diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9517464547444561,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 5.8577,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 5.2479,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 4.2581,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 4.1526,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 4.0411,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-05,
+      "loss": 3.4283,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-05,
+      "loss": 3.833,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-05,
+      "loss": 3.7164,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-05,
+      "loss": 3.715,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001,
+      "loss": 3.559,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.996817415104547e-05,
+      "loss": 2.9796,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.993634830209096e-05,
+      "loss": 3.2296,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.990452245313644e-05,
+      "loss": 2.6348,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.987269660418193e-05,
+      "loss": 3.2406,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98408707552274e-05,
+      "loss": 2.8193,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.980904490627288e-05,
+      "loss": 3.0848,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.977721905731836e-05,
+      "loss": 2.8889,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.974539320836383e-05,
+      "loss": 3.2759,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.971356735940932e-05,
+      "loss": 2.7277,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.96817415104548e-05,
+      "loss": 2.5394,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.964991566150027e-05,
+      "loss": 2.9889,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.961808981254575e-05,
+      "loss": 2.8291,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.958626396359122e-05,
+      "loss": 2.9713,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.955443811463672e-05,
+      "loss": 2.8579,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.952261226568219e-05,
+      "loss": 2.6235,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.949078641672766e-05,
+      "loss": 2.7682,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.945896056777315e-05,
+      "loss": 2.5362,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.942713471881863e-05,
+      "loss": 2.5971,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.939530886986411e-05,
+      "loss": 2.89,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.936348302090958e-05,
+      "loss": 2.4912,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.933165717195507e-05,
+      "loss": 2.8739,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.929983132300055e-05,
+      "loss": 2.6042,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.926800547404602e-05,
+      "loss": 2.3941,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.92361796250915e-05,
+      "loss": 2.5959,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.920435377613699e-05,
+      "loss": 2.6849,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.917252792718246e-05,
+      "loss": 2.8489,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.914070207822794e-05,
+      "loss": 2.6433,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.910887622927341e-05,
+      "loss": 2.5534,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.907705038031891e-05,
+      "loss": 2.3736,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.904522453136438e-05,
+      "loss": 2.8226,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.901339868240985e-05,
+      "loss": 2.468,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.898157283345533e-05,
+      "loss": 2.298,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.894974698450082e-05,
+      "loss": 2.5418,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.89179211355463e-05,
+      "loss": 2.3139,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.888609528659177e-05,
+      "loss": 2.5035,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.885426943763725e-05,
+      "loss": 2.8296,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.882244358868274e-05,
+      "loss": 2.1434,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.879061773972821e-05,
+      "loss": 2.2989,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.875879189077369e-05,
+      "loss": 2.7385,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.872696604181917e-05,
+      "loss": 2.5565,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.869514019286466e-05,
+      "loss": 2.4335,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.866331434391013e-05,
+      "loss": 2.9056,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.86314884949556e-05,
+      "loss": 2.392,
+      "step": 530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85996626460011e-05,
+      "loss": 2.8014,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.856783679704657e-05,
+      "loss": 2.1574,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.853601094809205e-05,
+      "loss": 2.3031,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.850418509913752e-05,
+      "loss": 2.3997,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.8472359250183e-05,
+      "loss": 2.4631,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.844053340122849e-05,
+      "loss": 2.228,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.840870755227396e-05,
+      "loss": 2.375,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.837688170331944e-05,
+      "loss": 2.4447,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.834505585436492e-05,
+      "loss": 2.587,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.83132300054104e-05,
+      "loss": 2.1196,
+      "step": 630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.828140415645588e-05,
+      "loss": 2.5909,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.824957830750136e-05,
+      "loss": 2.4393,
+      "step": 650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.821775245854684e-05,
+      "loss": 2.8212,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.818592660959231e-05,
+      "loss": 2.4206,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.815410076063778e-05,
+      "loss": 2.2936,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.812227491168328e-05,
+      "loss": 2.3386,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.809044906272875e-05,
+      "loss": 1.9638,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.805862321377424e-05,
+      "loss": 2.5946,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.80267973648197e-05,
+      "loss": 2.4551,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.799497151586519e-05,
+      "loss": 2.6737,
+      "step": 730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.796314566691067e-05,
