028e-06,
+      "loss": 0.3396,
+      "step": 8130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.952558843403633e-06,
+      "loss": 0.3007,
+      "step": 8160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.943473238901383e-06,
+      "loss": 0.3001,
+      "step": 8190
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.934437455001497e-06,
+      "loss": 0.3292,
+      "step": 8220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.925451038046182e-06,
+      "loss": 0.2772,
+      "step": 8250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.916513540139955e-06,
+      "loss": 0.3337,
+      "step": 8280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.907624519055888e-06,
+      "loss": 0.2756,
+      "step": 8310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.898783538143701e-06,
+      "loss": 0.3178,
+      "step": 8340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.889990166239664e-06,
+      "loss": 0.321,
+      "step": 8370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.881243977578285e-06,
+      "loss": 0.3156,
+      "step": 8400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.872544551705718e-06,
+      "loss": 0.2833,
+      "step": 8430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8638914733948715e-06,
+      "loss": 0.2803,
+      "step": 8460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.85528433256216e-06,
+      "loss": 0.2551,
+      "step": 8490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8467227241858775e-06,
+      "loss": 0.2435,
+      "step": 8520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.838206248226147e-06,
+      "loss": 0.2682,
+      "step": 8550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.829734509546414e-06,
+      "loss": 0.3036,
+      "step": 8580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.821307117836431e-06,
+      "loss": 0.2704,
+      "step": 8610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.8129236875367506e-06,
+      "loss": 0.299,
+      "step": 8640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.804583837764616e-06,
+      "loss": 0.2679,
+      "step": 8670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.796287192241285e-06,
+      "loss": 0.2749,
+      "step": 8700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.788033379220728e-06,
+      "loss": 0.2858,
+      "step": 8730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.779822031419661e-06,
+      "loss": 0.2544,
+      "step": 8760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.771652785948902e-06,
+      "loss": 0.3259,
+      "step": 8790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.763525284246008e-06,
+      "loss": 0.3,
+      "step": 8820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.755439172009178e-06,
+      "loss": 0.2682,
+      "step": 8850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.7473940991323755e-06,
+      "loss": 0.2911,
+      "step": 8880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.739389719641665e-06,
+      "loss": 0.2617,
+      "step": 8910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.731425691632717e-06,
+      "loss": 0.2739,
+      "step": 8940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.723501677209465e-06,
+      "loss": 0.259,
+      "step": 8970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.715617342423896e-06,
+      "loss": 0.3113,
+      "step": 9000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.707772357216934e-06,
+      "loss": 0.2924,
+      "step": 9030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.69996639536041e-06,
+      "loss": 0.2422,
+      "step": 9060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.692199134400076e-06,
+      "loss": 0.2919,
+      "step": 9090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.684470255599664e-06,
+      "loss": 0.2697,
+      "step": 9120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.676779443885949e-06,
+      "loss": 0.2681,
+      "step": 9150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.669126387794794e-06,
+      "loss": 0.2753,
+      "step": 9180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.6615107794181814e-06,
+      "loss": 0.2761,
+      "step": 9210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.653932314352172e-06,
+      "loss": 0.2615,
+      "step": 9240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.646390691645805e-06,
+      "loss": 0.2758,
+      "step": 9270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.638885613750897e-06,
+      "loss": 0.2362,
+      "step": 9300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.6314167864727375e-06,
+      "loss": 0.2778,
+      "step": 9330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.62398391892165e-06,
+      "loss": 0.2577,
+      "step": 9360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.616586723465408e-06,
+      "loss": 0.2716,
+      "step": 9390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.609224915682485e-06,
+      "loss": 0.2567,
+      "step": 9420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.601898214316119e-06,
+      "loss": 0.2956,
+      "step": 9450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.5946063412291834e-06,
+      "loss": 0.2766,
+      "step": 9480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.587349021359836e-06,
+      "loss": 0.2743,
+      "step": 9510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.580125982677943e-06,
+      "loss": 0.2914,
+      "step": 9540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.572936956142248e-06,
+      "loss": 0.2849,
+      "step": 9570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.565781675658282e-06,
+      "loss": 0.2763,
+      "step": 9600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5586598780369954e-06,
+      "loss": 0.2888,
+      "step": 9630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5515713029541005e-06,
+      "loss": 0.2379,
+      "step": 9660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5445156929101e-06,
+      "loss": 0.2716,
+      "step": 9690
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.537492793190998e-06,
+      "loss": 0.2842,
+      "step": 9720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.530502351829687e-06,
+      "loss": 0.2865,
+      "step": 9750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.5235441195679665e-06,
+      "loss": 0.2819,
+      "step": 9780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.51661784981922e-06,
+      "loss": 0.2526,
+      "step": 9810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.509723298631711e-06,
+      "loss": 0.274,
+      "step": 9840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.5028602246524934e-06,
+      "loss": 0.2711,
+      "step": 9870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.496028389091924e-06,
+      "loss": 0.2126,
+      "step": 9900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.489227555688767e-06,
+      "loss": 0.2575,
+      "step": 9930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.482457490675879e-06,
+      "loss": 0.2622,
+      "step": 9960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.475717962746456e-06,
+      "loss": 0.2528,
+      "step": 9990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.469008743020842e-06,
+      "loss": 0.2992,
+      "step": 10020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.462329605013882e-06,
+      "loss": 0.3006,
+      "step": 10050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.455680324602811e-06,
+      "loss": 0.2729,
+      "step": 10080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.4490606799956615e-06,
+      "loss": 0.2525,
+      "step": 10110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.442470451700195e-06,
+      "loss": 0.215,
+      "step": 10140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.435909422493333e-06,
+      "loss": 0.2413,
+      "step": 10170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.429377377391085e-06,
+      "loss": 0.2817,
+      "step": 10200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.42287410361896e-06,
+      "loss": 0.3017,
+      "step": 10230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.416399390582857e-06,
+      "loss": 0.2611,
+      "step": 10260
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.409953029840417e-06,
+      "loss": 0.3096,
+      "step": 10290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.403534815072848e-06,
+      "loss": 0.2704,
+      "step": 10320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.397144542057179e-06,
+      "loss": 0.2775,
+      "step": 10350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.390782008638974e-06,
+      "loss": 0.3019,
+      "step": 10380
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3844470147054704e-06,
+      "loss": 0.2804,
+      "step": 10410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.378139362159141e-06,
+      "loss": 0.274,
+      "step": 10440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.371858854891681e-06,
+      "loss": 0.2692,
+      "step": 10470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.365605298758394e-06,
+      "loss": 0.2608,
+      "step": 10500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.359378501552986e-06,
+      "loss": 0.245,
+      "step": 10530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353178272982758e-06,
+      "loss": 0.2776,
+      "step": 10560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347004424644176e-06,
+      "loss": 0.2599,
+      "step": 10590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.340856769998823e-06,
+      "loss": 0.2604,
+      "step": 10620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3347351243497375e-06,
+      "loss": 0.2437,
+      "step": 10650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3286393048180905e-06,
+      "loss": 0.2808,
+      "step": 10680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.322569130320256e-06,
+      "loss": 0.2593,
+      "step": 10710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.316524421545204e-06,
+      "loss": 0.278,
+      "step": 10740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.310505000932264e-06,
+      "loss": 0.242,
+      "step": 10770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.304510692649209e-06,
+      "loss": 0.2666,
+      "step": 10800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.298541322570686e-06,
+      "loss": 0.2539,
+      "step": 10830
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.2925967182569775e-06,
+      "loss": 0.2709,
+      "step": 10860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.286676708933067e-06,
+      "loss": 0.2264,
+      "step": 10890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.280781125468046e-06,
+      "loss": 0.2364,
+      "step": 10920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.274909800354809e-06,
+      "loss": 0.2534,
+      "step": 10950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.269062567690073e-06,
+      "loss": 0.248,
+      "step": 10980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.263239263154682e-06,
+      "loss": 0.2808,
+      "step": 11010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.257439723994211e-06,
+      "loss": 0.2815,
+      "step": 11040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.251663788999866e-06,
+      "loss": 0.2839,
+      "step": 11070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.245911298489661e-06,
+      "loss": 0.2275,
+      "step": 11100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.24018209428987e-06,
+      "loss": 0.2606,
+      "step": 11130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.234476019716772e-06,
+      "loss": 0.2492,
+      "step": 11160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.228792919558642e-06,
+      "loss": 0.244,
+      "step": 11190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.2233209499878915e-06,
+      "loss": 0.2441,
+      "step": 11220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.217682585651912e-06,
+      "loss": 0.2401,
+      "step": 11250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.212066743759049e-06,
+      "loss": 0.3018,
+      "step": 11280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.206473274764352e-06,
+      "loss": 0.2571,
+      "step": 11310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.200902030509323e-06,
+      "loss": 0.2204,
+      "step": 11340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.195352864205429e-06,
+      "loss": 0.2832,
+      "step": 11370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.18982563041786e-06,
+      "loss": 0.2632,
+      "step": 11400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.184320185049522e-06,
+      "loss": 0.2718,
+      "step": 11430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.178836385325257e-06,
+      "loss": 0.2863,
+      "step": 11460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.173374089776288e-06,
+      "loss": 0.221,
+      "step": 11490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.167933158224896e-06,
+      "loss": 0.2486,
+      "step": 11520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.1625134517693085e-06,
+      "loss": 0.2539,
+      "step": 11550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.157114832768805e-06,
+      "loss": 0.2283,
+      "step": 11580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.151737164829035e-06,
+      "loss": 0.2161,
+      "step": 11610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.146380312787547e-06,
+      "loss": 0.2378,
+      "step": 11640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.141044142699519e-06,
+      "loss": 0.2391,
+      "step": 11670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.135728521823688e-06,
+      "loss": 0.2288,
+      "step": 11700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.130433318608488e-06,
+      "loss": 0.2284,
+      "step": 11730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.125158402678369e-06,
+      "loss": 0.2769,
+      "step": 11760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.11990364482032e-06,
+      "loss": 0.2132,
+      "step": 11790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.114668916970569e-06,
+      "loss": 0.2419,
+      "step": 11820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.109454092201482e-06,
+      "loss": 0.2575,
+      "step": 11850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.104259044708631e-06,
+      "loss": 0.242,
+      "step": 11880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.099083649798046e-06,
+      "loss": 0.2375,
+      "step": 11910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0939277838736505e-06,
+      "loss": 0.2461,
+      "step": 11940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0887913244248565e-06,
+      "loss": 0.2147,
+      "step": 11970
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.083674150014344e-06,
+      "loss": 0.2503,
+      "step": 12000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.078576140266e-06,
+      "loss": 0.2648,
+      "step": 12030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.073497175853025e-06,
+      "loss": 0.227,
+      "step": 12060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.068437138486202e-06,
+      "loss": 0.2633,
+      "step": 12090
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0633959109023236e-06,
+      "loss": 0.2051,
+      "step": 12120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.0583733768527826e-06,
+      "loss": 0.2756,
+      "step": 12150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.053369421092312e-06,
+      "loss": 0.258,
+      "step": 12180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.048383929367871e-06,
+      "loss": 0.2311,
+      "step": 12210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.043416788407701e-06,
+      "loss": 0.2362,
+      "step": 12240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.038467885910504e-06,
+      "loss": 0.2256,
+      "step": 12270
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.03353711053479e-06,
+      "loss": 0.2869,
+      "step": 12300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.0286243518883425e-06,
+      "loss": 0.2358,
+      "step": 12330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.023729500517853e-06,
+      "loss": 0.2447,
+      "step": 12360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.018852447898667e-06,
+      "loss": 0.2623,
+      "step": 12390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.013993086424682e-06,
+      "loss": 0.2221,
+      "step": 12420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.009151309398378e-06,
+      "loss": 0.2294,
+      "step": 12450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.004327011020975e-06,
+      "loss": 0.2331,
+      "step": 12480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.999520086382724e-06,
+      "loss": 0.2695,
+      "step": 12510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.994730431453324e-06,
+      "loss": 0.2675,
+      "step": 12540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.989957943072466e-06,
+      "loss": 0.2398,
+      "step": 12570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.985202518940505e-06,
+      "loss": 0.2182,
+      "step": 12600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.980464057609243e-06,
+      "loss": 0.2384,
+      "step": 12630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.975742458472844e-06,
+      "loss": 0.2626,
+      "step": 12660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.971037621758861e-06,
+      "loss": 0.2213,
+      "step": 12690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.966349448519381e-06,
+      "loss": 0.2146,
+      "step": 12720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.961677840622287e-06,
+      "loss": 0.2705,
+      "step": 12750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.957022700742623e-06,
+      "loss": 0.2623,
+      "step": 12780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.952383932354089e-06,
+      "loss": 0.2269,
+      "step": 12810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.947761439720625e-06,
+      "loss": 0.2925,
+      "step": 12840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.94315512788812e-06,
+      "loss": 0.1932,
+      "step": 12870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9385649026762125e-06,
+      "loss": 0.2326,
+      "step": 12900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.9339906706702065e-06,
+      "loss": 0.205,
+      "step": 12930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.929432339213089e-06,
+      "loss": 0.2248,
+      "step": 12960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.92488981639764e-06,
+      "loss": 0.2375,
+      "step": 12990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.920363011058658e-06,
+      "loss": 0.2357,
+      "step": 13020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.91585183276527e-06,
+      "loss": 0.2548,
+      "step": 13050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.911356191813345e-06,
+      "loss": 0.2319,
+      "step": 13080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.906875999218003e-06,
+      "loss": 0.2322,
+      "step": 13110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.902411166706214e-06,
+      "loss": 0.2329,
+      "step": 13140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.897961606709499e-06,
+      "loss": 0.2551,
+      "step": 13170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.893527232356712e-06,
+      "loss": 0.2108,
+      "step": 13200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.8891079574669165e-06,
+      "loss": 0.2038,
+      "step": 13230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.884703696542354e-06,
+      "loss": 0.2323,
+      "step": 13260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.8803143647614936e-06,
+      "loss": 0.2142,
+      "step": 13290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.87593987797217e-06,
+      "loss": 0.2314,
+      "step": 13320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.871580152684815e-06,
+      "loss": 0.2061,
+      "step": 13350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.867235106065752e-06,
+      "loss": 0.2245,
+      "step": 13380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.862904655930595e-06,
+      "loss": 0.2586,
+      "step": 13410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.858588720737721e-06,
+      "loss": 0.2408,
+      "step": 13440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.854287219581815e-06,
+      "loss": 0.2,
+      "step": 13470
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.850000072187502e-06,
+      "loss": 0.225,
+      "step": 13500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.8457271989030575e-06,
+      "loss": 0.2431,
+      "step": 13530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.841468520694188e-06,
+      "loss": 0.2341,
+      "step": 13560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.837223959137896e-06,
+      "loss": 0.2108,
+      "step": 13590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.832993436416406e-06,
+      "loss": 0.206,
+      "step": 13620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.828776875311186e-06,
+      "loss": 0.2474,
+      "step": 13650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.824574199197011e-06,
+      "loss": 0.2402,
+      "step": 13680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.820524739262436e-06,
+      "loss": 0.2061,
+      "step": 13710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.8163491490251235e-06,
+      "loss": 0.2226,
+      "step": 13740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.8121872198917364e-06,
+      "loss": 0.1776,
+      "step": 13770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.8080388775336093e-06,
+      "loss": 0.2507,
+      "step": 13800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.8039040481870244e-06,
+      "loss": 0.1796,
+      "step": 13830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7997826586477104e-06,
+      "loss": 0.2571,
+      "step": 13860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.795674636265392e-06,
+      "loss": 0.2295,
+      "step": 13890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7915799089384153e-06,
+      "loss": 0.2282,
+      "step": 13920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7874984051084297e-06,
+      "loss": 0.2232,
+      "step": 13950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7834300537551376e-06,
+      "loss": 0.2261,
+      "step": 13980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7793747843911037e-06,
+      "loss": 0.2474,
+      "step": 14010
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.775332527056626e-06,
+      "loss": 0.2066,
+      "step": 14040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7713032123146647e-06,
+      "loss": 0.2008,
+      "step": 14070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7672867712458343e-06,
+      "loss": 0.2688,
+      "step": 14100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.763283135443452e-06,
+      "loss": 0.2385,
+      "step": 14130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7592922370086455e-06,
+      "loss": 0.24,
+      "step": 14160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7553140085455136e-06,
+      "loss": 0.1846,
+      "step": 14190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7513483831563473e-06,
+      "loss": 0.2544,
+      "step": 14220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7473952944369068e-06,
+      "loss": 0.1966,
+      "step": 14250
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7434546764717492e-06,
+      "loss": 0.2499,
+      "step": 14280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7395264638296123e-06,
+      "loss": 0.2198,
+      "step": 14310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.735610591558852e-06,
+      "loss": 0.227,
+      "step": 14340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.7318369179611134e-06,
+      "loss": 0.2223,
+      "step": 14370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.7279451274374513e-06,
+      "loss": 0.2398,
+      "step": 14400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.7240654873736155e-06,
+      "loss": 0.2662,
+      "step": 14430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.720197934676515e-06,
+      "loss": 0.2502,
+      "step": 14460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.7163424067107786e-06,
+      "loss": 0.2303,
+      "step": 14490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.712498841294488e-06,
+      "loss": 0.2081,
+      "step": 14520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.7086671766949745e-06,
+      "loss": 0.2286,
+      "step": 14550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.7048473516246474e-06,
+      "loss": 0.2246,
+      "step": 14580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.7010393052368808e-06,
+      "loss": 0.2142,
+      "step": 14610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.697242977121942e-06,
+      "loss": 0.2154,
+      "step": 14640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.693458307302965e-06,
+      "loss": 0.2299,
+      "step": 14670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6896852362319716e-06,
+      "loss": 0.1961,
+      "step": 14700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.685923704785937e-06,
+      "loss": 0.2212,
+      "step": 14730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6821736542628983e-06,
+      "loss": 0.2012,
+      "step": 14760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6784350263781066e-06,
+      "loss": 0.1914,
+      "step": 14790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.674707763260221e-06,
+      "loss": 0.2305,
+      "step": 14820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.670991807447548e-06,
+      "loss": 0.2325,
+      "step": 14850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.6672871018843207e-06,
+      "loss": 0.222,
+      "step": 14880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.663593589917019e-06,
+      "loss": 0.2182,
+      "step": 14910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.65991121529073e-06,
+      "loss": 0.256,
+      "step": 14940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.656239922145549e-06,
+      "loss": 0.1909,
+      "step": 14970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.652579655013025e-06,
+      "loss": 0.1799,
+      "step": 15000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6489303588126325e-06,
+      "loss": 0.2382,
+      "step": 15030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.645291978848299e-06,
+      "loss": 0.2181,
+      "step": 15060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.641664460804956e-06,
+      "loss": 0.2192,
+      "step": 15090
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.6380477507451357e-06,
+      "loss": 0.2245,
+      "step": 15120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6344417951056025e-06,
+      "loss": 0.2237,
+      "step": 15150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6308465406940214e-06,
+      "loss": 0.2335,
+      "step": 15180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.62726193468566e-06,
+      "loss": 0.2129,
+      "step": 15210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.623687924620131e-06,
+      "loss": 0.1975,
+      "step": 15240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6201244583981697e-06,
+      "loss": 0.2239,
+      "step": 15270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6165714842784393e-06,
+      "loss": 0.2059,
+      "step": 15300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6130289508743783e-06,
+      "loss": 0.22,
+      "step": 15330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.609496807151077e-06,
+      "loss": 0.2146,
+      "step": 15360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.6059750024221956e-06,
+      "loss": 0.2378,
+      "step": 15390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.6024634863468995e-06,
+      "loss": 0.2177,
+      "step": 15420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.598962208926845e-06,
+      "loss": 0.2528,
+      "step": 15450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5954711205031867e-06,
+      "loss": 0.2179,
+      "step": 15480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5919901717536176e-06,
+      "loss": 0.225,
+      "step": 15510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.588519313689446e-06,
+      "loss": 0.2327,
+      "step": 15540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.585058497652696e-06,
+      "loss": 0.2267,
+      "step": 15570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5816076753132444e-06,
+      "loss": 0.1847,
+      "step": 15600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.5781667986659885e-06,
+      "loss": 0.218,
+      "step": 15630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.574735820028035e-06,
+      "loss": 0.2386,
+      "step": 15660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.57131469203593e-06,
+      "loss": 0.2043,
+      "step": 15690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5679033676429114e-06,
+      "loss": 0.2702,
+      "step": 15720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5645018001161923e-06,
+      "loss": 0.2302,
+      "step": 15750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5611099430342683e-06,
+      "loss": 0.2406,
+      "step": 15780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5577277502842616e-06,
+      "loss": 0.2107,
+      "step": 15810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5543551760592854e-06,
+      "loss": 0.2309,
+      "step": 15840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5509921748558384e-06,
+      "loss": 0.2256,
+      "step": 15870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.547638701471227e-06,
+      "loss": 0.2328,
+      "step": 15900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5442947110010123e-06,
+      "loss": 0.2417,
+      "step": 15930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5409601588364854e-06,
+      "loss": 0.197,
+      "step": 15960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5376350006621686e-06,
+      "loss": 0.2551,
+      "step": 15990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.53431919245334e-06,
+      "loss": 0.1884,
+      "step": 16020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.531012690473587e-06,
+      "loss": 0.2096,
+      "step": 16050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5277154512723837e-06,
+      "loss": 0.1941,
+      "step": 16080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5244274316826905e-06,
+      "loss": 0.2387,
+      "step": 16110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5211485888185817e-06,
+      "loss": 0.1942,
+      "step": 16140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5178788800728976e-06,
+      "loss": 0.2117,
+      "step": 16170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5146182631149166e-06,
+      "loss": 0.2085,
+      "step": 16200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.511366695888056e-06,
+      "loss": 0.2429,
+      "step": 16230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.50812413660759e-06,
+      "loss": 0.2289,
+      "step": 16260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.504890543758399e-06,
+      "loss": 0.2241,
+      "step": 16290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5016658760927338e-06,
+      "loss": 0.195,
+      "step": 16320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4984500926280053e-06,
+      "loss": 0.2107,
+      "step": 16350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4952431526446017e-06,
+      "loss": 0.1986,
+      "step": 16380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.492045015683717e-06,
+      "loss": 0.2294,
+      "step": 16410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4888556415452106e-06,
+      "loss": 0.2219,
+      "step": 16440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.485674990285484e-06,
+      "loss": 0.2011,
+      "step": 16470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.48250302221538e-06,
+      "loss": 0.2119,
+      "step": 16500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4793396978981016e-06,
+      "loss": 0.2136,
+      "step": 16530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4761849781471552e-06,
+      "loss": 0.1887,
+      "step": 16560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4730388240243084e-06,
+      "loss": 0.2112,
+      "step": 16590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.469901196837573e-06,
+      "loss": 0.1988,
+      "step": 16620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4667720581392095e-06,
+      "loss": 0.1944,
+      "step": 16650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4636513697237406e-06,
+      "loss": 0.208,
+      "step": 16680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.460539093626001e-06,
+      "loss": 0.1934,
+      "step": 16710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.45743519211919e-06,
+      "loss": 0.175,
+      "step": 16740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4543396277129578e-06,
+      "loss": 0.2059,
+      "step": 16770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.451252363151495e-06,
+      "loss": 0.2617,
+      "step": 16800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4481733614116573e-06,
+      "loss": 0.1825,
+      "step": 16830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4451025857010983e-06,
+      "loss": 0.2182,
+      "step": 16860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4420399994564192e-06,
+      "loss": 0.1792,
+      "step": 16890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.438985566341346e-06,
+      "loss": 0.2609,
+      "step": 16920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4359392502449156e-06,
+      "loss": 0.2008,
+      "step": 16950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4329010152796816e-06,
+      "loss": 0.1738,
+      "step": 16980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.429870825779942e-06,
+      "loss": 0.2357,
+      "step": 17010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.426848646299979e-06,
+      "loss": 0.2309,
+      "step": 17040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.423834441612316e-06,
+      "loss": 0.2102,
+      "step": 17070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4208281767059957e-06,
+      "loss": 0.1933,
+      "step": 17100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4178298167848716e-06,
+      "loss": 0.1512,
+      "step": 17130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.414839327265915e-06,
+      "loss": 0.1976,
+      "step": 17160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4118566737775383e-06,
+      "loss": 0.2193,
+      "step": 17190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4088818221579424e-06,
+      "loss": 0.2124,
+      "step": 17220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4059147384534635e-06,
+      "loss": 0.2116,
+      "step": 17250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4029553889169558e-06,
+      "loss": 0.1893,
+      "step": 17280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4000037400061716e-06,
+      "loss": 0.2162,
+      "step": 17310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3970597583821664e-06,
+      "loss": 0.2022,
+      "step": 17340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.394123410907721e-06,
+      "loss": 0.2171,
+      "step": 17370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3911946646457707e-06,
+      "loss": 0.2321,
+      "step": 17400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3882734868578534e-06,
+      "loss": 0.2312,
+      "step": 17430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3853598450025764e-06,
+      "loss": 0.1934,
+      "step": 17460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3824537067340902e-06,
+      "loss": 0.177,
+      "step": 17490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.379555039900584e-06,
+      "loss": 0.193,
+      "step": 17520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3766638125427857e-06,
+      "loss": 0.2091,
+      "step": 17550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.373779992892488e-06,
+      "loss": 0.2174,
+      "step": 17580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3709035493710807e-06,
+      "loss": 0.2565,
+      "step": 17610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.368034450588095e-06,
+      "loss": 0.2153,
+      "step": 17640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3651726653397697e-06,
+      "loss": 0.2146,
+      "step": 17670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3623181626076225e-06,
+      "loss": 0.1969,
+      "step": 17700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.359470911557039e-06,
+      "loss": 0.2297,
+      "step": 17730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3566308815358723e-06,
+      "loss": 0.2054,
+      "step": 17760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3537980420730624e-06,
+      "loss": 0.2202,
+      "step": 17790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.3509723628772543e-06,
+      "loss": 0.2301,
+      "step": 17820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.348153813835445e-06,
+      "loss": 0.2316,
+      "step": 17850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3453423650116334e-06,
+      "loss": 0.2068,
+      "step": 17880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.342537986645483e-06,
+      "loss": 0.2173,
+      "step": 17910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3397406491510005e-06,
+      "loss": 0.1801,
+      "step": 17940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.336950323115227e-06,
+      "loss": 0.2013,
+      "step": 17970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3341669792969324e-06,
+      "loss": 0.2041,
+      "step": 18000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3313905886253366e-06,
+      "loss": 0.178,
+      "step": 18030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.3286211221988295e-06,
+      "loss": 0.1897,
+      "step": 18060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.325858551283706e-06,
+      "loss": 0.2258,
+      "step": 18090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3231028473129184e-06,
+      "loss": 0.191,
+      "step": 18120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3203539818848333e-06,
+      "loss": 0.1796,
+      "step": 18150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3176119267620007e-06,
+      "loss": 0.2318,
+      "step": 18180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3148766538699385e-06,
+      "loss": 0.1636,
+      "step": 18210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3121481352959235e-06,
+      "loss": 0.1984,
+      "step": 18240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3094263432877955e-06,
+      "loss": 0.1771,
+      "step": 18270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.306711250252772e-06,
+      "loss": 0.1933,
+      "step": 18300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.304002828756273e-06,
+      "loss": 0.1874,
+      "step": 18330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.3013010515207573e-06,
+      "loss": 0.2102,
+      "step": 18360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.298605891424567e-06,
+      "loss": 0.1955,
+      "step": 18390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2959173215007867e-06,
+      "loss": 0.1924,
+      "step": 18420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.293235314936108e-06,
+      "loss": 0.1786,
+      "step": 18450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2905598450697064e-06,
+      "loss": 0.2073,
+      "step": 18480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2878908853921286e-06,
+      "loss": 0.1909,
+      "step": 18510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2852284095441923e-06,
+      "loss": 0.1895,
+      "step": 18540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.282572391315887e-06,
+      "loss": 0.1972,
+      "step": 18570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.279922804645294e-06,
+      "loss": 0.1673,
+      "step": 18600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.2772796236175144e-06,
+      "loss": 0.225,
+      "step": 18630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.274642822463597e-06,
+      "loss": 0.2317,
+      "step": 18660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2720999550184424e-06,
+      "loss": 0.2248,
+      "step": 18690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.269475626333966e-06,
+      "loss": 0.2053,
+      "step": 18720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.266857601924006e-06,
+      "loss": 0.2262,
+      "step": 18750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2642458565882563e-06,
+      "loss": 0.2002,
+      "step": 18780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2616403652672114e-06,
+      "loss": 0.1788,
+      "step": 18810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2590411030411596e-06,
+      "loss": 0.1958,
+      "step": 18840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.2564480451291814e-06,
+      "loss": 0.1836,
+      "step": 18870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.253861166888155e-06,
+      "loss": 0.2255,
+      "step": 18900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2512804438117756e-06,
+      "loss": 0.1828,
+      "step": 18930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.24870585152958e-06,
+      "loss": 0.1978,
+      "step": 18960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2461373658059773e-06,
+      "loss": 0.1853,
+      "step": 18990
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.243574962539294e-06,
+      "loss": 0.1664,
+      "step": 19020
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.241018617760822e-06,
+      "loss": 0.1852,
+      "step": 19050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.2384683076338774e-06,
+      "loss": 0.193,
+      "step": 19080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.235924008452868e-06,
+      "loss": 0.2052,
+      "step": 19110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.233385696642366e-06,
+      "loss": 0.1822,
+      "step": 19140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.230853348756193e-06,
+      "loss": 0.2126,
+      "step": 19170
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2283269414765067e-06,
+      "loss": 0.1973,
+      "step": 19200
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.2046,
+      "step": 19230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2232918561015214e-06,
+      "loss": 0.2206,
+      "step": 19260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2207831320041542e-06,
+      "loss": 0.1786,
+      "step": 19290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.2182802565073746e-06,
+      "loss": 0.1932,
+      "step": 19320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.215783206921658e-06,
+      "loss": 0.1883,
+      "step": 19350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.213291960680524e-06,
+      "loss": 0.2078,
+      "step": 19380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.210806495339678e-06,
+      "loss": 0.1932,
+      "step": 19410
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.208326788576161e-06,
+      "loss": 0.1897,
+      "step": 19440
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.205852818187507e-06,
+      "loss": 0.1865,
+      "step": 19470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2033845620909114e-06,
+      "loss": 0.1811,
+      "step": 19500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.2009219983224e-06,
+      "loss": 0.1725,
+      "step": 19530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1984651050360064e-06,
+      "loss": 0.2187,
+      "step": 19560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1960138605029663e-06,
+      "loss": 0.2321,
+      "step": 19590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1935682431108977e-06,
+      "loss": 0.2007,
+      "step": 19620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.1911282313630144e-06,
+      "loss": 0.1824,
+      "step": 19650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.18869380387732e-06,
+      "loss": 0.2498,
+      "step": 19680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1862649393858314e-06,
+      "loss": 0.2345,
+      "step": 19710
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.183841616733788e-06,
+      "loss": 0.202,
+      "step": 19740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.181423814878889e-06,
+      "loss": 0.2071,
+      "step": 19770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1790115128905165e-06,
+      "loss": 0.1641,
+      "step": 19800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.17660468994898e-06,
+      "loss": 0.2057,
+      "step": 19830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1742033253447586e-06,
+      "loss": 0.1941,
+      "step": 19860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.1718073984777564e-06,
+      "loss": 0.2097,
+      "step": 19890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.169416888856555e-06,
+      "loss": 0.1986,
+      "step": 19920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.167031776097682e-06,
+      "loss": 0.1843,
+      "step": 19950
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1646520399248797e-06,
+      "loss": 0.2038,
+      "step": 19980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1622776601683796e-06,
+      "loss": 0.1745,
+      "step": 20010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159908616764186e-06,
+      "loss": 0.199,
+      "step": 20040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1575448897533633e-06,
+      "loss": 0.1845,
+      "step": 20070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.15518645928133e-06,
+      "loss": 0.206,
+      "step": 20100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1528333055971584e-06,
+      "loss": 0.2374,
+      "step": 20130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1504854090528795e-06,
+      "loss": 0.2106,
+      "step": 20160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1481427501027932e-06,
+      "loss": 0.1568,
+      "step": 20190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1458053093027873e-06,
+      "loss": 0.1677,
+      "step": 20220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1434730673096573e-06,
+      "loss": 0.187,
+      "step": 20250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.141146004880436e-06,
+      "loss": 0.1736,
+      "step": 20280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1388241028717226e-06,
+      "loss": 0.2169,
+      "step": 20310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.136507342239028e-06,
+      "loss": 0.1903,
+      "step": 20340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1341957040361138e-06,
+      "loss": 0.1915,
+      "step": 20370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.131889169414341e-06,
+      "loss": 0.1775,
+      "step": 20400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.1295877196220293e-06,
+      "loss": 0.2063,
+      "step": 20430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.127291336003811e-06,
+      "loss": 0.1973,
+      "step": 20460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.125e-06,
+      "loss": 0.1864,
+      "step": 20490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1227136931459613e-06,
+      "loss": 0.1893,
+      "step": 20520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1204323970714836e-06,
+      "loss": 0.23,
+      "step": 20550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.118156093500161e-06,
+      "loss": 0.2014,
+      "step": 20580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1158847642487794e-06,
+      "loss": 0.1873,
+      "step": 20610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1136183912267038e-06,
+      "loss": 0.2029,
+      "step": 20640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.111356956435275e-06,
+      "loss": 0.1946,
+      "step": 20670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1091004419672094e-06,
+      "loss": 0.2022,
+      "step": 20700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1068488300060003e-06,
+      "loss": 0.1916,
+      "step": 20730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.1046021028253316e-06,
+      "loss": 0.1954,
+      "step": 20760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.1023602427884887e-06,
+      "loss": 0.1604,
+      "step": 20790
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.1001232323477775e-06,
+      "loss": 0.2243,
+      "step": 20820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0978910540439495e-06,
+      "loss": 0.214,
+      "step": 20850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.095663690505624e-06,
+      "loss": 0.1776,
+      "step": 20880
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.09344112444873e-06,
+      "loss": 0.203,
+      "step": 20910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0912233386759318e-06,
+      "loss": 0.185,
+      "step": 20940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0890103160760777e-06,
+      "loss": 0.1801,
+      "step": 20970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0868020396236425e-06,
+      "loss": 0.1748,
+      "step": 21000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.0845984923781787e-06,
+      "loss": 0.2048,
+      "step": 21030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0823996574837696e-06,
+      "loss": 0.1797,
+      "step": 21060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0802055181684876e-06,
+      "loss": 0.1694,
+      "step": 21090
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0780889645514934e-06,
+      "loss": 0.1713,
+      "step": 21120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.075904011269121e-06,
+      "loss": 0.1855,
+      "step": 21150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0737237042981292e-06,
+      "loss": 0.2117,
+      "step": 21180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0715480271945e-06,
+      "loss": 0.1906,
+      "step": 21210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0693769635955745e-06,
+      "loss": 0.248,
+      "step": 21240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.067210497219538e-06,
+      "loss": 0.1815,
+      "step": 21270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.0650486118649127e-06,
+      "loss": 0.1695,
+      "step": 21300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0628912914100383e-06,
+      "loss": 0.1733,
+      "step": 21330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0607385198125768e-06,
+      "loss": 0.187,
+      "step": 21360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0585902811090056e-06,
+      "loss": 0.1775,
+      "step": 21390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.056446559414125e-06,
+      "loss": 0.2013,
+      "step": 21420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0543073389205587e-06,
+      "loss": 0.1936,
+      "step": 21450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.05217260389827e-06,
+      "loss": 0.1781,
+      "step": 21480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0500423386940733e-06,
+      "loss": 0.1818,
+      "step": 21510
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0479165277311516e-06,
+      "loss": 0.1745,
+      "step": 21540
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.0457951555085815e-06,
+      "loss": 0.1871,
+      "step": 21570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0436782066008522e-06,
+      "loss": 0.2433,
+      "step": 21600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0415656656574006e-06,
+      "loss": 0.1986,
+      "step": 21630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0394575174021406e-06,
+      "loss": 0.176,
+      "step": 21660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0373537466330005e-06,
+      "loss": 0.2019,
+      "step": 21690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0352543382214616e-06,
+      "loss": 0.2233,
+      "step": 21720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0332290426009857e-06,
+      "loss": 0.2413,
+      "step": 21750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0311381696410036e-06,
+      "loss": 0.1946,
+      "step": 21780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0290516145856953e-06,
+      "loss": 0.1893,
+      "step": 21810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.0269693625938358e-06,
+      "loss": 0.2216,
+      "step": 21840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.0248913988955188e-06,
+      "loss": 0.1971,
+      "step": 21870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.0228177087917153e-06,
+      "loss": 0.1776,
+      "step": 21900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.0207482776538377e-06,
+      "loss": 0.1729,
+      "step": 21930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.0186830909233056e-06,
+      "loss": 0.2129,
+      "step": 21960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.0166221341111156e-06,
+      "loss": 0.1983,
+      "step": 21990
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.014565392797416e-06,
+      "loss": 0.1759,
+      "step": 22020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.01251285263108e-06,
+      "loss": 0.1762,
+      "step": 22050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.01046449932929e-06,
+      "loss": 0.1912,
+      "step": 22080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.008420318677114e-06,
+      "loss": 0.1635,
+      "step": 22110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.006380296527096e-06,
+      "loss": 0.1975,
+      "step": 22140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.004344418798843e-06,
+      "loss": 0.1715,
+      "step": 22170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0023126714786185e-06,
+      "loss": 0.172,
+      "step": 22200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.0002850406189315e-06,
+      "loss": 0.1759,
+      "step": 22230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9982615123381435e-06,
+      "loss": 0.1836,
+      "step": 22260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.9962420728200638e-06,
+      "loss": 0.1822,
+      "step": 22290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.99422670831355e-06,
+      "loss": 0.1924,
+      "step": 22320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.992215405132124e-06,
+      "loss": 0.1958,
+      "step": 22350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.990208149653574e-06,
+      "loss": 0.1966,
+      "step": 22380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.988204928319569e-06,
+      "loss": 0.1612,
+      "step": 22410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9862057276352764e-06,
+      "loss": 0.1816,
+      "step": 22440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.984210534168976e-06,
+      "loss": 0.2078,
+      "step": 22470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.982219334551684e-06,
+      "loss": 0.1877,
+      "step": 22500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9802321154767783e-06,
+      "loss": 0.1862,
+      "step": 22530
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.978248863699621e-06,
+      "loss": 0.2263,
+      "step": 22560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9762695660371887e-06,
+      "loss": 0.1911,
+      "step": 22590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9742942093677073e-06,
+      "loss": 0.1784,
+      "step": 22620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9723227806302822e-06,
+      "loss": 0.2086,
+      "step": 22650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9703552668245405e-06,
+      "loss": 0.2191,
+      "step": 22680
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9683916550102658e-06,
+      "loss": 0.1633,
+      "step": 22710
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9664319323070455e-06,
+      "loss": 0.181,
+      "step": 22740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.964476085893913e-06,
+      "loss": 0.2081,
+      "step": 22770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9625241030089954e-06,
+      "loss": 0.1903,
+      "step": 22800
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9605759709491647e-06,
+      "loss": 0.1801,
+      "step": 22830
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9586316770696934e-06,
+      "loss": 0.223,
+      "step": 22860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9566912087839018e-06,
+      "loss": 0.1891,
+      "step": 22890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.954754553562824e-06,
+      "loss": 0.1841,
+      "step": 22920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9528216989348657e-06,
+      "loss": 0.2035,
+      "step": 22950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9508926324854625e-06,
+      "loss": 0.1673,
+      "step": 22980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9489673418567527e-06,
+      "loss": 0.1781,
+      "step": 23010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.947045814747238e-06,
+      "loss": 0.1882,
+      "step": 23040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9451280389114573e-06,
+      "loss": 0.1896,
+      "step": 23070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9432140021596566e-06,
+      "loss": 0.1775,
+      "step": 23100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.941303692357466e-06,
+      "loss": 0.1693,
+      "step": 23130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9393970974255757e-06,
+      "loss": 0.2118,
+      "step": 23160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.937494205339412e-06,
+      "loss": 0.1826,
+      "step": 23190
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9355950041288252e-06,
+      "loss": 0.1622,
+      "step": 23220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9336994818777668e-06,
+      "loss": 0.1661,
+      "step": 23250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.93180762672398e-06,
+      "loss": 0.1776,
+      "step": 23280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9299194268586878e-06,
+      "loss": 0.1617,
+      "step": 23310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9280348705262775e-06,
+      "loss": 0.165,
+      "step": 23340
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.926153946024002e-06,
+      "loss": 0.1841,
+      "step": 23370
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9242766417016677e-06,
+      "loss": 0.1686,
+      "step": 23400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.922402945961337e-06,
+      "loss": 0.179,
+      "step": 23430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9205328472570204e-06,
+      "loss": 0.1653,
+      "step": 23460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9186663340943855e-06,
+      "loss": 0.212,
+      "step": 23490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9168033950304554e-06,
+      "loss": 0.1896,
+      "step": 23520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9149440186733125e-06,
+      "loss": 0.1886,
+      "step": 23550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9130881936818123e-06,
+      "loss": 0.1959,
+      "step": 23580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9112359087652875e-06,
+      "loss": 0.211,
+      "step": 23610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.909387152683261e-06,
+      "loss": 0.1904,
+      "step": 23640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9075419142451583e-06,
+      "loss": 0.1705,
+      "step": 23670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.9057001823100285e-06,
+      "loss": 0.1779,
+      "step": 23700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.903861945786251e-06,
+      "loss": 0.1967,
+      "step": 23730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9020271936312684e-06,
+      "loss": 0.1746,
+      "step": 23760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9001959148512965e-06,
+      "loss": 0.2198,
+      "step": 23790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.898368098501052e-06,
+      "loss": 0.1712,
+      "step": 23820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.89654373368348e-06,
+      "loss": 0.218,
+      "step": 23850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.894722809549479e-06,
+      "loss": 0.1932,
+      "step": 23880
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8929053152976272e-06,
+      "loss": 0.1703,
+      "step": 23910
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8910912401739185e-06,
+      "loss": 0.2124,
+      "step": 23940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8892805734714926e-06,
+      "loss": 0.2129,
+      "step": 23970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8874733045303684e-06,
+      "loss": 0.1994,
+      "step": 24000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.885669422737183e-06,
+      "loss": 0.1716,
+      "step": 24030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8838689175249286e-06,
+      "loss": 0.2043,
+      "step": 24060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8820717783726926e-06,
+      "loss": 0.1575,
+      "step": 24090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8802779948053998e-06,
+      "loss": 0.1846,
+      "step": 24120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.878487556393554e-06,
+      "loss": 0.2028,
+      "step": 24150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8767004527529884e-06,
+      "loss": 0.1681,
+      "step": 24180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8749166735446054e-06,
+      "loss": 0.2006,
+      "step": 24210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.873136208474131e-06,
+      "loss": 0.1828,
+      "step": 24240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8713590472918633e-06,
+      "loss": 0.1757,
+      "step": 24270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8695851797924253e-06,
+      "loss": 0.1763,
+      "step": 24300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8678145958145158e-06,
+      "loss": 0.2044,
+      "step": 24330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8660472852406714e-06,
+      "loss": 0.2031,
+      "step": 24360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8642832379970157e-06,
+      "loss": 0.2166,
+      "step": 24390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.862522444053026e-06,
+      "loss": 0.1725,
+      "step": 24420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.860764893421287e-06,
+      "loss": 0.1614,
+      "step": 24450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.859010576157256e-06,
+      "loss": 0.1866,
+      "step": 24480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.857259482359027e-06,
+      "loss": 0.1888,
+      "step": 24510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.855511602167096e-06,
+      "loss": 0.1799,
+      "step": 24540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8537669257641252e-06,
+      "loss": 0.1852,
+      "step": 24570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8520254433747156e-06,
+      "loss": 0.1932,
+      "step": 24600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8502871452651733e-06,
+      "loss": 0.1782,
+      "step": 24630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8485520217432816e-06,
+      "loss": 0.1932,
+      "step": 24660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8468200631580763e-06,
+      "loss": 0.1474,
+      "step": 24690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8450912598996183e-06,
+      "loss": 0.1396,
+      "step": 24720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8433656023987705e-06,
+      "loss": 0.1527,
+      "step": 24750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841643081126973e-06,
+      "loss": 0.2015,
+      "step": 24780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8399236865960273e-06,
+      "loss": 0.1824,
+      "step": 24810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8382074093578704e-06,
+      "loss": 0.1834,
+      "step": 24840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.836494240004362e-06,
+      "loss": 0.1714,
+      "step": 24870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8347841691670646e-06,
+      "loss": 0.1673,
+      "step": 24900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.833077187517031e-06,
+      "loss": 0.1669,
+      "step": 24930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8313732857645867e-06,
+      "loss": 0.2079,
+      "step": 24960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.829672454659122e-06,
+      "loss": 0.1795,
+      "step": 24990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.827974684988878e-06,
+      "loss": 0.1923,
+      "step": 25020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.826279967580738e-06,
+      "loss": 0.1613,
+      "step": 25050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8245882933000197e-06,
+      "loss": 0.2025,
+      "step": 25080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8228996530502693e-06,
+      "loss": 0.2004,
+      "step": 25110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8212140377730524e-06,
+      "loss": 0.1627,
+      "step": 25140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.819531438447754e-06,
+      "loss": 0.1889,
+      "step": 25170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8178518460913747e-06,
+      "loss": 0.1689,
+      "step": 25200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8161752517583257e-06,
+      "loss": 0.1703,
+      "step": 25230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8145016465402352e-06,
+      "loss": 0.1798,
+      "step": 25260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.812831021565744e-06,
+      "loss": 0.2057,
+      "step": 25290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.811163368000311e-06,
+      "loss": 0.1988,
+      "step": 25320
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.809498677046015e-06,
+      "loss": 0.1747,
+      "step": 25350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.807836939941363e-06,
+      "loss": 0.1707,
+      "step": 25380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.806178147961091e-06,
+      "loss": 0.1702,
+      "step": 25410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8045222924159783e-06,
+      "loss": 0.1531,
+      "step": 25440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8028693646526505e-06,
+      "loss": 0.1773,
+      "step": 25470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8012193560533934e-06,
+      "loss": 0.2086,
+      "step": 25500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.799572258035962e-06,
+      "loss": 0.1826,
+      "step": 25530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.797928062053395e-06,
+      "loss": 0.1524,
+      "step": 25560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7962867595938266e-06,
+      "loss": 0.1954,
+      "step": 25590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.794648342180302e-06,
+      "loss": 0.1594,
+      "step": 25620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7930128013705946e-06,
+      "loss": 0.1492,
+      "step": 25650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.791380128757023e-06,
+      "loss": 0.1868,
+      "step": 25680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.789750315966267e-06,
+      "loss": 0.1852,
+      "step": 25710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7881233546591905e-06,
+      "loss": 0.1662,
+      "step": 25740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.786499236530662e-06,
+      "loss": 0.188,
+      "step": 25770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7848779533093735e-06,
+      "loss": 0.1579,
+      "step": 25800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.783259496757666e-06,
+      "loss": 0.1935,
+      "step": 25830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.7816438586713548e-06,
+      "loss": 0.1776,
+      "step": 25860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7800310308795516e-06,
+      "loss": 0.182,
+      "step": 25890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7784210052444933e-06,
+      "loss": 0.1845,
+      "step": 25920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.77681377366137e-06,
+      "loss": 0.1834,
+      "step": 25950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.775209328058151e-06,
+      "loss": 0.1745,
+      "step": 25980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7736076603954173e-06,
+      "loss": 0.1624,
+      "step": 26010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.772062014714772e-06,
+      "loss": 0.1656,
+      "step": 26040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7704657870071948e-06,
+      "loss": 0.1633,
+      "step": 26070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7689253850700936e-06,
+      "loss": 0.1738,
+      "step": 26100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7673345665931835e-06,
+      "loss": 0.1711,
+      "step": 26130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7657464868672066e-06,
+      "loss": 0.173,
+      "step": 26160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.764161138042791e-06,
+      "loss": 0.1626,
+      "step": 26190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.762578512302018e-06,
+      "loss": 0.182,
+      "step": 26220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.760998601858272e-06,
+      "loss": 0.1748,
+      "step": 26250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7594213989560703e-06,
+      "loss": 0.1944,
+      "step": 26280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7578468958709083e-06,
+      "loss": 0.1825,
+      "step": 26310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7562750849090968e-06,
+      "loss": 0.1798,
+      "step": 26340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7547059584076095e-06,
+      "loss": 0.1606,
+      "step": 26370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.753139508733921e-06,
+      "loss": 0.1856,
+      "step": 26400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7515757282858534e-06,
+      "loss": 0.1706,
+      "step": 26430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.75001460949142e-06,
+      "loss": 0.1915,
+      "step": 26460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7484561448086723e-06,
+      "loss": 0.195,
+      "step": 26490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.746900326725547e-06,
+      "loss": 0.1873,
+      "step": 26520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7453471477597134e-06,
+      "loss": 0.1907,
+      "step": 26550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.743796600458422e-06,
+      "loss": 0.1844,
+      "step": 26580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.742248677398357e-06,
+      "loss": 0.1556,
+      "step": 26610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7407033711854815e-06,
+      "loss": 0.1819,
+      "step": 26640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.739160674454895e-06,
+      "loss": 0.1591,
+      "step": 26670
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7376205798706813e-06,
+      "loss": 0.1474,
+      "step": 26700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7360830801257666e-06,
+      "loss": 0.1874,
+      "step": 26730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.73454816794177e-06,
+      "loss": 0.1631,
+      "step": 26760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7330158360688597e-06,
+      "loss": 0.1639,
+      "step": 26790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7314860772856113e-06,
+      "loss": 0.2042,
+      "step": 26820
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.729958884398861e-06,
+      "loss": 0.1806,
+      "step": 26850
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7284342502435685e-06,
+      "loss": 0.183,
+      "step": 26880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7269121676826703e-06,
+      "loss": 0.219,
+      "step": 26910
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7253926296069453e-06,
+      "loss": 0.1616,
+      "step": 26940
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.723875628934869e-06,
+      "loss": 0.1491,
+      "step": 26970
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7223611586124805e-06,
+      "loss": 0.2203,
+      "step": 27000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7208492116132414e-06,
+      "loss": 0.1466,
+      "step": 27030
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.719339780937899e-06,
+      "loss": 0.1725,
+      "step": 27060
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.717832859614352e-06,
+      "loss": 0.1656,
+      "step": 27090
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.716328440697516e-06,
+      "loss": 0.198,
+      "step": 27120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.714826517269184e-06,
+      "loss": 0.1988,
+      "step": 27150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7133270824378975e-06,
+      "loss": 0.1754,
+      "step": 27180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.711830129338813e-06,
+      "loss": 0.1788,
+      "step": 27210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7103356511335694e-06,
+      "loss": 0.1858,
+      "step": 27240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7088436410101547e-06,
+      "loss": 0.1464,
+      "step": 27270
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.70735409218278e-06,
+      "loss": 0.1739,
+      "step": 27300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7058669978917464e-06,
+      "loss": 0.1741,
+      "step": 27330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7043823514033168e-06,
+      "loss": 0.1454,
+      "step": 27360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7029001460095887e-06,
+      "loss": 0.1555,
+      "step": 27390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.701420375028368e-06,
+      "loss": 0.1698,
+      "step": 27420
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.6999430318030373e-06,
+      "loss": 0.1675,
+      "step": 27450
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.698468109702438e-06,
+      "loss": 0.1998,
+      "step": 27480
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.696995602120738e-06,
+      "loss": 0.2141,
+      "step": 27510
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.695525502477311e-06,
+      "loss": 0.1873,
+      "step": 27540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.6940578042166143e-06,
+      "loss": 0.1848,
+      "step": 27570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.692592500808063e-06,
+      "loss": 0.1503,
+      "step": 27600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.691129585745908e-06,
+      "loss": 0.1921,
+      "step": 27630
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.689669052549118e-06,
+      "loss": 0.1928,
+      "step": 27660
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.6882108947612555e-06,
+      "loss": 0.1454,
+      "step": 27690
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.6867551059503586e-06,
+      "loss": 0.1938,
+      "step": 27720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.685301679708821e-06,
+      "loss": 0.1702,
+      "step": 27750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6838506096532747e-06,
+      "loss": 0.161,
+      "step": 27780
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.682401889424471e-06,
+      "loss": 0.1699,
+      "step": 27810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6809555126871633e-06,
+      "loss": 0.1529,
+      "step": 27840
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6795114731299932e-06,
+      "loss": 0.1697,
+      "step": 27870
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6780697644653724e-06,
+      "loss": 0.1936,
+      "step": 27900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.676630380429367e-06,
+      "loss": 0.1625,
+      "step": 27930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.675193314781585e-06,
+      "loss": 0.1762,
+      "step": 27960
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.6737585613050653e-06,
+      "loss": 0.173,
+      "step": 27990
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.672326113806157e-06,
+      "loss": 0.181,
+      "step": 28020
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.670895966114414e-06,
+      "loss": 0.1551,
+      "step": 28050
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6694681120824823e-06,
+      "loss": 0.1743,
+      "step": 28080
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.668042545585986e-06,
+      "loss": 0.1613,
+      "step": 28110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6666192605234194e-06,
+      "loss": 0.1659,
+      "step": 28140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.665198250816037e-06,
+      "loss": 0.1945,
+      "step": 28170
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.663779510407744e-06,
+      "loss": 0.1741,
+      "step": 28200
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.662363033264988e-06,
+      "loss": 0.1546,
+      "step": 28230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6609488133766516e-06,
+      "loss": 0.1615,
+      "step": 28260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.659536844753944e-06,
+      "loss": 0.1553,
+      "step": 28290
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.6581271214302974e-06,
+      "loss": 0.1362,
+      "step": 28320
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.656719637461258e-06,
+      "loss": 0.1575,
+      "step": 28350
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.6553143869243816e-06,
+      "loss": 0.1635,
+      "step": 28380
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.653911363919129e-06,
+      "loss": 0.1524,
+      "step": 28410
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.652510562566764e-06,
+      "loss": 0.1629,
+      "step": 28440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.651111977010246e-06,
+      "loss": 0.1578,
+      "step": 28470
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.6497156014141305e-06,
+      "loss": 0.1817,
+      "step": 28500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.6483214299644663e-06,
+      "loss": 0.1375,
+      "step": 28530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6469294568686905e-06,
+      "loss": 0.1616,
+      "step": 28560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6455396763555318e-06,
+      "loss": 0.1864,
+      "step": 28590
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6441520826749096e-06,
+      "loss": 0.1538,
+      "step": 28620
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6427666700978293e-06,
+      "loss": 0.1763,
+      "step": 28650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.64138343291629e-06,
+      "loss": 0.171,
+      "step": 28680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6400023654431796e-06,
+      "loss": 0.1688,
+      "step": 28710
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6386234620121797e-06,
+      "loss": 0.1436,
+      "step": 28740
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6372467169776683e-06,
+      "loss": 0.1494,
+      "step": 28770
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6358721247146202e-06,
+      "loss": 0.1806,
+      "step": 28800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.634499679618514e-06,
+      "loss": 0.1994,
+      "step": 28830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6331293761052316e-06,
+      "loss": 0.1753,
+      "step": 28860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.631761208610969e-06,
+      "loss": 0.1863,
+      "step": 28890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.630395171592134e-06,
+      "loss": 0.1742,
+      "step": 28920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6290312595252576e-06,
+      "loss": 0.1663,
+      "step": 28950
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6276694669068985e-06,
+      "loss": 0.1601,
+      "step": 28980
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6263097882535484e-06,
+      "loss": 0.1424,
+      "step": 29010
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.6249522181015406e-06,
+      "loss": 0.1756,
+      "step": 29040
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.623596751006959e-06,
+      "loss": 0.1582,
+      "step": 29070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6222433815455445e-06,
+      "loss": 0.1803,
+      "step": 29100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6208921043126025e-06,
+      "loss": 0.1791,
+      "step": 29130
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6195429139229154e-06,
+      "loss": 0.1579,
+      "step": 29160
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.618195805010653e-06,
+      "loss": 0.1633,
+      "step": 29190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.616850772229277e-06,
+      "loss": 0.1399,
+      "step": 29220
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6155078102514603e-06,
+      "loss": 0.1731,
+      "step": 29250
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.6141669137689908e-06,
+      "loss": 0.1989,
+      "step": 29280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.612828077492688e-06,
+      "loss": 0.1508,
+      "step": 29310
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.611491296152313e-06,
+      "loss": 0.1514,
+      "step": 29340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6101565644964837e-06,
+      "loss": 0.1835,
+      "step": 29370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6088238772925843e-06,
+      "loss": 0.1823,
+      "step": 29400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.607493229326684e-06,
+      "loss": 0.1761,
+      "step": 29430
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.606164615403447e-06,
+      "loss": 0.1523,
+      "step": 29460
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6048380303460504e-06,
+      "loss": 0.167,
+      "step": 29490
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6035134689960962e-06,
+      "loss": 0.1705,
+      "step": 29520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.602190926213532e-06,
+      "loss": 0.185,
+      "step": 29550
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.6008703968765623e-06,
+      "loss": 0.1641,
+      "step": 29580
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.599551875881568e-06,
+      "loss": 0.1599,
+      "step": 29610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.598235358143022e-06,
+      "loss": 0.1412,
+      "step": 29640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5969208385934085e-06,
+      "loss": 0.1463,
+      "step": 29670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5956083121831394e-06,
+      "loss": 0.1586,
+      "step": 29700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5942977738804735e-06,
+      "loss": 0.1557,
+      "step": 29730
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5929892186714367e-06,
+      "loss": 0.157,
+      "step": 29760
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5916826415597386e-06,
+      "loss": 0.1673,
+      "step": 29790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.590378037566695e-06,
+      "loss": 0.1452,
+      "step": 29820
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.589075401731149e-06,
+      "loss": 0.1613,
+      "step": 29850
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.5877747291093864e-06,
+      "loss": 0.1588,
+      "step": 29880
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.586476014775063e-06,
+      "loss": 0.2077,
+      "step": 29910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.5851792538191226e-06,
+      "loss": 0.1625,
+      "step": 29940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.5838844413497223e-06,
+      "loss": 0.1911,
+      "step": 29970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.5825915724921503e-06,
+      "loss": 0.1794,
+      "step": 30000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.581300642388753e-06,
+      "loss": 0.148,
+      "step": 30030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.5800116461988566e-06,
+      "loss": 0.1765,
+      "step": 30060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.578724579098691e-06,
+      "loss": 0.1541,
+      "step": 30090
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.577439436281313e-06,
+      "loss": 0.1802,
+      "step": 30120
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.576156212956535e-06,
+      "loss": 0.181,
+      "step": 30150
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.574874904350842e-06,
+      "loss": 0.1844,
+      "step": 30180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.573595505707327e-06,
+      "loss": 0.1545,
+      "step": 30210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.5723180122856094e-06,
+      "loss": 0.1621,
+      "step": 30240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.5710424193617634e-06,
+      "loss": 0.1829,
+      "step": 30270
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.569768722228245e-06,
+      "loss": 0.1689,
+      "step": 30300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.56849691619382e-06,
+      "loss": 0.1457,
+      "step": 30330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.5672269965834895e-06,
+      "loss": 0.1635,
+      "step": 30360
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.5659589587384176e-06,
+      "loss": 0.1363,
+      "step": 30390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.564692798015863e-06,
+      "loss": 0.161,
+      "step": 30420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.5634285097891025e-06,
+      "loss": 0.1627,
+      "step": 30450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.5621660894473655e-06,
+      "loss": 0.1606,
+      "step": 30480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.560905532395757e-06,
+      "loss": 0.1622,
+      "step": 30510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.559646834055194e-06,
+      "loss": 0.1622,
+      "step": 30540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.5583899898623316e-06,
+      "loss": 0.1603,
+      "step": 30570
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.557134995269493e-06,
+      "loss": 0.1517,
+      "step": 30600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.5558818457446043e-06,
+      "loss": 0.169,
+      "step": 30630
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.554630536771122e-06,
+      "loss": 0.1671,
+      "step": 30660
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.553381063847966e-06,
+      "loss": 0.1823,
+      "step": 30690
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5521334224894508e-06,
+      "loss": 0.1486,
+      "step": 30720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5508876082252192e-06,
+      "loss": 0.1592,
+      "step": 30750
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5496436166001757e-06,
+      "loss": 0.1751,
+      "step": 30780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5484014431744175e-06,
+      "loss": 0.1687,
+      "step": 30810
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5471610835231664e-06,
+      "loss": 0.1749,
+      "step": 30840
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5459225332367084e-06,
+      "loss": 0.1721,
+      "step": 30870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5446857879203222e-06,
+      "loss": 0.1674,
+      "step": 30900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5434508431942174e-06,
+      "loss": 0.1674,
+      "step": 30930
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.5422176946934667e-06,
+      "loss": 0.1527,
+      "step": 30960
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5409863380679436e-06,
+      "loss": 0.1803,
+      "step": 30990
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.539756768982254e-06,
+      "loss": 0.1536,
+      "step": 31020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5385289831156773e-06,
+      "loss": 0.1405,
+      "step": 31050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5373029761620997e-06,
+      "loss": 0.1691,
+      "step": 31080
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5360787438299507e-06,
+      "loss": 0.1617,
+      "step": 31110
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5348562818421386e-06,
+      "loss": 0.158,
+      "step": 31140
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5336355859359934e-06,
+      "loss": 0.1632,
+      "step": 31170
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.5324166518631965e-06,
+      "loss": 0.175,
+      "step": 31200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.531199475389726e-06,
+      "loss": 0.1579,
+      "step": 31230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5299840522957895e-06,
+      "loss": 0.157,
+      "step": 31260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5287703783757656e-06,
+      "loss": 0.1626,
+      "step": 31290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5275584494381406e-06,
+      "loss": 0.1753,
+      "step": 31320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5263482613054506e-06,
+      "loss": 0.1429,
+      "step": 31350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.525139809814219e-06,
+      "loss": 0.1551,
+      "step": 31380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.523933090814896e-06,
+      "loss": 0.1516,
+      "step": 31410
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5227281001718014e-06,
+      "loss": 0.1365,
+      "step": 31440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.52152483376306e-06,
+      "loss": 0.1482,
+      "step": 31470
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.5203633113534356e-06,
+      "loss": 0.1979,
+      "step": 31500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.519163423967481e-06,
+      "loss": 0.1598,
+      "step": 31530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.5179652486683264e-06,
+      "loss": 0.153,
+      "step": 31560
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.5167687813882913e-06,
+      "loss": 0.1523,
+      "step": 31590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.5155740180732113e-06,
+      "loss": 0.1552,
+      "step": 31620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.514380954682381e-06,
+      "loss": 0.1824,
+      "step": 31650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.5131895871884982e-06,
+      "loss": 0.2012,
+      "step": 31680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.511999911577605e-06,
+      "loss": 0.2118,
+      "step": 31710
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.5108119238490314e-06,
+      "loss": 0.1636,
+      "step": 31740
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.50962562001534e-06,
+      "loss": 0.153,
+      "step": 31770
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.508440996102268e-06,
+      "loss": 0.1606,
+      "step": 31800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.507258048148672e-06,
+      "loss": 0.1909,
+      "step": 31830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.506076772206472e-06,
+      "loss": 0.1504,
+      "step": 31860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.5048971643405985e-06,
+      "loss": 0.1589,
+      "step": 31890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.5037192206289322e-06,
+      "loss": 0.1534,
+      "step": 31920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.5025429371622568e-06,
+      "loss": 0.1991,
+      "step": 31950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.5013683100441947e-06,
+      "loss": 0.1461,
+      "step": 31980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.5001953353911644e-06,
+      "loss": 0.1715,
+      "step": 32010
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.4990240093323155e-06,
+      "loss": 0.1519,
+      "step": 32040
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4978543280094823e-06,
+      "loss": 0.1527,
+      "step": 32070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4966862875771277e-06,
+      "loss": 0.1782,
+      "step": 32100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4955198842022903e-06,
+      "loss": 0.1457,
+      "step": 32130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.494355114064532e-06,
+      "loss": 0.1619,
+      "step": 32160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.493191973355886e-06,
+      "loss": 0.1677,
+      "step": 32190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4920304582808026e-06,
+      "loss": 0.185,
+      "step": 32220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.4908705650560973e-06,
+      "loss": 0.1631,
+      "step": 32250
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.489712289910904e-06,
+      "loss": 0.1768,
+      "step": 32280
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.488555629086615e-06,
+      "loss": 0.1695,
+      "step": 32310
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.487400578836838e-06,
+      "loss": 0.1578,
+      "step": 32340
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.4862471354273387e-06,
+      "loss": 0.1481,
+      "step": 32370
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.4851336640235753e-06,
+      "loss": 0.17,
+      "step": 32400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.483983369886367e-06,
+      "loss": 0.1522,
+      "step": 32430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.482834671582272e-06,
+      "loss": 0.1716,
+      "step": 32460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.481687565424798e-06,
+      "loss": 0.1523,
+      "step": 32490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.4805420477393657e-06,
+      "loss": 0.1714,
+      "step": 32520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.479398114863255e-06,
+      "loss": 0.1585,
+      "step": 32550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.4782557631455603e-06,
+      "loss": 0.1632,
+      "step": 32580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.47711498894714e-06,
+      "loss": 0.1372,
+      "step": 32610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.4759757886405678e-06,
+      "loss": 0.1625,
+      "step": 32640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.474838158610084e-06,
+      "loss": 0.15,
+      "step": 32670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.4737020952515466e-06,
+      "loss": 0.1996,
+      "step": 32700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.4725675949723856e-06,
+      "loss": 0.1648,
+      "step": 32730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.471434654191553e-06,
+      "loss": 0.1608,
+      "step": 32760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.4703032693394766e-06,
+      "loss": 0.1501,
+      "step": 32790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.4691734368580124e-06,
+      "loss": 0.2002,
+      "step": 32820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.4680451532003975e-06,
+      "loss": 0.1717,
+      "step": 32850
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.4669184148312046e-06,
+      "loss": 0.1575,
+      "step": 32880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.465793218226291e-06,
+      "loss": 0.191,
+      "step": 32910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.4646695598727596e-06,
+      "loss": 0.156,
+      "step": 32940
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.463547436268907e-06,
+      "loss": 0.1589,
+      "step": 32970
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.4624268439241786e-06,
+      "loss": 0.1618,
+      "step": 33000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.4613077793591255e-06,
+      "loss": 0.1438,
+      "step": 33030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.460190239105358e-06,
+      "loss": 0.171,
+      "step": 33060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.459074219705497e-06,
+      "loss": 0.1617,
+      "step": 33090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.457959717713135e-06,
+      "loss": 0.1846,
+      "step": 33120
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.456846729692788e-06,
+      "loss": 0.1812,
+      "step": 33150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.4557352522198512e-06,
+      "loss": 0.1568,
+      "step": 33180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.454625281880554e-06,
+      "loss": 0.1354,
+      "step": 33210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.4535168152719188e-06,
+      "loss": 0.185,
+      "step": 33240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.4524098490017136e-06,
+      "loss": 0.1512,
+      "step": 33270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.4513043796884104e-06,
+      "loss": 0.1976,
+      "step": 33300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.4502004039611437e-06,
+      "loss": 0.1595,
+      "step": 33330
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.4490979184596634e-06,
+      "loss": 0.1411,
+      "step": 33360
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.447996919834293e-06,
+      "loss": 0.131,
+      "step": 33390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.44689740474589e-06,
+      "loss": 0.1429,
+      "step": 33420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.445799369865799e-06,
+      "loss": 0.1504,
+      "step": 33450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.4447028118758115e-06,
+      "loss": 0.1438,
+      "step": 33480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.443607727468123e-06,
+      "loss": 0.1468,
+      "step": 33510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.4425141133452936e-06,
+      "loss": 0.1323,
+      "step": 33540
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.441421966220202e-06,
+      "loss": 0.1852,
+      "step": 33570
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.440331282816008e-06,
+      "loss": 0.1523,
+      "step": 33600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.4392420598661103e-06,
+      "loss": 0.1491,
+      "step": 33630
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.4381542941141006e-06,
+      "loss": 0.1689,
+      "step": 33660
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.4370679823137313e-06,
+      "loss": 0.1446,
+      "step": 33690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.4359831212288684e-06,
+      "loss": 0.1563,
+      "step": 33720
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.434899707633453e-06,
+      "loss": 0.1501,
+      "step": 33750
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.4338177383114604e-06,
+      "loss": 0.1654,
+      "step": 33780
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.432737210056862e-06,
+      "loss": 0.1693,
+      "step": 33810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.4316581196735815e-06,
+      "loss": 0.16,
+      "step": 33840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.4305804639754597e-06,
+      "loss": 0.1636,
+      "step": 33870
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.4295042397862126e-06,
+      "loss": 0.1367,
+      "step": 33900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.428429443939391e-06,
+      "loss": 0.1561,
+      "step": 33930
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.427356073278345e-06,
+      "loss": 0.1421,
+      "step": 33960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.42628412465618e-06,
+      "loss": 0.1553,
+      "step": 33990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.425213594935723e-06,
+      "loss": 0.2056,
+      "step": 34020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.424144480989481e-06,
+      "loss": 0.1346,
+      "step": 34050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.423076779699603e-06,
+      "loss": 0.1716,
+      "step": 34080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.422010487957844e-06,
+      "loss": 0.1501,
+      "step": 34110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.420945602665522e-06,
+      "loss": 0.1449,
+      "step": 34140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.419882120733486e-06,
+      "loss": 0.1514,
+      "step": 34170
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.4188200390820758e-06,
+      "loss": 0.1622,
+      "step": 34200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.4177593546410825e-06,
+      "loss": 0.1641,
+      "step": 34230
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.4167000643497152e-06,
+      "loss": 0.1417,
+      "step": 34260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.4156421651565615e-06,
+      "loss": 0.1399,
+      "step": 34290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.4145856540195495e-06,
+      "loss": 0.1595,
+      "step": 34320
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.413530527905915e-06,
+      "loss": 0.155,
+      "step": 34350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.4124767837921604e-06,
+      "loss": 0.1723,
+      "step": 34380
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.411424418664022e-06,
+      "loss": 0.1505,
+      "step": 34410
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.4103734295164312e-06,
+      "loss": 0.1443,
+      "step": 34440
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.40932381335348e-06,
+      "loss": 0.1555,
+      "step": 34470
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.408275567188384e-06,
+      "loss": 0.1472,
+      "step": 34500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.407228688043447e-06,
+      "loss": 0.1674,
+      "step": 34530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.4061831729500282e-06,
+      "loss": 0.1735,
+      "step": 34560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.4051390189485015e-06,
+      "loss": 0.1368,
+      "step": 34590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.404096223088225e-06,
+      "loss": 0.1499,
+      "step": 34620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.403054782427503e-06,
+      "loss": 0.171,
+      "step": 34650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.4020146940335533e-06,
+      "loss": 0.1371,
+      "step": 34680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.400975954982471e-06,
+      "loss": 0.1348,
+      "step": 34710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3999385623591958e-06,
+      "loss": 0.1607,
+      "step": 34740
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3989025132574736e-06,
+      "loss": 0.1845,
+      "step": 34770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3978678047798286e-06,
+      "loss": 0.1799,
+      "step": 34800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.396834434037523e-06,
+      "loss": 0.1409,
+      "step": 34830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3958023981505267e-06,
+      "loss": 0.1556,
+      "step": 34860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3947716942474835e-06,
+      "loss": 0.161,
+      "step": 34890
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3937423194656766e-06,
+      "loss": 0.1834,
+      "step": 34920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3927142709509947e-06,
+      "loss": 0.1527,
+      "step": 34950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.3916875458579e-06,
+      "loss": 0.1612,
+      "step": 34980
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.3906621413493943e-06,
+      "loss": 0.1413,
+      "step": 35010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.3896380545969873e-06,
+      "loss": 0.2112,
+      "step": 35040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.388615282780661e-06,
+      "loss": 0.1706,
+      "step": 35070
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.3875938230888414e-06,
+      "loss": 0.1461,
+      "step": 35100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.386573672718362e-06,
+      "loss": 0.1477,
+      "step": 35130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.385554828874434e-06,
+      "loss": 0.1657,
+      "step": 35160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.384537288770612e-06,
+      "loss": 0.1558,
+      "step": 35190
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.3835210496287646e-06,
+      "loss": 0.1707,
+      "step": 35220
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.3825061086790407e-06,
+      "loss": 0.1314,
+      "step": 35250
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3814924631598384e-06,
+      "loss": 0.1539,
+      "step": 35280
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3804801103177737e-06,
+      "loss": 0.2101,
+      "step": 35310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3794690474076476e-06,
+      "loss": 0.2028,
+      "step": 35340
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3784592716924168e-06,
+      "loss": 0.1903,
+      "step": 35370
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.377450780443162e-06,
+      "loss": 0.1499,
+      "step": 35400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3764435709390556e-06,
+      "loss": 0.1607,
+      "step": 35430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3754376404673334e-06,
+      "loss": 0.1493,
+      "step": 35460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3744329863232613e-06,
+      "loss": 0.1685,
+      "step": 35490
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.3734296058101067e-06,
+      "loss": 0.1396,
+      "step": 35520
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.372427496239106e-06,
+      "loss": 0.148,
+      "step": 35550
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.371426654929437e-06,
+      "loss": 0.1632,
+      "step": 35580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.3704270792081874e-06,
+      "loss": 0.1703,
+      "step": 35610
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.3694287664103236e-06,
+      "loss": 0.1319,
+      "step": 35640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.368431713878663e-06,
+      "loss": 0.1504,
+      "step": 35670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.3674359189638443e-06,
+      "loss": 0.1268,
+      "step": 35700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.3664413790242944e-06,
+      "loss": 0.2007,
+      "step": 35730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.3654480914262044e-06,
+      "loss": 0.1817,
+      "step": 35760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.3644560535434954e-06,
+      "loss": 0.1459,
+      "step": 35790
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.3634652627577933e-06,
+      "loss": 0.1827,
+      "step": 35820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.3624757164583963e-06,
+      "loss": 0.181,
+      "step": 35850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.3614874120422486e-06,
+      "loss": 0.1485,
+      "step": 35880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.36050034691391e-06,
+      "loss": 0.1573,
+      "step": 35910
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.3595145184855286e-06,
+      "loss": 0.15,
+      "step": 35940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.35852992417681e-06,
+      "loss": 0.2165,
+      "step": 35970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.357546561414994e-06,
+      "loss": 0.1448,
+      "step": 36000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.3565644276348197e-06,
+      "loss": 0.1556,
+      "step": 36030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.3555835202785018e-06,
+      "loss": 0.1416,
+      "step": 36060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.354603836795701e-06,
+      "loss": 0.1705,
+      "step": 36090
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3536253746434974e-06,
+      "loss": 0.1615,
+      "step": 36120
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.352648131286362e-06,
+      "loss": 0.1561,
+      "step": 36150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3516721041961274e-06,
+      "loss": 0.1637,
+      "step": 36180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3506972908519647e-06,
+      "loss": 0.1616,
+      "step": 36210
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3497236887403514e-06,
+      "loss": 0.1653,
+      "step": 36240
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.348751295355046e-06,
+      "loss": 0.1632,
+      "step": 36270
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.347780108197064e-06,
+      "loss": 0.1437,
+      "step": 36300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.3468101247746447e-06,
+      "loss": 0.1745,
+      "step": 36330
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3458413426032292e-06,
+      "loss": 0.1505,
+      "step": 36360
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3448737592054326e-06,
+      "loss": 0.1477,
+      "step": 36390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3439073721110166e-06,
+      "loss": 0.1484,
+      "step": 36420
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3429421788568627e-06,
+      "loss": 0.149,
+      "step": 36450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3419781769869478e-06,
+      "loss": 0.138,
+      "step": 36480
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.341015364052316e-06,
+      "loss": 0.1717,
+      "step": 36510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3400537376110534e-06,
+      "loss": 0.1377,
+      "step": 36540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.3390932952282607e-06,
+      "loss": 0.1655,
+      "step": 36570
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.33813403447603e-06,
+      "loss": 0.1842,
+      "step": 36600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.3371759529334183e-06,
+      "loss": 0.1865,
+      "step": 36630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.336219048186418e-06,
+      "loss": 0.1335,
+      "step": 36660
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.3352633178279355e-06,
+      "loss": 0.1753,
+      "step": 36690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.3343087594577666e-06,
+      "loss": 0.176,
+      "step": 36720
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.3333553706825667e-06,
+      "loss": 0.1251,
+      "step": 36750
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.33240314911583e-06,
+      "loss": 0.162,
+      "step": 36780
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.3314520923778613e-06,
+      "loss": 0.144,
+      "step": 36810
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.330533842535043e-06,
+      "loss": 0.1405,
+      "step": 36840
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.329585069711041e-06,
+      "loss": 0.1436,
+      "step": 36870
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.328637454695883e-06,
+      "loss": 0.1458,
+      "step": 36900
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.32769099513665e-06,
+      "loss": 0.1645,
+      "step": 36930
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.32674568868711e-06,
+      "loss": 0.1415,
+      "step": 36960
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.3258015330076976e-06,
+      "loss": 0.126,
+      "step": 36990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.324858525765487e-06,
+      "loss": 0.1576,
+      "step": 37020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.3239166646341674e-06,
+      "loss": 0.143,
+      "step": 37050
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.322975947294021e-06,
+      "loss": 0.1905,
+      "step": 37080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.3220363714318965e-06,
+      "loss": 0.1584,
+      "step": 37110
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.3210979347411877e-06,
+      "loss": 0.1583,
+      "step": 37140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.320160634921808e-06,
+      "loss": 0.1629,
+      "step": 37170
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.3192244696801673e-06,
+      "loss": 0.1418,
+      "step": 37200
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.318289436729148e-06,
+      "loss": 0.1481,
+      "step": 37230
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.3173555337880833e-06,
+      "loss": 0.1596,
+      "step": 37260
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.3164227585827304e-06,
+      "loss": 0.1332,
+      "step": 37290
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.3154911088452513e-06,
+      "loss": 0.1482,
+      "step": 37320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.314560582314186e-06,
+      "loss": 0.1413,
+      "step": 37350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.313631176734432e-06,
+      "loss": 0.1742,
+      "step": 37380
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.3127028898572203e-06,
+      "loss": 0.1686,
+      "step": 37410
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.311775719440093e-06,
+      "loss": 0.1497,
+      "step": 37440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.310849663246879e-06,
+      "loss": 0.1518,
+      "step": 37470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.309924719047674e-06,
+      "loss": 0.1568,
+      "step": 37500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.3090008846188165e-06,
+      "loss": 0.1332,
+      "step": 37530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.308078157742863e-06,
+      "loss": 0.1531,
+      "step": 37560
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.307156536208571e-06,
+      "loss": 0.1472,
+      "step": 37590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.3062360178108724e-06,
+      "loss": 0.1369,
+      "step": 37620
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.3053166003508527e-06,
+      "loss": 0.1482,
+      "step": 37650
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.30439828163573e-06,
+      "loss": 0.149,
+      "step": 37680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.3034810594788295e-06,
+      "loss": 0.1617,
+      "step": 37710
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.3025649316995668e-06,
+      "loss": 0.1573,
+      "step": 37740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.3016498961234214e-06,
+      "loss": 0.1497,
+      "step": 37770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.300735950581918e-06,
+      "loss": 0.1744,
+      "step": 37800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2998230929126046e-06,
+      "loss": 0.139,
+      "step": 37830
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.29891132095903e-06,
+      "loss": 0.1647,
+      "step": 37860
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2980006325707215e-06,
+      "loss": 0.1512,
+      "step": 37890
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.2970910256031658e-06,
+      "loss": 0.1802,
+      "step": 37920
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.296182497917788e-06,
+      "loss": 0.1654,
+      "step": 37950
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2952750473819264e-06,
+      "loss": 0.1639,
+      "step": 37980
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2943686718688178e-06,
+      "loss": 0.153,
+      "step": 38010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.293463369257572e-06,
+      "loss": 0.1656,
+      "step": 38040
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2925591374331504e-06,
+      "loss": 0.1373,
+      "step": 38070
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.291655974286349e-06,
+      "loss": 0.1618,
+      "step": 38100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2907538777137752e-06,
+      "loss": 0.1632,
+      "step": 38130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.2898528456178286e-06,
+      "loss": 0.148,
+      "step": 38160
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.288952875906677e-06,
+      "loss": 0.1529,
+      "step": 38190
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2880539664942414e-06,
+      "loss": 0.1352,
+      "step": 38220
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.287156115300173e-06,
+      "loss": 0.1447,
+      "step": 38250
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2862593202498297e-06,
+      "loss": 0.194,
+      "step": 38280
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2853635792742636e-06,
+      "loss": 0.1346,
+      "step": 38310
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.284468890310193e-06,
+      "loss": 0.1342,
+      "step": 38340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2835752512999883e-06,
+      "loss": 0.1408,
+      "step": 38370
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.282682660191648e-06,
+      "loss": 0.1545,
+      "step": 38400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2817911149387828e-06,
+      "loss": 0.1597,
+      "step": 38430
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.2809006135005923e-06,
+      "loss": 0.1433,
+      "step": 38460
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.280011153841847e-06,
+      "loss": 0.14,
+      "step": 38490
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.2791227339328693e-06,
+      "loss": 0.1647,
+      "step": 38520
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.2782353517495133e-06,
+      "loss": 0.1659,
+      "step": 38550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.2773490052731452e-06,
+      "loss": 0.1298,
+      "step": 38580
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.2764636924906245e-06,
+      "loss": 0.1773,
+      "step": 38610
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.27560887083038e-06,
+      "loss": 0.1642,
+      "step": 38640
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.2747255851274107e-06,
+      "loss": 0.1638,
+      "step": 38670
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.2738433271781068e-06,
+      "loss": 0.1954,
+      "step": 38700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.272962094990928e-06,
+      "loss": 0.1598,
+      "step": 38730
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.272081886579736e-06,
+      "loss": 0.1741,
+      "step": 38760
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2712026999637678e-06,
+      "loss": 0.1589,
+      "step": 38790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.270324533167624e-06,
+      "loss": 0.1637,
+      "step": 38820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2694473842212473e-06,
+      "loss": 0.197,
+      "step": 38850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2685712511599043e-06,
+      "loss": 0.1334,
+      "step": 38880
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2676961320241665e-06,
+      "loss": 0.1649,
+      "step": 38910
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.266822024859894e-06,
+      "loss": 0.1752,
+      "step": 38940
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2659489277182127e-06,
+      "loss": 0.1423,
+      "step": 38970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.2650768386555003e-06,
+      "loss": 0.172,
+      "step": 39000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.264205755733369e-06,
+      "loss": 0.1372,
+      "step": 39030
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.2633356770186404e-06,
+      "loss": 0.1646,
+      "step": 39060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.2624666005833367e-06,
+      "loss": 0.1595,
+      "step": 39090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.2615985245046557e-06,
+      "loss": 0.166,
+      "step": 39120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.2607314468649563e-06,
+      "loss": 0.1402,
+      "step": 39150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.25986536575174e-06,
+      "loss": 0.1519,
+      "step": 39180
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.259000279257632e-06,
+      "loss": 0.1585,
+      "step": 39210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.2581361854803667e-06,
+      "loss": 0.1605,
+      "step": 39240
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.2572730825227662e-06,
+      "loss": 0.139,
+      "step": 39270
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.256410968492726e-06,
+      "loss": 0.1307,
+      "step": 39300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2555498415031953e-06,
+      "loss": 0.1381,
+      "step": 39330
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2546896996721607e-06,
+      "loss": 0.1494,
+      "step": 39360
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2538305411226297e-06,
+      "loss": 0.1582,
+      "step": 39390
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2529723639826117e-06,
+      "loss": 0.1613,
+      "step": 39420
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2521151663851024e-06,
+      "loss": 0.166,
+      "step": 39450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.251258946468066e-06,
+      "loss": 0.1548,
+      "step": 39480
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2504037023744184e-06,
+      "loss": 0.1402,
+      "step": 39510
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.2495494322520092e-06,
+      "loss": 0.1515,
+      "step": 39540
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.248696134253608e-06,
+      "loss": 0.1395,
+      "step": 39570
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.247843806536884e-06,
+      "loss": 0.1625,
+      "step": 39600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.2469924472643904e-06,
+      "loss": 0.147,
+      "step": 39630
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.2461420546035494e-06,
+      "loss": 0.1336,
+      "step": 39660
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.245292626726635e-06,
+      "loss": 0.1217,
+      "step": 39690
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.2444441618107533e-06,
+      "loss": 0.1453,
+      "step": 39720
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.243596658037831e-06,
+      "loss": 0.1382,
+      "step": 39750
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.2427501135945952e-06,
+      "loss": 0.1632,
+      "step": 39780
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.2419045266725605e-06,
+      "loss": 0.1615,
+      "step": 39810
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.241059895468009e-06,
+      "loss": 0.1506,
+      "step": 39840
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.2402162181819777e-06,