+      "loss": 2.1145,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.793131981795614e-05,
+      "loss": 2.1091,
+      "step": 750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.789949396900163e-05,
+      "loss": 2.7712,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.786766812004711e-05,
+      "loss": 2.1544,
+      "step": 770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.783584227109258e-05,
+      "loss": 2.3773,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.780401642213806e-05,
+      "loss": 2.1907,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.777219057318355e-05,
+      "loss": 2.6661,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.774036472422903e-05,
+      "loss": 2.238,
+      "step": 810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.77085388752745e-05,
+      "loss": 1.7985,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.767671302631997e-05,
+      "loss": 2.1975,
+      "step": 830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.764488717736547e-05,
+      "loss": 2.1023,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.761306132841094e-05,
+      "loss": 2.4209,
+      "step": 850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.758123547945642e-05,
+      "loss": 2.4297,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.754940963050189e-05,
+      "loss": 1.9522,
+      "step": 870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.751758378154738e-05,
+      "loss": 2.3907,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.748575793259286e-05,
+      "loss": 2.206,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.745393208363833e-05,
+      "loss": 1.9964,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.742210623468381e-05,
+      "loss": 2.6844,
+      "step": 910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.73902803857293e-05,
+      "loss": 2.2864,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.735845453677477e-05,
+      "loss": 2.4212,
+      "step": 930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.732662868782025e-05,
+      "loss": 2.463,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.729480283886573e-05,
+      "loss": 2.2846,
+      "step": 950
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.726297698991122e-05,
+      "loss": 2.3168,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.723115114095669e-05,
+      "loss": 2.205,
+      "step": 970
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.719932529200216e-05,
+      "loss": 2.6097,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.716749944304766e-05,
+      "loss": 2.5665,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.713567359409313e-05,
+      "loss": 2.0707,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.710384774513861e-05,
+      "loss": 2.3376,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.707202189618408e-05,
+      "loss": 2.1752,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.704019604722956e-05,
+      "loss": 1.9965,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.700837019827505e-05,
+      "loss": 2.3768,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.697654434932052e-05,
+      "loss": 1.8873,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.6944718500366e-05,
+      "loss": 2.1792,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.691289265141148e-05,
+      "loss": 2.0115,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.688106680245697e-05,
+      "loss": 2.3,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.684924095350244e-05,
+      "loss": 2.3574,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.681741510454792e-05,
+      "loss": 2.2998,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.67855892555934e-05,
+      "loss": 2.2051,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.675376340663888e-05,
+      "loss": 2.5525,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.672193755768436e-05,
+      "loss": 2.5324,
+      "step": 1130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.669011170872984e-05,
+      "loss": 2.4017,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.665828585977531e-05,
+      "loss": 2.2123,
+      "step": 1150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.66264600108208e-05,
+      "loss": 2.0349,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.659463416186627e-05,
+      "loss": 1.7593,
+      "step": 1170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.656280831291175e-05,
+      "loss": 2.4243,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.653098246395723e-05,
+      "loss": 2.3423,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.64991566150027e-05,
+      "loss": 1.8123,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.646733076604819e-05,
+      "loss": 2.0992,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.643550491709367e-05,
+      "loss": 2.5563,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.640367906813915e-05,
+      "loss": 2.3245,
+      "step": 1230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.637185321918462e-05,
+      "loss": 1.7916,
+      "step": 1240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.63400273702301e-05,
+      "loss": 1.4747,
+      "step": 1250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.630820152127559e-05,
+      "loss": 2.0067,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.627637567232106e-05,
+      "loss": 1.9419,
+      "step": 1270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.624454982336655e-05,
+      "loss": 2.2326,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.621272397441202e-05,
+      "loss": 2.4721,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.61808981254575e-05,
+      "loss": 1.9018,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.614907227650298e-05,
+      "loss": 2.0016,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.611724642754845e-05,
+      "loss": 1.8423,
+      "step": 1320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608542057859394e-05,
+      "loss": 2.4551,
+      "step": 1330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605359472963942e-05,
+      "loss": 2.1678,
+      "step": 1340
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.602176888068489e-05,
+      "loss": 2.2086,
+      "step": 1350
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.598994303173037e-05,
+      "loss": 2.5018,
+      "step": 1360
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.595811718277586e-05,
+      "loss": 1.8025,
+      "step": 1370
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.592629133382134e-05,
+      "loss": 2.0562,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.589446548486681e-05,
+      "loss": 2.4341,
+      "step": 1390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.586263963591228e-05,
+      "loss": 2.0408,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.583081378695778e-05,
+      "loss": 1.923,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.579898793800325e-05,
+      "loss": 2.232,