+      "loss": 0.1484,
+      "step": 39870
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.2393734930202386e-06,
+      "loss": 0.1555,
+      "step": 39900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.238531718193287e-06,
+      "loss": 0.1427,
+      "step": 39930
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.2376908919163214e-06,
+      "loss": 0.1651,
+      "step": 39960
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.236851012409231e-06,
+      "loss": 0.1755,
+      "step": 39990
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.236012077896579e-06,
+      "loss": 0.1369,
+      "step": 40020
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.235174086607584e-06,
+      "loss": 0.175,
+      "step": 40050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.2343370367761084e-06,
+      "loss": 0.1441,
+      "step": 40080
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2335009266406403e-06,
+      "loss": 0.1731,
+      "step": 40110
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2326657544442797e-06,
+      "loss": 0.1675,
+      "step": 40140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.231831518434719e-06,
+      "loss": 0.1612,
+      "step": 40170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2309982168642326e-06,
+      "loss": 0.1538,
+      "step": 40200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2301658479896607e-06,
+      "loss": 0.1631,
+      "step": 40230
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.229334410072389e-06,
+      "loss": 0.155,
+      "step": 40260
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2285039013783404e-06,
+      "loss": 0.135,
+      "step": 40290
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2276743201779543e-06,
+      "loss": 0.1491,
+      "step": 40320
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.2268456647461743e-06,
+      "loss": 0.1364,
+      "step": 40350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.226017933362433e-06,
+      "loss": 0.1611,
+      "step": 40380
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2251911243106365e-06,
+      "loss": 0.1172,
+      "step": 40410
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.224365235879149e-06,
+      "loss": 0.1509,
+      "step": 40440
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2235402663607773e-06,
+      "loss": 0.1661,
+      "step": 40470
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2227162140527596e-06,
+      "loss": 0.15,
+      "step": 40500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2218930772567463e-06,
+      "loss": 0.1425,
+      "step": 40530
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.221070854278788e-06,
+      "loss": 0.1721,
+      "step": 40560
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.2202495434293184e-06,
+      "loss": 0.1377,
+      "step": 40590
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.219429143023144e-06,
+      "loss": 0.1455,
+      "step": 40620
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2186096513794256e-06,
+      "loss": 0.1348,
+      "step": 40650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2177910668216645e-06,
+      "loss": 0.1614,
+      "step": 40680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2169733876776897e-06,
+      "loss": 0.1431,
+      "step": 40710
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2161566122796414e-06,
+      "loss": 0.1521,
+      "step": 40740
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.2153407389639607e-06,
+      "loss": 0.175,
+      "step": 40770
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.21452576607137e-06,
+      "loss": 0.1389,
+      "step": 40800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.213711691946862e-06,
+      "loss": 0.1679,
+      "step": 40830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.212898514939685e-06,
+      "loss": 0.1289,
+      "step": 40860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.212086233403329e-06,
+      "loss": 0.1763,
+      "step": 40890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2112748456955123e-06,
+      "loss": 0.1648,
+      "step": 40920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2104643501781657e-06,
+      "loss": 0.1724,
+      "step": 40950
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.20965474521742e-06,
+      "loss": 0.1341,
+      "step": 40980
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.208846029183593e-06,
+      "loss": 0.155,
+      "step": 41010
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2080382004511725e-06,
+      "loss": 0.1591,
+      "step": 41040
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.207231257398806e-06,
+      "loss": 0.1678,
+      "step": 41070
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2064251984092854e-06,
+      "loss": 0.145,
+      "step": 41100
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.2056200218695346e-06,
+      "loss": 0.1617,
+      "step": 41130
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2048157261705934e-06,
+      "loss": 0.1493,
+      "step": 41160
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.204012309707607e-06,
+      "loss": 0.1345,
+      "step": 41190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.203209770879811e-06,
+      "loss": 0.1389,
+      "step": 41220
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.202408108090517e-06,
+      "loss": 0.1593,
+      "step": 41250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2016073197471015e-06,
+      "loss": 0.1582,
+      "step": 41280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.200807404260992e-06,
+      "loss": 0.1529,
+      "step": 41310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.2000083600476527e-06,
+      "loss": 0.1601,
+      "step": 41340
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.199210185526571e-06,
+      "loss": 0.1604,
+      "step": 41370
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.198412879121248e-06,
+      "loss": 0.1478,
+      "step": 41400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.1976164392591793e-06,
+      "loss": 0.1689,
+      "step": 41430
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.1968208643718476e-06,
+      "loss": 0.1644,
+      "step": 41460
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.196026152894708e-06,
+      "loss": 0.1742,
+      "step": 41490
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.195232303267173e-06,
+      "loss": 0.1596,
+      "step": 41520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.194439313932602e-06,
+      "loss": 0.1316,
+      "step": 41550
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.193647183338287e-06,
+      "loss": 0.1481,
+      "step": 41580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.1928559099354423e-06,
+      "loss": 0.1735,
+      "step": 41610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.1920654921791885e-06,
+      "loss": 0.1617,
+      "step": 41640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.191275928528542e-06,
+      "loss": 0.15,
+      "step": 41670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.190487217446401e-06,
+      "loss": 0.1473,
+      "step": 41700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.189699357399535e-06,
+      "loss": 0.1206,
+      "step": 41730
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.188912346858569e-06,
+      "loss": 0.1352,
+      "step": 41760
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.188126184297976e-06,
+      "loss": 0.173,
+      "step": 41790
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.1873408681960583e-06,
+      "loss": 0.1624,
+      "step": 41820
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.1865563970349406e-06,
+      "loss": 0.1448,
+      "step": 41850
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.185772769300554e-06,
+      "loss": 0.1501,
+      "step": 41880
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.1849899834826275e-06,
+      "loss": 0.1603,
+      "step": 41910
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.184208038074671e-06,
+      "loss": 0.1447,
+      "step": 41940
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1834269315739657e-06,
+      "loss": 0.1551,
+      "step": 41970
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.182646662481554e-06,
+      "loss": 0.1428,
+      "step": 42000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1818672293022237e-06,
+      "loss": 0.1594,
+      "step": 42030
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1810886305444976e-06,
+      "loss": 0.1461,
+      "step": 42060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.180310864720622e-06,
+      "loss": 0.1699,
+      "step": 42090
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.1795339303465547e-06,
+      "loss": 0.158,
+      "step": 42120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.178757825941951e-06,
+      "loss": 0.1227,
+      "step": 42150
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.177982550030154e-06,
+      "loss": 0.1795,
+      "step": 42180
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.177208101138184e-06,
+      "loss": 0.1337,
+      "step": 42210
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1764344777967235e-06,
+      "loss": 0.1626,
+      "step": 42240
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1756616785401066e-06,
+      "loss": 0.129,
+      "step": 42270
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1748897019063095e-06,
+      "loss": 0.1494,
+      "step": 42300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1741185464369354e-06,
+      "loss": 0.1537,
+      "step": 42330
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1733482106772056e-06,
+      "loss": 0.1555,
+      "step": 42360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1725786931759464e-06,
+      "loss": 0.1368,
+      "step": 42390
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1718099924855798e-06,
+      "loss": 0.1489,
+      "step": 42420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.171042107162109e-06,
+      "loss": 0.1486,
+      "step": 42450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.1702750357651087e-06,
+      "loss": 0.18,
+      "step": 42480
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1695087768577147e-06,
+      "loss": 0.1532,
+      "step": 42510
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.16874332900661e-06,
+      "loss": 0.147,
+      "step": 42540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1679786907820158e-06,
+      "loss": 0.1383,
+      "step": 42570
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1672148607576797e-06,
+      "loss": 0.1606,
+      "step": 42600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1664518375108646e-06,
+      "loss": 0.1814,
+      "step": 42630
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1656896196223354e-06,
+      "loss": 0.1362,
+      "step": 42660
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1649282056763532e-06,
+      "loss": 0.1501,
+      "step": 42690
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1641675942606574e-06,
+      "loss": 0.1471,
+      "step": 42720
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.1634077839664602e-06,
+      "loss": 0.1501,
+      "step": 42750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.162648773388433e-06,
+      "loss": 0.1738,
+      "step": 42780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.161890561124696e-06,
+      "loss": 0.1404,
+      "step": 42810
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.161133145776808e-06,
+      "loss": 0.2075,
+      "step": 42840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.160376525949755e-06,
+      "loss": 0.151,
+      "step": 42870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1596207002519383e-06,
+      "loss": 0.1608,
+      "step": 42900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.158865667295166e-06,
+      "loss": 0.1798,
+      "step": 42930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1581114256946403e-06,
+      "loss": 0.1514,
+      "step": 42960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1573579740689486e-06,
+      "loss": 0.1378,
+      "step": 42990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.1566053110400517e-06,
+      "loss": 0.1434,
+      "step": 43020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1558534352332726e-06,
+      "loss": 0.1566,
+      "step": 43050
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1551023452772875e-06,
+      "loss": 0.1509,
+      "step": 43080
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.154352039804115e-06,
+      "loss": 0.137,
+      "step": 43110
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1536025174491042e-06,
+      "loss": 0.1861,
+      "step": 43140
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1528537768509265e-06,
+      "loss": 0.1358,
+      "step": 43170
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1521058166515628e-06,
+      "loss": 0.1538,
+      "step": 43200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1513586354962963e-06,
+      "loss": 0.1644,
+      "step": 43230
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1506122320336976e-06,
+      "loss": 0.1482,
+      "step": 43260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.1498666049156193e-06,
+      "loss": 0.1492,
+      "step": 43290
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.149121752797183e-06,
+      "loss": 0.1353,
+      "step": 43320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.148377674336769e-06,
+      "loss": 0.1671,
+      "step": 43350
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1476343681960082e-06,
+      "loss": 0.1606,
+      "step": 43380
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.1468918330397695e-06,
+      "loss": 0.1622,
+      "step": 43410
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.146174780667022e-06,
+      "loss": 0.1372,
+      "step": 43440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.145433757897916e-06,
+      "loss": 0.1529,
+      "step": 43470
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.144693502171406e-06,
+      "loss": 0.1499,
+      "step": 43500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.14395401216511e-06,
+      "loss": 0.1556,
+      "step": 43530
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.143215286559837e-06,
+      "loss": 0.15,
+      "step": 43560
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1424773240395752e-06,
+      "loss": 0.1706,
+      "step": 43590
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.141740123291482e-06,
+      "loss": 0.1412,
+      "step": 43620
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.141003683005876e-06,
+      "loss": 0.1537,
+      "step": 43650
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1402680018762254e-06,
+      "loss": 0.164,
+      "step": 43680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1395330785991377e-06,
+      "loss": 0.1404,
+      "step": 43710
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1387989118743533e-06,
+      "loss": 0.1549,
+      "step": 43740
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1380655004047324e-06,
+      "loss": 0.1339,
+      "step": 43770
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.1373328428962457e-06,
+      "loss": 0.1393,
+      "step": 43800
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.136600938057969e-06,
+      "loss": 0.1585,
+      "step": 43830
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1358697846020666e-06,
+      "loss": 0.1746,
+      "step": 43860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.135139381243788e-06,
+      "loss": 0.1542,
+      "step": 43890
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1344097267014553e-06,
+      "loss": 0.1699,
+      "step": 43920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1336808196964546e-06,
+      "loss": 0.1372,
+      "step": 43950
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1329526589532265e-06,
+      "loss": 0.1553,
+      "step": 43980
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1322252431992567e-06,
+      "loss": 0.1484,
+      "step": 44010
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.131498571165066e-06,
+      "loss": 0.1472,
+      "step": 44040
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.130772641584203e-06,
+      "loss": 0.161,
+      "step": 44070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.1300474531932315e-06,
+      "loss": 0.1696,
+      "step": 44100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1293230047317256e-06,
+      "loss": 0.145,
+      "step": 44130
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.128599294942257e-06,
+      "loss": 0.1463,
+      "step": 44160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.127876322570386e-06,
+      "loss": 0.1505,
+      "step": 44190
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1271540863646543e-06,
+      "loss": 0.1658,
+      "step": 44220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1264325850765765e-06,
+      "loss": 0.1551,
+      "step": 44250
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.125711817460626e-06,
+      "loss": 0.1468,
+      "step": 44280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.124991782274232e-06,
+      "loss": 0.1521,
+      "step": 44310
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1242724782777676e-06,
+      "loss": 0.1355,
+      "step": 44340
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.1235539042345404e-06,
+      "loss": 0.1431,
+      "step": 44370
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1228360589107853e-06,
+      "loss": 0.1589,
+      "step": 44400
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1221189410756536e-06,
+      "loss": 0.171,
+      "step": 44430
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1214025495012067e-06,
+      "loss": 0.1383,
+      "step": 44460
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1206868829624043e-06,
+      "loss": 0.157,
+      "step": 44490
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1199719402370975e-06,
+      "loss": 0.1388,
+      "step": 44520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.11925772010602e-06,
+      "loss": 0.1638,
+      "step": 44550
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.118544221352778e-06,
+      "loss": 0.1473,
+      "step": 44580
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.117831442763844e-06,
+      "loss": 0.1358,
+      "step": 44610
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.1171193831285454e-06,
+      "loss": 0.1704,
+      "step": 44640
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1164080412390573e-06,
+      "loss": 0.1302,
+      "step": 44670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.115697415890394e-06,
+      "loss": 0.1402,
+      "step": 44700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1149875058803997e-06,
+      "loss": 0.1471,
+      "step": 44730
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.114278310009741e-06,
+      "loss": 0.1479,
+      "step": 44760
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.113569827081896e-06,
+      "loss": 0.1359,
+      "step": 44790
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.11286205590315e-06,
+      "loss": 0.1965,
+      "step": 44820
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1121549952825822e-06,
+      "loss": 0.1311,
+      "step": 44850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.1114486440320624e-06,
+      "loss": 0.1513,
+      "step": 44880
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.110743000966237e-06,
+      "loss": 0.1503,
+      "step": 44910
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.110038064902525e-06,
+      "loss": 0.1677,
+      "step": 44940
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.109333834661109e-06,
+      "loss": 0.1297,
+      "step": 44970
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1086303090649252e-06,
+      "loss": 0.1588,
+      "step": 45000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1079274869396553e-06,
+      "loss": 0.144,
+      "step": 45030
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1072253671137207e-06,
+      "loss": 0.1601,
+      "step": 45060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.106523948418271e-06,
+      "loss": 0.1327,
+      "step": 45090
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1058232296871785e-06,
+      "loss": 0.1346,
+      "step": 45120
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1051232097570295e-06,
+      "loss": 0.1722,
+      "step": 45150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1044238874671146e-06,
+      "loss": 0.1664,
+      "step": 45180
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.103725261659422e-06,
+      "loss": 0.1612,
+      "step": 45210
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.103027331178629e-06,
+      "loss": 0.1759,
+      "step": 45240
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1023300948720953e-06,
+      "loss": 0.1522,
+      "step": 45270
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1016335515898515e-06,
+      "loss": 0.1648,
+      "step": 45300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.100937700184596e-06,
+      "loss": 0.141,
+      "step": 45330
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.100242539511684e-06,
+      "loss": 0.1659,
+      "step": 45360
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.099548068429119e-06,
+      "loss": 0.131,
+      "step": 45390
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.098854285797546e-06,
+      "loss": 0.1419,
+      "step": 45420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.0981611904802458e-06,
+      "loss": 0.1415,
+      "step": 45450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.097468781343123e-06,
+      "loss": 0.1619,
+      "step": 45480
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.096800103699349e-06,
+      "loss": 0.1505,
+      "step": 45510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.0961090407515926e-06,
+      "loss": 0.1329,
+      "step": 45540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.095418660634908e-06,
+      "loss": 0.1501,
+      "step": 45570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.0947289622255397e-06,
+      "loss": 0.1388,
+      "step": 45600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.0940399444023193e-06,
+      "loss": 0.1288,
+      "step": 45630
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.0933516060466585e-06,
+      "loss": 0.1495,
+      "step": 45660
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.092663946042543e-06,
+      "loss": 0.1207,
+      "step": 45690
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.09197696327652e-06,
+      "loss": 0.1552,
+      "step": 45720
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.0912906566376966e-06,
+      "loss": 0.1263,
+      "step": 45750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.090605025017727e-06,
+      "loss": 0.124,
+      "step": 45780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.0899200673108096e-06,
+      "loss": 0.1456,
+      "step": 45810
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.0892357824136746e-06,
+      "loss": 0.1448,
+      "step": 45840
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.0885521692255816e-06,
+      "loss": 0.1373,
+      "step": 45870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.0878692266483083e-06,
+      "loss": 0.1591,
+      "step": 45900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.0871869535861456e-06,
+      "loss": 0.151,
+      "step": 45930
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.0865053489458874e-06,
+      "loss": 0.1493,
+      "step": 45960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0858244116368275e-06,
+      "loss": 0.1613,
+      "step": 45990
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.085144140570748e-06,
+      "loss": 0.1259,
+      "step": 46020
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0844645346619135e-06,
+      "loss": 0.1247,
+      "step": 46050
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.083785592827065e-06,
+      "loss": 0.1386,
+      "step": 46080
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.083107313985412e-06,
+      "loss": 0.1597,
+      "step": 46110
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0824296970586242e-06,
+      "loss": 0.1377,
+      "step": 46140
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0817527409708258e-06,
+      "loss": 0.1499,
+      "step": 46170
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0810764446485862e-06,
+      "loss": 0.1699,
+      "step": 46200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.0804008070209163e-06,
+      "loss": 0.1369,
+      "step": 46230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.0797258270192577e-06,
+      "loss": 0.1405,
+      "step": 46260
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.079051503577478e-06,
+      "loss": 0.1308,
+      "step": 46290
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.0783778356318633e-06,
+      "loss": 0.1678,
+      "step": 46320
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.0777048221211104e-06,
+      "loss": 0.1471,
+      "step": 46350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.0770324619863196e-06,
+      "loss": 0.1736,
+      "step": 46380
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.07636075417099e-06,
+      "loss": 0.1684,
+      "step": 46410
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.075689697621009e-06,
+      "loss": 0.144,
+      "step": 46440
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.075019291284648e-06,
+      "loss": 0.1527,
+      "step": 46470
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.074349534112556e-06,
+      "loss": 0.1447,
+      "step": 46500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0736804250577488e-06,
+      "loss": 0.1627,
+      "step": 46530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.073011963075606e-06,
+      "loss": 0.1689,
+      "step": 46560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.072344147123863e-06,
+      "loss": 0.1459,
+      "step": 46590
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0716769761626044e-06,
+      "loss": 0.1646,
+      "step": 46620
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0710104491542556e-06,
+      "loss": 0.1403,
+      "step": 46650
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.070344565063579e-06,
+      "loss": 0.1408,
+      "step": 46680
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.069679322857663e-06,
+      "loss": 0.1511,
+      "step": 46710
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.06901472150592e-06,
+      "loss": 0.1633,
+      "step": 46740
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0683507599800766e-06,
+      "loss": 0.132,
+      "step": 46770
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.067687437254168e-06,
+      "loss": 0.137,
+      "step": 46800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.067024752304531e-06,
+      "loss": 0.1452,
+      "step": 46830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.066362704109797e-06,
+      "loss": 0.136,
+      "step": 46860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.065701291650887e-06,
+      "loss": 0.1621,
+      "step": 46890
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0650405139110026e-06,
+      "loss": 0.1482,
+      "step": 46920
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.064380369875622e-06,
+      "loss": 0.1495,
+      "step": 46950
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0637208585324905e-06,
+      "loss": 0.1801,
+      "step": 46980
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0630619788716174e-06,
+      "loss": 0.1441,
+      "step": 47010
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0624037298852674e-06,
+      "loss": 0.124,
+      "step": 47040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0617461105679532e-06,
+      "loss": 0.1459,
+      "step": 47070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0610891199164325e-06,
+      "loss": 0.1501,
+      "step": 47100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.060432756929697e-06,
+      "loss": 0.1366,
+      "step": 47130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0597770206089703e-06,
+      "loss": 0.1484,
+      "step": 47160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0591219099576987e-06,
+      "loss": 0.1414,
+      "step": 47190
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.058467423981546e-06,
+      "loss": 0.1673,
+      "step": 47220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0578135616883866e-06,
+      "loss": 0.1479,
+      "step": 47250
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0571603220883005e-06,
+      "loss": 0.1706,
+      "step": 47280
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0565077041935645e-06,
+      "loss": 0.1624,
+      "step": 47310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.055855707018649e-06,
+      "loss": 0.1562,
+      "step": 47340
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0552043295802093e-06,
+      "loss": 0.1392,
+      "step": 47370
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0545535708970804e-06,
+      "loss": 0.1177,
+      "step": 47400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0539034299902704e-06,
+      "loss": 0.1239,
+      "step": 47430
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0532539058829546e-06,
+      "loss": 0.1797,
+      "step": 47460
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0526049976004704e-06,
+      "loss": 0.1431,
+      "step": 47490
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0519567041703083e-06,
+      "loss": 0.1416,
+      "step": 47520
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.051309024622109e-06,
+      "loss": 0.1593,
+      "step": 47550
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.050661957987655e-06,
+      "loss": 0.1251,
+      "step": 47580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.050015503300866e-06,
+      "loss": 0.1457,
+      "step": 47610
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0493696595977914e-06,
+      "loss": 0.1571,
+      "step": 47640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0487244259166056e-06,
+      "loss": 0.1269,
+      "step": 47670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0480798012976015e-06,
+      "loss": 0.1328,
+      "step": 47700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0474357847831843e-06,
+      "loss": 0.1706,
+      "step": 47730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0467923754178647e-06,
+      "loss": 0.1635,
+      "step": 47760
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0461495722482565e-06,
+      "loss": 0.1581,
+      "step": 47790
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.045507374323066e-06,
+      "loss": 0.1475,
+      "step": 47820
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.044865780693088e-06,
+      "loss": 0.1588,
+      "step": 47850
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.044224790411201e-06,
+      "loss": 0.1508,
+      "step": 47880
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.043605739099571e-06,
+      "loss": 0.1639,
+      "step": 47910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0429659326473045e-06,
+      "loss": 0.1608,
+      "step": 47940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0423267267455322e-06,
+      "loss": 0.1413,
+      "step": 47970
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.041688120455337e-06,
+      "loss": 0.1471,
+      "step": 48000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.041050112839854e-06,
+      "loss": 0.1402,
+      "step": 48030
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0404127029642688e-06,
+      "loss": 0.1442,
+      "step": 48060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0397758898958064e-06,
+      "loss": 0.1605,
+      "step": 48090
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0391396727037307e-06,
+      "loss": 0.1374,
+      "step": 48120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.038504050459336e-06,
+      "loss": 0.1727,
+      "step": 48150
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0378690222359403e-06,
+      "loss": 0.1639,
+      "step": 48180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0372345871088826e-06,
+      "loss": 0.1575,
+      "step": 48210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.036600744155515e-06,
+      "loss": 0.1342,
+      "step": 48240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.035967492455198e-06,
+      "loss": 0.1381,
+      "step": 48270
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0353348310892955e-06,
+      "loss": 0.158,
+      "step": 48300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.034702759141167e-06,
+      "loss": 0.1511,
+      "step": 48330
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.034071275696164e-06,
+      "loss": 0.1147,
+      "step": 48360
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.033440379841623e-06,
+      "loss": 0.1487,
+      "step": 48390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.032810070666863e-06,
+      "loss": 0.1717,
+      "step": 48420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0321803472631754e-06,
+      "loss": 0.1458,
+      "step": 48450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0315512087238225e-06,
+      "loss": 0.1381,
+      "step": 48480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.030922654144029e-06,
+      "loss": 0.1421,
+      "step": 48510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.03029468262098e-06,
+      "loss": 0.152,
+      "step": 48540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0296672932538105e-06,
+      "loss": 0.1784,
+      "step": 48570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0290404851436057e-06,
+      "loss": 0.147,
+      "step": 48600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.028414257393391e-06,
+      "loss": 0.1303,
+      "step": 48630
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.0277886091081305e-06,
+      "loss": 0.1226,
+      "step": 48660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0271635393947167e-06,
+      "loss": 0.1633,
+      "step": 48690
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0265390473619705e-06,
+      "loss": 0.1397,
+      "step": 48720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.025915132120632e-06,
+      "loss": 0.1613,
+      "step": 48750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0252917927833574e-06,
+      "loss": 0.1628,
+      "step": 48780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.024669028464713e-06,
+      "loss": 0.1448,
+      "step": 48810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0240468382811682e-06,
+      "loss": 0.1142,
+      "step": 48840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0234252213510938e-06,
+      "loss": 0.1803,
+      "step": 48870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0228041767947535e-06,
+      "loss": 0.1561,
+      "step": 48900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.0221837037343006e-06,
+      "loss": 0.1309,
+      "step": 48930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0215638012937715e-06,
+      "loss": 0.1456,
+      "step": 48960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.020944468599082e-06,
+      "loss": 0.1509,
+      "step": 48990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.02032570477802e-06,
+      "loss": 0.1118,
+      "step": 49020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.019707508960244e-06,
+      "loss": 0.1519,
+      "step": 49050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0190898802772724e-06,
+      "loss": 0.1298,
+      "step": 49080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0184728178624835e-06,
+      "loss": 0.1334,
+      "step": 49110
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.017856320851108e-06,
+      "loss": 0.144,
+      "step": 49140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0172403883802254e-06,
+      "loss": 0.1516,
+      "step": 49170
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.0166250195887554e-06,
+      "loss": 0.1373,
+      "step": 49200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.0160102136174573e-06,
+      "loss": 0.1762,
+      "step": 49230
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.015395969608922e-06,
+      "loss": 0.1707,
+      "step": 49260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.01478228670757e-06,
+      "loss": 0.1251,
+      "step": 49290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.014169164059641e-06,
+      "loss": 0.1271,
+      "step": 49320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.013556600813194e-06,
+      "loss": 0.1403,
+      "step": 49350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.0129445961181016e-06,
+      "loss": 0.1334,
+      "step": 49380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.0123331491260424e-06,
+      "loss": 0.1512,
+      "step": 49410
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.0117222589904985e-06,
+      "loss": 0.1416,
+      "step": 49440
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.011111924866749e-06,
+      "loss": 0.1468,
+      "step": 49470
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0105021459118675e-06,
+      "loss": 0.1453,
+      "step": 49500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0098929212847133e-06,
+      "loss": 0.1572,
+      "step": 49530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0092842501459305e-06,
+      "loss": 0.1549,
+      "step": 49560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0086761316579416e-06,
+      "loss": 0.1335,
+      "step": 49590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.008068564984941e-06,
+      "loss": 0.1544,
+      "step": 49620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0074615492928926e-06,
+      "loss": 0.1223,
+      "step": 49650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0068550837495242e-06,
+      "loss": 0.1296,
+      "step": 49680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.0062491675243216e-06,
+      "loss": 0.1567,
+      "step": 49710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0056437997885276e-06,
+      "loss": 0.1718,
+      "step": 49740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.00503897971513e-06,
+      "loss": 0.1537,
+      "step": 49770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.004434706478865e-06,
+      "loss": 0.1418,
+      "step": 49800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0038309792562076e-06,
+      "loss": 0.1969,
+      "step": 49830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.003227797225367e-06,
+      "loss": 0.1665,
+      "step": 49860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0026251595662847e-06,
+      "loss": 0.1401,
+      "step": 49890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.002023065460627e-06,
+      "loss": 0.1375,
+      "step": 49920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.001421514091782e-06,
+      "loss": 0.1325,
+      "step": 49950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0008205046448528e-06,
+      "loss": 0.1449,
+      "step": 49980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0002200363066566e-06,
+      "loss": 0.1495,
+      "step": 50010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9996400971708495e-06,
+      "loss": 0.1364,
+      "step": 50040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9990406906475044e-06,
+      "loss": 0.1351,
+      "step": 50070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9984418228304744e-06,
+      "loss": 0.143,
+      "step": 50100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9978434929133213e-06,
+      "loss": 0.1434,
+      "step": 50130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.997245700091295e-06,
+      "loss": 0.1609,
+      "step": 50160
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.996648443561332e-06,
+      "loss": 0.137,
+      "step": 50190
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.996051722522046e-06,
+      "loss": 0.1427,
+      "step": 50220
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9954555361737284e-06,
+      "loss": 0.1237,
+      "step": 50250
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9948598837183403e-06,
+      "loss": 0.1591,
+      "step": 50280
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.99426476435951e-06,
+      "loss": 0.1496,
+      "step": 50310
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9936701773025286e-06,
+      "loss": 0.1364,
+      "step": 50340
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9930761217543435e-06,
+      "loss": 0.1425,
+      "step": 50370
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9924825969235547e-06,
+      "loss": 0.165,
+      "step": 50400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9918896020204136e-06,
+      "loss": 0.1356,
+      "step": 50430
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9912971362568133e-06,
+      "loss": 0.1314,
+      "step": 50460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.9907051988462883e-06,
+      "loss": 0.1481,
+      "step": 50490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.990113789004008e-06,
+      "loss": 0.1468,
+      "step": 50520
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.989522905946773e-06,
+      "loss": 0.1308,
+      "step": 50550
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9889325488930104e-06,
+      "loss": 0.1535,
+      "step": 50580
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9883427170627706e-06,
+      "loss": 0.1222,
+      "step": 50610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9877534096777218e-06,
+      "loss": 0.1415,
+      "step": 50640
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9871646259611458e-06,
+      "loss": 0.1625,
+      "step": 50670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9865763651379345e-06,
+      "loss": 0.1266,
+      "step": 50700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9859886264345835e-06,
+      "loss": 0.146,
+      "step": 50730
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.985420974599683e-06,
+      "loss": 0.1416,
+      "step": 50760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9848342604817485e-06,
+      "loss": 0.1516,
+      "step": 50790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9842480661983527e-06,
+      "loss": 0.1389,
+      "step": 50820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.983662390982317e-06,
+      "loss": 0.1129,
+      "step": 50850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.983077234068048e-06,
+      "loss": 0.1344,
+      "step": 50880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9824925946915318e-06,
+      "loss": 0.1372,
+      "step": 50910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.98190847209033e-06,
+      "loss": 0.142,
+      "step": 50940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.981324865503577e-06,
+      "loss": 0.1483,
+      "step": 50970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.980741774171974e-06,
+      "loss": 0.1239,
+      "step": 51000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9801591973377847e-06,
+      "loss": 0.1557,
+      "step": 51030
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.979577134244833e-06,
+      "loss": 0.1366,
+      "step": 51060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.978995584138498e-06,
+      "loss": 0.1379,
+      "step": 51090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.978414546265708e-06,
+      "loss": 0.1274,
+      "step": 51120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9778340198749395e-06,
+      "loss": 0.148,
+      "step": 51150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9772540042162125e-06,
+      "loss": 0.1374,
+      "step": 51180
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.976674498541084e-06,
+      "loss": 0.1523,
+      "step": 51210
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9760955021026456e-06,
+      "loss": 0.1177,
+      "step": 51240
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9755170141555212e-06,
+      "loss": 0.1533,
+      "step": 51270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9749390339558593e-06,
+      "loss": 0.1248,
+      "step": 51300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9743615607613323e-06,
+      "loss": 0.158,
+      "step": 51330
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9737845938311307e-06,
+      "loss": 0.1513,
+      "step": 51360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.973208132425959e-06,
+      "loss": 0.1404,
+      "step": 51390
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.972632175808033e-06,
+      "loss": 0.171,
+      "step": 51420
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.972056723241075e-06,
+      "loss": 0.1785,
+      "step": 51450
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9714817739903085e-06,
+      "loss": 0.1428,
+      "step": 51480
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9709073273224575e-06,
+      "loss": 0.1473,
+      "step": 51510
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9703333825057403e-06,
+      "loss": 0.1568,
+      "step": 51540
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9697599388098655e-06,
+      "loss": 0.1389,
+      "step": 51570
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.9691869955060294e-06,
+      "loss": 0.1472,
+      "step": 51600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9686145518669107e-06,
+      "loss": 0.1402,
+      "step": 51630
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.968042607166667e-06,
+      "loss": 0.1313,
+      "step": 51660
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.967471160680932e-06,
+      "loss": 0.1738,
+      "step": 51690
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.966900211686811e-06,
+      "loss": 0.1363,
+      "step": 51720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9663297594628757e-06,
+      "loss": 0.14,
+      "step": 51750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.965759803289163e-06,
+      "loss": 0.1745,
+      "step": 51780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9651903424471683e-06,
+      "loss": 0.1279,
+      "step": 51810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9646213762198447e-06,
+      "loss": 0.1601,
+      "step": 51840
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.9640529038915965e-06,
+      "loss": 0.1329,
+      "step": 51870
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.963484924748277e-06,
+      "loss": 0.1545,
+      "step": 51900
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9629174380771844e-06,
+      "loss": 0.1474,
+      "step": 51930
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9623504431670577e-06,
+      "loss": 0.1261,
+      "step": 51960
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.961783939308074e-06,
+      "loss": 0.1548,
+      "step": 51990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9612179257918425e-06,
+      "loss": 0.1457,
+      "step": 52020
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.960652401911404e-06,
+      "loss": 0.1263,
+      "step": 52050
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9600873669612238e-06,
+      "loss": 0.1153,
+      "step": 52080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.959522820237191e-06,
+      "loss": 0.1425,
+      "step": 52110
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.958958761036613e-06,
+      "loss": 0.1344,
+      "step": 52140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.958395188658212e-06,
+      "loss": 0.1372,
+      "step": 52170
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.957832102402122e-06,
+      "loss": 0.1396,
+      "step": 52200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9572695015698843e-06,
+      "loss": 0.127,
+      "step": 52230
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9567073854644454e-06,
+      "loss": 0.132,
+      "step": 52260
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9561457533901515e-06,
+      "loss": 0.137,
+      "step": 52290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.9555846046527456e-06,
+      "loss": 0.1722,
+      "step": 52320
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.955023938559364e-06,
+      "loss": 0.1586,
+      "step": 52350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.954463754418534e-06,
+      "loss": 0.153,
+      "step": 52380
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.953904051540167e-06,
+      "loss": 0.1555,
+      "step": 52410
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9533448292355585e-06,
+      "loss": 0.1491,
+      "step": 52440
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9527860868173816e-06,
+      "loss": 0.1267,
+      "step": 52470
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9522278235996874e-06,
+      "loss": 0.1533,
+      "step": 52500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.951670038897896e-06,
+      "loss": 0.1299,
+      "step": 52530
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9511127320287964e-06,
+      "loss": 0.1336,
+      "step": 52560
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9505559023105446e-06,
+      "loss": 0.1581,
+      "step": 52590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9499995490626562e-06,
+      "loss": 0.1434,
+      "step": 52620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9494436716060055e-06,
+      "loss": 0.1276,
+      "step": 52650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.9488882692628197e-06,
+      "loss": 0.1335,
+      "step": 52680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9483333413566786e-06,
+      "loss": 0.1384,
+      "step": 52710
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9477788872125087e-06,
+      "loss": 0.1393,
+      "step": 52740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.94722490615658e-06,
+      "loss": 0.1604,
+      "step": 52770
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.946671397516505e-06,
+      "loss": 0.1486,
+      "step": 52800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.946118360621231e-06,
+      "loss": 0.1258,
+      "step": 52830
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9455657948010406e-06,
+      "loss": 0.1385,
+      "step": 52860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.945013699387546e-06,
+      "loss": 0.1514,
+      "step": 52890
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9444620737136872e-06,
+      "loss": 0.14,
+      "step": 52920
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.9439109171137267e-06,
+      "loss": 0.1281,
+      "step": 52950
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.943360228923247e-06,
+      "loss": 0.1233,
+      "step": 52980
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.942810008479148e-06,
+      "loss": 0.143,
+      "step": 53010
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.942260255119644e-06,
+      "loss": 0.1697,
+      "step": 53040
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.941710968184258e-06,
+      "loss": 0.1515,
+      "step": 53070
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9411621470138204e-06,
+      "loss": 0.1349,
+      "step": 53100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.940613790950464e-06,
+      "loss": 0.139,
+      "step": 53130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.940065899337623e-06,
+      "loss": 0.1279,
+      "step": 53160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.939518471520029e-06,
+      "loss": 0.1355,
+      "step": 53190
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9389715068437056e-06,
+      "loss": 0.1538,
+      "step": 53220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.938425004655967e-06,
+      "loss": 0.1466,
+      "step": 53250
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9378789643054147e-06,
+      "loss": 0.1248,
+      "step": 53280
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.937333385141934e-06,
+      "loss": 0.1291,
+      "step": 53310
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9367882665166908e-06,
+      "loss": 0.1363,
+      "step": 53340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9362436077821283e-06,
+      "loss": 0.1325,
+      "step": 53370
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.935699408291963e-06,
+      "loss": 0.1311,
+      "step": 53400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9351556674011827e-06,
+      "loss": 0.1375,
+      "step": 53430
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9346123844660423e-06,
+      "loss": 0.159,
+      "step": 53460
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.934069558844063e-06,
+      "loss": 0.1449,
+      "step": 53490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.933527189894024e-06,
+      "loss": 0.1417,
+      "step": 53520
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.932985276975965e-06,
+      "loss": 0.111,
+      "step": 53550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.93244381945118e-06,
+      "loss": 0.163,
+      "step": 53580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9319028166822136e-06,
+      "loss": 0.1297,
+      "step": 53610
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9313622680328613e-06,
+      "loss": 0.1178,
+      "step": 53640
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9308221728681602e-06,
+      "loss": 0.1362,
+      "step": 53670
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9302825305543936e-06,
+      "loss": 0.1115,
+      "step": 53700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.929743340459082e-06,
+      "loss": 0.1427,
+      "step": 53730
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.929204601950982e-06,
+      "loss": 0.143,
+      "step": 53760
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9286663144000823e-06,
+      "loss": 0.1393,
+      "step": 53790
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.928128477177604e-06,
+      "loss": 0.1221,
+      "step": 53820
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.927591089655992e-06,
+      "loss": 0.1452,
+      "step": 53850
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9270541512089164e-06,
+      "loss": 0.1481,
+      "step": 53880
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9265176612112675e-06,
+      "loss": 0.1434,
+      "step": 53910
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9259816190391543e-06,
+      "loss": 0.1411,
+      "step": 53940
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.925446024069898e-06,
+      "loss": 0.1442,
+      "step": 53970
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.9249108756820324e-06,
+      "loss": 0.129,
+      "step": 54000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9243761732553013e-06,
+      "loss": 0.1403,
+      "step": 54030
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9238419161706506e-06,
+      "loss": 0.1284,
+      "step": 54060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.923308103810231e-06,
+      "loss": 0.1248,
+      "step": 54090
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.922774735557393e-06,
+      "loss": 0.1347,
+      "step": 54120
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9222418107966813e-06,
+      "loss": 0.1383,
+      "step": 54150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.921709328913835e-06,
+      "loss": 0.1221,
+      "step": 54180
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.921177289295785e-06,
+      "loss": 0.1909,
+      "step": 54210
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9206456913306475e-06,
+      "loss": 0.145,
+      "step": 54240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.9201145344077244e-06,
+      "loss": 0.1358,
+      "step": 54270
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9195838179174996e-06,
+      "loss": 0.1371,
+      "step": 54300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9190535412516335e-06,
+      "loss": 0.1221,
+      "step": 54330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9185237038029657e-06,
+      "loss": 0.1306,
+      "step": 54360
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9179943049655055e-06,
+      "loss": 0.1412,
+      "step": 54390
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.917465344134433e-06,
+      "loss": 0.1251,
+      "step": 54420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.916936820706096e-06,
+      "loss": 0.1566,
+      "step": 54450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.916408734078005e-06,
+      "loss": 0.161,
+      "step": 54480
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.915881083648834e-06,
+      "loss": 0.1591,
+      "step": 54510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.9153538688184124e-06,
+      "loss": 0.1193,
+      "step": 54540
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9148270889877275e-06,
+      "loss": 0.157,
+      "step": 54570
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.914300743558917e-06,
+      "loss": 0.1461,
+      "step": 54600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.913774831935271e-06,
+      "loss": 0.1418,
+      "step": 54630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9132493535212246e-06,
+      "loss": 0.1584,
+      "step": 54660
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9127243077223566e-06,
+      "loss": 0.156,
+      "step": 54690
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9121996939453884e-06,
+      "loss": 0.1312,
+      "step": 54720
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9116755115981794e-06,
+      "loss": 0.1259,
+      "step": 54750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.9111692115382816e-06,
+      "loss": 0.1549,
+      "step": 54780
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.910645875946571e-06,
+      "loss": 0.14,
+      "step": 54810
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.910122970034606e-06,
+      "loss": 0.1441,
+      "step": 54840
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9096004932147346e-06,
+      "loss": 0.1485,
+      "step": 54870
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.909078444900429e-06,
+      "loss": 0.1247,
+      "step": 54900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9085568245062825e-06,
+      "loss": 0.1509,
+      "step": 54930
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.908035631448009e-06,
+      "loss": 0.1342,
+      "step": 54960
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.907514865142436e-06,
+      "loss": 0.1619,
+      "step": 54990
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9069945250075084e-06,
+      "loss": 0.1242,
+      "step": 55020
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9064746104622783e-06,
+      "loss": 0.1585,
+      "step": 55050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9059551209269078e-06,
+      "loss": 0.1655,
+      "step": 55080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9054360558226643e-06,
+      "loss": 0.1182,
+      "step": 55110
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9049174145719173e-06,
+      "loss": 0.129,
+      "step": 55140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9043991965981363e-06,
+      "loss": 0.1676,
+      "step": 55170
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9038814013258886e-06,
+      "loss": 0.1417,
+      "step": 55200
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9033640281808363e-06,
+      "loss": 0.1112,
+      "step": 55230
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.902847076589732e-06,
+      "loss": 0.1362,
+      "step": 55260
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.9023305459804192e-06,
+      "loss": 0.1654,
+      "step": 55290
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.901814435781827e-06,
+      "loss": 0.1577,
+      "step": 55320
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.901298745423969e-06,
+      "loss": 0.1288,
+      "step": 55350
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9007834743379393e-06,
+      "loss": 0.1274,
+      "step": 55380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9002686219559125e-06,
+      "loss": 0.1204,
+      "step": 55410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8997541877111366e-06,
+      "loss": 0.1297,
+      "step": 55440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.899240171037935e-06,
+      "loss": 0.1634,
+      "step": 55470
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8987265713717016e-06,
+      "loss": 0.1315,
+      "step": 55500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8982133881488975e-06,
+      "loss": 0.1278,
+      "step": 55530
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8977006208070509e-06,
+      "loss": 0.1762,
+      "step": 55560
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8971882687847517e-06,
+      "loss": 0.1379,
+      "step": 55590
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8966763315216514e-06,
+      "loss": 0.1479,
+      "step": 55620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8961648084584578e-06,
+      "loss": 0.1305,
+      "step": 55650
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8956536990369353e-06,
+      "loss": 0.1345,
+      "step": 55680
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8951430026999016e-06,
+      "loss": 0.1128,
+      "step": 55710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.894632718891222e-06,
+      "loss": 0.1565,
+      "step": 55740
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8941228470558122e-06,
+      "loss": 0.1482,
+      "step": 55770
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8936133866396317e-06,
+      "loss": 0.1325,
+      "step": 55800
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8931043370896827e-06,
+      "loss": 0.1463,
+      "step": 55830
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8925956978540075e-06,
+      "loss": 0.1328,
+      "step": 55860
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8920874683816859e-06,
+      "loss": 0.1245,
+      "step": 55890
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.891579648122833e-06,
+      "loss": 0.1185,
+      "step": 55920
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8910722365285963e-06,
+      "loss": 0.1188,
+      "step": 55950
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.890565233051153e-06,
+      "loss": 0.1299,
+      "step": 55980
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8900586371437086e-06,
+      "loss": 0.1448,
+      "step": 56010
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8895524482604927e-06,
+      "loss": 0.1581,
+      "step": 56040
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8890466658567582e-06,
+      "loss": 0.1294,
+      "step": 56070
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8885412893887783e-06,
+      "loss": 0.1452,
+      "step": 56100
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8880363183138433e-06,
+      "loss": 0.1344,
+      "step": 56130
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8875317520902595e-06,
+      "loss": 0.1482,
+      "step": 56160
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.887027590177345e-06,
+      "loss": 0.1595,
+      "step": 56190
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.886523832035429e-06,
+      "loss": 0.1496,
+      "step": 56220
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8860204771258493e-06,
+      "loss": 0.1285,
+      "step": 56250
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.885517524910947e-06,
+      "loss": 0.1363,
+      "step": 56280
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8850149748540687e-06,
+      "loss": 0.1447,
+      "step": 56310
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8845128264195603e-06,
+      "loss": 0.1389,
+      "step": 56340
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.884011079072767e-06,
+      "loss": 0.1175,
+      "step": 56370
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.883509732280029e-06,
+      "loss": 0.1539,
+      "step": 56400
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8830087855086806e-06,
+      "loss": 0.1483,
+      "step": 56430
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8825082382270468e-06,
+      "loss": 0.1369,
+      "step": 56460
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.882008089904442e-06,
+      "loss": 0.1516,
+      "step": 56490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8815083400111672e-06,
+      "loss": 0.1518,
+      "step": 56520
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8810089880185063e-06,
+      "loss": 0.16,
+      "step": 56550
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8805100333987256e-06,
+      "loss": 0.1144,
+      "step": 56580
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8800114756250712e-06,
+      "loss": 0.1198,
+      "step": 56610
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8795133141717656e-06,
+      "loss": 0.1455,
+      "step": 56640
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8790155485140063e-06,
+      "loss": 0.1119,
+      "step": 56670
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8785181781279636e-06,
+      "loss": 0.1224,
+      "step": 56700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8780212024907767e-06,
+      "loss": 0.1444,
+      "step": 56730
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.877524621080554e-06,
+      "loss": 0.1615,
+      "step": 56760
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8770284333763686e-06,
+      "loss": 0.1363,
+      "step": 56790
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8765326388582567e-06,
+      "loss": 0.1387,
+      "step": 56820
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8760372370072152e-06,
+      "loss": 0.1401,
+      "step": 56850
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8755422273052006e-06,
+      "loss": 0.128,
+      "step": 56880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.875047609235125e-06,
+      "loss": 0.1281,
+      "step": 56910
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8745533822808548e-06,
+      "loss": 0.1453,
+      "step": 56940
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8740595459272077e-06,
+      "loss": 0.1508,
+      "step": 56970
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8735660996599512e-06,
+      "loss": 0.1184,
+      "step": 57000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8730730429658007e-06,
+      "loss": 0.1293,
+      "step": 57030
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8725803753324157e-06,
+      "loss": 0.1484,
+      "step": 57060
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8720880962483992e-06,
+      "loss": 0.1366,
+      "step": 57090
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8715962052032946e-06,
+      "loss": 0.1456,
+      "step": 57120
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8711047016875836e-06,
+      "loss": 0.14,
+      "step": 57150
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8706135851926837e-06,
+      "loss": 0.1289,
+      "step": 57180
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.870122855210947e-06,
+      "loss": 0.1278,
+      "step": 57210
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8696325112356574e-06,
+      "loss": 0.1643,
+      "step": 57240
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.869142552761028e-06,
+      "loss": 0.1741,
+      "step": 57270
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8686529792821986e-06,
+      "loss": 0.1421,
+      "step": 57300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.868163790295235e-06,
+      "loss": 0.1173,
+      "step": 57330
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8676749852971257e-06,
+      "loss": 0.1389,
+      "step": 57360
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8671865637857806e-06,
+      "loss": 0.1341,
+      "step": 57390
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.866698525260027e-06,
+      "loss": 0.1396,
+      "step": 57420
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8662108692196096e-06,
+      "loss": 0.1264,
+      "step": 57450
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8657235951651874e-06,
+      "loss": 0.1288,
+      "step": 57480
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8652367025983317e-06,
+      "loss": 0.1226,
+      "step": 57510
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8647501910215233e-06,
+      "loss": 0.1149,
+      "step": 57540
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8642640599381507e-06,
+      "loss": 0.1227,
+      "step": 57570
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8637783088525087e-06,
+      "loss": 0.1211,
+      "step": 57600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8632929372697966e-06,
+      "loss": 0.1226,
+      "step": 57630
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8628079446961137e-06,
+      "loss": 0.1482,
+      "step": 57660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8623233306384588e-06,
+      "loss": 0.1585,
+      "step": 57690
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8618390946047293e-06,
+      "loss": 0.1692,
+      "step": 57720
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8613552361037172e-06,
+      "loss": 0.1351,
+      "step": 57750
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.860871754645107e-06,
+      "loss": 0.112,
+      "step": 57780
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8603886497394748e-06,
+      "loss": 0.1435,
+      "step": 57810
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.859905920898286e-06,
+      "loss": 0.1509,
+      "step": 57840
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8594235676338921e-06,
+      "loss": 0.1242,
+      "step": 57870
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8589415894595303e-06,
+      "loss": 0.134,
+      "step": 57900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8584599858893192e-06,
+      "loss": 0.1328,
+      "step": 57930
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.85797875643826e-06,
+      "loss": 0.1295,
+      "step": 57960
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8575139231348627e-06,
+      "loss": 0.1351,
+      "step": 57990
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8570334280400031e-06,
+      "loss": 0.1316,
+      "step": 58020
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.856553305630624e-06,
+      "loss": 0.1225,
+      "step": 58050
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.856073555425199e-06,
+      "loss": 0.1474,
+      "step": 58080
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8555941769430718e-06,
+      "loss": 0.1281,
+      "step": 58110
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8551151697044554e-06,
+      "loss": 0.1203,
+      "step": 58140
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8546365332304288e-06,
+      "loss": 0.1301,
+      "step": 58170
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8541582670429356e-06,
+      "loss": 0.141,
+      "step": 58200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8536803706647816e-06,
+      "loss": 0.1162,
+      "step": 58230
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.853202843619634e-06,
+      "loss": 0.1177,
+      "step": 58260
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8527256854320173e-06,
+      "loss": 0.1242,
+      "step": 58290
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8522488956273137e-06,
+      "loss": 0.1218,
+      "step": 58320
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8517724737317594e-06,
+      "loss": 0.1214,
+      "step": 58350
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8512964192724428e-06,
+      "loss": 0.1419,
+      "step": 58380
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8508207317773033e-06,
+      "loss": 0.1471,
+      "step": 58410
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8503454107751295e-06,
+      "loss": 0.1293,
+      "step": 58440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.849870455795555e-06,
+      "loss": 0.1229,
+      "step": 58470
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8493958663690606e-06,
+      "loss": 0.1366,
+      "step": 58500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8489216420269671e-06,
+      "loss": 0.131,
+      "step": 58530
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8484477823014386e-06,
+      "loss": 0.1232,
+      "step": 58560
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8479742867254763e-06,
+      "loss": 0.1155,
+      "step": 58590
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8475011548329189e-06,
+      "loss": 0.1409,
+      "step": 58620
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.847028386158441e-06,
+      "loss": 0.1163,
+      "step": 58650
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8465559802375483e-06,
+      "loss": 0.1438,
+      "step": 58680
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8460839366065802e-06,
+      "loss": 0.1332,
+      "step": 58710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8456122548027036e-06,
+      "loss": 0.1116,
+      "step": 58740
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8451409343639128e-06,
+      "loss": 0.1541,
+      "step": 58770
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8446699748290289e-06,
+      "loss": 0.1484,
+      "step": 58800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.844199375737695e-06,
+      "loss": 0.1485,
+      "step": 58830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.843729136630377e-06,
+      "loss": 0.1469,
+      "step": 58860
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8432592570483596e-06,
+      "loss": 0.113,
+      "step": 58890
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8427897365337466e-06,
+      "loss": 0.151,
+      "step": 58920
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8423205746294573e-06,
+      "loss": 0.1283,
+      "step": 58950
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.841851770879224e-06,
+      "loss": 0.1135,
+      "step": 58980
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8413833248275939e-06,
+      "loss": 0.1451,
+      "step": 59010
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.840915236019922e-06,
+      "loss": 0.1232,
+      "step": 59040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.840447504002374e-06,
+      "loss": 0.1312,
+      "step": 59070
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8399801283219207e-06,
+      "loss": 0.1322,
+      "step": 59100
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8395131085263387e-06,
+      "loss": 0.1463,
+      "step": 59130
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8390464441642077e-06,
+      "loss": 0.1342,
+      "step": 59160
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8385801347849084e-06,
+      "loss": 0.1137,
+      "step": 59190
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8381141799386204e-06,
+      "loss": 0.1086,
+      "step": 59220
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8376485791763221e-06,
+      "loss": 0.1405,
+      "step": 59250
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8371833320497865e-06,
+      "loss": 0.1248,
+      "step": 59280
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8367184381115817e-06,
+      "loss": 0.1268,
+      "step": 59310
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8362538969150664e-06,
+      "loss": 0.1336,
+      "step": 59340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8357897080143913e-06,
+      "loss": 0.1585,
+      "step": 59370
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8353258709644942e-06,
+      "loss": 0.1658,
+      "step": 59400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.834862385321101e-06,
+      "loss": 0.1299,
+      "step": 59430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8343992506407218e-06,
+      "loss": 0.1415,
+      "step": 59460
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8339364664806493e-06,
+      "loss": 0.1314,
+      "step": 59490
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8334740323989596e-06,
+      "loss": 0.1507,
+      "step": 59520
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8330119479545063e-06,
+      "loss": 0.1176,
+      "step": 59550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.832550212706922e-06,
+      "loss": 0.1788,
+      "step": 59580
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8320888262166148e-06,
+      "loss": 0.134,
+      "step": 59610
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8316277880447678e-06,
+      "loss": 0.1447,
+      "step": 59640
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8311670977533358e-06,
+      "loss": 0.1302,
+      "step": 59670
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8307067549050465e-06,
+      "loss": 0.1325,
+      "step": 59700
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8302467590633934e-06,
+      "loss": 0.1361,
+      "step": 59730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8297871097926407e-06,
+      "loss": 0.1315,
+      "step": 59760
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8293278066578158e-06,
+      "loss": 0.1069,
+      "step": 59790
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8288688492247114e-06,
+      "loss": 0.1539,
+      "step": 59820
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8284102370598822e-06,
+      "loss": 0.1699,
+      "step": 59850
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8279519697306424e-06,
+      "loss": 0.1171,
+      "step": 59880
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8274940468050661e-06,
+      "loss": 0.1478,
+      "step": 59910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8270364678519841e-06,
+      "loss": 0.1364,
+      "step": 59940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8265792324409817e-06,
+      "loss": 0.1339,
+      "step": 59970
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8261223401423995e-06,
+      "loss": 0.1459,
+      "step": 60000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8256657905273283e-06,
+      "loss": 0.1115,
+      "step": 60030
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8252095831676105e-06,
+      "loss": 0.1448,
+      "step": 60060
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8247537176358363e-06,
+      "loss": 0.1464,
+      "step": 60090
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.824298193505343e-06,
+      "loss": 0.1243,
+      "step": 60120
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8238430103502131e-06,
+      "loss": 0.1311,
+      "step": 60150
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8233881677452723e-06,
+      "loss": 0.1542,
+      "step": 60180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8229336652660886e-06,
+      "loss": 0.1457,
+      "step": 60210
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8224795024889702e-06,
+      "loss": 0.1127,
+      "step": 60240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8220256789909638e-06,
+      "loss": 0.1399,
+      "step": 60270
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8215721943498523e-06,
+      "loss": 0.1329,
+      "step": 60300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8211190481441551e-06,
+      "loss": 0.1228,
+      "step": 60330
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.820666239953124e-06,
+      "loss": 0.1176,
+      "step": 60360
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.820213769356744e-06,
+      "loss": 0.1104,
+      "step": 60390
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8197616359357285e-06,
+      "loss": 0.0979,
+      "step": 60420
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8193098392715216e-06,
+      "loss": 0.1361,
+      "step": 60450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8188583789462934e-06,
+      "loss": 0.1128,
+      "step": 60480
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8184072545429393e-06,
+      "loss": 0.1394,
+      "step": 60510
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8179564656450797e-06,
+      "loss": 0.1262,
+      "step": 60540
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.817506011837055e-06,
+      "loss": 0.164,
+      "step": 60570
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8170558927039284e-06,
+      "loss": 0.1167,
+      "step": 60600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8166061078314813e-06,
+      "loss": 0.1234,
+      "step": 60630
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8161566568062119e-06,
+      "loss": 0.1186,
+      "step": 60660
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8157075392153348e-06,
+      "loss": 0.1331,
+      "step": 60690
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.815273708771384e-06,
+      "loss": 0.1283,
+      "step": 60720
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8148252457333782e-06,
+      "loss": 0.1571,
+      "step": 60750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8143771149093633e-06,
+      "loss": 0.1229,
+      "step": 60780
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.813929315889378e-06,
+      "loss": 0.1486,
+      "step": 60810
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.813481848264168e-06,
+      "loss": 0.132,
+      "step": 60840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8130347116251856e-06,
+      "loss": 0.1496,
+      "step": 60870
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8125879055645884e-06,
+      "loss": 0.1164,
+      "step": 60900
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8121414296752363e-06,
+      "loss": 0.1095,
+      "step": 60930
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8116952835506918e-06,
+      "loss": 0.1405,
+      "step": 60960
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8112494667852163e-06,
+      "loss": 0.1231,
+      "step": 60990
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8108039789737708e-06,
+      "loss": 0.1263,
+      "step": 61020
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8103588197120128e-06,
+      "loss": 0.125,
+      "step": 61050
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8099139885962961e-06,
+      "loss": 0.138,
+      "step": 61080
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8094694852236673e-06,
+      "loss": 0.129,
+      "step": 61110
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8090253091918654e-06,
+      "loss": 0.1303,
+      "step": 61140
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8085814600993218e-06,
+      "loss": 0.1323,
+      "step": 61170
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8081379375451555e-06,
+      "loss": 0.1063,
+      "step": 61200
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.807694741129174e-06,
+      "loss": 0.0986,
+      "step": 61230
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8072518704518715e-06,
+      "loss": 0.1299,
+      "step": 61260
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8068093251144268e-06,
+      "loss": 0.1464,
+      "step": 61290
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.806367104718701e-06,
+      "loss": 0.1312,
+      "step": 61320
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.805925208867238e-06,
+      "loss": 0.1306,
+      "step": 61350
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8054836371632622e-06,
+      "loss": 0.1175,
+      "step": 61380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8050423892106764e-06,
+      "loss": 0.1434,
+      "step": 61410
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8046014646140606e-06,
+      "loss": 0.1421,
+      "step": 61440
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.80416086297867e-06,
+      "loss": 0.117,
+      "step": 61470
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8037205839104353e-06,
+      "loss": 0.1353,
+      "step": 61500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8032806270159595e-06,
+      "loss": 0.1557,
+      "step": 61530
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8028409919025168e-06,
+      "loss": 0.1324,
+      "step": 61560
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.802401678178052e-06,
+      "loss": 0.1611,
+      "step": 61590
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8019626854511773e-06,
+      "loss": 0.1378,
+      "step": 61620
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8015240133311723e-06,
+      "loss": 0.1334,
+      "step": 61650
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8010856614279822e-06,
+      "loss": 0.1509,
+      "step": 61680
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8006476293522169e-06,
+      "loss": 0.1196,
+      "step": 61710
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8002099167151475e-06,
+      "loss": 0.1628,
+      "step": 61740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7997725231287063e-06,
+      "loss": 0.1287,
+      "step": 61770
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7993354482054874e-06,
+      "loss": 0.1063,
+      "step": 61800
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7988986915587407e-06,
+      "loss": 0.1428,
+      "step": 61830
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.798462252802374e-06,
+      "loss": 0.1239,
+      "step": 61860
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7980261315509503e-06,
+      "loss": 0.1267,
+      "step": 61890
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7975903274196872e-06,
+      "loss": 0.1421,
+      "step": 61920
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7971548400244536e-06,
+      "loss": 0.1186,
+      "step": 61950
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7967196689817703e-06,
+      "loss": 0.1345,
+      "step": 61980
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7962848139088083e-06,
+      "loss": 0.127,
+      "step": 62010
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7958502744233853e-06,
+      "loss": 0.12,
+      "step": 62040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7954160501439674e-06,
+      "loss": 0.1149,
+      "step": 62070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.794982140689665e-06,
+      "loss": 0.1273,
+      "step": 62100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7945485456802333e-06,
+      "loss": 0.1217,
+      "step": 62130
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7941152647360701e-06,
+      "loss": 0.1355,
+      "step": 62160
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7936822974782144e-06,
+      "loss": 0.1411,
+      "step": 62190
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7932496435283447e-06,
+      "loss": 0.1448,
+      "step": 62220
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7928173025087783e-06,
+      "loss": 0.1502,
+      "step": 62250
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7923852740424697e-06,
+      "loss": 0.1166,
+      "step": 62280
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7919535577530084e-06,
+      "loss": 0.1408,
+      "step": 62310
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7915221532646194e-06,
+      "loss": 0.1304,
+      "step": 62340
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7910910602021597e-06,
+      "loss": 0.1646,
+      "step": 62370
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7906602781911183e-06,
+      "loss": 0.117,
+      "step": 62400
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7902298068576147e-06,
+      "loss": 0.1251,
+      "step": 62430
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.789799645828396e-06,
+      "loss": 0.134,
+      "step": 62460
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7893697947308383e-06,
+      "loss": 0.1444,
+      "step": 62490
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.788940253192943e-06,
+      "loss": 0.1239,
+      "step": 62520
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.788511020843337e-06,
+      "loss": 0.1473,
+      "step": 62550
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.78808209731127e-06,
+      "loss": 0.1201,
+      "step": 62580
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7876534822266136e-06,
+      "loss": 0.1015,
+      "step": 62610
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7872251752198607e-06,
+      "loss": 0.1279,
+      "step": 62640
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.786797175922124e-06,
+      "loss": 0.1349,
+      "step": 62670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7863694839651327e-06,
+      "loss": 0.1354,
+      "step": 62700
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7859420989812349e-06,
+      "loss": 0.1395,
+      "step": 62730
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7855150206033921e-06,
+      "loss": 0.1443,
+      "step": 62760
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7850882484651816e-06,
+      "loss": 0.1144,
+      "step": 62790
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.784661782200792e-06,
+      "loss": 0.1383,
+      "step": 62820
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7842356214450247e-06,
+      "loss": 0.1158,
+      "step": 62850
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.78380976583329e-06,
+      "loss": 0.1306,
+      "step": 62880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7833983951228248e-06,
+      "loss": 0.1603,
+      "step": 62910
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7829731385664483e-06,
+      "loss": 0.1303,
+      "step": 62940
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7825481860760678e-06,
+      "loss": 0.1474,
+      "step": 62970
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7821235372895012e-06,
+      "loss": 0.142,
+      "step": 63000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7816991918451711e-06,
+      "loss": 0.1243,
+      "step": 63030
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7812751493821018e-06,
+      "loss": 0.1289,
+      "step": 63060
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7808514095399185e-06,
+      "loss": 0.1412,
+      "step": 63090
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7804279719588469e-06,
+      "loss": 0.1573,
+      "step": 63120
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7800048362797105e-06,
+      "loss": 0.1378,
+      "step": 63150
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7795820021439302e-06,
+      "loss": 0.1555,
+      "step": 63180
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7791594691935238e-06,
+      "loss": 0.1078,
+      "step": 63210
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7787372370711024e-06,
+      "loss": 0.1395,
+      "step": 63240
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.778315305419871e-06,
+      "loss": 0.1537,
+      "step": 63270
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7778936738836273e-06,
+      "loss": 0.1295,
+      "step": 63300
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7774723421067596e-06,
+      "loss": 0.1323,
+      "step": 63330
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7770513097342456e-06,
+      "loss": 0.1363,
+      "step": 63360
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7766305764116515e-06,
+      "loss": 0.1299,
+      "step": 63390
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.776210141785131e-06,
+      "loss": 0.1398,
+      "step": 63420
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7757900055014226e-06,
+      "loss": 0.1318,
+      "step": 63450
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7753701672078514e-06,
+      "loss": 0.1221,
+      "step": 63480
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7749506265523237e-06,
+      "loss": 0.1205,
+      "step": 63510
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7745313831833294e-06,
+      "loss": 0.1196,
+      "step": 63540
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7741124367499385e-06,
+      "loss": 0.1355,
+      "step": 63570
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7736937869018014e-06,
+      "loss": 0.1209,
+      "step": 63600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7732754332891463e-06,
+      "loss": 0.1121,
+      "step": 63630
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7728573755627793e-06,
+      "loss": 0.1288,
+      "step": 63660
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7724396133740815e-06,
+      "loss": 0.1344,
+      "step": 63690
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7720221463750099e-06,
+      "loss": 0.1287,
+      "step": 63720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.771604974218094e-06,
+      "loss": 0.1217,
+      "step": 63750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.771188096556437e-06,
+      "loss": 0.1205,
+      "step": 63780
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7707715130437115e-06,
+      "loss": 0.1208,
+      "step": 63810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7703552233341614e-06,
+      "loss": 0.1376,
+      "step": 63840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7699392270825987e-06,
+      "loss": 0.124,
+      "step": 63870
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7695235239444037e-06,
+      "loss": 0.122,
+      "step": 63900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7691081135755214e-06,
+      "loss": 0.1829,
+      "step": 63930
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7686929956324636e-06,
+      "loss": 0.1384,
+      "step": 63960
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7682781697723047e-06,
+      "loss": 0.1203,
+      "step": 63990
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7678636356526833e-06,
+      "loss": 0.1392,
+      "step": 64020
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7674493929317981e-06,
+      "loss": 0.145,
+      "step": 64050
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7670354412684088e-06,
+      "loss": 0.1608,
+      "step": 64080
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.766621780321835e-06,
+      "loss": 0.1315,
+      "step": 64110
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7662084097519527e-06,
+      "loss": 0.1225,
+      "step": 64140
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.765795329219196e-06,
+      "loss": 0.1484,
+      "step": 64170
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.765382538384554e-06,
+      "loss": 0.1474,
+      "step": 64200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7649700369095715e-06,
+      "loss": 0.1302,
+      "step": 64230
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7645578244563446e-06,
+      "loss": 0.1335,
+      "step": 64260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7641459006875235e-06,
+      "loss": 0.1331,
+      "step": 64290
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7637342652663083e-06,
+      "loss": 0.1407,
+      "step": 64320
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7633229178564493e-06,
+      "loss": 0.1179,
+      "step": 64350
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7629118581222452e-06,
+      "loss": 0.1337,
+      "step": 64380
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7625010857285424e-06,
+      "loss": 0.1431,
+      "step": 64410
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7620906003407343e-06,
+      "loss": 0.121,
+      "step": 64440
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7616804016247588e-06,
+      "loss": 0.1222,
+      "step": 64470
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7612704892470981e-06,
+      "loss": 0.1214,
+      "step": 64500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7608608628747776e-06,
+      "loss": 0.1377,
+      "step": 64530
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7604515221753638e-06,
+      "loss": 0.138,
+      "step": 64560
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7600424668169652e-06,
+      "loss": 0.1408,
+      "step": 64590
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.759633696468229e-06,
+      "loss": 0.1168,
+      "step": 64620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7592252107983406e-06,
+      "loss": 0.1144,
+      "step": 64650
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7588170094770235e-06,
+      "loss": 0.1349,
+      "step": 64680
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.758409092174537e-06,
+      "loss": 0.133,
+      "step": 64710
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7580014585616753e-06,
+      "loss": 0.1372,
+      "step": 64740
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7575941083097669e-06,
+      "loss": 0.1289,
+      "step": 64770
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.757187041090673e-06,
+      "loss": 0.1299,
+      "step": 64800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.756780256576786e-06,
+      "loss": 0.135,
+      "step": 64830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.756373754441031e-06,
+      "loss": 0.1343,
+      "step": 64860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7559675343568594e-06,
+      "loss": 0.1417,
+      "step": 64890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7555615959982538e-06,
+      "loss": 0.1317,
+      "step": 64920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7551559390397228e-06,
+      "loss": 0.1475,
+      "step": 64950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7547505631563013e-06,
+      "loss": 0.1257,
+      "step": 64980
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.75434546802355e-06,
+      "loss": 0.1235,
+      "step": 65010
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.753954142626517e-06,
+      "loss": 0.1328,
+      "step": 65040
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7535495986923073e-06,
+      "loss": 0.1292,
+      "step": 65070
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7531453345493427e-06,
+      "loss": 0.1148,
+      "step": 65100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7527548115319052e-06,
+      "loss": 0.1447,
+      "step": 65130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7523510967051263e-06,
+      "loss": 0.1528,
+      "step": 65160
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.751947660714751e-06,
+      "loss": 0.1356,
+      "step": 65190
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7515445032399503e-06,
+      "loss": 0.1249,
+      "step": 65220
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7511416239604125e-06,
+      "loss": 0.1283,
+      "step": 65250
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7507390225563407e-06,
+      "loss": 0.1367,
+      "step": 65280
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.750336698708453e-06,
+      "loss": 0.1744,
+      "step": 65310
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7499346520979806e-06,
+      "loss": 0.13,
+      "step": 65340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.749532882406668e-06,
+      "loss": 0.1157,
+      "step": 65370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.749131389316768e-06,
+      "loss": 0.1433,
+      "step": 65400
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.748730172511047e-06,
+      "loss": 0.1172,
+      "step": 65430
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7483292316727791e-06,
+      "loss": 0.133,
+      "step": 65460
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7479285664857462e-06,
+      "loss": 0.1475,
+      "step": 65490
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7475281766342374e-06,
+      "loss": 0.1359,
+      "step": 65520
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7471280618030487e-06,
+      "loss": 0.1233,
+      "step": 65550
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7467282216774797e-06,
+      "loss": 0.1339,
+      "step": 65580
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7463286559433349e-06,
+      "loss": 0.1281,
+      "step": 65610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7459293642869206e-06,
+      "loss": 0.1236,
+      "step": 65640
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7455303463950469e-06,
+      "loss": 0.1084,
+      "step": 65670
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7451316019550222e-06,
+      "loss": 0.1366,
+      "step": 65700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7447331306546573e-06,
+      "loss": 0.1339,
+      "step": 65730
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7443349321822589e-06,
+      "loss": 0.1312,
+      "step": 65760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.743937006226634e-06,
+      "loss": 0.148,
+      "step": 65790
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7435393524770849e-06,
+      "loss": 0.1354,
+      "step": 65820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7431419706234102e-06,
+      "loss": 0.1295,
+      "step": 65850
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7427448603559022e-06,
+      "loss": 0.1206,
+      "step": 65880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7423480213653484e-06,
+      "loss": 0.1504,
+      "step": 65910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.741951453343028e-06,
+      "loss": 0.1348,
+      "step": 65940
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7415551559807117e-06,
+      "loss": 0.1427,
+      "step": 65970
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7411591289706613e-06,
+      "loss": 0.1443,
+      "step": 66000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.740763372005628e-06,
+      "loss": 0.1161,
+      "step": 66030
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.740367884778852e-06,
+      "loss": 0.1284,
+      "step": 66060
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.73997266698406e-06,
+      "loss": 0.1192,
+      "step": 66090
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.739577718315467e-06,
+      "loss": 0.1527,
+      "step": 66120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7391830384677721e-06,
+      "loss": 0.1139,
+      "step": 66150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7387886271361605e-06,
+      "loss": 0.1251,
+      "step": 66180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.738394484016299e-06,
+      "loss": 0.1388,
+      "step": 66210
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7380006088043402e-06,
+      "loss": 0.1317,
+      "step": 66240
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.737607001196915e-06,
+      "loss": 0.1217,
+      "step": 66270
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7372136608911374e-06,
+      "loss": 0.1292,
+      "step": 66300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7368205875846004e-06,
+      "loss": 0.1529,
+      "step": 66330
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.736427780975375e-06,
+      "loss": 0.1139,
+      "step": 66360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7360352407620109e-06,
+      "loss": 0.1318,
+      "step": 66390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7356429666435347e-06,
+      "loss": 0.1351,
+      "step": 66420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7352509583194484e-06,
+      "loss": 0.1298,
+      "step": 66450
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7348592154897293e-06,
+      "loss": 0.1234,
+      "step": 66480
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7344677378548272e-06,
+      "loss": 0.1144,
+      "step": 66510
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.734076525115667e-06,
+      "loss": 0.1293,
+      "step": 66540
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7336855769736436e-06,
+      "loss": 0.1262,
+      "step": 66570
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7332948931306248e-06,
+      "loss": 0.1397,
+      "step": 66600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7329044732889466e-06,
+      "loss": 0.1356,
+      "step": 66630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7325143171514161e-06,
+      "loss": 0.1253,
+      "step": 66660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7321244244213068e-06,
+      "loss": 0.1445,
+      "step": 66690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7317347948023603e-06,
+      "loss": 0.1342,
+      "step": 66720
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.731345427998784e-06,
+      "loss": 0.1199,
+      "step": 66750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7309563237152517e-06,
+      "loss": 0.1211,
+      "step": 66780
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7305674816569004e-06,
+      "loss": 0.1161,
+      "step": 66810
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7301789015293308e-06,
+      "loss": 0.1222,
+      "step": 66840
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7297905830386069e-06,
+      "loss": 0.1463,
+      "step": 66870
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7294025258912534e-06,
+      "loss": 0.1206,
+      "step": 66900
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.729014729794256e-06,
+      "loss": 0.1558,
+      "step": 66930
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.72862719445506e-06,
+      "loss": 0.1534,
+      "step": 66960
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7282399195815697e-06,
+      "loss": 0.1256,
+      "step": 66990
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7278529048821474e-06,
+      "loss": 0.1624,
+      "step": 67020
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.727466150065612e-06,
+      "loss": 0.1436,
+      "step": 67050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.727079654841238e-06,
+      "loss": 0.1193,
+      "step": 67080
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7266934189187565e-06,
+      "loss": 0.1208,
+      "step": 67110
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7263074420083515e-06,
+      "loss": 0.1451,
+      "step": 67140
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7259217238206608e-06,
+      "loss": 0.1378,
+      "step": 67170
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7255362640667738e-06,
+      "loss": 0.1341,
+      "step": 67200
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.725151062458233e-06,
+      "loss": 0.1367,
+      "step": 67230
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7247661187070297e-06,
+      "loss": 0.1245,
+      "step": 67260
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.724394251251618e-06,
+      "loss": 0.1341,
+      "step": 67290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.724009813781396e-06,
+      "loss": 0.1256,
+      "step": 67320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7236256333167308e-06,
+      "loss": 0.1293,
+      "step": 67350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7232417095713918e-06,
+      "loss": 0.1201,
+      "step": 67380
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7228580422595966e-06,
+      "loss": 0.132,
+      "step": 67410
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7224746310960061e-06,
+      "loss": 0.1452,
+      "step": 67440
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7220914757957266e-06,
+      "loss": 0.1442,
+      "step": 67470
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.721708576074308e-06,
+      "loss": 0.1237,
+      "step": 67500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.721325931647741e-06,
+      "loss": 0.1099,
+      "step": 67530
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.72094354223246e-06,
+      "loss": 0.1321,
+      "step": 67560
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7205614075453395e-06,
+      "loss": 0.1209,
+      "step": 67590
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.720179527303693e-06,
+      "loss": 0.1123,
+      "step": 67620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7197979012252745e-06,
+      "loss": 0.1219,
+      "step": 67650
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7194165290282747e-06,
+      "loss": 0.1198,
+      "step": 67680
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7190354104313222e-06,
+      "loss": 0.1125,
+      "step": 67710
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7186545451534831e-06,
+      "loss": 0.1147,
+      "step": 67740
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7182739329142563e-06,
+      "loss": 0.1279,
+      "step": 67770
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7178935734335775e-06,
+      "loss": 0.1507,
+      "step": 67800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7175134664318157e-06,
+      "loss": 0.1235,
+      "step": 67830
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.717133611629773e-06,
+      "loss": 0.111,
+      "step": 67860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7167540087486825e-06,
+      "loss": 0.111,
+      "step": 67890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.71637465751021e-06,
+      "loss": 0.1187,
+      "step": 67920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7159955576364506e-06,
+      "loss": 0.1197,
+      "step": 67950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7156167088499294e-06,
+      "loss": 0.1162,
+      "step": 67980
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7152381108735994e-06,
+      "loss": 0.1414,
+      "step": 68010
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.714859763430842e-06,
+      "loss": 0.1314,
+      "step": 68040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7144816662454657e-06,
+      "loss": 0.1337,
+      "step": 68070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7141038190417047e-06,
+      "loss": 0.1477,
+      "step": 68100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7137262215442186e-06,
+      "loss": 0.1408,
+      "step": 68130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.713348873478091e-06,
+      "loss": 0.1232,
+      "step": 68160
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7129717745688293e-06,
+      "loss": 0.1486,
+      "step": 68190
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.712594924542364e-06,
+      "loss": 0.1385,
+      "step": 68220
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7122183231250472e-06,
+      "loss": 0.1458,
+      "step": 68250
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7118419700436517e-06,
+      "loss": 0.1231,
+      "step": 68280
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.711465865025371e-06,
+      "loss": 0.1225,
+      "step": 68310
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7110900077978183e-06,
+      "loss": 0.13,
+      "step": 68340
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7107143980890237e-06,
+      "loss": 0.1255,
+      "step": 68370
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.710339035627437e-06,
+      "loss": 0.1142,
+      "step": 68400
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7099639201419239e-06,
+      "loss": 0.132,
+      "step": 68430
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7095890513617658e-06,
+      "loss": 0.1368,
+      "step": 68460
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.709214429016661e-06,
+      "loss": 0.1405,
+      "step": 68490
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7088400528367202e-06,
+      "loss": 0.1366,
+      "step": 68520
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.708465922552469e-06,
+      "loss": 0.1501,
+      "step": 68550
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7080920378948456e-06,
+      "loss": 0.1163,
+      "step": 68580
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7077183985952001e-06,
+      "loss": 0.1612,
+      "step": 68610
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7073450043852934e-06,
+      "loss": 0.1368,
+      "step": 68640
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7069718549972975e-06,
+      "loss": 0.1204,
+      "step": 68670
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.706598950163793e-06,
+      "loss": 0.1105,
+      "step": 68700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7062262896177706e-06,
+      "loss": 0.1209,
+      "step": 68730
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7058538730926278e-06,
+      "loss": 0.1472,
+      "step": 68760
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7054817003221697e-06,
+      "loss": 0.1081,
+      "step": 68790
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7051097710406078e-06,
+      "loss": 0.1353,
+      "step": 68820
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7047380849825593e-06,
+      "loss": 0.1516,
+      "step": 68850
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7043666418830458e-06,
+      "loss": 0.1594,
+      "step": 68880
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7039954414774928e-06,
+      "loss": 0.1348,
+      "step": 68910
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.703624483501729e-06,
+      "loss": 0.1525,
+      "step": 68940
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.703253767691987e-06,
+      "loss": 0.1299,
+      "step": 68970
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7028832937848983e-06,
+      "loss": 0.1177,
+      "step": 69000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7025130615174974e-06,
+      "loss": 0.129,
+      "step": 69030
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7021430706272177e-06,
+      "loss": 0.1141,
+      "step": 69060
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7017733208518927e-06,
+      "loss": 0.1381,
+      "step": 69090
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7014038119297534e-06,
+      "loss": 0.1344,
+      "step": 69120
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7010345435994293e-06,
+      "loss": 0.1369,
+      "step": 69150
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7006655155999465e-06,
+      "loss": 0.1376,
+      "step": 69180
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.700296727670727e-06,
+      "loss": 0.1413,
+      "step": 69210
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.699928179551589e-06,
+      "loss": 0.1448,
+      "step": 69240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6995598709827443e-06,
+      "loss": 0.159,
+      "step": 69270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.6991918017047987e-06,
+      "loss": 0.117,
+      "step": 69300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.698823971458752e-06,
+      "loss": 0.1308,
+      "step": 69330
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.6984563799859952e-06,
+      "loss": 0.1395,
+      "step": 69360
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.6980890270283112e-06,
+      "loss": 0.1099,
+      "step": 69390
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.697721912327874e-06,
+      "loss": 0.132,
+      "step": 69420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.6973550356272473e-06,
+      "loss": 0.1263,
+      "step": 69450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.6969883966693837e-06,
+      "loss": 0.1259,
+      "step": 69480
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.6966219951976253e-06,
+      "loss": 0.1129,
+      "step": 69510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.696255830955701e-06,
+      "loss": 0.1204,
+      "step": 69540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.695889903687727e-06,
+      "loss": 0.1258,
+      "step": 69570
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6955242131382063e-06,
+      "loss": 0.192,
+      "step": 69600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6951587590520263e-06,
+      "loss": 0.1266,
+      "step": 69630
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6947935411744596e-06,
+      "loss": 0.1154,
+      "step": 69660
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.694428559251163e-06,
+      "loss": 0.1221,
+      "step": 69690
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.694063813028177e-06,
+      "loss": 0.1197,
+      "step": 69720
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6936993022519233e-06,
+      "loss": 0.1561,
+      "step": 69750
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.693335026669207e-06,
+      "loss": 0.1296,
+      "step": 69780
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6929709860272125e-06,
+      "loss": 0.136,
+      "step": 69810
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.6926071800735058e-06,
+      "loss": 0.1368,
+      "step": 69840
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6922436085560317e-06,
+      "loss": 0.1193,
+      "step": 69870
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6918802712231148e-06,
+      "loss": 0.1259,
+      "step": 69900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6915171678234566e-06,
+      "loss": 0.1172,
+      "step": 69930
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6911542981061369e-06,
+      "loss": 0.142,
+      "step": 69960
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6907916618206113e-06,
+      "loss": 0.163,
+      "step": 69990
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.690429258716712e-06,
+      "loss": 0.1358,
+      "step": 70020
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6900670885446464e-06,
+      "loss": 0.1268,
+      "step": 70050
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6897051510549954e-06,
+      "loss": 0.1575,
+      "step": 70080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.6893434459987152e-06,
+      "loss": 0.1099,
+      "step": 70110
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.688981973127133e-06,
+      "loss": 0.1692,
+      "step": 70140
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6886207321919508e-06,
+      "loss": 0.1387,
+      "step": 70170
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6882597229452405e-06,
+      "loss": 0.1381,
+      "step": 70200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6878989451394446e-06,
+      "loss": 0.1406,
+      "step": 70230
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6875383985273766e-06,
+      "loss": 0.1322,
+      "step": 70260
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6871780828622196e-06,
+      "loss": 0.1357,
+      "step": 70290
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6868179978975248e-06,
+      "loss": 0.129,
+      "step": 70320
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6864581433872124e-06,
+      "loss": 0.136,
+      "step": 70350
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.6860985190855678e-06,
+      "loss": 0.1737,
+      "step": 70380
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6857391247472457e-06,
+      "loss": 0.1339,
+      "step": 70410
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6853799601272647e-06,
+      "loss": 0.1383,
+      "step": 70440
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6850210249810098e-06,
+      "loss": 0.1398,
+      "step": 70470
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.68466231906423e-06,
+      "loss": 0.133,
+      "step": 70500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6843038421330383e-06,
+      "loss": 0.1246,
+      "step": 70530
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.68394559394391e-06,
+      "loss": 0.1165,
+      "step": 70560
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6835875742536848e-06,
+      "loss": 0.1417,
+      "step": 70590
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.6832297828195618e-06,
+      "loss": 0.1204,
+      "step": 70620
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6828722193991028e-06,
+      "loss": 0.1198,
+      "step": 70650
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6825148837502292e-06,
+      "loss": 0.1255,
+      "step": 70680
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6821577756312225e-06,
+      "loss": 0.108,
+      "step": 70710
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6818008948007225e-06,
+      "loss": 0.1433,
+      "step": 70740
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.681444241017728e-06,
+      "loss": 0.1285,
+      "step": 70770
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6810878140415945e-06,
+      "loss": 0.148,
+      "step": 70800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.680731613632036e-06,
+      "loss": 0.1284,
+      "step": 70830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6803756395491212e-06,
+      "loss": 0.1459,
+      "step": 70860
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.6800198915532748e-06,
+      "loss": 0.132,
+      "step": 70890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.679664369405277e-06,
+      "loss": 0.137,
+      "step": 70920
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6793090728662615e-06,
+      "loss": 0.1349,
+      "step": 70950
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6789540016977157e-06,
+      "loss": 0.1346,
+      "step": 70980
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.67859915566148e-06,
+      "loss": 0.1209,
+      "step": 71010
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.678244534519747e-06,
+      "loss": 0.1226,
+      "step": 71040
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6778901380350609e-06,
+      "loss": 0.1397,
+      "step": 71070
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6775359659703164e-06,
+      "loss": 0.1666,
+      "step": 71100
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.677182018088759e-06,
+      "loss": 0.1219,
+      "step": 71130
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.6768282941539827e-06,
+      "loss": 0.1102,
+      "step": 71160
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6764747939299313e-06,
+      "loss": 0.1129,
+      "step": 71190
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.676121517180897e-06,
+      "loss": 0.1192,
+      "step": 71220
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6757684636715186e-06,
+      "loss": 0.1267,
+      "step": 71250
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.675415633166782e-06,
+      "loss": 0.1253,
+      "step": 71280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6750630254320201e-06,
+      "loss": 0.1526,
+      "step": 71310
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6747106402329109e-06,
+      "loss": 0.1352,
+      "step": 71340
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6743584773354766e-06,
+      "loss": 0.124,
+      "step": 71370
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6740065365060848e-06,
+      "loss": 0.1198,
+      "step": 71400
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.6736548175114462e-06,
+      "loss": 0.119,
+      "step": 71430
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6733033201186144e-06,
+      "loss": 0.1203,
+      "step": 71460