+      "step": 1420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.576716208904873e-05,
+      "loss": 1.8952,
+      "step": 1430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.57353362400942e-05,
+      "loss": 2.0362,
+      "step": 1440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.570351039113969e-05,
+      "loss": 2.3579,
+      "step": 1450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.567168454218517e-05,
+      "loss": 2.3189,
+      "step": 1460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563985869323064e-05,
+      "loss": 2.0279,
+      "step": 1470
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.560803284427612e-05,
+      "loss": 2.206,
+      "step": 1480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557620699532161e-05,
+      "loss": 2.0405,
+      "step": 1490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554438114636708e-05,
+      "loss": 2.0556,
+      "step": 1500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.551255529741256e-05,
+      "loss": 2.1394,
+      "step": 1510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.548072944845804e-05,
+      "loss": 2.0859,
+      "step": 1520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.544890359950353e-05,
+      "loss": 1.9614,
+      "step": 1530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.5417077750549e-05,
+      "loss": 2.0116,
+      "step": 1540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.538525190159447e-05,
+      "loss": 1.694,
+      "step": 1550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.535342605263997e-05,
+      "loss": 2.0689,
+      "step": 1560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.532160020368544e-05,
+      "loss": 2.4515,
+      "step": 1570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.528977435473092e-05,
+      "loss": 2.0836,
+      "step": 1580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.525794850577639e-05,
+      "loss": 2.1763,
+      "step": 1590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.522612265682187e-05,
+      "loss": 2.3392,
+      "step": 1600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.519429680786736e-05,
+      "loss": 2.2552,
+      "step": 1610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.516247095891283e-05,
+      "loss": 1.7769,
+      "step": 1620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.513064510995831e-05,
+      "loss": 2.3103,
+      "step": 1630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.50988192610038e-05,
+      "loss": 2.1365,
+      "step": 1640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.506699341204926e-05,
+      "loss": 2.4357,
+      "step": 1650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.503516756309475e-05,
+      "loss": 1.8809,
+      "step": 1660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.500334171414023e-05,
+      "loss": 2.2216,
+      "step": 1670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.497151586518572e-05,
+      "loss": 2.1124,
+      "step": 1680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.493969001623119e-05,
+      "loss": 1.9944,
+      "step": 1690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.490786416727667e-05,
+      "loss": 2.2696,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.487603831832215e-05,
+      "loss": 1.9576,
+      "step": 1710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.484421246936762e-05,
+      "loss": 1.8947,
+      "step": 1720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.48123866204131e-05,
+      "loss": 2.0351,
+      "step": 1730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478056077145858e-05,
+      "loss": 2.291,
+      "step": 1740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474873492250407e-05,
+      "loss": 1.915,
+      "step": 1750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471690907354954e-05,
+      "loss": 2.1551,
+      "step": 1760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468508322459501e-05,
+      "loss": 1.8664,
+      "step": 1770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46532573756405e-05,
+      "loss": 2.5013,
+      "step": 1780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462143152668598e-05,
+      "loss": 2.2391,
+      "step": 1790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.458960567773146e-05,
+      "loss": 1.9795,
+      "step": 1800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.455777982877693e-05,
+      "loss": 1.8689,
+      "step": 1810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.452595397982242e-05,
+      "loss": 2.4631,
+      "step": 1820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.44941281308679e-05,
+      "loss": 2.1215,
+      "step": 1830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446230228191337e-05,
+      "loss": 2.169,
+      "step": 1840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443047643295886e-05,
+      "loss": 1.9249,
+      "step": 1850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439865058400434e-05,
+      "loss": 2.1031,
+      "step": 1860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436682473504981e-05,
+      "loss": 2.0699,
+      "step": 1870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433499888609529e-05,
+      "loss": 2.0062,
+      "step": 1880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430317303714076e-05,
+      "loss": 1.988,
+      "step": 1890
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427134718818626e-05,
+      "loss": 1.7972,
+      "step": 1900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423952133923173e-05,
+      "loss": 1.7263,
+      "step": 1910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42076954902772e-05,
+      "loss": 1.7781,
+      "step": 1920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417586964132268e-05,
+      "loss": 1.784,
+      "step": 1930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414404379236817e-05,
+      "loss": 2.1332,
+      "step": 1940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.411221794341365e-05,
+      "loss": 2.1236,
+      "step": 1950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.408039209445912e-05,
+      "loss": 2.1588,
+      "step": 1960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.40485662455046e-05,
+      "loss": 2.117,
+      "step": 1970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.401674039655009e-05,
+      "loss": 1.7931,
+      "step": 1980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.398491454759556e-05,
+      "loss": 2.1229,
+      "step": 1990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.395308869864104e-05,
+      "loss": 1.772,
+      "step": 2000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.392126284968653e-05,
+      "loss": 1.8689,
+      "step": 2010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3889437000732e-05,
+      "loss": 2.1535,
+      "step": 2020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.385761115177748e-05,