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6729520440949857e-06,
+      "loss": 0.1236,
+      "step": 71490
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6726009892082972e-06,
+      "loss": 0.137,
+      "step": 71520
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.672250155226628e-06,
+      "loss": 0.1126,
+      "step": 71550
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6718995419183973e-06,
+      "loss": 0.1264,
+      "step": 71580
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6715491490523637e-06,
+      "loss": 0.1052,
+      "step": 71610
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6711989763976253e-06,
+      "loss": 0.1434,
+      "step": 71640
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6708606852710505e-06,
+      "loss": 0.164,
+      "step": 71670
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.6705109450262383e-06,
+      "loss": 0.1292,
+      "step": 71700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6701614243097006e-06,
+      "loss": 0.1275,
+      "step": 71730
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6698121228918743e-06,
+      "loss": 0.1123,
+      "step": 71760
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6694630405435322e-06,
+      "loss": 0.1452,
+      "step": 71790
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6691141770357822e-06,
+      "loss": 0.1189,
+      "step": 71820
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6687655321400668e-06,
+      "loss": 0.1196,
+      "step": 71850
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.668417105628163e-06,
+      "loss": 0.1111,
+      "step": 71880
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6680688972721806e-06,
+      "loss": 0.1195,
+      "step": 71910
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.667720906844563e-06,
+      "loss": 0.1359,
+      "step": 71940
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.6673731341180851e-06,
+      "loss": 0.1419,
+      "step": 71970
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6670255788658535e-06,
+      "loss": 0.1338,
+      "step": 72000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6666782408613056e-06,
+      "loss": 0.1178,
+      "step": 72030
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6663311198782094e-06,
+      "loss": 0.1335,
+      "step": 72060
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6659842156906629e-06,
+      "loss": 0.1428,
+      "step": 72090
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.665637528073092e-06,
+      "loss": 0.1141,
+      "step": 72120
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6652910568002523e-06,
+      "loss": 0.1324,
+      "step": 72150
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6649448016472265e-06,
+      "loss": 0.1105,
+      "step": 72180
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.6645987623894243e-06,
+      "loss": 0.1171,
+      "step": 72210
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.664252938802584e-06,
+      "loss": 0.1304,
+      "step": 72240
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6639188474653653e-06,
+      "loss": 0.1418,
+      "step": 72270
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6635734473784454e-06,
+      "loss": 0.1655,
+      "step": 72300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6632282622990878e-06,
+      "loss": 0.1287,
+      "step": 72330
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.662883292004319e-06,
+      "loss": 0.1168,
+      "step": 72360
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6625385362714889e-06,
+      "loss": 0.1226,
+      "step": 72390
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6621939948782702e-06,
+      "loss": 0.114,
+      "step": 72420
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6618496676026586e-06,
+      "loss": 0.1378,
+      "step": 72450
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.661505554222971e-06,
+      "loss": 0.1486,
+      "step": 72480
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.6611616545178456e-06,
+      "loss": 0.166,
+      "step": 72510
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6608179682662409e-06,
+      "loss": 0.127,
+      "step": 72540
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6604744952474365e-06,
+      "loss": 0.132,
+      "step": 72570
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6601312352410301e-06,
+      "loss": 0.1409,
+      "step": 72600
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6597881880269391e-06,
+      "loss": 0.1251,
+      "step": 72630
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6594453533853988e-06,
+      "loss": 0.1233,
+      "step": 72660
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6591027310969617e-06,
+      "loss": 0.1673,
+      "step": 72690
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6587603209424982e-06,
+      "loss": 0.1573,
+      "step": 72720
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6584181227031944e-06,
+      "loss": 0.1496,
+      "step": 72750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.6580761361605527e-06,
+      "loss": 0.1375,
+      "step": 72780
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.657734361096391e-06,
+      "loss": 0.1405,
+      "step": 72810
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.65739279729284e-06,
+      "loss": 0.1278,
+      "step": 72840
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6570514445323473e-06,
+      "loss": 0.1159,
+      "step": 72870
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6567103025976724e-06,
+      "loss": 0.1367,
+      "step": 72900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6563693712718881e-06,
+      "loss": 0.1064,
+      "step": 72930
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6560286503383794e-06,
+      "loss": 0.1339,
+      "step": 72960
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6556881395808428e-06,
+      "loss": 0.1459,
+      "step": 72990
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6553478387832867e-06,
+      "loss": 0.1296,
+      "step": 73020
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.6550077477300302e-06,
+      "loss": 0.1061,
+      "step": 73050
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6546678662057017e-06,
+      "loss": 0.1268,
+      "step": 73080
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.654328193995239e-06,
+      "loss": 0.1635,
+      "step": 73110
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.65398873088389e-06,
+      "loss": 0.1153,
+      "step": 73140
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.65364947665721e-06,
+      "loss": 0.1259,
+      "step": 73170
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.653310431101062e-06,
+      "loss": 0.1366,
+      "step": 73200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.652971594001617e-06,
+      "loss": 0.1298,
+      "step": 73230
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6526329651453512e-06,
+      "loss": 0.125,
+      "step": 73260
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6522945443190482e-06,
+      "loss": 0.1218,
+      "step": 73290
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.6519563313097975e-06,
+      "loss": 0.1287,
+      "step": 73320
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6516183259049915e-06,
+      "loss": 0.1331,
+      "step": 73350
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6512805278923288e-06,
+      "loss": 0.1167,
+      "step": 73380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6509429370598115e-06,
+      "loss": 0.1302,
+      "step": 73410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6506055531957439e-06,
+      "loss": 0.1303,
+      "step": 73440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6502683760887347e-06,
+      "loss": 0.1249,
+      "step": 73470
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6499314055276932e-06,
+      "loss": 0.1246,
+      "step": 73500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6495946413018313e-06,
+      "loss": 0.1197,
+      "step": 73530
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6492580832006616e-06,
+      "loss": 0.1325,
+      "step": 73560
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.6489217310139972e-06,
+      "loss": 0.1582,
+      "step": 73590
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6485855845319507e-06,
+      "loss": 0.1284,
+      "step": 73620
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6482496435449352e-06,
+      "loss": 0.1203,
+      "step": 73650
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6479139078436612e-06,
+      "loss": 0.1214,
+      "step": 73680
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6475783772191392e-06,
+      "loss": 0.116,
+      "step": 73710
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6472430514626755e-06,
+      "loss": 0.1251,
+      "step": 73740
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6469079303658753e-06,
+      "loss": 0.141,
+      "step": 73770
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6465730137206392e-06,
+      "loss": 0.1403,
+      "step": 73800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.6462383013191651e-06,
+      "loss": 0.145,
+      "step": 73830
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.645903792953945e-06,
+      "loss": 0.1302,
+      "step": 73860
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.645569488417768e-06,
+      "loss": 0.1324,
+      "step": 73890
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6452353875037153e-06,
+      "loss": 0.1237,
+      "step": 73920
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6449014900051635e-06,
+      "loss": 0.1183,
+      "step": 73950
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.644567795715783e-06,
+      "loss": 0.1438,
+      "step": 73980
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6442343044295355e-06,
+      "loss": 0.0983,
+      "step": 74010
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6439010159406765e-06,
+      "loss": 0.125,
+      "step": 74040
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6435679300437526e-06,
+      "loss": 0.1353,
+      "step": 74070
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6432350465336022e-06,
+      "loss": 0.1112,
+      "step": 74100
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.6429023652053536e-06,
+      "loss": 0.1583,
+      "step": 74130
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.642569885854426e-06,
+      "loss": 0.1328,
+      "step": 74160
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6422376082765283e-06,
+      "loss": 0.129,
+      "step": 74190
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.641905532267658e-06,
+      "loss": 0.136,
+      "step": 74220
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6415736576241027e-06,
+      "loss": 0.1275,
+      "step": 74250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6412419841424357e-06,
+      "loss": 0.1138,
+      "step": 74280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6409105116195198e-06,
+      "loss": 0.1318,
+      "step": 74310
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6405792398525046e-06,
+      "loss": 0.134,
+      "step": 74340
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6402481686388257e-06,
+      "loss": 0.1285,
+      "step": 74370
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6399172977762055e-06,
+      "loss": 0.1327,
+      "step": 74400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6395866270626506e-06,
+      "loss": 0.1371,
+      "step": 74430
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6392561562964538e-06,
+      "loss": 0.1307,
+      "step": 74460
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.638925885276192e-06,
+      "loss": 0.1189,
+      "step": 74490
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6385958138007257e-06,
+      "loss": 0.1392,
+      "step": 74520
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6382659416692e-06,
+      "loss": 0.1304,
+      "step": 74550
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6379362686810413e-06,
+      "loss": 0.1366,
+      "step": 74580
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6376067946359594e-06,
+      "loss": 0.122,
+      "step": 74610
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.637277519333946e-06,
+      "loss": 0.117,
+      "step": 74640
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6369484425752744e-06,
+      "loss": 0.1312,
+      "step": 74670
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6366195641604977e-06,
+      "loss": 0.1446,
+      "step": 74700
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6362908838904503e-06,
+      "loss": 0.1355,
+      "step": 74730
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6359624015662468e-06,
+      "loss": 0.112,
+      "step": 74760
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.63563411698928e-06,
+      "loss": 0.1381,
+      "step": 74790
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6353060299612225e-06,
+      "loss": 0.1352,
+      "step": 74820
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6349781402840249e-06,
+      "loss": 0.1279,
+      "step": 74850
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.634650447759916e-06,
+      "loss": 0.1435,
+      "step": 74880
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6343229521914015e-06,
+      "loss": 0.1055,
+      "step": 74910
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6339956533812637e-06,
+      "loss": 0.1178,
+      "step": 74940
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6336685511325629e-06,
+      "loss": 0.1602,
+      "step": 74970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.633341645248633e-06,
+      "loss": 0.1273,
+      "step": 75000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6330149355330848e-06,
+      "loss": 0.1194,
+      "step": 75030
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6326884217898035e-06,
+      "loss": 0.1113,
+      "step": 75060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6323621038229485e-06,
+      "loss": 0.1342,
+      "step": 75090
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6320359814369536e-06,
+      "loss": 0.1232,
+      "step": 75120
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6317100544365252e-06,
+      "loss": 0.1481,
+      "step": 75150
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6313843226266435e-06,
+      "loss": 0.1225,
+      "step": 75180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6310587858125606e-06,
+      "loss": 0.1109,
+      "step": 75210
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6307334437998005e-06,
+      "loss": 0.128,
+      "step": 75240
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6304082963941586e-06,
+      "loss": 0.1299,
+      "step": 75270
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6300833434017019e-06,
+      "loss": 0.1102,
+      "step": 75300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6297585846287667e-06,
+      "loss": 0.1546,
+      "step": 75330
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.62943401988196e-06,
+      "loss": 0.1139,
+      "step": 75360
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6291096489681585e-06,
+      "loss": 0.1585,
+      "step": 75390
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6287854716945076e-06,
+      "loss": 0.1309,
+      "step": 75420
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6284614878684206e-06,
+      "loss": 0.1177,
+      "step": 75450
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6281376972975807e-06,
+      "loss": 0.1354,
+      "step": 75480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.627814099789936e-06,
+      "loss": 0.1256,
+      "step": 75510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6274906951537042e-06,
+      "loss": 0.1198,
+      "step": 75540
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6271674831973684e-06,
+      "loss": 0.1113,
+      "step": 75570
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6268444637296777e-06,
+      "loss": 0.1301,
+      "step": 75600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.626521636559647e-06,
+      "loss": 0.13,
+      "step": 75630
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6261990014965568e-06,
+      "loss": 0.1195,
+      "step": 75660
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.625876558349952e-06,
+      "loss": 0.1487,
+      "step": 75690
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6255543069296417e-06,
+      "loss": 0.1401,
+      "step": 75720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6252322470456988e-06,
+      "loss": 0.1356,
+      "step": 75750
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.62491037850846e-06,
+      "loss": 0.1466,
+      "step": 75780
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6245887011285238e-06,
+      "loss": 0.14,
+      "step": 75810
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6242672147167518e-06,
+      "loss": 0.1438,
+      "step": 75840
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6239459190842678e-06,
+      "loss": 0.1164,
+      "step": 75870
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6236248140424554e-06,
+      "loss": 0.1274,
+      "step": 75900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6233038994029616e-06,
+      "loss": 0.1271,
+      "step": 75930
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6229831749776917e-06,
+      "loss": 0.1332,
+      "step": 75960
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6226626405788123e-06,
+      "loss": 0.1383,
+      "step": 75990
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6223422960187493e-06,
+      "loss": 0.1406,
+      "step": 76020
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6220221411101878e-06,
+      "loss": 0.1296,
+      "step": 76050
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.621702175666071e-06,
+      "loss": 0.1257,
+      "step": 76080
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6213823994996006e-06,
+      "loss": 0.1331,
+      "step": 76110
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6210628124242367e-06,
+      "loss": 0.1133,
+      "step": 76140
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6207434142536954e-06,
+      "loss": 0.1475,
+      "step": 76170
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6204242048019508e-06,
+      "loss": 0.1178,
+      "step": 76200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.620105183883233e-06,
+      "loss": 0.146,
+      "step": 76230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6197863513120277e-06,
+      "loss": 0.1307,
+      "step": 76260
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6194677069030761e-06,
+      "loss": 0.1189,
+      "step": 76290
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6191492504713747e-06,
+      "loss": 0.1067,
+      "step": 76320
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6188415877632176e-06,
+      "loss": 0.1414,
+      "step": 76350
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6185235004813931e-06,
+      "loss": 0.1605,
+      "step": 76380
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6182056006294704e-06,
+      "loss": 0.1265,
+      "step": 76410
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6178878880234536e-06,
+      "loss": 0.1211,
+      "step": 76440
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6175703624795988e-06,
+      "loss": 0.1318,
+      "step": 76470
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.617253023814414e-06,
+      "loss": 0.1321,
+      "step": 76500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6169358718446608e-06,
+      "loss": 0.1517,
+      "step": 76530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6166189063873493e-06,
+      "loss": 0.1042,
+      "step": 76560
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6163021272597446e-06,
+      "loss": 0.1061,
+      "step": 76590
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6159855342793588e-06,
+      "loss": 0.1301,
+      "step": 76620
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6156691272639551e-06,
+      "loss": 0.112,
+      "step": 76650
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6153529060315482e-06,
+      "loss": 0.1173,
+      "step": 76680
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6150368704003994e-06,
+      "loss": 0.1286,
+      "step": 76710
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6147210201890205e-06,
+      "loss": 0.1261,
+      "step": 76740
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6144053552161706e-06,
+      "loss": 0.1122,
+      "step": 76770
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.614089875300858e-06,
+      "loss": 0.1155,
+      "step": 76800
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.613774580262337e-06,
+      "loss": 0.1299,
+      "step": 76830
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.61345946992011e-06,
+      "loss": 0.1319,
+      "step": 76860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.613144544093926e-06,
+      "loss": 0.141,
+      "step": 76890
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6128298026037798e-06,
+      "loss": 0.1458,
+      "step": 76920
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6125152452699114e-06,
+      "loss": 0.1293,
+      "step": 76950
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6122008719128075e-06,
+      "loss": 0.1072,
+      "step": 76980
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.611886682353198e-06,
+      "loss": 0.1663,
+      "step": 77010
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6115726764120597e-06,
+      "loss": 0.1185,
+      "step": 77040
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6112588539106105e-06,
+      "loss": 0.1152,
+      "step": 77070
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6109452146703141e-06,
+      "loss": 0.1354,
+      "step": 77100
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6106317585128764e-06,
+      "loss": 0.1285,
+      "step": 77130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6103184852602463e-06,
+      "loss": 0.1265,
+      "step": 77160
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6100053947346149e-06,
+      "loss": 0.1186,
+      "step": 77190
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6096924867584152e-06,
+      "loss": 0.1134,
+      "step": 77220
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6093797611543222e-06,
+      "loss": 0.1244,
+      "step": 77250
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6090672177452509e-06,
+      "loss": 0.1286,
+      "step": 77280
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6087548563543582e-06,
+      "loss": 0.1144,
+      "step": 77310
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6084426768050402e-06,
+      "loss": 0.1276,
+      "step": 77340
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.608130678920933e-06,
+      "loss": 0.1158,
+      "step": 77370
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6078188625259127e-06,
+      "loss": 0.148,
+      "step": 77400
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6075072274440942e-06,
+      "loss": 0.1105,
+      "step": 77430
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6071957734998298e-06,
+      "loss": 0.1044,
+      "step": 77460
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6068845005177113e-06,
+      "loss": 0.1486,
+      "step": 77490
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6065734083225676e-06,
+      "loss": 0.1365,
+      "step": 77520
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.606262496739465e-06,
+      "loss": 0.1277,
+      "step": 77550
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6059517655937066e-06,
+      "loss": 0.1579,
+      "step": 77580
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.605641214710832e-06,
+      "loss": 0.1188,
+      "step": 77610
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6053308439166174e-06,
+      "loss": 0.117,
+      "step": 77640
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6050206530370737e-06,
+      "loss": 0.1237,
+      "step": 77670
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6047106418984479e-06,
+      "loss": 0.1474,
+      "step": 77700
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.604400810327221e-06,
+      "loss": 0.1049,
+      "step": 77730
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.604091158150109e-06,
+      "loss": 0.1182,
+      "step": 77760
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6037816851940622e-06,
+      "loss": 0.1306,
+      "step": 77790
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6034723912862638e-06,
+      "loss": 0.1245,
+      "step": 77820
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6031632762541306e-06,
+      "loss": 0.1207,
+      "step": 77850
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.602854339925312e-06,
+      "loss": 0.1281,
+      "step": 77880
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6025455821276895e-06,
+      "loss": 0.1307,
+      "step": 77910
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.602237002689378e-06,
+      "loss": 0.1328,
+      "step": 77940
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6019286014387222e-06,
+      "loss": 0.1234,
+      "step": 77970
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6016203782042993e-06,
+      "loss": 0.1513,
+      "step": 78000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6013123328149165e-06,
+      "loss": 0.1261,
+      "step": 78030
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6010044650996121e-06,
+      "loss": 0.1743,
+      "step": 78060
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6006967748876534e-06,
+      "loss": 0.1145,
+      "step": 78090
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6003892620085383e-06,
+      "loss": 0.1329,
+      "step": 78120
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.600081926291993e-06,
+      "loss": 0.1111,
+      "step": 78150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5997747675679737e-06,
+      "loss": 0.1281,
+      "step": 78180
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5994677856666639e-06,
+      "loss": 0.1112,
+      "step": 78210
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5991609804184753e-06,
+      "loss": 0.1142,
+      "step": 78240
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.598854351654048e-06,
+      "loss": 0.149,
+      "step": 78270
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5985478992042483e-06,
+      "loss": 0.1226,
+      "step": 78300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5982416229001699e-06,
+      "loss": 0.1297,
+      "step": 78330
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.597935522573133e-06,
+      "loss": 0.1264,
+      "step": 78360
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5976295980546836e-06,
+      "loss": 0.144,
+      "step": 78390
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.5973238491765936e-06,
+      "loss": 0.1262,
+      "step": 78420
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5970182757708593e-06,
+      "loss": 0.1319,
+      "step": 78450
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5967128776697042e-06,
+      "loss": 0.1319,
+      "step": 78480
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5964076547055735e-06,
+      "loss": 0.1383,
+      "step": 78510
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5961026067111377e-06,
+      "loss": 0.1365,
+      "step": 78540
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5957977335192917e-06,
+      "loss": 0.1186,
+      "step": 78570
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.595493034963153e-06,
+      "loss": 0.1287,
+      "step": 78600
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5951885108760623e-06,
+      "loss": 0.1259,
+      "step": 78630
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5948841610915825e-06,
+      "loss": 0.1007,
+      "step": 78660
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5945799854434987e-06,
+      "loss": 0.1322,
+      "step": 78690
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.594275983765819e-06,
+      "loss": 0.1383,
+      "step": 78720
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5939721558927715e-06,
+      "loss": 0.1543,
+      "step": 78750
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5936685016588058e-06,
+      "loss": 0.1241,
+      "step": 78780
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5933650208985923e-06,
+      "loss": 0.1312,
+      "step": 78810
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5930617134470222e-06,
+      "loss": 0.1462,
+      "step": 78840
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5927686808283344e-06,
+      "loss": 0.1193,
+      "step": 78870
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5924657137362828e-06,
+      "loss": 0.1061,
+      "step": 78900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5921629194643422e-06,
+      "loss": 0.126,
+      "step": 78930
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.591860297848274e-06,
+      "loss": 0.1174,
+      "step": 78960
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5915578487240574e-06,
+      "loss": 0.1343,
+      "step": 78990
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.591255571927889e-06,
+      "loss": 0.1251,
+      "step": 79020
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5909534672961845e-06,
+      "loss": 0.104,
+      "step": 79050
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.590651534665576e-06,
+      "loss": 0.1217,
+      "step": 79080
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5903497738729133e-06,
+      "loss": 0.1201,
+      "step": 79110
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5900481847552615e-06,
+      "loss": 0.1156,
+      "step": 79140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5897467671499027e-06,
+      "loss": 0.1092,
+      "step": 79170
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5894455208943354e-06,
+      "loss": 0.1086,
+      "step": 79200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5891444458262723e-06,
+      "loss": 0.1487,
+      "step": 79230
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5888435417836425e-06,
+      "loss": 0.1268,
+      "step": 79260
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.588542808604589e-06,
+      "loss": 0.1304,
+      "step": 79290
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5882422461274692e-06,
+      "loss": 0.12,
+      "step": 79320
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5879418541908546e-06,
+      "loss": 0.1192,
+      "step": 79350
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.587641632633531e-06,
+      "loss": 0.1381,
+      "step": 79380
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.587341581294496e-06,
+      "loss": 0.1398,
+      "step": 79410
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5870417000129618e-06,
+      "loss": 0.1147,
+      "step": 79440
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5867419886283522e-06,
+      "loss": 0.1287,
+      "step": 79470
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5864424469803034e-06,
+      "loss": 0.1311,
+      "step": 79500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5861430749086632e-06,
+      "loss": 0.1002,
+      "step": 79530
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5858438722534912e-06,
+      "loss": 0.121,
+      "step": 79560
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5855448388550579e-06,
+      "loss": 0.1285,
+      "step": 79590
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5852459745538446e-06,
+      "loss": 0.124,
+      "step": 79620
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5849472791905432e-06,
+      "loss": 0.1084,
+      "step": 79650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.584648752606055e-06,
+      "loss": 0.1275,
+      "step": 79680
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5843503946414926e-06,
+      "loss": 0.1033,
+      "step": 79710
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5840522051381754e-06,
+      "loss": 0.1243,
+      "step": 79740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5837541839376345e-06,
+      "loss": 0.1235,
+      "step": 79770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.583456330881607e-06,
+      "loss": 0.1354,
+      "step": 79800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5831586458120407e-06,
+      "loss": 0.133,
+      "step": 79830
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5828611285710897e-06,
+      "loss": 0.1308,
+      "step": 79860
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5825637790011156e-06,
+      "loss": 0.1371,
+      "step": 79890
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5822665969446885e-06,
+      "loss": 0.1257,
+      "step": 79920
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5819695822445844e-06,
+      "loss": 0.1381,
+      "step": 79950
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.581672734743786e-06,
+      "loss": 0.1247,
+      "step": 79980
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5813760542854823e-06,
+      "loss": 0.1273,
+      "step": 80010
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.581079540713068e-06,
+      "loss": 0.1241,
+      "step": 80040
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5807831938701429e-06,
+      "loss": 0.1138,
+      "step": 80070
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5804870136005133e-06,
+      "loss": 0.1325,
+      "step": 80100
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5801909997481882e-06,
+      "loss": 0.1183,
+      "step": 80130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5798951521573824e-06,
+      "loss": 0.1255,
+      "step": 80160
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5795994706725152e-06,
+      "loss": 0.1126,
+      "step": 80190
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5793039551382084e-06,
+      "loss": 0.1329,
+      "step": 80220
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5790086053992875e-06,
+      "loss": 0.1352,
+      "step": 80250
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5787134213007817e-06,
+      "loss": 0.1207,
+      "step": 80280
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.578418402687923e-06,
+      "loss": 0.1156,
+      "step": 80310
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5781235494061444e-06,
+      "loss": 0.1152,
+      "step": 80340
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5778288613010825e-06,
+      "loss": 0.1198,
+      "step": 80370
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.577534338218575e-06,
+      "loss": 0.1269,
+      "step": 80400
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5772399800046603e-06,
+      "loss": 0.1328,
+      "step": 80430
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.576945786505579e-06,
+      "loss": 0.1224,
+      "step": 80460
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5766517575677718e-06,
+      "loss": 0.1253,
+      "step": 80490
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5763578930378794e-06,
+      "loss": 0.1651,
+      "step": 80520
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5760641927627434e-06,
+      "loss": 0.0942,
+      "step": 80550
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5757706565894044e-06,
+      "loss": 0.1177,
+      "step": 80580
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5754772843651023e-06,
+      "loss": 0.1323,
+      "step": 80610
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5751840759372764e-06,
+      "loss": 0.1054,
+      "step": 80640
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5748910311535644e-06,
+      "loss": 0.129,
+      "step": 80670
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5745981498618028e-06,
+      "loss": 0.1538,
+      "step": 80700
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5743054319100251e-06,
+      "loss": 0.1182,
+      "step": 80730
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5740128771464635e-06,
+      "loss": 0.1425,
+      "step": 80760
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5737204854195472e-06,
+      "loss": 0.1119,
+      "step": 80790
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5734282565779023e-06,
+      "loss": 0.1081,
+      "step": 80820
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.573136190470352e-06,
+      "loss": 0.1015,
+      "step": 80850
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5728442869459149e-06,
+      "loss": 0.1261,
+      "step": 80880
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5725525458538072e-06,
+      "loss": 0.1303,
+      "step": 80910
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5722609670434387e-06,
+      "loss": 0.1189,
+      "step": 80940
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5719695503644169e-06,
+      "loss": 0.1273,
+      "step": 80970
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.571678295666543e-06,
+      "loss": 0.1485,
+      "step": 81000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5713872027998126e-06,
+      "loss": 0.1234,
+      "step": 81030
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5711059667173108e-06,
+      "loss": 0.1268,
+      "step": 81060
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5708151916816517e-06,
+      "loss": 0.1171,
+      "step": 81090
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.570524578033268e-06,
+      "loss": 0.1064,
+      "step": 81120
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5702341256229242e-06,
+      "loss": 0.1265,
+      "step": 81150
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5699438343015793e-06,
+      "loss": 0.1359,
+      "step": 81180
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.569653703920385e-06,
+      "loss": 0.1267,
+      "step": 81210
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5693637343306845e-06,
+      "loss": 0.1095,
+      "step": 81240
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.569073925384014e-06,
+      "loss": 0.1374,
+      "step": 81270
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5687842769321014e-06,
+      "loss": 0.1208,
+      "step": 81300
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.568494788826866e-06,
+      "loss": 0.1166,
+      "step": 81330
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5682054609204182e-06,
+      "loss": 0.1039,
+      "step": 81360
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5679162930650595e-06,
+      "loss": 0.1325,
+      "step": 81390
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5676272851132824e-06,
+      "loss": 0.1569,
+      "step": 81420
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5673384369177683e-06,
+      "loss": 0.1288,
+      "step": 81450
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.56704974833139e-06,
+      "loss": 0.1136,
+      "step": 81480
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5667612192072093e-06,
+      "loss": 0.1078,
+      "step": 81510
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5664728493984767e-06,
+      "loss": 0.133,
+      "step": 81540
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5661846387586333e-06,
+      "loss": 0.1134,
+      "step": 81570
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5658965871413072e-06,
+      "loss": 0.1342,
+      "step": 81600
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.565608694400316e-06,
+      "loss": 0.1197,
+      "step": 81630
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5653209603896647e-06,
+      "loss": 0.134,
+      "step": 81660
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5650333849635466e-06,
+      "loss": 0.124,
+      "step": 81690
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5647459679763419e-06,
+      "loss": 0.1352,
+      "step": 81720
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5644587092826183e-06,
+      "loss": 0.1432,
+      "step": 81750
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5641716087371303e-06,
+      "loss": 0.1214,
+      "step": 81780
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5638846661948185e-06,
+      "loss": 0.1189,
+      "step": 81810
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5635978815108104e-06,
+      "loss": 0.139,
+      "step": 81840
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.563311254540419e-06,
+      "loss": 0.141,
+      "step": 81870
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5630247851391424e-06,
+      "loss": 0.1134,
+      "step": 81900
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5627384731626653e-06,
+      "loss": 0.1101,
+      "step": 81930
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.562452318466856e-06,
+      "loss": 0.1219,
+      "step": 81960
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5621663209077684e-06,
+      "loss": 0.1193,
+      "step": 81990
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.56188048034164e-06,
+      "loss": 0.1315,
+      "step": 82020
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5615947966248937e-06,
+      "loss": 0.1027,
+      "step": 82050
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5613092696141343e-06,
+      "loss": 0.1191,
+      "step": 82080
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5610238991661516e-06,
+      "loss": 0.1327,
+      "step": 82110
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5607386851379177e-06,
+      "loss": 0.1149,
+      "step": 82140
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5604536273865881e-06,
+      "loss": 0.1186,
+      "step": 82170
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5601687257695005e-06,
+      "loss": 0.1243,
+      "step": 82200
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5598839801441752e-06,
+      "loss": 0.1272,
+      "step": 82230
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5595993903683138e-06,
+      "loss": 0.1184,
+      "step": 82260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5593149562998e-06,
+      "loss": 0.1334,
+      "step": 82290
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5590306777966993e-06,
+      "loss": 0.129,
+      "step": 82320
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5587465547172577e-06,
+      "loss": 0.1391,
+      "step": 82350
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.558462586919902e-06,
+      "loss": 0.1281,
+      "step": 82380
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5581787742632399e-06,
+      "loss": 0.1231,
+      "step": 82410
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.5578951166060582e-06,
+      "loss": 0.1002,
+      "step": 82440
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.557611613807325e-06,
+      "loss": 0.1116,
+      "step": 82470
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.5573282657261874e-06,
+      "loss": 0.1374,
+      "step": 82500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.5570450722219712e-06,
+      "loss": 0.12,
+      "step": 82530
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.556762033154182e-06,
+      "loss": 0.1218,
+      "step": 82560
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.556479148382504e-06,
+      "loss": 0.1386,
+      "step": 82590
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.5561964177667995e-06,
+      "loss": 0.1384,
+      "step": 82620
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.5559138411671087e-06,
+      "loss": 0.1249,
+      "step": 82650
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.5556314184436507e-06,
+      "loss": 0.1146,
+      "step": 82680
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.5553491494568212e-06,
+      "loss": 0.1199,
+      "step": 82710
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.555067034067193e-06,
+      "loss": 0.1184,
+      "step": 82740
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.554785072135517e-06,
+      "loss": 0.108,
+      "step": 82770
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.5545032635227195e-06,
+      "loss": 0.1385,
+      "step": 82800
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.554221608089904e-06,
+      "loss": 0.1237,
+      "step": 82830
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.5539401056983492e-06,
+      "loss": 0.136,
+      "step": 82860
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.553658756209511e-06,
+      "loss": 0.1085,
+      "step": 82890
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.55337755948502e-06,
+      "loss": 0.1238,
+      "step": 82920
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.5530965153866817e-06,
+      "loss": 0.1196,
+      "step": 82950
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.5528156237764777e-06,
+      "loss": 0.0952,
+      "step": 82980
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.5525348845165627e-06,
+      "loss": 0.1253,
+      "step": 83010
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.5522542974692667e-06,
+      "loss": 0.144,
+      "step": 83040
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.5519738624970942e-06,
+      "loss": 0.1095,
+      "step": 83070
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.551693579462723e-06,
+      "loss": 0.1236,
+      "step": 83100
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.5514134482290044e-06,
+      "loss": 0.1229,
+      "step": 83130
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.5511334686589628e-06,
+      "loss": 0.129,
+      "step": 83160
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.550853640615796e-06,
+      "loss": 0.1211,
+      "step": 83190
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.5505739639628744e-06,
+      "loss": 0.1231,
+      "step": 83220
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.5502944385637406e-06,
+      "loss": 0.1098,
+      "step": 83250
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5500150642821094e-06,
+      "loss": 0.1199,
+      "step": 83280
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5497358409818678e-06,
+      "loss": 0.1487,
+      "step": 83310
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5494567685270737e-06,
+      "loss": 0.1068,
+      "step": 83340
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5491778467819573e-06,
+      "loss": 0.118,
+      "step": 83370
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5488990756109185e-06,
+      "loss": 0.1085,
+      "step": 83400
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5486204548785288e-06,
+      "loss": 0.1082,
+      "step": 83430
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5483419844495307e-06,
+      "loss": 0.1041,
+      "step": 83460
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.5480636641888352e-06,
+      "loss": 0.1008,
+      "step": 83490
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.547785493961525e-06,
+      "loss": 0.1108,
+      "step": 83520
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5475074736328517e-06,
+      "loss": 0.1091,
+      "step": 83550
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5472296030682354e-06,
+      "loss": 0.1135,
+      "step": 83580
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5469518821332668e-06,
+      "loss": 0.1369,
+      "step": 83610
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5466743106937049e-06,
+      "loss": 0.1298,
+      "step": 83640
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5463968886154769e-06,
+      "loss": 0.1157,
+      "step": 83670
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5461196157646783e-06,
+      "loss": 0.1205,
+      "step": 83700
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.545842492007573e-06,
+      "loss": 0.1166,
+      "step": 83730
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.5455655172105921e-06,
+      "loss": 0.1373,
+      "step": 83760
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.545288691240335e-06,
+      "loss": 0.1065,
+      "step": 83790
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5450120139635676e-06,
+      "loss": 0.1403,
+      "step": 83820
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5447354852472227e-06,
+      "loss": 0.1516,
+      "step": 83850
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5444591049584e-06,
+      "loss": 0.1211,
+      "step": 83880
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.544182872964366e-06,
+      "loss": 0.1235,
+      "step": 83910
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.543906789132552e-06,
+      "loss": 0.1217,
+      "step": 83940
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.543630853330557e-06,
+      "loss": 0.1362,
+      "step": 83970
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5433550654261445e-06,
+      "loss": 0.1273,
+      "step": 84000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.5430794252872425e-06,
+      "loss": 0.1416,
+      "step": 84030
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5428039327819458e-06,
+      "loss": 0.1232,
+      "step": 84060
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.542528587778513e-06,
+      "loss": 0.1246,
+      "step": 84090
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.542253390145367e-06,
+      "loss": 0.1087,
+      "step": 84120
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5419783397510962e-06,
+      "loss": 0.1087,
+      "step": 84150
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5417034364644514e-06,
+      "loss": 0.1325,
+      "step": 84180
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5414286801543485e-06,
+      "loss": 0.1321,
+      "step": 84210
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5411540706898656e-06,
+      "loss": 0.116,
+      "step": 84240
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5408796079402449e-06,
+      "loss": 0.1356,
+      "step": 84270
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5406052917748917e-06,
+      "loss": 0.1228,
+      "step": 84300
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5403311220633734e-06,
+      "loss": 0.1235,
+      "step": 84330
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.54005709867542e-06,
+      "loss": 0.1147,
+      "step": 84360
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5397832214809234e-06,
+      "loss": 0.1158,
+      "step": 84390
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.539509490349938e-06,
+      "loss": 0.1112,
+      "step": 84420
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5392359051526795e-06,
+      "loss": 0.118,
+      "step": 84450
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5389624657595257e-06,
+      "loss": 0.1281,
+      "step": 84480
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5386891720410136e-06,
+      "loss": 0.099,
+      "step": 84510
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.538416023867843e-06,
+      "loss": 0.1043,
+      "step": 84540
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5381430211108741e-06,
+      "loss": 0.1287,
+      "step": 84570
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5378701636411262e-06,
+      "loss": 0.1269,
+      "step": 84600
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5375974513297803e-06,
+      "loss": 0.1099,
+      "step": 84630
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5373248840481758e-06,
+      "loss": 0.1497,
+      "step": 84660
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5370524616678132e-06,
+      "loss": 0.1104,
+      "step": 84690
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5367801840603504e-06,
+      "loss": 0.1058,
+      "step": 84720
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5365080510976067e-06,
+      "loss": 0.1078,
+      "step": 84750
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5362360626515582e-06,
+      "loss": 0.114,
+      "step": 84780
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5359642185943406e-06,
+      "loss": 0.0988,
+      "step": 84810
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.535692518798248e-06,
+      "loss": 0.1161,
+      "step": 84840
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5354209631357322e-06,
+      "loss": 0.1342,
+      "step": 84870
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5351495514794033e-06,
+      "loss": 0.1202,
+      "step": 84900
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5348782837020273e-06,
+      "loss": 0.1082,
+      "step": 84930
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5346071596765304e-06,
+      "loss": 0.1103,
+      "step": 84960
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5343361792759937e-06,
+      "loss": 0.1361,
+      "step": 84990
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5340653423736556e-06,
+      "loss": 0.1208,
+      "step": 85020
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5337946488429114e-06,
+      "loss": 0.1119,
+      "step": 85050
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5335240985573125e-06,
+      "loss": 0.1237,
+      "step": 85080
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5332536913905663e-06,
+      "loss": 0.1237,
+      "step": 85110
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5329924337198969e-06,
+      "loss": 0.1091,
+      "step": 85140
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5327223076524054e-06,
+      "loss": 0.1237,
+      "step": 85170
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5324523243300148e-06,
+      "loss": 0.1296,
+      "step": 85200
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5321824836270496e-06,
+      "loss": 0.1202,
+      "step": 85230
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5319127854179885e-06,
+      "loss": 0.146,
+      "step": 85260
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5316432295774654e-06,
+      "loss": 0.1462,
+      "step": 85290
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5313738159802683e-06,
+      "loss": 0.1368,
+      "step": 85320
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5311045445013391e-06,
+      "loss": 0.1462,
+      "step": 85350
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.530835415015774e-06,
+      "loss": 0.1282,
+      "step": 85380
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5305664273988224e-06,
+      "loss": 0.1152,
+      "step": 85410
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5302975815258878e-06,
+      "loss": 0.1182,
+      "step": 85440
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5300378318006353e-06,
+      "loss": 0.1259,
+      "step": 85470
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5297692643280468e-06,
+      "loss": 0.1203,
+      "step": 85500
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5295008382307369e-06,
+      "loss": 0.102,
+      "step": 85530
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5292325533847157e-06,
+      "loss": 0.1313,
+      "step": 85560
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5289644096661436e-06,
+      "loss": 0.1287,
+      "step": 85590
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.528696406951334e-06,
+      "loss": 0.138,
+      "step": 85620
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5284285451167524e-06,
+      "loss": 0.1381,
+      "step": 85650
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5281608240390144e-06,
+      "loss": 0.1392,
+      "step": 85680
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5278932435948881e-06,
+      "loss": 0.1166,
+      "step": 85710
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.527625803661292e-06,
+      "loss": 0.1184,
+      "step": 85740
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5273585041152956e-06,
+      "loss": 0.1142,
+      "step": 85770
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5270913448341186e-06,
+      "loss": 0.1084,
+      "step": 85800
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.526824325695131e-06,
+      "loss": 0.1404,
+      "step": 85830
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5265574465758534e-06,
+      "loss": 0.1101,
+      "step": 85860
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5262907073539553e-06,
+      "loss": 0.1247,
+      "step": 85890
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5260241079072572e-06,
+      "loss": 0.1084,
+      "step": 85920
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5257576481137268e-06,
+      "loss": 0.1379,
+      "step": 85950
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5254913278514829e-06,
+      "loss": 0.1223,
+      "step": 85980
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5252251469987929e-06,
+      "loss": 0.1168,
+      "step": 86010
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.524959105434071e-06,
+      "loss": 0.1237,
+      "step": 86040
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5246932030358824e-06,
+      "loss": 0.1205,
+      "step": 86070
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5244274396829384e-06,
+      "loss": 0.118,
+      "step": 86100
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5241618152540993e-06,
+      "loss": 0.1071,
+      "step": 86130
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.523896329628373e-06,
+      "loss": 0.1027,
+      "step": 86160
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5236309826849143e-06,
+      "loss": 0.1233,
+      "step": 86190
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5233657743030265e-06,
+      "loss": 0.1153,
+      "step": 86220
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5231007043621581e-06,
+      "loss": 0.1201,
+      "step": 86250
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.522835772741906e-06,
+      "loss": 0.1336,
+      "step": 86280
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5225709793220126e-06,
+      "loss": 0.1331,
+      "step": 86310
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5223063239823674e-06,
+      "loss": 0.1216,
+      "step": 86340
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5220418066030055e-06,
+      "loss": 0.1039,
+      "step": 86370
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5217774270641078e-06,
+      "loss": 0.1173,
+      "step": 86400
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5215131852460013e-06,
+      "loss": 0.153,
+      "step": 86430
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.521249081029158e-06,
+      "loss": 0.1116,
+      "step": 86460
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5209851142941952e-06,
+      "loss": 0.1253,
+      "step": 86490
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5207212849218755e-06,
+      "loss": 0.1241,
+      "step": 86520
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5204575927931055e-06,
+      "loss": 0.1126,
+      "step": 86550
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.520194037788937e-06,
+      "loss": 0.1449,
+      "step": 86580
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5199306197905658e-06,
+      "loss": 0.1137,
+      "step": 86610
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5196673386793316e-06,
+      "loss": 0.1225,
+      "step": 86640
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5194041943367178e-06,
+      "loss": 0.1255,
+      "step": 86670
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.519141186644352e-06,
+      "loss": 0.1401,
+      "step": 86700
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.518878315484005e-06,
+      "loss": 0.1344,
+      "step": 86730
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5186155807375907e-06,
+      "loss": 0.1046,
+      "step": 86760
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5183529822871654e-06,
+      "loss": 0.1374,
+      "step": 86790
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.518090520014929e-06,
+      "loss": 0.1191,
+      "step": 86820
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5178281938032233e-06,
+      "loss": 0.1181,
+      "step": 86850
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5175660035345323e-06,
+      "loss": 0.1411,
+      "step": 86880
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5173039490914826e-06,
+      "loss": 0.1249,
+      "step": 86910
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5170420303568423e-06,
+      "loss": 0.1299,
+      "step": 86940
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5167802472135212e-06,
+      "loss": 0.1245,
+      "step": 86970
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5165185995445702e-06,
+      "loss": 0.1229,
+      "step": 87000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5162570872331817e-06,
+      "loss": 0.1035,
+      "step": 87030
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5159957101626888e-06,
+      "loss": 0.1268,
+      "step": 87060
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5157344682165656e-06,
+      "loss": 0.1357,
+      "step": 87090
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.515473361278426e-06,
+      "loss": 0.1802,
+      "step": 87120
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5152123892320259e-06,
+      "loss": 0.1332,
+      "step": 87150
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5149515519612596e-06,
+      "loss": 0.1239,
+      "step": 87180
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5146908493501613e-06,
+      "loss": 0.1301,
+      "step": 87210
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5144302812829061e-06,
+      "loss": 0.1008,
+      "step": 87240
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5141698476438073e-06,
+      "loss": 0.1131,
+      "step": 87270
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.513909548317318e-06,
+      "loss": 0.1224,
+      "step": 87300
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5136493831880306e-06,
+      "loss": 0.1287,
+      "step": 87330
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5133893521406753e-06,
+      "loss": 0.1115,
+      "step": 87360
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5131294550601217e-06,
+      "loss": 0.1134,
+      "step": 87390
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5128696918313772e-06,
+      "loss": 0.1424,
+      "step": 87420
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5126100623395884e-06,
+      "loss": 0.1172,
+      "step": 87450
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.512350566470038e-06,
+      "loss": 0.107,
+      "step": 87480
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5120912041081485e-06,
+      "loss": 0.0991,
+      "step": 87510
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5118319751394783e-06,
+      "loss": 0.1176,
+      "step": 87540
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5115728794497237e-06,
+      "loss": 0.108,
+      "step": 87570
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5113139169247178e-06,
+      "loss": 0.1178,
+      "step": 87600
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.511063712957183e-06,
+      "loss": 0.111,
+      "step": 87630
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5108050119903283e-06,
+      "loss": 0.111,
+      "step": 87660
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5105464438503328e-06,
+      "loss": 0.1297,
+      "step": 87690
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5102880084235716e-06,
+      "loss": 0.1191,
+      "step": 87720
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5100297055965558e-06,
+      "loss": 0.1325,
+      "step": 87750
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.5097715352559318e-06,
+      "loss": 0.109,
+      "step": 87780
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5095134972884821e-06,
+      "loss": 0.0999,
+      "step": 87810
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5092555915811246e-06,
+      "loss": 0.1436,
+      "step": 87840
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5089978180209128e-06,
+      "loss": 0.1361,
+      "step": 87870
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5087401764950343e-06,
+      "loss": 0.1119,
+      "step": 87900
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5084826668908123e-06,
+      "loss": 0.1274,
+      "step": 87930
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5082252890957039e-06,
+      "loss": 0.1391,
+      "step": 87960
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5079680429973012e-06,
+      "loss": 0.1352,
+      "step": 87990
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5077109284833297e-06,
+      "loss": 0.1133,
+      "step": 88020
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5074539454416498e-06,
+      "loss": 0.1059,
+      "step": 88050
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5071970937602551e-06,
+      "loss": 0.0964,
+      "step": 88080
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5069403733272725e-06,
+      "loss": 0.1188,
+      "step": 88110
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5066837840309626e-06,
+      "loss": 0.1084,
+      "step": 88140
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5064273257597195e-06,
+      "loss": 0.1163,
+      "step": 88170
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5061709984020693e-06,
+      "loss": 0.1232,
+      "step": 88200
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5059148018466718e-06,
+      "loss": 0.1235,
+      "step": 88230
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5056587359823183e-06,
+      "loss": 0.1132,
+      "step": 88260
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5054028006979338e-06,
+      "loss": 0.1157,
+      "step": 88290
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5051469958825737e-06,
+      "loss": 0.1168,
+      "step": 88320
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.504891321425427e-06,
+      "loss": 0.1184,
+      "step": 88350
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5046357772158132e-06,
+      "loss": 0.1309,
+      "step": 88380
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5043803631431841e-06,
+      "loss": 0.11,
+      "step": 88410
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5041250790971224e-06,
+      "loss": 0.132,
+      "step": 88440
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5038699249673416e-06,
+      "loss": 0.1204,
+      "step": 88470
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.503614900643687e-06,
+      "loss": 0.1047,
+      "step": 88500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5033600060161343e-06,
+      "loss": 0.1349,
+      "step": 88530
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5031052409747888e-06,
+      "loss": 0.1237,
+      "step": 88560
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5028506054098876e-06,
+      "loss": 0.1142,
+      "step": 88590
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.502596099211797e-06,
+      "loss": 0.1132,
+      "step": 88620
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5023417222710132e-06,
+      "loss": 0.1193,
+      "step": 88650
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5020874744781625e-06,
+      "loss": 0.1266,
+      "step": 88680
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5018333557240003e-06,
+      "loss": 0.1131,
+      "step": 88710
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.501579365899412e-06,
+      "loss": 0.1058,
+      "step": 88740
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5013255048954111e-06,
+      "loss": 0.1074,
+      "step": 88770
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5010717726031415e-06,
+      "loss": 0.1117,
+      "step": 88800
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5008181689138746e-06,
+      "loss": 0.1114,
+      "step": 88830
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5005646937190105e-06,
+      "loss": 0.1221,
+      "step": 88860
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.500311346910078e-06,
+      "loss": 0.117,
+      "step": 88890
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.5000581283787338e-06,
+      "loss": 0.1257,
+      "step": 88920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.499805038016763e-06,
+      "loss": 0.1129,
+      "step": 88950
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4995520757160784e-06,
+      "loss": 0.1134,
+      "step": 88980
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4992992413687193e-06,
+      "loss": 0.1142,
+      "step": 89010
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.499046534866854e-06,
+      "loss": 0.1165,
+      "step": 89040
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4987939561027765e-06,
+      "loss": 0.0977,
+      "step": 89070
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4985415049689088e-06,
+      "loss": 0.1337,
+      "step": 89100
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4982891813577996e-06,
+      "loss": 0.142,
+      "step": 89130
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4980369851621237e-06,
+      "loss": 0.1403,
+      "step": 89160
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4977849162746824e-06,
+      "loss": 0.1387,
+      "step": 89190
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4975329745884038e-06,
+      "loss": 0.1118,
+      "step": 89220
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4972811599963413e-06,
+      "loss": 0.1307,
+      "step": 89250
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4970294723916748e-06,
+      "loss": 0.1178,
+      "step": 89280
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4967779116677095e-06,
+      "loss": 0.1113,
+      "step": 89310
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4965264777178759e-06,
+      "loss": 0.1112,
+      "step": 89340
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4962751704357302e-06,
+      "loss": 0.1064,
+      "step": 89370
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.496023989714953e-06,
+      "loss": 0.1224,
+      "step": 89400
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4957729354493509e-06,
+      "loss": 0.106,
+      "step": 89430
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4955220075328538e-06,
+      "loss": 0.1232,
+      "step": 89460
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4952712058595175e-06,
+      "loss": 0.1527,
+      "step": 89490
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.495028884143625e-06,
+      "loss": 0.1414,
+      "step": 89520
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4947783304399193e-06,
+      "loss": 0.1012,
+      "step": 89550
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4945279026658018e-06,
+      "loss": 0.139,
+      "step": 89580
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.49427760071582e-06,
+      "loss": 0.1174,
+      "step": 89610
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4940274244846445e-06,
+      "loss": 0.1127,
+      "step": 89640
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4937773738670697e-06,
+      "loss": 0.1122,
+      "step": 89670
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.4935274487580126e-06,
+      "loss": 0.1302,
+      "step": 89700
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.4932776490525144e-06,
+      "loss": 0.1179,
+      "step": 89730
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.4930279746457379e-06,
+      "loss": 0.1215,
+      "step": 89760
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.492778425432969e-06,
+      "loss": 0.1021,
+      "step": 89790
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.4925290013096165e-06,
+      "loss": 0.1068,
+      "step": 89820
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.492279702171211e-06,
+      "loss": 0.1374,
+      "step": 89850
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.4920305279134051e-06,
+      "loss": 0.1252,
+      "step": 89880
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.4917814784319737e-06,
+      "loss": 0.1192,
+      "step": 89910
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.4915325536228134e-06,
+      "loss": 0.1032,
+      "step": 89940
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.4912837533819422e-06,
+      "loss": 0.1232,
+      "step": 89970
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.4910350776054995e-06,
+      "loss": 0.1362,
+      "step": 90000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.4907865261897456e-06,
+      "loss": 0.117,
+      "step": 90030
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.4905380990310628e-06,
+      "loss": 0.1541,
+      "step": 90060
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.490289796025953e-06,
+      "loss": 0.1077,
+      "step": 90090
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.49004161707104e-06,
+      "loss": 0.1133,
+      "step": 90120
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.4897935620630663e-06,
+      "loss": 0.1163,
+      "step": 90150
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.4895456308988964e-06,
+      "loss": 0.1588,
+      "step": 90180
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.4892978234755149e-06,
+      "loss": 0.1337,
+      "step": 90210
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4890501396900245e-06,
+      "loss": 0.1156,
+      "step": 90240
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4888025794396497e-06,
+      "loss": 0.1172,
+      "step": 90270
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4885551426217331e-06,
+      "loss": 0.1159,
+      "step": 90300
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.488307829133738e-06,
+      "loss": 0.1138,
+      "step": 90330
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4880606388732456e-06,
+      "loss": 0.121,
+      "step": 90360
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4878135717379569e-06,
+      "loss": 0.1292,
+      "step": 90390
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.487566627625692e-06,
+      "loss": 0.1111,
+      "step": 90420
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4873198064343886e-06,
+      "loss": 0.1321,
+      "step": 90450
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.4870731080621042e-06,
+      "loss": 0.1313,
+      "step": 90480
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4868265324070142e-06,
+      "loss": 0.1153,
+      "step": 90510
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4865800793674113e-06,
+      "loss": 0.1199,
+      "step": 90540
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4863337488417067e-06,
+      "loss": 0.1136,
+      "step": 90570
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4860875407284305e-06,
+      "loss": 0.1175,
+      "step": 90600
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4858414549262287e-06,
+      "loss": 0.1414,
+      "step": 90630
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4855954913338655e-06,
+      "loss": 0.1195,
+      "step": 90660
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4853496498502227e-06,
+      "loss": 0.1175,
+      "step": 90690
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4851039303742986e-06,
+      "loss": 0.1238,
+      "step": 90720
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.4848583328052088e-06,
+      "loss": 0.1193,
+      "step": 90750
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4846128570421855e-06,
+      "loss": 0.1214,
+      "step": 90780
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4843675029845775e-06,
+      "loss": 0.1141,
+      "step": 90810
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.48412227053185e-06,
+      "loss": 0.1322,
+      "step": 90840
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4838771595835847e-06,
+      "loss": 0.1243,
+      "step": 90870
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4836321700394784e-06,
+      "loss": 0.1155,
+      "step": 90900
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.483387301799345e-06,
+      "loss": 0.1255,
+      "step": 90930
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4831425547631134e-06,
+      "loss": 0.0969,
+      "step": 90960
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.4828979288308285e-06,
+      "loss": 0.1114,
+      "step": 90990
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.48265342390265e-06,
+      "loss": 0.1182,
+      "step": 91020
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4824090398788529e-06,
+      "loss": 0.114,
+      "step": 91050
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.482164776659828e-06,
+      "loss": 0.1142,
+      "step": 91080
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4819206341460796e-06,
+      "loss": 0.1193,
+      "step": 91110
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4816766122382278e-06,
+      "loss": 0.1243,
+      "step": 91140
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4814327108370068e-06,
+      "loss": 0.1066,
+      "step": 91170
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4811889298432654e-06,
+      "loss": 0.1011,
+      "step": 91200
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4809452691579664e-06,
+      "loss": 0.1181,
+      "step": 91230
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4807017286821866e-06,
+      "loss": 0.1294,
+      "step": 91260
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.4804583083171158e-06,
+      "loss": 0.119,
+      "step": 91290
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4802150079640592e-06,
+      "loss": 0.1297,
+      "step": 91320
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4799718275244343e-06,
+      "loss": 0.0969,
+      "step": 91350
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4797287668997723e-06,
+      "loss": 0.1048,
+      "step": 91380
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4794858259917175e-06,
+      "loss": 0.1058,
+      "step": 91410
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4792430047020267e-06,
+      "loss": 0.1071,
+      "step": 91440
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4790003029325705e-06,
+      "loss": 0.1385,
+      "step": 91470
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4787577205853318e-06,
+      "loss": 0.0996,
+      "step": 91500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4785152575624056e-06,
+      "loss": 0.1223,
+      "step": 91530
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.4782729137659996e-06,
+      "loss": 0.1278,
+      "step": 91560
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4780306890984334e-06,
+      "loss": 0.12,
+      "step": 91590
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4777885834621391e-06,
+      "loss": 0.1116,
+      "step": 91620
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.47754659675966e-06,
+      "loss": 0.1276,
+      "step": 91650
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4773047288936515e-06,
+      "loss": 0.1446,
+      "step": 91680
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4770629797668806e-06,
+      "loss": 0.1144,
+      "step": 91710
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4768213492822253e-06,
+      "loss": 0.133,
+      "step": 91740
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4765798373426748e-06,
+      "loss": 0.1227,
+      "step": 91770
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.4763384438513297e-06,
+      "loss": 0.1087,
+      "step": 91800
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.476097168711401e-06,
+      "loss": 0.1146,
+      "step": 91830
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.4758560118262105e-06,
+      "loss": 0.1266,
+      "step": 91860
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.4756149730991908e-06,
+      "loss": 0.1054,
+      "step": 91890
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.4753740524338848e-06,
+      "loss": 0.1123,
+      "step": 91920
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.4751332497339452e-06,
+      "loss": 0.1227,
+      "step": 91950
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.474892564903135e-06,
+      "loss": 0.1042,
+      "step": 91980
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.4746519978453271e-06,
+      "loss": 0.1297,
+      "step": 92010
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.4744115484645044e-06,
+      "loss": 0.1177,
+      "step": 92040
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.474171216664759e-06,
+      "loss": 0.1245,
+      "step": 92070
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.473931002350292e-06,
+      "loss": 0.104,
+      "step": 92100
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.473690905425415e-06,
+      "loss": 0.1007,
+      "step": 92130
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.473450925794547e-06,
+      "loss": 0.0941,
+      "step": 92160
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.4732110633622177e-06,
+      "loss": 0.1019,
+      "step": 92190
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.472971318033064e-06,
+      "loss": 0.1402,
+      "step": 92220
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.472731689711832e-06,
+      "loss": 0.1147,
+      "step": 92250
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.4724921783033766e-06,
+      "loss": 0.128,
+      "step": 92280
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.472252783712661e-06,
+      "loss": 0.1135,
+      "step": 92310
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.472013505844755e-06,
+      "loss": 0.137,
+      "step": 92340
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.471774344604839e-06,
+      "loss": 0.124,
+      "step": 92370
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.471535299898199e-06,
+      "loss": 0.1294,
+      "step": 92400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.4712963716302299e-06,
+      "loss": 0.1158,
+      "step": 92430
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.4710575597064326e-06,
+      "loss": 0.1321,
+      "step": 92460
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.4708188640324172e-06,
+      "loss": 0.1243,
+      "step": 92490
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.4705802845138998e-06,
+      "loss": 0.1155,
+      "step": 92520
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.4703418210567037e-06,
+      "loss": 0.1176,
+      "step": 92550
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.4701034735667596e-06,
+      "loss": 0.1017,
+      "step": 92580
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.469865241950104e-06,
+      "loss": 0.1216,
+      "step": 92610
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.469627126112881e-06,
+      "loss": 0.1047,
+      "step": 92640
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.4693891259613402e-06,
+      "loss": 0.1072,
+      "step": 92670
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.4691512414018378e-06,
+      "loss": 0.1367,
+      "step": 92700
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.468913472340836e-06,
+      "loss": 0.1179,
+      "step": 92730
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.4686758186849038e-06,
+      "loss": 0.1153,
+      "step": 92760
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.4684382803407139e-06,
+      "loss": 0.109,
+      "step": 92790
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.4682008572150469e-06,
+      "loss": 0.1358,
+      "step": 92820
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.4679635492147872e-06,
+      "loss": 0.0986,
+      "step": 92850
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.467726356246926e-06,
+      "loss": 0.1167,
+      "step": 92880
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.4674892782185584e-06,
+      "loss": 0.1173,
+      "step": 92910
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.467252315036885e-06,
+      "loss": 0.1071,
+      "step": 92940
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.4670154666092112e-06,
+      "loss": 0.1257,
+      "step": 92970
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.4667787328429475e-06,
+      "loss": 0.1203,
+      "step": 93000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.4665421136456081e-06,
+      "loss": 0.1233,
+      "step": 93030
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.4663056089248127e-06,
+      "loss": 0.1099,
+      "step": 93060
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.4660692185882838e-06,
+      "loss": 0.1145,
+      "step": 93090
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.46583294254385e-06,
+      "loss": 0.1229,
+      "step": 93120
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.4655967806994417e-06,
+      "loss": 0.1132,
+      "step": 93150
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.4653607329630951e-06,
+      "loss": 0.1076,
+      "step": 93180
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.465124799242948e-06,
+      "loss": 0.1297,
+      "step": 93210
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.4648889794472434e-06,
+      "loss": 0.1107,
+      "step": 93240
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.4646532734843268e-06,
+      "loss": 0.1041,
+      "step": 93270
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.4644176812626475e-06,
+      "loss": 0.1077,
+      "step": 93300
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.464182202690757e-06,
+      "loss": 0.1062,
+      "step": 93330
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.4639468376773105e-06,
+      "loss": 0.1202,
+      "step": 93360
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.4637115861310652e-06,
+      "loss": 0.1397,
+      "step": 93390
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.4634764479608819e-06,
+      "loss": 0.1304,
+      "step": 93420
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.4632414230757227e-06,
+      "loss": 0.1259,
+      "step": 93450
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.463006511384653e-06,
+      "loss": 0.1204,
+      "step": 93480
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.4627717127968393e-06,
+      "loss": 0.1246,
+      "step": 93510
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.4625370272215516e-06,
+      "loss": 0.1108,
+      "step": 93540
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.4623024545681609e-06,
+      "loss": 0.1497,
+      "step": 93570
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.462067994746139e-06,
+      "loss": 0.1127,
+      "step": 93600
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.4618414574190027e-06,
+      "loss": 0.1123,
+      "step": 93630
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.4616072192349775e-06,
+      "loss": 0.1436,
+      "step": 93660
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.4613730936143547e-06,
+      "loss": 0.135,
+      "step": 93690
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.4611390804670088e-06,
+      "loss": 0.1123,
+      "step": 93720
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.460905179702916e-06,
+      "loss": 0.1224,
+      "step": 93750
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.4606713912321522e-06,
+      "loss": 0.1134,
+      "step": 93780
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.460437714964895e-06,
+      "loss": 0.1188,
+      "step": 93810
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.4602041508114229e-06,
+      "loss": 0.0994,
+      "step": 93840
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.4599706986821135e-06,
+      "loss": 0.1052,
+      "step": 93870
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.4597373584874452e-06,
+      "loss": 0.1548,
+      "step": 93900
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.459504130137997e-06,
+      "loss": 0.103,
+      "step": 93930
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.4592710135444477e-06,
+      "loss": 0.0989,
+      "step": 93960
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4590380086175755e-06,
+      "loss": 0.1272,
+      "step": 93990
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4588051152682587e-06,
+      "loss": 0.1303,
+      "step": 94020
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4585723334074747e-06,
+      "loss": 0.1212,
+      "step": 94050
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4583396629463011e-06,
+      "loss": 0.1169,
+      "step": 94080
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4581071037959144e-06,
+      "loss": 0.1174,
+      "step": 94110
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4578746558675893e-06,
+      "loss": 0.1132,
+      "step": 94140
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.457642319072701e-06,
+      "loss": 0.1436,
+      "step": 94170
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4574100933227226e-06,
+      "loss": 0.1252,
+      "step": 94200
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.4571779785292261e-06,
+      "loss": 0.114,
+      "step": 94230
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4569459746038823e-06,
+      "loss": 0.1098,
+      "step": 94260
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4567140814584597e-06,
+      "loss": 0.1197,
+      "step": 94290
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4564822990048258e-06,
+      "loss": 0.1026,
+      "step": 94320
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4562506271549461e-06,
+      "loss": 0.1173,
+      "step": 94350
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.456019065820884e-06,
+      "loss": 0.1308,
+      "step": 94380
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4557876149148007e-06,
+      "loss": 0.1223,
+      "step": 94410
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4555562743489552e-06,
+      "loss": 0.112,
+      "step": 94440
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4553250440357034e-06,
+      "loss": 0.1283,
+      "step": 94470
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.4550939238875e-06,
+      "loss": 0.1201,
+      "step": 94500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.454862913816896e-06,
+      "loss": 0.1092,
+      "step": 94530
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4546320137365396e-06,
+      "loss": 0.1174,
+      "step": 94560
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.454401223559176e-06,
+      "loss": 0.1355,
+      "step": 94590
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.454170543197648e-06,
+      "loss": 0.1051,
+      "step": 94620
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4539399725648944e-06,
+      "loss": 0.1139,
+      "step": 94650
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.453709511573951e-06,
+      "loss": 0.1283,
+      "step": 94680
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4534791601379494e-06,
+      "loss": 0.1038,
+      "step": 94710
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.4532489181701182e-06,
+      "loss": 0.1181,
+      "step": 94740
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.453018785583782e-06,
+      "loss": 0.1246,
+      "step": 94770
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4527887622923615e-06,
+      "loss": 0.0963,
+      "step": 94800
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4525588482093733e-06,
+      "loss": 0.14,
+      "step": 94830
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4523290432484293e-06,
+      "loss": 0.1527,
+      "step": 94860
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.452099347323238e-06,
+      "loss": 0.1099,
+      "step": 94890
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4518697603476026e-06,
+      "loss": 0.1439,
+      "step": 94920
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4516402822354216e-06,
+      "loss": 0.1077,
+      "step": 94950
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4514109129006895e-06,
+      "loss": 0.1217,
+      "step": 94980
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4511816522574956e-06,
+      "loss": 0.1292,
+      "step": 95010
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4509525002200234e-06,
+      "loss": 0.1279,
+      "step": 95040
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4507234567025525e-06,
+      "loss": 0.1096,
+      "step": 95070
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4504945216194558e-06,
+      "loss": 0.1203,
+      "step": 95100
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4502656948852023e-06,
+      "loss": 0.1156,
+      "step": 95130
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4500369764143537e-06,
+      "loss": 0.1166,
+      "step": 95160
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.449808366121568e-06,
+      "loss": 0.13,
+      "step": 95190
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4495798639215955e-06,
+      "loss": 0.1074,
+      "step": 95220
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4493514697292815e-06,
+      "loss": 0.1208,
+      "step": 95250
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4491231834595649e-06,
+      "loss": 0.1303,
+      "step": 95280
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4488950050274784e-06,
+      "loss": 0.0958,
+      "step": 95310
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4486669343481484e-06,
+      "loss": 0.1031,
+      "step": 95340
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.448438971336795e-06,
+      "loss": 0.121,
+      "step": 95370
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4482111159087308e-06,
+      "loss": 0.1065,
+      "step": 95400
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4479833679793626e-06,
+      "loss": 0.1094,
+      "step": 95430
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.44775572746419e-06,
+      "loss": 0.1047,
+      "step": 95460
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4475281942788052e-06,
+      "loss": 0.1358,
+      "step": 95490
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.447300768338894e-06,
+      "loss": 0.0914,
+      "step": 95520
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.447073449560234e-06,
+      "loss": 0.119,
+      "step": 95550
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.446846237858696e-06,
+      "loss": 0.1437,
+      "step": 95580
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4466191331502428e-06,
+      "loss": 0.1209,
+      "step": 95610
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4463921353509293e-06,
+      "loss": 0.1104,
+      "step": 95640
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4461652443769042e-06,
+      "loss": 0.1094,
+      "step": 95670
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.445938460144406e-06,
+      "loss": 0.1296,
+      "step": 95700
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4457117825697664e-06,
+      "loss": 0.1246,
+      "step": 95730
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.445485211569409e-06,
+      "loss": 0.1162,
+      "step": 95760
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4452587470598483e-06,
+      "loss": 0.1259,
+      "step": 95790
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4450323889576911e-06,
+      "loss": 0.1168,
+      "step": 95820
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4448061371796352e-06,
+      "loss": 0.1107,
+      "step": 95850
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4445799916424695e-06,
+      "loss": 0.1137,
+      "step": 95880
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.444353952263074e-06,
+      "loss": 0.1077,
+      "step": 95910
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4441280189584204e-06,
+      "loss": 0.1215,
+      "step": 95940
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4439021916455707e-06,
+      "loss": 0.1089,
+      "step": 95970
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.443676470241678e-06,
+      "loss": 0.0999,
+      "step": 96000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4434508546639858e-06,
+      "loss": 0.1198,
+      "step": 96030
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4432253448298278e-06,
+      "loss": 0.1172,
+      "step": 96060
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4429999406566286e-06,
+      "loss": 0.1174,
+      "step": 96090
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4427746420619032e-06,
+      "loss": 0.1225,
+      "step": 96120
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.442549448963256e-06,
+      "loss": 0.1261,
+      "step": 96150
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.442324361278382e-06,
+      "loss": 0.1238,
+      "step": 96180
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4420993789250654e-06,
+      "loss": 0.1185,
+      "step": 96210
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4418745018211811e-06,
+      "loss": 0.1178,
+      "step": 96240
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4416497298846933e-06,
+      "loss": 0.1304,
+      "step": 96270
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4414250630336546e-06,
+      "loss": 0.1275,
+      "step": 96300
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4412005011862088e-06,
+      "loss": 0.1186,
+      "step": 96330
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4409760442605878e-06,
+      "loss": 0.1224,
+      "step": 96360
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4407516921751121e-06,
+      "loss": 0.1124,
+      "step": 96390
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.440527444848193e-06,
+      "loss": 0.1042,
+      "step": 96420
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.440303302198329e-06,
+      "loss": 0.1279,
+      "step": 96450
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4400792641441078e-06,
+      "loss": 0.1308,
+      "step": 96480
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4398553306042063e-06,
+      "loss": 0.1229,
+      "step": 96510
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4396315014973892e-06,
+      "loss": 0.1241,
+      "step": 96540
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4394077767425097e-06,
+      "loss": 0.1101,
+      "step": 96570
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4391841562585096e-06,
+      "loss": 0.1157,
+      "step": 96600
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4389606399644181e-06,
+      "loss": 0.1147,
+      "step": 96630
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4387372277793537e-06,
+      "loss": 0.1013,
+      "step": 96660
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4385139196225214e-06,
+      "loss": 0.126,
+      "step": 96690
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4382907154132144e-06,
+      "loss": 0.1176,
+      "step": 96720
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4380676150708138e-06,
+      "loss": 0.1106,
+      "step": 96750
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4378446185147884e-06,
+      "loss": 0.13,
+      "step": 96780
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4376217256646936e-06,
+      "loss": 0.1368,
+      "step": 96810
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4373989364401727e-06,
+      "loss": 0.1175,
+      "step": 96840
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4371762507609556e-06,
+      "loss": 0.1321,
+      "step": 96870
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.43695366854686e-06,
+      "loss": 0.1229,
+      "step": 96900
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.43673118971779e-06,
+      "loss": 0.1182,
+      "step": 96930
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.436508814193736e-06,
+      "loss": 0.1125,
+      "step": 96960
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4362865418947767e-06,
+      "loss": 0.1194,
+      "step": 96990
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4360643727410753e-06,
+      "loss": 0.1089,
+      "step": 97020
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4358423066528826e-06,
+      "loss": 0.1206,
+      "step": 97050
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.435620343550536e-06,
+      "loss": 0.1121,
+      "step": 97080
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4353984833544576e-06,
+      "loss": 0.1156,
+      "step": 97110
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4351767259851573e-06,
+      "loss": 0.1002,
+      "step": 97140
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4349550713632295e-06,
+      "loss": 0.1131,
+      "step": 97170
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.434733519409356e-06,
+      "loss": 0.1187,
+      "step": 97200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.434512070044302e-06,
+      "loss": 0.1085,
+      "step": 97230
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4342907231889208e-06,
+      "loss": 0.1583,
+      "step": 97260
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4340694787641495e-06,
+      "loss": 0.1241,
+      "step": 97290
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4338483366910113e-06,
+      "loss": 0.151,
+      "step": 97320
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4336272968906137e-06,
+      "loss": 0.1152,
+      "step": 97350
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.433406359284151e-06,
+      "loss": 0.1052,
+      "step": 97380
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4331855237929008e-06,
+      "loss": 0.172,
+      "step": 97410
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.432964790338226e-06,
+      "loss": 0.1167,
+      "step": 97440
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.432744158841575e-06,
+      "loss": 0.1048,
+      "step": 97470
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4325236292244808e-06,
+      "loss": 0.1389,
+      "step": 97500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4323032014085592e-06,
+      "loss": 0.1215,
+      "step": 97530
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4320828753155122e-06,
+      "loss": 0.131,
+      "step": 97560
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4318626508671255e-06,
+      "loss": 0.1223,
+      "step": 97590
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4316425279852693e-06,
+      "loss": 0.1169,
+      "step": 97620
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4314225065918974e-06,
+      "loss": 0.131,
+      "step": 97650
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4312025866090474e-06,
+      "loss": 0.1271,
+      "step": 97680
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.430990093615422e-06,
+      "loss": 0.1121,
+      "step": 97710
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4307703728461559e-06,
+      "loss": 0.1135,
+      "step": 97740
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.430550753256617e-06,
+      "loss": 0.1015,
+      "step": 97770
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4303312347691753e-06,
+      "loss": 0.1113,
+      "step": 97800
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4301118173062832e-06,
+      "loss": 0.1095,
+      "step": 97830
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4298925007904774e-06,
+      "loss": 0.1054,
+      "step": 97860
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4296732851443772e-06,
+      "loss": 0.108,
+      "step": 97890
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4294614724960801e-06,
+      "loss": 0.115,
+      "step": 97920
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4292424550016517e-06,
+      "loss": 0.1127,
+      "step": 97950
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.4290235381478545e-06,
+      "loss": 0.0992,
+      "step": 97980
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4288047218576368e-06,
+      "loss": 0.1295,
+      "step": 98010
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.428586006054029e-06,
+      "loss": 0.1439,
+      "step": 98040
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4283673906601443e-06,
+      "loss": 0.0981,
+      "step": 98070
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4281488755991779e-06,
+      "loss": 0.1126,
+      "step": 98100
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4279304607944077e-06,
+      "loss": 0.1362,
+      "step": 98130
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4277121461691937e-06,
+      "loss": 0.1153,
+      "step": 98160
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4274939316469769e-06,
+      "loss": 0.1056,
+      "step": 98190
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.427275817151281e-06,
+      "loss": 0.1356,
+      "step": 98220
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4270578026057119e-06,
+      "loss": 0.1119,
+      "step": 98250
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4268398879339565e-06,
+      "loss": 0.0941,
+      "step": 98280
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.426622073059783e-06,
+      "loss": 0.1,
+      "step": 98310
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4264043579070419e-06,
+      "loss": 0.1155,
+      "step": 98340
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4261867423996644e-06,
+      "loss": 0.1066,
+      "step": 98370
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4259692264616632e-06,
+      "loss": 0.1106,
+      "step": 98400
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4257518100171313e-06,
+      "loss": 0.1162,
+      "step": 98430
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4255344929902443e-06,
+      "loss": 0.1214,
+      "step": 98460
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4253172753052571e-06,
+      "loss": 0.1344,
+      "step": 98490
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.425100156886506e-06,
+      "loss": 0.1302,
+      "step": 98520
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4248831376584085e-06,
+      "loss": 0.1253,
+      "step": 98550
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4246662175454615e-06,
+      "loss": 0.11,
+      "step": 98580
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.424449396472243e-06,
+      "loss": 0.1244,
+      "step": 98610
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4242326743634117e-06,
+      "loss": 0.117,
+      "step": 98640
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4240160511437056e-06,
+      "loss": 0.1077,
+      "step": 98670
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4237995267379438e-06,
+      "loss": 0.109,
+      "step": 98700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4235831010710242e-06,
+      "loss": 0.0924,
+      "step": 98730
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4233667740679259e-06,
+      "loss": 0.1209,
+      "step": 98760
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4231505456537064e-06,
+      "loss": 0.1183,
+      "step": 98790
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4229344157535044e-06,
+      "loss": 0.1203,
+      "step": 98820
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4227183842925372e-06,
+      "loss": 0.1105,
+      "step": 98850
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4225024511961013e-06,
+      "loss": 0.1109,
+      "step": 98880
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4222866163895734e-06,
+      "loss": 0.1084,
+      "step": 98910
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4220708797984086e-06,
+      "loss": 0.1072,
+      "step": 98940
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.421855241348142e-06,
+      "loss": 0.1183,
+      "step": 98970
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4216397009643871e-06,
+      "loss": 0.0961,
+      "step": 99000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.421424258572836e-06,
+      "loss": 0.1152,
+      "step": 99030
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4212089140992608e-06,
+      "loss": 0.1305,
+      "step": 99060
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.4209936674695108e-06,
+      "loss": 0.1238,
+      "step": 99090
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.420778518609515e-06,
+      "loss": 0.1217,
+      "step": 99120
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.4205634674452803e-06,
+      "loss": 0.1249,
+      "step": 99150
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.4203485139028927e-06,
+      "loss": 0.14,
+      "step": 99180
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.4201336579085157e-06,
+      "loss": 0.0928,
+      "step": 99210
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.419918899388391e-06,
+      "loss": 0.1117,
+      "step": 99240
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.419704238268839e-06,
+      "loss": 0.1062,
+      "step": 99270
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.4194896744762571e-06,
+      "loss": 0.1059,
+      "step": 99300
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.4192752079371213e-06,
+      "loss": 0.1109,
+      "step": 99330
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4190608385779853e-06,
+      "loss": 0.1194,
+      "step": 99360
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.41884656632548e-06,
+      "loss": 0.1202,
+      "step": 99390
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4186323911063138e-06,
+      "loss": 0.1218,
+      "step": 99420
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4184183128472735e-06,
+      "loss": 0.1167,
+      "step": 99450
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4182043314752219e-06,
+      "loss": 0.1251,
+      "step": 99480
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4179904469170998e-06,
+      "loss": 0.1127,
+      "step": 99510
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4177766590999248e-06,
+      "loss": 0.1076,
+      "step": 99540
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4175629679507922e-06,
+      "loss": 0.1442,
+      "step": 99570
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4173493733968724e-06,
+      "loss": 0.1303,
+      "step": 99600
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4171358753654148e-06,
+      "loss": 0.1263,
+      "step": 99630
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4169224737837442e-06,
+      "loss": 0.1221,
+      "step": 99660
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4167091685792625e-06,
+      "loss": 0.1102,
+      "step": 99690
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4164959596794472e-06,
+      "loss": 0.1012,
+      "step": 99720
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4162828470118536e-06,
+      "loss": 0.1182,
+      "step": 99750
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4160698305041118e-06,
+      "loss": 0.1022,
+      "step": 99780
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4158569100839291e-06,
+      "loss": 0.1242,
+      "step": 99810
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4156440856790887e-06,
+      "loss": 0.1134,
+      "step": 99840
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4154313572174494e-06,
+      "loss": 0.1051,
+      "step": 99870
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.415218724626946e-06,
+      "loss": 0.0987,
+      "step": 99900
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4150061878355892e-06,
+      "loss": 0.1139,
+      "step": 99930
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4147937467714653e-06,
+      "loss": 0.1376,
+      "step": 99960
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4145814013627363e-06,
+      "loss": 0.1143,
+      "step": 99990
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4143691515376389e-06,
+      "loss": 0.1214,
+      "step": 100020
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4141569972244865e-06,
+      "loss": 0.1126,
+      "step": 100050
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4139449383516666e-06,
+      "loss": 0.1227,
+      "step": 100080
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4137329748476418e-06,
+      "loss": 0.1061,
+      "step": 100110
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.4135211066409513e-06,
+      "loss": 0.101,
+      "step": 100140
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4133093336602069e-06,
+      "loss": 0.1301,
+      "step": 100170
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4130976558340972e-06,
+      "loss": 0.1159,
+      "step": 100200
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4128860730913845e-06,
+      "loss": 0.0994,
+      "step": 100230
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4126745853609064e-06,
+      "loss": 0.1119,
+      "step": 100260
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4124631925715744e-06,
+      "loss": 0.1187,
+      "step": 100290
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4122518946523744e-06,
+      "loss": 0.1317,
+      "step": 100320
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4120406915323675e-06,
+      "loss": 0.1247,
+      "step": 100350
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4118295831406886e-06,
+      "loss": 0.1153,
+      "step": 100380
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4116185694065465e-06,
+      "loss": 0.0972,
+      "step": 100410
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4114076502592238e-06,
+      "loss": 0.1053,
+      "step": 100440
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.411196825628078e-06,
+      "loss": 0.1179,
+      "step": 100470
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4109860954425395e-06,
+      "loss": 0.1218,
+      "step": 100500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.410775459632113e-06,
+      "loss": 0.1105,
+      "step": 100530
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4105649181263764e-06,
+      "loss": 0.122,
+      "step": 100560
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4103544708549818e-06,
+      "loss": 0.1452,
+      "step": 100590
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4101441177476543e-06,
+      "loss": 0.0935,
+      "step": 100620
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4099338587341923e-06,
+      "loss": 0.1175,
+      "step": 100650
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4097236937444675e-06,
+      "loss": 0.1094,
+      "step": 100680
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.409513622708425e-06,
+      "loss": 0.1027,
+      "step": 100710
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4093036455560824e-06,
+      "loss": 0.1261,
+      "step": 100740
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4090937622175312e-06,
+      "loss": 0.1162,
+      "step": 100770
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4088839726229348e-06,
+      "loss": 0.117,
+      "step": 100800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4086742767025294e-06,
+      "loss": 0.1106,
+      "step": 100830
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.408464674386625e-06,
+      "loss": 0.1228,
+      "step": 100860
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4082551656056025e-06,
+      "loss": 0.1195,
+      "step": 100890
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.408045750289917e-06,
+      "loss": 0.1171,
+      "step": 100920
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4078364283700944e-06,
+      "loss": 0.1187,
+      "step": 100950
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4076271997767338e-06,
+      "loss": 0.0992,
+      "step": 100980
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4074180644405066e-06,
+      "loss": 0.114,
+      "step": 101010
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4072090222921556e-06,
+      "loss": 0.1496,
+      "step": 101040
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.407000073262496e-06,
+      "loss": 0.106,
+      "step": 101070
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4067912172824144e-06,
+      "loss": 0.125,
+      "step": 101100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4065824542828704e-06,
+      "loss": 0.1184,
+      "step": 101130
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.406373784194894e-06,
+      "loss": 0.101,
+      "step": 101160
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4061652069495874e-06,
+      "loss": 0.1235,
+      "step": 101190
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4059567224781244e-06,
+      "loss": 0.1172,
+      "step": 101220
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4057483307117492e-06,
+      "loss": 0.1236,
+      "step": 101250
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4055469733943611e-06,
+      "loss": 0.1012,
+      "step": 101280
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.405338763747695e-06,
+      "loss": 0.1181,
+      "step": 101310
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4051306466025635e-06,
+      "loss": 0.1481,
+      "step": 101340
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4049226218904939e-06,
+      "loss": 0.1125,
+      "step": 101370
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4047146895430846e-06,
+      "loss": 0.0995,
+      "step": 101400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4045068494920044e-06,
+      "loss": 0.1279,
+      "step": 101430
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4042991016689937e-06,
+      "loss": 0.1084,
+      "step": 101460
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4040914460058624e-06,
+      "loss": 0.1215,
+      "step": 101490
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4038838824344921e-06,
+      "loss": 0.1227,
+      "step": 101520
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4036764108868342e-06,
+      "loss": 0.1338,
+      "step": 101550
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4034690312949106e-06,
+      "loss": 0.1117,
+      "step": 101580
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4032617435908135e-06,
+      "loss": 0.1473,
+      "step": 101610
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4030545477067056e-06,
+      "loss": 0.1147,
+      "step": 101640
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.402847443574819e-06,
+      "loss": 0.0933,
+      "step": 101670
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4026404311274564e-06,
+      "loss": 0.101,
+      "step": 101700
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.40243351029699e-06,
+      "loss": 0.111,
+      "step": 101730
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4022266810158624e-06,
+      "loss": 0.1315,
+      "step": 101760
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4020199432165854e-06,
+      "loss": 0.1191,
+      "step": 101790
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4018132968317404e-06,
+      "loss": 0.1174,
+      "step": 101820
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4016067417939789e-06,
+      "loss": 0.1159,
+      "step": 101850
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.401400278036021e-06,
+      "loss": 0.1581,
+      "step": 101880
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.401193905490657e-06,
+      "loss": 0.1141,
+      "step": 101910
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4009876240907458e-06,
+      "loss": 0.1043,
+      "step": 101940
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4007814337692157e-06,
+      "loss": 0.1014,
+      "step": 101970
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4005753344590648e-06,
+      "loss": 0.107,
+      "step": 102000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4003693260933584e-06,
+      "loss": 0.1048,
+      "step": 102030
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.400163408605233e-06,
+      "loss": 0.1149,
+      "step": 102060
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.3999575819278919e-06,
+      "loss": 0.0958,
+      "step": 102090
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.399751845994608e-06,
+      "loss": 0.1199,
+      "step": 102120
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.399546200738723e-06,
+      "loss": 0.1058,
+      "step": 102150
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.399340646093647e-06,
+      "loss": 0.1048,
+      "step": 102180
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.399135181992858e-06,
+      "loss": 0.1233,
+      "step": 102210
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.3989298083699029e-06,
+      "loss": 0.0979,
+      "step": 102240
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.3987245251583972e-06,
+      "loss": 0.1196,
+      "step": 102270
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3985193322920238e-06,
+      "loss": 0.0984,
+      "step": 102300
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3983142297045335e-06,
+      "loss": 0.1046,
+      "step": 102330
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3981092173297464e-06,
+      "loss": 0.1106,
+      "step": 102360
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3979042951015493e-06,
+      "loss": 0.132,
+      "step": 102390
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.397699462953897e-06,
+      "loss": 0.1121,
+      "step": 102420
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3974947208208125e-06,
+      "loss": 0.1055,
+      "step": 102450
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3972900686363861e-06,
+      "loss": 0.119,
+      "step": 102480
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3970855063347757e-06,
+      "loss": 0.1229,
+      "step": 102510
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3968810338502064e-06,
+      "loss": 0.1116,
+      "step": 102540
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3966766511169712e-06,
+      "loss": 0.1086,
+      "step": 102570
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.39647235806943e-06,
+      "loss": 0.1059,
+      "step": 102600
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3962681546420103e-06,
+      "loss": 0.1363,
+      "step": 102630
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3960640407692055e-06,
+      "loss": 0.1313,
+      "step": 102660
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.395860016385578e-06,
+      "loss": 0.1198,
+      "step": 102690
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3956560814257553e-06,
+      "loss": 0.1155,
+      "step": 102720
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3954522358244327e-06,
+      "loss": 0.0942,
+      "step": 102750
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3952484795163714e-06,
+      "loss": 0.0958,
+      "step": 102780
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.395044812436401e-06,
+      "loss": 0.1206,
+      "step": 102810
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3948412345194157e-06,
+      "loss": 0.1209,
+      "step": 102840
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3946377457003774e-06,
+      "loss": 0.0968,
+      "step": 102870
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3944343459143137e-06,
+      "loss": 0.1192,
+      "step": 102900
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.394231035096319e-06,
+      "loss": 0.1358,
+      "step": 102930
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3940278131815536e-06,
+      "loss": 0.1137,
+      "step": 102960
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3938246801052445e-06,
+      "loss": 0.1127,
+      "step": 102990
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3936216358026841e-06,
+      "loss": 0.1176,
+      "step": 103020
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3934186802092306e-06,
+      "loss": 0.0936,
+      "step": 103050
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3932158132603095e-06,
+      "loss": 0.1183,
+      "step": 103080
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3930130348914097e-06,
+      "loss": 0.1232,
+      "step": 103110
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3928103450380883e-06,
+      "loss": 0.106,
+      "step": 103140
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3926077436359668e-06,
+      "loss": 0.1055,
+      "step": 103170
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3924052306207317e-06,
+      "loss": 0.1181,
+      "step": 103200
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3922028059281358e-06,
+      "loss": 0.117,
+      "step": 103230
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3920004694939975e-06,
+      "loss": 0.1069,
+      "step": 103260
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3917982212541996e-06,
+      "loss": 0.1076,
+      "step": 103290
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3915960611446907e-06,
+      "loss": 0.103,
+      "step": 103320
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3913939891014843e-06,
+      "loss": 0.1132,
+      "step": 103350
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3911920050606586e-06,
+      "loss": 0.1139,
+      "step": 103380
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3909901089583574e-06,
+      "loss": 0.1523,
+      "step": 103410
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3907883007307891e-06,
+      "loss": 0.1019,
+      "step": 103440
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3905865803142269e-06,
+      "loss": 0.1363,
+      "step": 103470
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.390384947645008e-06,
+      "loss": 0.1106,
+      "step": 103500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3901834026595352e-06,
+      "loss": 0.1135,
+      "step": 103530
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.389988659128795e-06,
+      "loss": 0.1244,
+      "step": 103560
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.389787286402743e-06,
+      "loss": 0.1028,
+      "step": 103590
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3895860011721435e-06,
+      "loss": 0.1134,
+      "step": 103620
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3893848033736538e-06,
+      "loss": 0.1133,
+      "step": 103650
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.389183692943996e-06,
+      "loss": 0.1096,
+      "step": 103680
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3889826698199566e-06,
+      "loss": 0.1113,
+      "step": 103710
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3887817339383853e-06,
+      "loss": 0.1163,
+      "step": 103740
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.388580885236196e-06,
+      "loss": 0.1146,
+      "step": 103770
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3883801236503666e-06,
+      "loss": 0.1152,
+      "step": 103800
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3881794491179394e-06,
+      "loss": 0.1135,
+      "step": 103830
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3879788615760188e-06,
+      "loss": 0.1387,
+      "step": 103860
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.3877783609617737e-06,
+      "loss": 0.1048,
+      "step": 103890
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3875779472124371e-06,
+      "loss": 0.1132,
+      "step": 103920
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3873776202653045e-06,
+      "loss": 0.1212,
+      "step": 103950
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3871773800577352e-06,
+      "loss": 0.1054,
+      "step": 103980
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3869772265271516e-06,
+      "loss": 0.1036,
+      "step": 104010
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3867771596110397e-06,
+      "loss": 0.1176,
+      "step": 104040
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3865771792469473e-06,
+      "loss": 0.1169,
+      "step": 104070
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3863772853724872e-06,
+      "loss": 0.1286,
+      "step": 104100
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.3861774779253334e-06,
+      "loss": 0.1002,
+      "step": 104130
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.385977756843224e-06,
+      "loss": 0.1055,
+      "step": 104160
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3857781220639588e-06,
+      "loss": 0.1271,
+      "step": 104190
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3855785735254016e-06,
+      "loss": 0.111,
+      "step": 104220
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.385379111165477e-06,
+      "loss": 0.1064,
+      "step": 104250
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3851797349221744e-06,
+      "loss": 0.1118,
+      "step": 104280
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3849804447335435e-06,
+      "loss": 0.1082,
+      "step": 104310
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3847812405376978e-06,
+      "loss": 0.1194,
+      "step": 104340
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3845821222728126e-06,
+      "loss": 0.1165,
+      "step": 104370
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.384383089877125e-06,
+      "loss": 0.1352,
+      "step": 104400
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3841841432889352e-06,
+      "loss": 0.1224,
+      "step": 104430
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3839852824466047e-06,
+      "loss": 0.1184,
+      "step": 104460
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.383786507288557e-06,
+      "loss": 0.095,
+      "step": 104490
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3835878177532778e-06,
+      "loss": 0.114,
+      "step": 104520
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3833892137793143e-06,
+      "loss": 0.1229,
+      "step": 104550
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.383190695305276e-06,
+      "loss": 0.1303,
+      "step": 104580
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.382992262269833e-06,
+      "loss": 0.1409,
+      "step": 104610
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3827939146117186e-06,
+      "loss": 0.1269,
+      "step": 104640
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3825956522697256e-06,
+      "loss": 0.0983,
+      "step": 104670
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3823974751827096e-06,
+      "loss": 0.1142,
+      "step": 104700
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3821993832895875e-06,
+      "loss": 0.1333,
+      "step": 104730
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3820013765293367e-06,
+      "loss": 0.1081,
+      "step": 104760
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3818034548409964e-06,
+      "loss": 0.0999,
+      "step": 104790
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3816056181636663e-06,
+      "loss": 0.1191,
+      "step": 104820
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.381407866436508e-06,
+      "loss": 0.1172,
+      "step": 104850
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3812101995987435e-06,
+      "loss": 0.1165,
+      "step": 104880
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3810126175896551e-06,
+      "loss": 0.1119,
+      "step": 104910
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.380815120348587e-06,
+      "loss": 0.0985,
+      "step": 104940
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3806177078149435e-06,
+      "loss": 0.1227,
+      "step": 104970
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3804203799281896e-06,
+      "loss": 0.1505,
+      "step": 105000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3802231366278507e-06,
+      "loss": 0.1282,
+      "step": 105030
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.380025977853513e-06,
+      "loss": 0.1181,
+      "step": 105060
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3798289035448228e-06,
+      "loss": 0.1274,
+      "step": 105090
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.379631913641487e-06,
+      "loss": 0.0914,
+      "step": 105120
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3794350080832725e-06,
+      "loss": 0.1216,
+      "step": 105150
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3792381868100063e-06,
+      "loss": 0.1029,
+      "step": 105180
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3790414497615755e-06,
+      "loss": 0.134,
+      "step": 105210
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3788447968779277e-06,
+      "loss": 0.1104,
+      "step": 105240
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.37864822809907e-06,
+      "loss": 0.1341,
+      "step": 105270
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3784582915027789e-06,
+      "loss": 0.1317,
+      "step": 105300
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3782618879552259e-06,
+      "loss": 0.1131,
+      "step": 105330
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3780655683348362e-06,
+      "loss": 0.1161,
+      "step": 105360
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3778693325818535e-06,
+      "loss": 0.111,
+      "step": 105390
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3776731806365815e-06,
+      "loss": 0.1064,
+      "step": 105420
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3774771124393832e-06,
+      "loss": 0.1345,
+      "step": 105450
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3772811279306814e-06,
+      "loss": 0.1034,
+      "step": 105480
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3770852270509577e-06,
+      "loss": 0.1205,
+      "step": 105510
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.376889409740753e-06,
+      "loss": 0.1096,
+      "step": 105540
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3766936759406681e-06,
+      "loss": 0.1226,
+      "step": 105570
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.376498025591362e-06,
+      "loss": 0.1108,
+      "step": 105600
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3763024586335532e-06,
+      "loss": 0.1035,
+      "step": 105630
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.376106975008019e-06,
+      "loss": 0.1237,
+      "step": 105660
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.375911574655595e-06,
+      "loss": 0.1138,
+      "step": 105690
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3757162575171767e-06,
+      "loss": 0.0998,
+      "step": 105720
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3755210235337177e-06,
+      "loss": 0.1323,
+      "step": 105750
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.37532587264623e-06,
+      "loss": 0.0974,
+      "step": 105780
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3751308047957848e-06,
+      "loss": 0.116,
+      "step": 105810
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3749358199235115e-06,
+      "loss": 0.1157,
+      "step": 105840
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3747409179705975e-06,
+      "loss": 0.1282,
+      "step": 105870
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3745460988782894e-06,
+      "loss": 0.1165,
+      "step": 105900
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.374351362587891e-06,
+      "loss": 0.102,
+      "step": 105930
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3741567090407655e-06,
+      "loss": 0.1212,
+      "step": 105960
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3739621381783332e-06,
+      "loss": 0.1073,
+      "step": 105990
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3737676499420727e-06,
+      "loss": 0.1018,
+      "step": 106020
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3735732442735211e-06,
+      "loss": 0.0942,
+      "step": 106050
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3733789211142725e-06,
+      "loss": 0.1314,
+      "step": 106080
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.37318468040598e-06,
+      "loss": 0.1032,
+      "step": 106110
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3729905220903536e-06,
+      "loss": 0.0956,
+      "step": 106140
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3727964461091606e-06,
+      "loss": 0.1217,
+      "step": 106170
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.372602452404227e-06,
+      "loss": 0.1134,
+      "step": 106200
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3724085409174356e-06,
+      "loss": 0.1021,
+      "step": 106230
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3722147115907266e-06,
+      "loss": 0.1069,
+      "step": 106260
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3720209643660984e-06,
+      "loss": 0.0964,
+      "step": 106290
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3718272991856054e-06,
+      "loss": 0.1349,
+      "step": 106320
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.371633715991361e-06,
+      "loss": 0.109,
+      "step": 106350
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3714402147255338e-06,
+      "loss": 0.1037,
+      "step": 106380
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3712467953303504e-06,
+      "loss": 0.1073,
+      "step": 106410
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3710534577480952e-06,
+      "loss": 0.1274,
+      "step": 106440
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3708602019211083e-06,
+      "loss": 0.1111,
+      "step": 106470
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3706670277917871e-06,
+      "loss": 0.0943,
+      "step": 106500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3704739353025865e-06,
+      "loss": 0.1143,
+      "step": 106530
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3702809243960166e-06,
+      "loss": 0.1126,
+      "step": 106560
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3700879950146456e-06,
+      "loss": 0.1193,
+      "step": 106590
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3698951471010977e-06,
+      "loss": 0.1067,
+      "step": 106620
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3697023805980537e-06,
+      "loss": 0.1251,
+      "step": 106650
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3695096954482504e-06,
+      "loss": 0.1187,
+      "step": 106680
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.369317091594482e-06,
+      "loss": 0.1151,
+      "step": 106710
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3691245689795982e-06,
+      "loss": 0.1155,
+      "step": 106740
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3689321275465047e-06,
+      "loss": 0.1101,
+      "step": 106770
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3687397672381643e-06,
+      "loss": 0.1228,
+      "step": 106800
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3685474879975953e-06,
+      "loss": 0.1129,
+      "step": 106830
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3683552897678718e-06,
+      "loss": 0.1096,
+      "step": 106860
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3681631724921244e-06,
+      "loss": 0.1081,
+      "step": 106890
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.367971136113539e-06,
+      "loss": 0.1173,
+      "step": 106920
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3677791805753576e-06,
+      "loss": 0.1289,
+      "step": 106950
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3675873058208785e-06,
+      "loss": 0.1104,
+      "step": 106980
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3673955117934549e-06,
+      "loss": 0.1059,
+      "step": 107010
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3672037984364952e-06,
+      "loss": 0.1283,
+      "step": 107040
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3670121656934646e-06,
+      "loss": 0.0906,
+      "step": 107070
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3668206135078825e-06,
+      "loss": 0.1177,
+      "step": 107100
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3666291418233252e-06,
+      "loss": 0.1346,
+      "step": 107130
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3664377505834228e-06,
+      "loss": 0.126,
+      "step": 107160
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.366246439731861e-06,
+      "loss": 0.144,
+      "step": 107190
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.366055209212381e-06,
+      "loss": 0.1125,
+      "step": 107220
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3658640589687793e-06,
+      "loss": 0.1121,
+      "step": 107250
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.365672988944907e-06,
+      "loss": 0.1044,
+      "step": 107280
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3654819990846702e-06,
+      "loss": 0.0992,
+      "step": 107310
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3652910893320304e-06,
+      "loss": 0.1141,
+      "step": 107340
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.3651002596310025e-06,
+      "loss": 0.1074,
+      "step": 107370
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3649095099256587e-06,
+      "loss": 0.1144,
+      "step": 107400
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3647188401601233e-06,
+      "loss": 0.1032,
+      "step": 107430
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3645282502785764e-06,
+      "loss": 0.1415,
+      "step": 107460
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3643377402252533e-06,
+      "loss": 0.1187,
+      "step": 107490
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3641473099444421e-06,
+      "loss": 0.1106,
+      "step": 107520
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3639569593804868e-06,
+      "loss": 0.1035,
+      "step": 107550
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3637666884777853e-06,
+      "loss": 0.0971,
+      "step": 107580
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3635764971807894e-06,
+      "loss": 0.1054,
+      "step": 107610
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3633863854340057e-06,
+      "loss": 0.1055,
+      "step": 107640
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.363196353181994e-06,
+      "loss": 0.0957,
+      "step": 107670
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3630064003693693e-06,
+      "loss": 0.1063,
+      "step": 107700
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3628165269408007e-06,
+      "loss": 0.1403,
+      "step": 107730
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3626267328410096e-06,
+      "loss": 0.1029,
+      "step": 107760
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.362437018014773e-06,
+      "loss": 0.1215,
+      "step": 107790
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3622473824069208e-06,
+      "loss": 0.1096,
+      "step": 107820
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3620578259623368e-06,
+      "loss": 0.1101,
+      "step": 107850
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.361868348625959e-06,
+      "loss": 0.1188,
+      "step": 107880
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3616789503427782e-06,
+      "loss": 0.1114,
+      "step": 107910
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3614896310578394e-06,
+      "loss": 0.1102,
+      "step": 107940
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3613003907162404e-06,
+      "loss": 0.1081,
+      "step": 107970
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3611112292631326e-06,
+      "loss": 0.0958,
+      "step": 108000
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3609221466437218e-06,
+      "loss": 0.1194,
+      "step": 108030
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3607331428032656e-06,
+      "loss": 0.1281,
+      "step": 108060
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3605442176870751e-06,
+      "loss": 0.1063,
+      "step": 108090
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.360355371240515e-06,
+      "loss": 0.0942,
+      "step": 108120
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3601666034090034e-06,
+      "loss": 0.1045,
+      "step": 108150
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.35997791413801e-06,
+      "loss": 0.1134,
+      "step": 108180
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3597893033730595e-06,
+      "loss": 0.1106,
+      "step": 108210
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3596007710597273e-06,
+      "loss": 0.1185,
+      "step": 108240
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.359412317143643e-06,
+      "loss": 0.122,
+      "step": 108270
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3592239415704883e-06,
+      "loss": 0.1212,
+      "step": 108300
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.359035644285998e-06,
+      "loss": 0.1086,
+      "step": 108330
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3588536979444226e-06,
+      "loss": 0.1204,
+      "step": 108360
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3586655544695377e-06,
+      "loss": 0.1009,
+      "step": 108390
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3584774891226393e-06,
+      "loss": 0.1142,
+      "step": 108420
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.35828950184967e-06,
+      "loss": 0.119,
+      "step": 108450
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3581015925966255e-06,
+      "loss": 0.1209,
+      "step": 108480
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.357913761309553e-06,
+      "loss": 0.1119,
+      "step": 108510
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3577260079345526e-06,
+      "loss": 0.1257,
+      "step": 108540
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3575383324177758e-06,
+      "loss": 0.1217,
+      "step": 108570
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3573507347054265e-06,
+      "loss": 0.1121,
+      "step": 108600
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3571632147437613e-06,
+      "loss": 0.1233,
+      "step": 108630
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3569757724790883e-06,
+      "loss": 0.1067,
+      "step": 108660
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3567884078577668e-06,
+      "loss": 0.1206,
+      "step": 108690
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3566011208262083e-06,
+      "loss": 0.1261,
+      "step": 108720
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3564139113308771e-06,
+      "loss": 0.1353,
+      "step": 108750
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.356226779318288e-06,
+      "loss": 0.113,
+      "step": 108780
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3560397247350075e-06,
+      "loss": 0.1146,
+      "step": 108810
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3558527475276545e-06,
+      "loss": 0.112,
+      "step": 108840
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3556658476428985e-06,
+      "loss": 0.1154,
+      "step": 108870
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3554790250274613e-06,
+      "loss": 0.1079,
+      "step": 108900
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3552922796281152e-06,
+      "loss": 0.115,
+      "step": 108930
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3551056113916844e-06,
+      "loss": 0.106,
+      "step": 108960
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3549190202650446e-06,
+      "loss": 0.1007,
+      "step": 108990
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3547325061951213e-06,
+      "loss": 0.1171,
+      "step": 109020
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3545460691288933e-06,
+      "loss": 0.1176,
+      "step": 109050
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3543597090133884e-06,
+      "loss": 0.1057,
+      "step": 109080
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3541734257956868e-06,
+      "loss": 0.1183,
+      "step": 109110
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3539872194229187e-06,
+      "loss": 0.0971,
+      "step": 109140
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3538010898422657e-06,
+      "loss": 0.1048,
+      "step": 109170
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3536150370009606e-06,
+      "loss": 0.1265,
+      "step": 109200
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.353429060846286e-06,
+      "loss": 0.1153,
+      "step": 109230
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3532431613255755e-06,
+      "loss": 0.1093,
+      "step": 109260
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3530573383862138e-06,
+      "loss": 0.1128,
+      "step": 109290
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3528715919756358e-06,
+      "loss": 0.1073,
+      "step": 109320
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.352685922041327e-06,
+      "loss": 0.1076,
+      "step": 109350
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.352500328530823e-06,
+      "loss": 0.1086,
+      "step": 109380
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.35231481139171e-06,
+      "loss": 0.1398,
+      "step": 109410
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3521293705716248e-06,
+      "loss": 0.1132,
+      "step": 109440
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3519440060182543e-06,
+      "loss": 0.1092,
+      "step": 109470
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3517587176793354e-06,
+      "loss": 0.104,
+      "step": 109500
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.351573505502655e-06,
+      "loss": 0.1141,
+      "step": 109530
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3513883694360502e-06,
+      "loss": 0.1259,
+      "step": 109560
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3512033094274085e-06,
+      "loss": 0.1059,
+      "step": 109590
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3510183254246667e-06,
+      "loss": 0.1101,
+      "step": 109620
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3508334173758125e-06,
+      "loss": 0.1061,
+      "step": 109650
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3506485852288816e-06,
+      "loss": 0.1091,
+      "step": 109680
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3504638289319617e-06,
+      "loss": 0.1138,
+      "step": 109710
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3502791484331885e-06,
+      "loss": 0.1002,
+      "step": 109740
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3500945436807477e-06,
+      "loss": 0.0967,
+      "step": 109770
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3499100146228755e-06,
+      "loss": 0.1197,
+      "step": 109800
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.349725561207856e-06,
+      "loss": 0.1062,
+      "step": 109830
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3495411833840244e-06,
+      "loss": 0.1048,
+      "step": 109860
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3493568810997643e-06,
+      "loss": 0.1209,
+      "step": 109890
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3491726543035087e-06,
+      "loss": 0.1055,
+      "step": 109920
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3489885029437399e-06,
+      "loss": 0.0939,
+      "step": 109950
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3488044269689899e-06,
+      "loss": 0.1263,
+      "step": 109980
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3486204263278393e-06,
+      "loss": 0.0975,
+      "step": 110010
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3484365009689182e-06,
+      "loss": 0.1174,
+      "step": 110040
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.348252650840905e-06,
+      "loss": 0.1112,
+      "step": 110070
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.348068875892528e-06,
+      "loss": 0.1009,
+      "step": 110100
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.347885176072564e-06,
+      "loss": 0.1322,
+      "step": 110130
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.347701551329839e-06,
+      "loss": 0.1163,
+      "step": 110160
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.347518001613226e-06,
+      "loss": 0.1042,
+      "step": 110190
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.3473345268716497e-06,
+      "loss": 0.1109,
+      "step": 110220
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.3471511270540812e-06,
+      "loss": 0.116,
+      "step": 110250
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.346967802109541e-06,
+      "loss": 0.1151,
+      "step": 110280
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.3467845519870978e-06,
+      "loss": 0.1289,
+      "step": 110310
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3466013766358696e-06,
+      "loss": 0.1383,
+      "step": 110340
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3464182760050218e-06,
+      "loss": 0.1085,
+      "step": 110370
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3462352500437688e-06,
+      "loss": 0.1116,
+      "step": 110400
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3460522987013739e-06,
+      "loss": 0.1335,
+      "step": 110430
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3458694219271468e-06,
+      "loss": 0.1253,
+      "step": 110460
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3456866196704468e-06,
+      "loss": 0.0908,
+      "step": 110490
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3455038918806814e-06,
+      "loss": 0.1125,
+      "step": 110520
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3453212385073058e-06,
+      "loss": 0.1235,
+      "step": 110550
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3451386594998233e-06,
+      "loss": 0.1163,
+      "step": 110580
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.344956154807785e-06,
+      "loss": 0.1028,
+      "step": 110610
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.3447737243807896e-06,
+      "loss": 0.1258,
+      "step": 110640
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.3445913681684844e-06,
+      "loss": 0.1037,
+      "step": 110670
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.344409086120564e-06,
+      "loss": 0.1141,
+      "step": 110700
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.3442268781867715e-06,
+      "loss": 0.1071,
+      "step": 110730
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.344044744316896e-06,
+      "loss": 0.1079,
+      "step": 110760
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.343862684460776e-06,
+      "loss": 0.1089,
+      "step": 110790
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.3436806985682964e-06,
+      "loss": 0.1047,
+      "step": 110820
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.3434987865893903e-06,
+      "loss": 0.0858,
+      "step": 110850
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.3433169484740375e-06,
+      "loss": 0.105,
+      "step": 110880
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.343135184172266e-06,
+      "loss": 0.096,
+      "step": 110910
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.3429534936341501e-06,
+      "loss": 0.0958,
+      "step": 110940
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.3427718768098128e-06,
+      "loss": 0.1155,
+      "step": 110970
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.3425903336494232e-06,
+      "loss": 0.0927,
+      "step": 111000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.3424088641031974e-06,
+      "loss": 0.1146,
+      "step": 111030
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.342227468121399e-06,
+      "loss": 0.1023,
+      "step": 111060
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.3420461456543393e-06,
+      "loss": 0.1133,
+      "step": 111090
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.3418648966523754e-06,
+      "loss": 0.1097,
+      "step": 111120
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3416837210659122e-06,
+      "loss": 0.12,
+      "step": 111150
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3415086544046006e-06,
+      "loss": 0.1087,
+      "step": 111180
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3413276230574544e-06,
+      "loss": 0.1138,
+      "step": 111210
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3411466649789512e-06,
+      "loss": 0.1045,
+      "step": 111240
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3409657801196808e-06,
+      "loss": 0.1208,
+      "step": 111270
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.34078496843028e-06,
+      "loss": 0.1282,
+      "step": 111300
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3406042298614324e-06,
+      "loss": 0.1103,
+      "step": 111330
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3404235643638672e-06,
+      "loss": 0.1066,
+      "step": 111360
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3402429718883612e-06,
+      "loss": 0.1165,
+      "step": 111390
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.3400624523857367e-06,
+      "loss": 0.117,
+      "step": 111420
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.3398820058068632e-06,
+      "loss": 0.1059,
+      "step": 111450
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.339701632102656e-06,
+      "loss": 0.1196,
+      "step": 111480
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.339521331224076e-06,
+      "loss": 0.1074,
+      "step": 111510
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.3393411031221322e-06,
+      "loss": 0.1078,
+      "step": 111540
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.339160947747877e-06,
+      "loss": 0.1048,
+      "step": 111570
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.3389808650524116e-06,
+      "loss": 0.1235,
+      "step": 111600
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.3388008549868816e-06,
+      "loss": 0.116,
+      "step": 111630
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.3386209175024792e-06,
+      "loss": 0.1006,
+      "step": 111660
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3384410525504416e-06,
+      "loss": 0.1163,
+      "step": 111690
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3382612600820532e-06,
+      "loss": 0.1175,
+      "step": 111720
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3380815400486433e-06,
+      "loss": 0.1113,
+      "step": 111750
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3379018924015868e-06,
+      "loss": 0.1203,
+      "step": 111780
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3377223170923053e-06,
+      "loss": 0.1267,
+      "step": 111810
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3375428140722648e-06,
+      "loss": 0.1165,
+      "step": 111840
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3373633832929783e-06,
+      "loss": 0.1114,
+      "step": 111870
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3371840247060025e-06,
+      "loss": 0.1183,
+      "step": 111900
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.337004738262941e-06,
+      "loss": 0.1023,
+      "step": 111930
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3368255239154419e-06,
+      "loss": 0.1172,
+      "step": 111960
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3366463816151999e-06,
+      "loss": 0.106,
+      "step": 111990
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3364673113139533e-06,
+      "loss": 0.1254,
+      "step": 112020
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3362883129634875e-06,
+      "loss": 0.102,
+      "step": 112050
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3361093865156313e-06,
+      "loss": 0.1292,
+      "step": 112080
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.33593053192226e-06,
+      "loss": 0.1158,
+      "step": 112110
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3357517491352932e-06,
+      "loss": 0.1191,
+      "step": 112140
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3355730381066959e-06,
+      "loss": 0.1157,
+      "step": 112170
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.335394398788478e-06,
+      "loss": 0.0981,
+      "step": 112200
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.335215831132694e-06,
+      "loss": 0.1214,
+      "step": 112230
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3350373350914441e-06,
+      "loss": 0.114,
+      "step": 112260
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3348589106168724e-06,
+      "loss": 0.1069,
+      "step": 112290
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3346805576611683e-06,
+      "loss": 0.0948,
+      "step": 112320
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3345022761765652e-06,
+      "loss": 0.1153,
+      "step": 112350
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3343240661153423e-06,
+      "loss": 0.1226,
+      "step": 112380
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3341459274298223e-06,
+      "loss": 0.0996,
+      "step": 112410
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3339678600723734e-06,
+      "loss": 0.1247,
+      "step": 112440
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3337898639954075e-06,
+      "loss": 0.1071,
+      "step": 112470
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3336119391513816e-06,
+      "loss": 0.1086,
+      "step": 112500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.333434085492796e-06,
+      "loss": 0.1173,
+      "step": 112530
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3332563029721968e-06,
+      "loss": 0.1388,
+      "step": 112560
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3330785915421732e-06,
+      "loss": 0.1097,
+      "step": 112590
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3329009511553593e-06,
+      "loss": 0.1164,
+      "step": 112620
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.332723381764433e-06,
+      "loss": 0.1041,
+      "step": 112650
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3325458833221163e-06,
+      "loss": 0.1056,
+      "step": 112680
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3323684557811759e-06,
+      "loss": 0.1168,
+      "step": 112710
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3321910990944214e-06,
+      "loss": 0.125,
+      "step": 112740
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3320138132147075e-06,
+      "loss": 0.1154,
+      "step": 112770
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3318365980949322e-06,
+      "loss": 0.0917,
+      "step": 112800
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3316594536880373e-06,
+      "loss": 0.1145,
+      "step": 112830
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3314823799470083e-06,
+      "loss": 0.1074,
+      "step": 112860
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.331305376824875e-06,
+      "loss": 0.1049,
+      "step": 112890
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.331128444274711e-06,
+      "loss": 0.1078,
+      "step": 112920
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3309515822496324e-06,
+      "loss": 0.1126,
+      "step": 112950
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3307747907028002e-06,
+      "loss": 0.1207,
+      "step": 112980
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3305980695874179e-06,
+      "loss": 0.1214,
+      "step": 113010
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3304214188567336e-06,
+      "loss": 0.1187,
+      "step": 113040
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3302448384640373e-06,
+      "loss": 0.1061,
+      "step": 113070
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3300683283626642e-06,
+      "loss": 0.1027,
+      "step": 113100
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3298918885059912e-06,
+      "loss": 0.1075,
+      "step": 113130
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3297155188474394e-06,
+      "loss": 0.1261,
+      "step": 113160
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3295392193404732e-06,
+      "loss": 0.109,
+      "step": 113190
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3293629899385998e-06,
+      "loss": 0.1053,
+      "step": 113220
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.3291868305953697e-06,
+      "loss": 0.1162,
+      "step": 113250
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.329010741264376e-06,
+      "loss": 0.1282,
+      "step": 113280
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.3288347218992556e-06,
+      "loss": 0.1163,
+      "step": 113310
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.3286587724536878e-06,
+      "loss": 0.113,
+      "step": 113340
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.3284828928813952e-06,
+      "loss": 0.1118,
+      "step": 113370
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.3283070831361434e-06,
+      "loss": 0.115,
+      "step": 113400
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.3281313431717397e-06,
+      "loss": 0.0957,
+      "step": 113430
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.3279556729420353e-06,
+      "loss": 0.1053,
+      "step": 113460
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.3277800724009243e-06,
+      "loss": 0.1032,
+      "step": 113490
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.3276045415023425e-06,
+      "loss": 0.1142,
+      "step": 113520
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.3274290802002687e-06,
+      "loss": 0.0996,
+      "step": 113550
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.3272536884487247e-06,
+      "loss": 0.1274,
+      "step": 113580
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.327078366201774e-06,
+      "loss": 0.0937,
+      "step": 113610
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.3269031134135237e-06,
+      "loss": 0.1127,
+      "step": 113640
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.3267279300381222e-06,
+      "loss": 0.1036,
+      "step": 113670
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.3265528160297608e-06,
+      "loss": 0.1173,
+      "step": 113700
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.3263777713426725e-06,
+      "loss": 0.1237,
+      "step": 113730
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.3262027959311342e-06,
+      "loss": 0.1113,
+      "step": 113760
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.3260278897494627e-06,
+      "loss": 0.1228,
+      "step": 113790
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.3258530527520192e-06,
+      "loss": 0.1077,
+      "step": 113820
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.3256782848932048e-06,
+      "loss": 0.1051,
+      "step": 113850
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.325503586127465e-06,
+      "loss": 0.1137,
+      "step": 113880
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.3253289564092853e-06,
+      "loss": 0.1075,
+      "step": 113910
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.3251543956931945e-06,
+      "loss": 0.0949,
+      "step": 113940
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.3249799039337626e-06,
+      "loss": 0.0969,
+      "step": 113970
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.3248054810856017e-06,
+      "loss": 0.0864,
+      "step": 114000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.3246311271033657e-06,
+      "loss": 0.142,
+      "step": 114030
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.3244568419417503e-06,
+      "loss": 0.0939,
+      "step": 114060
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.3242826255554928e-06,
+      "loss": 0.1117,
+      "step": 114090
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.3241084778993723e-06,
+      "loss": 0.0976,
+      "step": 114120
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.3239343989282094e-06,
+      "loss": 0.0982,
+      "step": 114150
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.3237603885968664e-06,
+      "loss": 0.1163,
+      "step": 114180
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.323586446860247e-06,
+      "loss": 0.1309,
+      "step": 114210
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.3234125736732962e-06,
+      "loss": 0.1001,
+      "step": 114240
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.3232387689910007e-06,
+      "loss": 0.1048,
+      "step": 114270
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.3230650327683888e-06,
+      "loss": 0.1139,
+      "step": 114300
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.3228913649605295e-06,
+      "loss": 0.1219,
+      "step": 114330
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.3227177655225338e-06,
+      "loss": 0.1055,
+      "step": 114360
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.322544234409553e-06,
+      "loss": 0.0944,
+      "step": 114390
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.32237077157678e-06,
+      "loss": 0.1249,
+      "step": 114420
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.3221973769794496e-06,
+      "loss": 0.1084,
+      "step": 114450
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.3220240505728365e-06,
+      "loss": 0.0998,
+      "step": 114480
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.3218507923122572e-06,
+      "loss": 0.1065,
+      "step": 114510
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.3216776021530689e-06,
+      "loss": 0.1051,
+      "step": 114540
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.321504480050669e-06,
+      "loss": 0.1216,
+      "step": 114570
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.3213314259604976e-06,
+      "loss": 0.097,
+      "step": 114600
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.321158439838034e-06,
+      "loss": 0.1079,
+      "step": 114630
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.3209855216387988e-06,
+      "loss": 0.108,
+      "step": 114660
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.3208126713183533e-06,
+      "loss": 0.1599,
+      "step": 114690
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.3206398888323003e-06,
+      "loss": 0.1108,
+      "step": 114720
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.3204671741362815e-06,
+      "loss": 0.1027,
+      "step": 114750
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.320294527185981e-06,
+      "loss": 0.1121,
+      "step": 114780
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.3201219479371225e-06,
+      "loss": 0.0957,
+      "step": 114810
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.3199494363454705e-06,
+      "loss": 0.0941,
+      "step": 114840
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.319776992366829e-06,
+      "loss": 0.1005,
+      "step": 114870
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.3196046159570445e-06,
+      "loss": 0.1123,
+      "step": 114900
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.3194323070720016e-06,
+      "loss": 0.11,
+      "step": 114930
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.3192600656676268e-06,
+      "loss": 0.1043,
+      "step": 114960
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.3190878916998856e-06,
+      "loss": 0.1014,
+      "step": 114990
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.3189157851247853e-06,
+      "loss": 0.1287,
+      "step": 115020
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.3187437458983718e-06,
+      "loss": 0.1203,
+      "step": 115050
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.318571773976732e-06,
+      "loss": 0.1252,
+      "step": 115080
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.3183998693159926e-06,
+      "loss": 0.1076,
+      "step": 115110
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.3182280318723206e-06,
+      "loss": 0.0897,
+      "step": 115140
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.3180562616019221e-06,
+      "loss": 0.102,
+      "step": 115170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.3178845584610447e-06,
+      "loss": 0.114,
+      "step": 115200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.3177129224059746e-06,
+      "loss": 0.127,
+      "step": 115230
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.3175413533930378e-06,
+      "loss": 0.1226,
+      "step": 115260
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.3173698513786012e-06,
+      "loss": 0.0991,
+      "step": 115290
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.3171984163190704e-06,
+      "loss": 0.1143,
+      "step": 115320
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.3170270481708914e-06,
+      "loss": 0.1015,
+      "step": 115350
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.3168557468905489e-06,
+      "loss": 0.1062,
+      "step": 115380
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.3166845124345683e-06,
+      "loss": 0.1077,
+      "step": 115410
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.316513344759514e-06,
+      "loss": 0.1142,
+      "step": 115440
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3163422438219902e-06,
+      "loss": 0.1156,
+      "step": 115470
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3161712095786402e-06,
+      "loss": 0.118,
+      "step": 115500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3160002419861468e-06,
+      "loss": 0.1225,
+      "step": 115530
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3158293410012324e-06,
+      "loss": 0.1199,
+      "step": 115560
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3156585065806587e-06,
+      "loss": 0.1145,
+      "step": 115590
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3154877386812268e-06,
+      "loss": 0.1273,
+      "step": 115620
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.315317037259776e-06,
+      "loss": 0.1092,
+      "step": 115650
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.3151464022731864e-06,
+      "loss": 0.1297,
+      "step": 115680
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3149758336783764e-06,
+      "loss": 0.1129,
+      "step": 115710
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.314805331432303e-06,
+      "loss": 0.098,
+      "step": 115740
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3146348954919639e-06,
+      "loss": 0.1029,
+      "step": 115770
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3144645258143936e-06,
+      "loss": 0.1019,
+      "step": 115800
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3142942223566676e-06,
+      "loss": 0.1293,
+      "step": 115830
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3141239850758991e-06,
+      "loss": 0.1258,
+      "step": 115860
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3139538139292404e-06,
+      "loss": 0.1157,
+      "step": 115890
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.313783708873883e-06,
+      "loss": 0.1181,
+      "step": 115920
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.3136136698670565e-06,
+      "loss": 0.1069,
+      "step": 115950
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3134436968660303e-06,
+      "loss": 0.1075,
+      "step": 115980
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3132737898281114e-06,
+      "loss": 0.0945,
+      "step": 116010
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3131039487106457e-06,
+      "loss": 0.1112,
+      "step": 116040
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3129341734710184e-06,
+      "loss": 0.1181,
+      "step": 116070
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3127644640666526e-06,
+      "loss": 0.0994,
+      "step": 116100
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3125948204550102e-06,
+      "loss": 0.0933,
+      "step": 116130
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.312425242593591e-06,
+      "loss": 0.1108,
+      "step": 116160
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3122557304399341e-06,
+      "loss": 0.0987,
+      "step": 116190
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.3120862839516165e-06,
+      "loss": 0.1096,
+      "step": 116220
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3119169030862539e-06,
+      "loss": 0.1133,
+      "step": 116250
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3117475878014992e-06,
+      "loss": 0.1315,
+      "step": 116280
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3115783380550451e-06,
+      "loss": 0.1295,
+      "step": 116310
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.311409153804621e-06,
+      "loss": 0.0983,
+      "step": 116340
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3112400350079959e-06,
+      "loss": 0.1068,
+      "step": 116370
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3110709816229757e-06,
+      "loss": 0.1147,
+      "step": 116400
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3109019936074052e-06,
+      "loss": 0.1137,
+      "step": 116430
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.3107330709191668e-06,
+      "loss": 0.1208,
+      "step": 116460
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.310564213516181e-06,
+      "loss": 0.1401,
+      "step": 116490
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3103954213564058e-06,
+      "loss": 0.1,
+      "step": 116520
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.310226694397838e-06,
+      "loss": 0.0975,
+      "step": 116550
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3100580325985118e-06,
+      "loss": 0.1046,
+      "step": 116580
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.309889435916499e-06,
+      "loss": 0.1025,
+      "step": 116610
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3097209043099094e-06,
+      "loss": 0.0954,
+      "step": 116640
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3095524377368902e-06,
+      "loss": 0.1025,
+      "step": 116670
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3093840361556264e-06,
+      "loss": 0.1078,
+      "step": 116700
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3092156995243415e-06,
+      "loss": 0.1226,
+      "step": 116730
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.3090474278012954e-06,
+      "loss": 0.1026,
+      "step": 116760
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.308879220944786e-06,
+      "loss": 0.104,
+      "step": 116790
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.3087110789131483e-06,
+      "loss": 0.0929,
+      "step": 116820
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.3085430016647556e-06,
+      "loss": 0.1205,
+      "step": 116850
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.308374989158018e-06,
+      "loss": 0.1218,
+      "step": 116880
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.3082070413513832e-06,
+      "loss": 0.0964,
+      "step": 116910
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.3080391582033357e-06,
+      "loss": 0.1315,
+      "step": 116940
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.3078713396723986e-06,
+      "loss": 0.1172,
+      "step": 116970
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.3077035857171309e-06,
+      "loss": 0.0963,
+      "step": 117000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.307535896296129e-06,
+      "loss": 0.1049,
+      "step": 117030
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.3073682713680269e-06,
+      "loss": 0.1218,
+      "step": 117060
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.3072007108914958e-06,
+      "loss": 0.0959,
+      "step": 117090
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.3070332148252434e-06,
+      "loss": 0.105,
+      "step": 117120
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.306865783128015e-06,
+      "loss": 0.1334,
+      "step": 117150
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.3066984157585923e-06,
+      "loss": 0.11,
+      "step": 117180
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.3065311126757948e-06,
+      "loss": 0.098,
+      "step": 117210
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.3063694474317565e-06,
+      "loss": 0.133,
+      "step": 117240
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.3062022706593288e-06,
+      "loss": 0.1123,
+      "step": 117270
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.3060351580515726e-06,
+      "loss": 0.1021,
+      "step": 117300
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.305868109567453e-06,
+      "loss": 0.1113,
+      "step": 117330
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.3057011251659715e-06,
+      "loss": 0.1077,
+      "step": 117360
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.3055342048061668e-06,
+      "loss": 0.1296,
+      "step": 117390
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.305367348447114e-06,
+      "loss": 0.1151,
+      "step": 117420
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.3052005560479244e-06,
+      "loss": 0.1305,
+      "step": 117450
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.305033827567747e-06,
+      "loss": 0.1127,
+      "step": 117480
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.3048671629657663e-06,
+      "loss": 0.1051,
+      "step": 117510
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.3047005622012037e-06,
+      "loss": 0.117,
+      "step": 117540
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.304534025233317e-06,
+      "loss": 0.0994,
+      "step": 117570
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.3043675520214e-06,
+      "loss": 0.1318,
+      "step": 117600
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.3042011425247836e-06,
+      "loss": 0.1427,
+      "step": 117630
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.3040347967028342e-06,
+      "loss": 0.0987,
+      "step": 117660
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.3038685145149548e-06,
+      "loss": 0.1065,
+      "step": 117690
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.3037022959205856e-06,
+      "loss": 0.1003,
+      "step": 117720
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.303536140879201e-06,
+      "loss": 0.1064,
+      "step": 117750
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.303370049350313e-06,
+      "loss": 0.0985,
+      "step": 117780
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.30320402129347e-06,
+      "loss": 0.1175,
+      "step": 117810
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.3030380566682546e-06,
+      "loss": 0.1028,
+      "step": 117840
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.302872155434287e-06,
+      "loss": 0.1041,
+      "step": 117870
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.3027063175512236e-06,
+      "loss": 0.12,
+      "step": 117900
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.3025405429787558e-06,
+      "loss": 0.1162,
+      "step": 117930
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.3023748316766109e-06,
+      "loss": 0.1014,
+      "step": 117960
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.3022091836045526e-06,
+      "loss": 0.1341,
+      "step": 117990
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.3020435987223801e-06,
+      "loss": 0.1052,
+      "step": 118020
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.3018780769899285e-06,
+      "loss": 0.0979,
+      "step": 118050
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.301712618367069e-06,
+      "loss": 0.1356,
+      "step": 118080
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.3015472228137074e-06,
+      "loss": 0.1122,
+      "step": 118110
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.3013818902897862e-06,
+      "loss": 0.1188,
+      "step": 118140
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.3012166207552833e-06,
+      "loss": 0.0868,
+      "step": 118170
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.3010514141702115e-06,
+      "loss": 0.1188,
+      "step": 118200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.30088627049462e-06,
+      "loss": 0.1018,
+      "step": 118230
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.3007211896885935e-06,
+      "loss": 0.1174,
+      "step": 118260
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.3005561717122512e-06,
+      "loss": 0.0993,
+      "step": 118290
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.3003912165257485e-06,
+      "loss": 0.1044,
+      "step": 118320
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.300226324089276e-06,
+      "loss": 0.1051,
+      "step": 118350
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.3000614943630596e-06,
+      "loss": 0.1175,
+      "step": 118380
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.2998967273073604e-06,
+      "loss": 0.1079,
+      "step": 118410
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.2997320228824746e-06,
+      "loss": 0.088,
+      "step": 118440
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.2995673810487344e-06,
+      "loss": 0.1007,
+      "step": 118470
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.2994028017665058e-06,
+      "loss": 0.1451,
+      "step": 118500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.2992382849961916e-06,
+      "loss": 0.1068,
+      "step": 118530
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.299073830698228e-06,
+      "loss": 0.0986,
+      "step": 118560
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.2989094388330875e-06,
+      "loss": 0.1178,
+      "step": 118590
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.2987451093612773e-06,
+      "loss": 0.1121,
+      "step": 118620
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.2985808422433387e-06,
+      "loss": 0.1215,
+      "step": 118650
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.2984166374398496e-06,
+      "loss": 0.1104,
+      "step": 118680
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.2982524949114212e-06,
+      "loss": 0.1247,
+      "step": 118710
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.2980884146187004e-06,
+      "loss": 0.1251,
+      "step": 118740
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.297924396522368e-06,
+      "loss": 0.1199,
+      "step": 118770
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.2977604405831415e-06,
+      "loss": 0.1071,
+      "step": 118800
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.297596546761771e-06,
+      "loss": 0.0898,
+      "step": 118830
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.2974327150190424e-06,
+      "loss": 0.1061,
+      "step": 118860
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.2972689453157762e-06,
+      "loss": 0.1134,
+      "step": 118890
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2971052376128267e-06,
+      "loss": 0.1091,
+      "step": 118920
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.296947045731291e-06,
+      "loss": 0.1343,
+      "step": 118950
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2967834598482373e-06,
+      "loss": 0.1069,
+      "step": 118980
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2966199358495734e-06,
+      "loss": 0.1203,
+      "step": 119010
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2964564736962915e-06,
+      "loss": 0.1226,
+      "step": 119040
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2962930733494176e-06,
+      "loss": 0.1056,
+      "step": 119070
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2961297347700125e-06,
+      "loss": 0.1027,
+      "step": 119100
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2959664579191709e-06,
+      "loss": 0.0915,
+      "step": 119130
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.295803242758022e-06,
+      "loss": 0.0957,
+      "step": 119160
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2956400892477298e-06,
+      "loss": 0.1198,
+      "step": 119190
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2954769973494915e-06,
+      "loss": 0.1063,
+      "step": 119220
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2953139670245392e-06,
+      "loss": 0.0955,
+      "step": 119250
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2951509982341392e-06,
+      "loss": 0.1139,
+      "step": 119280
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2949880909395915e-06,
+      "loss": 0.124,
+      "step": 119310
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2948252451022304e-06,
+      "loss": 0.1036,
+      "step": 119340
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2946624606834235e-06,
+      "loss": 0.0989,
+      "step": 119370
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.294499737644574e-06,
+      "loss": 0.135,
+      "step": 119400
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2943370759471175e-06,
+      "loss": 0.1173,
+      "step": 119430
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2941744755525243e-06,
+      "loss": 0.1293,
+      "step": 119460
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2940119364222983e-06,
+      "loss": 0.1396,
+      "step": 119490
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2938494585179772e-06,
+      "loss": 0.1242,
+      "step": 119520
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.293687041801133e-06,
+      "loss": 0.1183,
+      "step": 119550
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2935246862333709e-06,
+      "loss": 0.1215,
+      "step": 119580
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2933623917763297e-06,
+      "loss": 0.1166,
+      "step": 119610
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2932001583916825e-06,
+      "loss": 0.1092,
+      "step": 119640
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2930379860411354e-06,
+      "loss": 0.1093,
+      "step": 119670
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.292875874686429e-06,
+      "loss": 0.103,
+      "step": 119700
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2927138242893364e-06,
+      "loss": 0.1017,
+      "step": 119730
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2925518348116648e-06,
+      "loss": 0.1116,
+      "step": 119760
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2923899062152547e-06,
+      "loss": 0.1123,
+      "step": 119790
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2922280384619804e-06,
+      "loss": 0.1173,
+      "step": 119820
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2920662315137495e-06,
+      "loss": 0.0982,
+      "step": 119850
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2919044853325025e-06,
+      "loss": 0.1055,
+      "step": 119880
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.291742799880214e-06,
+      "loss": 0.0932,
+      "step": 119910
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.291581175118891e-06,
+      "loss": 0.1237,
+      "step": 119940
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2914196110105747e-06,
+      "loss": 0.1194,
+      "step": 119970
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2912581075173393e-06,
+      "loss": 0.1114,
+      "step": 120000
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2910966646012915e-06,
+      "loss": 0.1114,
+      "step": 120030
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2909352822245722e-06,
+      "loss": 0.1029,
+      "step": 120060
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.290773960349354e-06,
+      "loss": 0.1043,
+      "step": 120090
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2906126989378443e-06,
+      "loss": 0.1089,
+      "step": 120120
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2904514979522822e-06,
+      "loss": 0.09,
+      "step": 120150
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2902903573549407e-06,
+      "loss": 0.0915,
+      "step": 120180
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2901292771081256e-06,
+      "loss": 0.0993,
+      "step": 120210
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2899682571741744e-06,
+      "loss": 0.1236,
+      "step": 120240
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2898072975154596e-06,
+      "loss": 0.096,
+      "step": 120270
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.289646398094385e-06,
+      "loss": 0.1118,
+      "step": 120300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2894855588733873e-06,
+      "loss": 0.1015,
+      "step": 120330
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2893247798149368e-06,
+      "loss": 0.1416,
+      "step": 120360
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2891640608815364e-06,
+      "loss": 0.1075,
+      "step": 120390
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.289003402035721e-06,
+      "loss": 0.1143,
+      "step": 120420
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2888428032400585e-06,
+      "loss": 0.0982,
+      "step": 120450
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2886822644571501e-06,
+      "loss": 0.1061,
+      "step": 120480
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2885217856496285e-06,
+      "loss": 0.1185,
+      "step": 120510
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2883613667801595e-06,
+      "loss": 0.1238,
+      "step": 120540
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.288201007811442e-06,
+      "loss": 0.1035,
+      "step": 120570
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2880407087062064e-06,
+      "loss": 0.1049,
+      "step": 120600
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2878804694272156e-06,
+      "loss": 0.1034,
+      "step": 120630
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2877202899372663e-06,
+      "loss": 0.113,
+      "step": 120660
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.287560170199186e-06,
+      "loss": 0.1191,
+      "step": 120690
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.287400110175835e-06,
+      "loss": 0.1073,
+      "step": 120720
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2872401098301063e-06,
+      "loss": 0.113,
+      "step": 120750
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.287080169124925e-06,
+      "loss": 0.0863,
+      "step": 120780
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2869202880232477e-06,
+      "loss": 0.113,
+      "step": 120810
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2867604664880645e-06,
+      "loss": 0.1062,
+      "step": 120840
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2866007044823972e-06,
+      "loss": 0.1116,
+      "step": 120870
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.286441001969299e-06,
+      "loss": 0.0937,
+      "step": 120900
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2862813589118562e-06,
+      "loss": 0.1041,
+      "step": 120930
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.286121775273186e-06,
+      "loss": 0.0942,
+      "step": 120960
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.285962251016439e-06,
+      "loss": 0.1193,
+      "step": 120990
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2858027861047972e-06,
+      "loss": 0.0999,
+      "step": 121020
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.285643380501474e-06,
+      "loss": 0.1127,
+      "step": 121050
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2854840341697152e-06,
+      "loss": 0.1086,
+      "step": 121080
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.285324747072799e-06,
+      "loss": 0.1131,
+      "step": 121110
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2851655191740345e-06,
+      "loss": 0.1009,
+      "step": 121140
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.285006350436763e-06,
+      "loss": 0.119,
+      "step": 121170
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2848525435259235e-06,
+      "loss": 0.0867,
+      "step": 121200
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2846934910327693e-06,
+      "loss": 0.1177,
+      "step": 121230
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2845344975925409e-06,
+      "loss": 0.1146,
+      "step": 121260
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2843755631687056e-06,
+      "loss": 0.1109,
+      "step": 121290
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2842166877247616e-06,
+      "loss": 0.1045,
+      "step": 121320
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.28405787122424e-06,
+      "loss": 0.0967,
+      "step": 121350
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2838991136307022e-06,
+      "loss": 0.1022,
+      "step": 121380
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.283740414907742e-06,
+      "loss": 0.1271,
+      "step": 121410
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2835817750189845e-06,
+      "loss": 0.1612,
+      "step": 121440
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2834231939280856e-06,
+      "loss": 0.1181,
+      "step": 121470
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2832646715987334e-06,
+      "loss": 0.125,
+      "step": 121500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.283106207994647e-06,
+      "loss": 0.113,
+      "step": 121530
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2829478030795776e-06,
+      "loss": 0.109,
+      "step": 121560
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.282789456817306e-06,
+      "loss": 0.112,
+      "step": 121590
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.2826311691716465e-06,
+      "loss": 0.1436,
+      "step": 121620
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.282472940106443e-06,
+      "loss": 0.0931,
+      "step": 121650
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.2823147695855707e-06,
+      "loss": 0.1049,
+      "step": 121680
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.282156657572937e-06,
+      "loss": 0.0898,
+      "step": 121710
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.2819986040324803e-06,
+      "loss": 0.0961,
+      "step": 121740
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.2818406089281688e-06,
+      "loss": 0.1009,
+      "step": 121770
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.2816826722240031e-06,
+      "loss": 0.0844,
+      "step": 121800
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.2815247938840142e-06,
+      "loss": 0.1061,
+      "step": 121830
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.2813669738722644e-06,
+      "loss": 0.113,
+      "step": 121860
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.281209212152847e-06,
+      "loss": 0.098,
+      "step": 121890
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.2810515086898859e-06,
+      "loss": 0.1026,
+      "step": 121920
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.2808938634475362e-06,
+      "loss": 0.1044,
+      "step": 121950
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.2807362763899839e-06,
+      "loss": 0.1047,
+      "step": 121980
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.2805787474814456e-06,
+      "loss": 0.1011,
+      "step": 122010
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.280421276686169e-06,
+      "loss": 0.0955,
+      "step": 122040
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.2802638639684322e-06,
+      "loss": 0.0999,
+      "step": 122070
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.280106509292544e-06,
+      "loss": 0.1088,
+      "step": 122100
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.279949212622845e-06,
+      "loss": 0.1164,
+      "step": 122130
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.2797919739237045e-06,
+      "loss": 0.0978,
+      "step": 122160
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.2796347931595244e-06,
+      "loss": 0.1141,
+      "step": 122190
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.2794776702947356e-06,
+      "loss": 0.1212,
+      "step": 122220
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.2793206052938007e-06,
+      "loss": 0.1063,
+      "step": 122250
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.2791635981212126e-06,
+      "loss": 0.1107,
+      "step": 122280
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.2790066487414936e-06,
+      "loss": 0.1031,
+      "step": 122310
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.2788497571191988e-06,
+      "loss": 0.1171,
+      "step": 122340
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.2786929232189112e-06,
+      "loss": 0.1054,
+      "step": 122370
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.2785361470052457e-06,
+      "loss": 0.1144,
+      "step": 122400
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.2783794284428473e-06,
+      "loss": 0.0912,
+      "step": 122430
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.2782227674963907e-06,
+      "loss": 0.0941,
+      "step": 122460
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.2780661641305817e-06,
+      "loss": 0.1035,
+      "step": 122490
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.2779096183101562e-06,
+      "loss": 0.1032,
+      "step": 122520
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.27775312999988e-06,
+      "loss": 0.1078,
+      "step": 122550
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.2775966991645491e-06,
+      "loss": 0.0942,
+      "step": 122580
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.2774403257689902e-06,
+      "loss": 0.1285,
+      "step": 122610
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.2772840097780594e-06,
+      "loss": 0.0951,
+      "step": 122640
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.2771277511566435e-06,
+      "loss": 0.1057,
+      "step": 122670
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.276971549869659e-06,
+      "loss": 0.1358,
+      "step": 122700
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.2768154058820525e-06,
+      "loss": 0.1089,
+      "step": 122730
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.2766593191588008e-06,
+      "loss": 0.0994,
+      "step": 122760
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.2765032896649106e-06,
+      "loss": 0.0995,
+      "step": 122790
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.276347317365418e-06,
+      "loss": 0.1219,
+      "step": 122820
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.2761914022253899e-06,
+      "loss": 0.1226,
+      "step": 122850
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.2760355442099224e-06,
+      "loss": 0.1121,
+      "step": 122880
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.2758797432841417e-06,
+      "loss": 0.1174,
+      "step": 122910
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.275723999413203e-06,
+      "loss": 0.117,
+      "step": 122940
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.2755683125622934e-06,
+      "loss": 0.1104,
+      "step": 122970
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.2754126826966274e-06,
+      "loss": 0.122,
+      "step": 123000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.2752571097814504e-06,
+      "loss": 0.1165,
+      "step": 123030
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.2751015937820368e-06,
+      "loss": 0.1147,
+      "step": 123060
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.2749461346636915e-06,
+      "loss": 0.1127,
+      "step": 123090
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.2747907323917488e-06,
+      "loss": 0.1077,
+      "step": 123120
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.2746353869315714e-06,
+      "loss": 0.1128,
+      "step": 123150
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.274480098248553e-06,
+      "loss": 0.109,
+      "step": 123180
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.2743248663081165e-06,
+      "loss": 0.1067,
+      "step": 123210
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.2741696910757134e-06,
+      "loss": 0.1056,
+      "step": 123240
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.2740145725168258e-06,
+      "loss": 0.1165,
+      "step": 123270
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.2738595105969641e-06,
+      "loss": 0.1169,
+      "step": 123300
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.2737045052816693e-06,
+      "loss": 0.1161,
+      "step": 123330
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.2735495565365108e-06,
+      "loss": 0.1099,
+      "step": 123360
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.2733946643270871e-06,
+      "loss": 0.1046,
+      "step": 123390
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.2732398286190276e-06,
+      "loss": 0.11,
+      "step": 123420
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.2730850493779893e-06,
+      "loss": 0.1129,
+      "step": 123450
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.2729303265696585e-06,
+      "loss": 0.0978,
+      "step": 123480
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.2727756601597516e-06,
+      "loss": 0.1122,
+      "step": 123510
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.2726210501140138e-06,
+      "loss": 0.1107,
+      "step": 123540
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.272466496398219e-06,
+      "loss": 0.1242,
+      "step": 123570
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.2723119989781708e-06,
+      "loss": 0.1147,
+      "step": 123600
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.2721575578197013e-06,
+      "loss": 0.1031,
+      "step": 123630
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.2720031728886717e-06,
+      "loss": 0.118,
+      "step": 123660
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.271848844150973e-06,
+      "loss": 0.1143,
+      "step": 123690
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.2716945715725241e-06,
+      "loss": 0.1111,
+      "step": 123720
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.2715403551192732e-06,
+      "loss": 0.0922,
+      "step": 123750
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.2713861947571978e-06,
+      "loss": 0.1152,
+      "step": 123780
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.2712320904523037e-06,
+      "loss": 0.1104,
+      "step": 123810
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.2710780421706256e-06,
+      "loss": 0.1041,
+      "step": 123840
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.270924049878228e-06,
+      "loss": 0.1147,
+      "step": 123870
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.2707701135412023e-06,
+      "loss": 0.1253,
+      "step": 123900
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.27061623312567e-06,
+      "loss": 0.0983,
+      "step": 123930
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.2704624085977816e-06,
+      "loss": 0.1467,
+      "step": 123960
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.2703086399237149e-06,
+      "loss": 0.1096,
+      "step": 123990
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.2701549270696776e-06,
+      "loss": 0.1055,
+      "step": 124020
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.270001270001905e-06,
+      "loss": 0.0958,
+      "step": 124050
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.2698476686866623e-06,
+      "loss": 0.1173,
+      "step": 124080
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.2696941230902421e-06,
+      "loss": 0.1011,
+      "step": 124110
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.2695406331789656e-06,
+      "loss": 0.0969,
+      "step": 124140
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.2693871989191832e-06,
+      "loss": 0.1129,
+      "step": 124170
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.2692338202772736e-06,
+      "loss": 0.0921,
+      "step": 124200
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.2690804972196432e-06,
+      "loss": 0.1043,
+      "step": 124230
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.2689272297127274e-06,
+      "loss": 0.1239,
+      "step": 124260
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.26877401772299e-06,
+      "loss": 0.1071,
+      "step": 124290
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.2686208612169226e-06,
+      "loss": 0.106,
+      "step": 124320
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.268467760161046e-06,
+      "loss": 0.0871,
+      "step": 124350
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.2683147145219085e-06,
+      "loss": 0.0989,
+      "step": 124380
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.268161724266087e-06,
+      "loss": 0.1087,
+      "step": 124410
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.2680138862989871e-06,
+      "loss": 0.1111,
+      "step": 124440
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.2678610048662932e-06,
+      "loss": 0.1169,
+      "step": 124470
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.2677081787179258e-06,
+      "loss": 0.1131,
+      "step": 124500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.2675554078205732e-06,
+      "loss": 0.0931,
+      "step": 124530
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2674026921409524e-06,
+      "loss": 0.1193,
+      "step": 124560
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.267250031645808e-06,
+      "loss": 0.1002,
+      "step": 124590
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.267097426301913e-06,
+      "loss": 0.1363,
+      "step": 124620
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2669448760760672e-06,
+      "loss": 0.1128,
+      "step": 124650
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2667923809351006e-06,
+      "loss": 0.1391,
+      "step": 124680
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2666399408458692e-06,
+      "loss": 0.1102,
+      "step": 124710
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2664875557752578e-06,
+      "loss": 0.1066,
+      "step": 124740
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2663352256901788e-06,
+      "loss": 0.093,
+      "step": 124770
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2661829505575724e-06,
+      "loss": 0.1056,
+      "step": 124800
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.266030730344407e-06,
+      "loss": 0.1005,
+      "step": 124830
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.265878565017678e-06,
+      "loss": 0.1037,
+      "step": 124860
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2657264545444095e-06,
+      "loss": 0.1174,
+      "step": 124890
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2655743988916528e-06,
+      "loss": 0.1028,
+      "step": 124920
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2654223980264873e-06,
+      "loss": 0.1058,
+      "step": 124950
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2652704519160192e-06,
+      "loss": 0.1087,
+      "step": 124980
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2651185605273832e-06,
+      "loss": 0.1167,
+      "step": 125010
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2649667238277418e-06,
+      "loss": 0.1114,
+      "step": 125040
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2648149417842841e-06,
+      "loss": 0.088,
+      "step": 125070
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.2646632143642272e-06,
+      "loss": 0.1024,
+      "step": 125100
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.2645115415348164e-06,
+      "loss": 0.091,
+      "step": 125130
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.2643599232633234e-06,
+      "loss": 0.0915,
+      "step": 125160
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.2642083595170476e-06,
+      "loss": 0.0983,
+      "step": 125190
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.2640568502633168e-06,
+      "loss": 0.1039,
+      "step": 125220
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.263905395469485e-06,
+      "loss": 0.1241,
+      "step": 125250
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.263753995102934e-06,
+      "loss": 0.113,
+      "step": 125280
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.2636026491310733e-06,
+      "loss": 0.1267,
+      "step": 125310
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.263451357521339e-06,
+      "loss": 0.1088,
+      "step": 125340
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2633001202411954e-06,
+      "loss": 0.1209,
+      "step": 125370
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2631489372581335e-06,
+      "loss": 0.1221,
+      "step": 125400
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2630028452896997e-06,
+      "loss": 0.1268,
+      "step": 125430
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2628517689961675e-06,
+      "loss": 0.1132,
+      "step": 125460
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2627007469034332e-06,
+      "loss": 0.1092,
+      "step": 125490
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2625497789790967e-06,
+      "loss": 0.0943,
+      "step": 125520
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2623988651907834e-06,
+      "loss": 0.1242,
+      "step": 125550
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2622480055061464e-06,
+      "loss": 0.1133,
+      "step": 125580
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.262097199892866e-06,
+      "loss": 0.1141,
+      "step": 125610
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2619464483186494e-06,
+      "loss": 0.1023,
+      "step": 125640
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2617957507512305e-06,
+      "loss": 0.0932,
+      "step": 125670
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2616451071583708e-06,
+      "loss": 0.1173,
+      "step": 125700
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2614945175078577e-06,
+      "loss": 0.1018,
+      "step": 125730
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2613439817675066e-06,
+      "loss": 0.1226,
+      "step": 125760
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2611934999051594e-06,
+      "loss": 0.0986,
+      "step": 125790
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2610430718886849e-06,
+      "loss": 0.1193,
+      "step": 125820
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.260892697685978e-06,
+      "loss": 0.0996,
+      "step": 125850
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2607423772649616e-06,
+      "loss": 0.0975,
+      "step": 125880
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2605921105935848e-06,
+      "loss": 0.1183,
+      "step": 125910
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2604418976398225e-06,
+      "loss": 0.1177,
+      "step": 125940
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2602917383716784e-06,
+      "loss": 0.1219,
+      "step": 125970
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2601416327571808e-06,
+      "loss": 0.1232,
+      "step": 126000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2599915807643858e-06,
+      "loss": 0.108,
+      "step": 126030
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2598415823613763e-06,
+      "loss": 0.1135,
+      "step": 126060
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2596916375162605e-06,
+      "loss": 0.1191,
+      "step": 126090
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2595417461971747e-06,
+      "loss": 0.1177,
+      "step": 126120
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2593919083722806e-06,
+      "loss": 0.1073,
+      "step": 126150
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.2592421240097672e-06,
+      "loss": 0.1122,
+      "step": 126180
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.2590923930778489e-06,
+      "loss": 0.1189,
+      "step": 126210
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.2589427155447684e-06,
+      "loss": 0.1024,
+      "step": 126240
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.2587930913787927e-06,
+      "loss": 0.115,
+      "step": 126270
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.2586435205482167e-06,
+      "loss": 0.0978,
+      "step": 126300
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.258494003021361e-06,
+      "loss": 0.109,
+      "step": 126330
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.2583445387665727e-06,
+      "loss": 0.1314,
+      "step": 126360
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.2581951277522253e-06,
+      "loss": 0.1095,
+      "step": 126390
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.2580457699467181e-06,
+      "loss": 0.1293,
+      "step": 126420
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.2578964653184774e-06,
+      "loss": 0.1144,
+      "step": 126450
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.2577472138359553e-06,
+      "loss": 0.1177,
+      "step": 126480
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.25759801546763e-06,
+      "loss": 0.0867,
+      "step": 126510
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.2574488701820063e-06,
+      "loss": 0.0993,
+      "step": 126540
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.2572997779476142e-06,
+      "loss": 0.1046,
+      "step": 126570
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.2571507387330115e-06,
+      "loss": 0.1035,
+      "step": 126600
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.2570017525067802e-06,
+      "loss": 0.1182,
+      "step": 126630
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.2568528192375293e-06,
+      "loss": 0.0898,
+      "step": 126660
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.256703938893894e-06,
+      "loss": 0.1042,
+      "step": 126690
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.2565551114445355e-06,
+      "loss": 0.1312,
+      "step": 126720
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.25640633685814e-06,
+      "loss": 0.1011,
+      "step": 126750
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.2562576151034208e-06,
+      "loss": 0.0966,
+      "step": 126780
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.2561089461491165e-06,
+      "loss": 0.1087,
+      "step": 126810
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.2559603299639922e-06,
+      "loss": 0.1019,
+      "step": 126840
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.255811766516838e-06,
+      "loss": 0.1032,
+      "step": 126870
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.25566325577647e-06,
+      "loss": 0.1446,
+      "step": 126900
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.255514797711731e-06,
+      "loss": 0.085,
+      "step": 126930
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.2553663922914887e-06,
+      "loss": 0.1077,
+      "step": 126960
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.2552180394846366e-06,
+      "loss": 0.1019,
+      "step": 126990
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.2550697392600945e-06,
+      "loss": 0.1007,
+      "step": 127020
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.2549214915868075e-06,
+      "loss": 0.1203,
+      "step": 127050
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.254773296433746e-06,
+      "loss": 0.1135,
+      "step": 127080
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.2546251537699069e-06,
+      "loss": 0.0953,
+      "step": 127110
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.2544770635643121e-06,
+      "loss": 0.1147,
+      "step": 127140
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.254329025786009e-06,
+      "loss": 0.1283,
+      "step": 127170
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.254181040404071e-06,
+      "loss": 0.1249,
+      "step": 127200
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2540331073875973e-06,
+      "loss": 0.1066,
+      "step": 127230
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2538852267057116e-06,
+      "loss": 0.0965,
+      "step": 127260
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2537373983275636e-06,
+      "loss": 0.1096,
+      "step": 127290
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2535896222223288e-06,
+      "loss": 0.1008,
+      "step": 127320
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2534418983592075e-06,
+      "loss": 0.0941,
+      "step": 127350
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2532942267074264e-06,
+      "loss": 0.0951,
+      "step": 127380
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2531466072362361e-06,
+      "loss": 0.1065,
+      "step": 127410
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2529990399149137e-06,
+      "loss": 0.1239,
+      "step": 127440
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2528515247127609e-06,
+      "loss": 0.1015,
+      "step": 127470
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2527040615991059e-06,
+      "loss": 0.0923,
+      "step": 127500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2525566505433004e-06,
+      "loss": 0.1019,
+      "step": 127530
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2524092915147229e-06,
+      "loss": 0.1318,
+      "step": 127560
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.252261984482776e-06,
+      "loss": 0.0982,
+      "step": 127590
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.252114729416888e-06,
+      "loss": 0.1154,
+      "step": 127620
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2519675262865127e-06,
+      "loss": 0.1057,
+      "step": 127650
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2518203750611286e-06,
+      "loss": 0.1046,
+      "step": 127680
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2516732757102388e-06,
+      "loss": 0.1086,
+      "step": 127710
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2515262282033725e-06,
+      "loss": 0.1039,
+      "step": 127740
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.2513792325100835e-06,
+      "loss": 0.1005,
+      "step": 127770
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.2512322885999503e-06,
+      "loss": 0.1234,
+      "step": 127800
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.251085396442577e-06,
+      "loss": 0.1053,
+      "step": 127830
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.250938556007592e-06,
+      "loss": 0.1155,
+      "step": 127860
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.2507917672646498e-06,
+      "loss": 0.1063,
+      "step": 127890
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.2506450301834284e-06,
+      "loss": 0.096,
+      "step": 127920
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.2504983447336312e-06,
+      "loss": 0.1392,
+      "step": 127950
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.2503517108849869e-06,
+      "loss": 0.0902,
+      "step": 127980
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.2502051286072488e-06,
+      "loss": 0.1048,
+      "step": 128010
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.2500585978701952e-06,
+      "loss": 0.1352,
+      "step": 128040
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.2499121186436284e-06,
+      "loss": 0.103,
+      "step": 128070
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.249765690897376e-06,
+      "loss": 0.1082,
+      "step": 128100
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.2496193146012914e-06,
+      "loss": 0.1334,
+      "step": 128130
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.2494729897252504e-06,
+      "loss": 0.1062,
+      "step": 128160
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.2493267162391553e-06,
+      "loss": 0.1236,
+      "step": 128190
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.2491804941129326e-06,
+      "loss": 0.0906,
+      "step": 128220
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.2490343233165332e-06,
+      "loss": 0.1046,
+      "step": 128250
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.2488882038199327e-06,
+      "loss": 0.1001,
+      "step": 128280
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2487421355931315e-06,
+      "loss": 0.1148,
+      "step": 128310
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2485961186061542e-06,
+      "loss": 0.0994,
+      "step": 128340
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2484501528290504e-06,
+      "loss": 0.1428,
+      "step": 128370
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2483042382318938e-06,
+      "loss": 0.1143,
+      "step": 128400
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2481583747847825e-06,
+      "loss": 0.1171,
+      "step": 128430
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2480125624578395e-06,
+      "loss": 0.1124,
+      "step": 128460
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2478668012212119e-06,
+      "loss": 0.0794,
+      "step": 128490
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.247721091045071e-06,
+      "loss": 0.1156,
+      "step": 128520
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.2475754318996134e-06,
+      "loss": 0.095,
+      "step": 128550
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2474298237550587e-06,
+      "loss": 0.111,
+      "step": 128580
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2472842665816518e-06,
+      "loss": 0.0807,
+      "step": 128610
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.247138760349662e-06,
+      "loss": 0.0867,
+      "step": 128640
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2469933050293821e-06,
+      "loss": 0.1201,
+      "step": 128670
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2468479005911294e-06,
+      "loss": 0.0993,
+      "step": 128700
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.246702547005246e-06,
+      "loss": 0.0974,
+      "step": 128730
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2465572442420977e-06,
+      "loss": 0.122,
+      "step": 128760
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2464119922720743e-06,
+      "loss": 0.1179,
+      "step": 128790
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.2462667910655897e-06,
+      "loss": 0.1133,
+      "step": 128820
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.2461216405930831e-06,
+      "loss": 0.1158,
+      "step": 128850
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.245976540825016e-06,
+      "loss": 0.1173,
+      "step": 128880
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.2458314917318756e-06,
+      "loss": 0.1182,
+      "step": 128910
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.2456864932841722e-06,
+      "loss": 0.0994,
+      "step": 128940
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.24554154545244e-06,
+      "loss": 0.1251,
+      "step": 128970
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.2453966482072375e-06,
+      "loss": 0.1237,
+      "step": 129000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.2452518015191477e-06,
+      "loss": 0.1108,
+      "step": 129030
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.2451070053587767e-06,
+      "loss": 0.1188,
+      "step": 129060
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.2449622596967553e-06,
+      "loss": 0.0938,
+      "step": 129090
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.244817564503737e-06,
+      "loss": 0.1308,
+      "step": 129120
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.2446729197504007e-06,
+      "loss": 0.0957,
+      "step": 129150
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.2445283254074479e-06,
+      "loss": 0.1108,
+      "step": 129180
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.2443837814456042e-06,
+      "loss": 0.1098,
+      "step": 129210
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.24424410347837e-06,
+      "loss": 0.0849,
+      "step": 129240
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.2440996585140674e-06,
+      "loss": 0.1112,
+      "step": 129270
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.2439552638441673e-06,
+      "loss": 0.1154,
+      "step": 129300
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.24381091943949e-06,
+      "loss": 0.1224,
+      "step": 129330
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.2436666252708784e-06,
+      "loss": 0.1096,
+      "step": 129360
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.2435223813092007e-06,
+      "loss": 0.0951,
+      "step": 129390
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.2433781875253476e-06,
+      "loss": 0.1117,
+      "step": 129420
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.2432340438902337e-06,
+      "loss": 0.1054,
+      "step": 129450
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.2430899503747977e-06,
+      "loss": 0.114,
+      "step": 129480
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.242945906950001e-06,
+      "loss": 0.147,
+      "step": 129510
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.2428019135868293e-06,
+      "loss": 0.1143,
+      "step": 129540
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.2426579702562912e-06,
+      "loss": 0.1192,
+      "step": 129570
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.2425140769294192e-06,
+      "loss": 0.098,
+      "step": 129600
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.2423702335772691e-06,
+      "loss": 0.0906,
+      "step": 129630
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2422264401709206e-06,
+      "loss": 0.0879,
+      "step": 129660
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2420826966814762e-06,
+      "loss": 0.1079,
+      "step": 129690
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2419390030800618e-06,
+      "loss": 0.0955,
+      "step": 129720
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2417953593378274e-06,
+      "loss": 0.0878,
+      "step": 129750
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2416517654259457e-06,
+      "loss": 0.1031,
+      "step": 129780
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.241508221315613e-06,
+      "loss": 0.0887,
+      "step": 129810
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2413647269780489e-06,
+      "loss": 0.1279,
+      "step": 129840
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2412212823844956e-06,
+      "loss": 0.1016,
+      "step": 129870
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2410778875062197e-06,
+      "loss": 0.1015,
+      "step": 129900
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2409345423145103e-06,
+      "loss": 0.1133,
+      "step": 129930
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2407912467806803e-06,
+      "loss": 0.1121,
+      "step": 129960
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2406480008760647e-06,
+      "loss": 0.1025,
+      "step": 129990
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.240504804572023e-06,
+      "loss": 0.0929,
+      "step": 130020
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2403616578399369e-06,
+      "loss": 0.1031,
+      "step": 130050
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2402185606512112e-06,
+      "loss": 0.0915,
+      "step": 130080
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2400755129772745e-06,
+      "loss": 0.1082,
+      "step": 130110
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2399325147895776e-06,
+      "loss": 0.1257,
+      "step": 130140
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2397895660595954e-06,
+      "loss": 0.1182,
+      "step": 130170
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.2396466667588247e-06,
+      "loss": 0.1072,
+      "step": 130200
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.2395038168587853e-06,
+      "loss": 0.1104,
+      "step": 130230
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.239361016331022e-06,
+      "loss": 0.0867,
+      "step": 130260
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.2392182651470993e-06,
+      "loss": 0.118,
+      "step": 130290
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.2390755632786075e-06,
+      "loss": 0.1032,
+      "step": 130320
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.238932910697158e-06,
+      "loss": 0.1065,
+      "step": 130350
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.2387903073743856e-06,
+      "loss": 0.1119,
+      "step": 130380
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.2386477532819482e-06,
+      "loss": 0.1183,
+      "step": 130410
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.2385052483915269e-06,
+      "loss": 0.1116,
+      "step": 130440
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.2383627926748244e-06,
+      "loss": 0.1086,
+      "step": 130470
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.2382203861035668e-06,
+      "loss": 0.1153,
+      "step": 130500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.2380780286495036e-06,
+      "loss": 0.1252,
+      "step": 130530
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.2379357202844055e-06,
+      "loss": 0.1114,
+      "step": 130560
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.2377934609800675e-06,
+      "loss": 0.1164,
+      "step": 130590
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.2376512507083065e-06,
+      "loss": 0.1095,
+      "step": 130620
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.2375090894409624e-06,
+      "loss": 0.0931,
+      "step": 130650
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.237366977149897e-06,
+      "loss": 0.093,
+      "step": 130680
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2372249138069957e-06,
+      "loss": 0.122,
+      "step": 130710
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2370828993841653e-06,
+      "loss": 0.1082,
+      "step": 130740
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2369409338533365e-06,
+      "loss": 0.1006,
+      "step": 130770
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2367990171864617e-06,
+      "loss": 0.0942,
+      "step": 130800
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2366571493555165e-06,
+      "loss": 0.1071,
+      "step": 130830
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2365153303324978e-06,
+      "loss": 0.1091,
+      "step": 130860
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.236373560089426e-06,
+      "loss": 0.0992,
+      "step": 130890
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2362318385983439e-06,
+      "loss": 0.1205,
+      "step": 130920
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.2360901658313162e-06,
+      "loss": 0.1319,
+      "step": 130950
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2359485417604307e-06,
+      "loss": 0.1027,
+      "step": 130980
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2358069663577966e-06,
+      "loss": 0.1099,
+      "step": 131010
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2356654395955465e-06,
+      "loss": 0.1503,
+      "step": 131040
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2355239614458344e-06,
+      "loss": 0.1108,
+      "step": 131070
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2353825318808377e-06,
+      "loss": 0.1128,
+      "step": 131100
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.235241150872755e-06,
+      "loss": 0.1135,
+      "step": 131130
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2350998183938077e-06,
+      "loss": 0.1158,
+      "step": 131160
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2349585344162398e-06,
+      "loss": 0.0993,
+      "step": 131190
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2348172989123166e-06,
+      "loss": 0.1053,
+      "step": 131220
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.2346761118543262e-06,
+      "loss": 0.1223,
+      "step": 131250
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.2345349732145788e-06,
+      "loss": 0.1265,
+      "step": 131280
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.2343938829654069e-06,
+      "loss": 0.0982,
+      "step": 131310
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.2342528410791647e-06,
+      "loss": 0.0982,
+      "step": 131340
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.234111847528229e-06,
+      "loss": 0.1148,
+      "step": 131370
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.2339709022849983e-06,
+      "loss": 0.0997,
+      "step": 131400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.2338300053218935e-06,
+      "loss": 0.1387,
+      "step": 131430
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.2336891566113572e-06,
+      "loss": 0.1052,
+      "step": 131460
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.233548356125854e-06,
+      "loss": 0.1157,
+      "step": 131490
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.2334076038378707e-06,
+      "loss": 0.1167,
+      "step": 131520
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.2332668997199166e-06,
+      "loss": 0.0956,
+      "step": 131550
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.2331262437445215e-06,
+      "loss": 0.1308,
+      "step": 131580
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.2329856358842388e-06,
+      "loss": 0.1052,
+      "step": 131610
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.2328450761116422e-06,
+      "loss": 0.1155,
+      "step": 131640
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.2327045643993285e-06,
+      "loss": 0.1028,
+      "step": 131670
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.2325641007199162e-06,
+      "loss": 0.0988,
+      "step": 131700
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.232423685046045e-06,
+      "loss": 0.1303,
+      "step": 131730
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.2322833173503769e-06,
+      "loss": 0.1042,
+      "step": 131760
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2321429976055954e-06,
+      "loss": 0.1129,
+      "step": 131790
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2320027257844061e-06,
+      "loss": 0.1206,
+      "step": 131820
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2318625018595365e-06,
+      "loss": 0.1058,
+      "step": 131850
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.231722325803735e-06,
+      "loss": 0.1102,
+      "step": 131880
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2315821975897726e-06,
+      "loss": 0.1181,
+      "step": 131910
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2314421171904414e-06,
+      "loss": 0.1001,
+      "step": 131940
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2313020845785554e-06,
+      "loss": 0.101,
+      "step": 131970
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2311620997269502e-06,
+      "loss": 0.1221,
+      "step": 132000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.231022162608483e-06,
+      "loss": 0.0918,
+      "step": 132030
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2308822731960328e-06,
+      "loss": 0.1022,
+      "step": 132060
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2307424314624995e-06,
+      "loss": 0.1079,
+      "step": 132090
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2306026373808053e-06,
+      "loss": 0.1142,
+      "step": 132120
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.230462890923894e-06,
+      "loss": 0.1201,
+      "step": 132150
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2303231920647298e-06,
+      "loss": 0.1014,
+      "step": 132180
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2301835407762994e-06,
+      "loss": 0.1395,
+      "step": 132210
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2300439370316107e-06,
+      "loss": 0.1087,
+      "step": 132240
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2299043808036933e-06,
+      "loss": 0.1153,
+      "step": 132270
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.2297648720655975e-06,
+      "loss": 0.0962,
+      "step": 132300
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.229625410790396e-06,
+      "loss": 0.0934,
+      "step": 132330
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2294859969511816e-06,
+      "loss": 0.1272,
+      "step": 132360
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2293466305210697e-06,
+      "loss": 0.0931,
+      "step": 132390
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.229207311473196e-06,
+      "loss": 0.1055,
+      "step": 132420
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2290680397807183e-06,
+      "loss": 0.1143,
+      "step": 132450
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2289288154168154e-06,
+      "loss": 0.104,
+      "step": 132480
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2287896383546869e-06,
+      "loss": 0.1297,
+      "step": 132510
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2286505085675547e-06,
+      "loss": 0.1069,
+      "step": 132540
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.2285114260286608e-06,
+      "loss": 0.1035,
+      "step": 132570
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2283723907112692e-06,
+      "loss": 0.0854,
+      "step": 132600
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2282334025886644e-06,
+      "loss": 0.117,
+      "step": 132630
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2280944616341529e-06,
+      "loss": 0.121,
+      "step": 132660
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2279555678210612e-06,
+      "loss": 0.1162,
+      "step": 132690
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2278167211227382e-06,
+      "loss": 0.1071,
+      "step": 132720
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2276779215125527e-06,
+      "loss": 0.0994,
+      "step": 132750
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2275391689638956e-06,
+      "loss": 0.1016,
+      "step": 132780
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2274004634501783e-06,
+      "loss": 0.1168,
+      "step": 132810
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2272618049448327e-06,
+      "loss": 0.1099,
+      "step": 132840
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.2271231934213132e-06,
+      "loss": 0.1209,
+      "step": 132870
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.2269846288530937e-06,
+      "loss": 0.1109,
+      "step": 132900
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.2268461112136695e-06,
+      "loss": 0.1087,
+      "step": 132930
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.2267076404765577e-06,
+      "loss": 0.1185,
+      "step": 132960
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.2265692166152949e-06,
+      "loss": 0.1128,
+      "step": 132990
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.2264308396034396e-06,
+      "loss": 0.1114,
+      "step": 133020
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.226292509414571e-06,
+      "loss": 0.1145,
+      "step": 133050
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.2261542260222886e-06,
+      "loss": 0.1106,
+      "step": 133080
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.2260159894002136e-06,
+      "loss": 0.1064,
+      "step": 133110
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2258777995219873e-06,
+      "loss": 0.1026,
+      "step": 133140
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2257396563612719e-06,
+      "loss": 0.1053,
+      "step": 133170
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2256015598917514e-06,
+      "loss": 0.1187,
+      "step": 133200
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.225463510087129e-06,
+      "loss": 0.1068,
+      "step": 133230
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2253255069211288e-06,
+      "loss": 0.0971,
+      "step": 133260
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2251875503674974e-06,
+      "loss": 0.0877,
+      "step": 133290
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2250496403999999e-06,
+      "loss": 0.1098,
+      "step": 133320
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2249117769924233e-06,
+      "loss": 0.1212,
+      "step": 133350
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2247739601185745e-06,
+      "loss": 0.1139,
+      "step": 133380
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2246361897522823e-06,
+      "loss": 0.1067,
+      "step": 133410
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2244984658673947e-06,
+      "loss": 0.1118,
+      "step": 133440
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2243607884377809e-06,
+      "loss": 0.0962,
+      "step": 133470
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2242231574373308e-06,
+      "loss": 0.1155,
+      "step": 133500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.224085572839954e-06,
+      "loss": 0.0926,
+      "step": 133530
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2239480346195822e-06,
+      "loss": 0.0914,
+      "step": 133560
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2238105427501658e-06,
+      "loss": 0.1052,
+      "step": 133590
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.2236730972056771e-06,
+      "loss": 0.0977,
+      "step": 133620
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.223535697960108e-06,
+      "loss": 0.0894,
+      "step": 133650
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2233983449874714e-06,
+      "loss": 0.1005,
+      "step": 133680
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2232610382618002e-06,
+      "loss": 0.1008,
+      "step": 133710
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2231237777571477e-06,
+      "loss": 0.115,
+      "step": 133740
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.222986563447588e-06,
+      "loss": 0.1066,
+      "step": 133770
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2228493953072147e-06,
+      "loss": 0.1144,
+      "step": 133800
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2227122733101433e-06,
+      "loss": 0.0956,
+      "step": 133830
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.222575197430508e-06,
+      "loss": 0.0942,
+      "step": 133860
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2224427345597513e-06,
+      "loss": 0.1127,
+      "step": 133890
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.222305749302364e-06,
+      "loss": 0.1057,
+      "step": 133920
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.222168810085799e-06,
+      "loss": 0.1057,
+      "step": 133950
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.2220319168842711e-06,
+      "loss": 0.1067,
+      "step": 133980
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.2218950696720162e-06,
+      "loss": 0.1108,
+      "step": 134010
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.2217582684232896e-06,
+      "loss": 0.1227,
+      "step": 134040
+    },
+    {
+      "epoch": 5.0,
+      "step": 134060,
+      "total_flos": 1.587534895646056e+21,
+      "train_loss": 0.1688916235028426,
+      "train_runtime": 101673.2407,
+      "train_samples_per_second": 10.548,
+      "train_steps_per_second": 1.319
+    }
+  ],
+  "logging_steps": 30,
+  "max_steps": 134060,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 1.587534895646056e+21,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}