+      "loss": 1.9323,
+      "step": 2030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382578530282295e-05,
+      "loss": 2.2067,
+      "step": 2040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.379395945386845e-05,
+      "loss": 2.1581,
+      "step": 2050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.376213360491392e-05,
+      "loss": 2.0287,
+      "step": 2060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.373030775595939e-05,
+      "loss": 2.1503,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.369848190700487e-05,
+      "loss": 1.8953,
+      "step": 2080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.366665605805035e-05,
+      "loss": 2.4307,
+      "step": 2090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.363483020909584e-05,
+      "loss": 2.0636,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.360300436014131e-05,
+      "loss": 2.0434,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.357117851118679e-05,
+      "loss": 1.9077,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.353935266223228e-05,
+      "loss": 2.2045,
+      "step": 2130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.350752681327775e-05,
+      "loss": 1.8585,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.347570096432323e-05,
+      "loss": 1.6934,
+      "step": 2150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.344387511536871e-05,
+      "loss": 1.8759,
+      "step": 2160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.341204926641418e-05,
+      "loss": 2.2663,
+      "step": 2170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.338022341745967e-05,
+      "loss": 1.7734,
+      "step": 2180
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.334839756850514e-05,
+      "loss": 1.941,
+      "step": 2190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.331657171955063e-05,
+      "loss": 2.2826,
+      "step": 2200
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.32847458705961e-05,
+      "loss": 2.4277,
+      "step": 2210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.325292002164157e-05,
+      "loss": 2.0279,
+      "step": 2220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.322109417268706e-05,
+      "loss": 1.738,
+      "step": 2230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.318926832373254e-05,
+      "loss": 2.3177,
+      "step": 2240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.315744247477803e-05,
+      "loss": 2.0615,
+      "step": 2250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.31256166258235e-05,
+      "loss": 1.788,
+      "step": 2260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.309379077686898e-05,
+      "loss": 2.0254,
+      "step": 2270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.306196492791446e-05,
+      "loss": 1.9799,
+      "step": 2280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.303013907895993e-05,
+      "loss": 1.9327,
+      "step": 2290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.299831323000542e-05,
+      "loss": 2.0215,
+      "step": 2300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.296648738105089e-05,
+      "loss": 1.7893,
+      "step": 2310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.293466153209637e-05,
+      "loss": 1.5906,
+      "step": 2320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.290283568314185e-05,
+      "loss": 1.9828,
+      "step": 2330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.287100983418732e-05,
+      "loss": 1.8949,
+      "step": 2340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.283918398523281e-05,
+      "loss": 2.0422,
+      "step": 2350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.280735813627829e-05,
+      "loss": 1.9592,
+      "step": 2360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.277553228732377e-05,
+      "loss": 2.0711,
+      "step": 2370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.274370643836924e-05,
+      "loss": 1.9374,
+      "step": 2380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.271188058941473e-05,
+      "loss": 2.0565,
+      "step": 2390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.268005474046021e-05,
+      "loss": 2.1121,
+      "step": 2400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.264822889150568e-05,
+      "loss": 2.263,
+      "step": 2410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.261640304255117e-05,
+      "loss": 1.8936,
+      "step": 2420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.258457719359665e-05,
+      "loss": 2.4082,
+      "step": 2430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.255275134464212e-05,
+      "loss": 2.1083,
+      "step": 2440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.25209254956876e-05,
+      "loss": 1.7903,
+      "step": 2450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.248909964673307e-05,
+      "loss": 2.2809,
+      "step": 2460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.245727379777857e-05,
+      "loss": 1.9134,
+      "step": 2470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.242544794882404e-05,
+      "loss": 1.8962,
+      "step": 2480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.239362209986951e-05,
+      "loss": 1.9613,
+      "step": 2490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.2361796250915e-05,
+      "loss": 2.0229,
+      "step": 2500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.232997040196048e-05,
+      "loss": 2.4332,
+      "step": 2510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.229814455300596e-05,
+      "loss": 2.3423,
+      "step": 2520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.226631870405143e-05,
+      "loss": 1.9032,
+      "step": 2530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.223449285509692e-05,
+      "loss": 1.7481,
+      "step": 2540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.22026670061424e-05,
+      "loss": 1.9051,
+      "step": 2550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.217084115718787e-05,
+      "loss": 1.7612,
+      "step": 2560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.213901530823335e-05,
+      "loss": 2.3907,
+      "step": 2570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.210718945927884e-05,
+      "loss": 1.8452,
+      "step": 2580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.20753636103243e-05,
+      "loss": 2.0806,
+      "step": 2590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.204353776136979e-05,
+      "loss": 1.6906,
+      "step": 2600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.201171191241526e-05,
+      "loss": 2.0094,
+      "step": 2610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.197988606346076e-05,
+      "loss": 2.2129,
+      "step": 2620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.194806021450623e-05,
+      "loss": 2.297,
+      "step": 2630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.19162343655517e-05,
+      "loss": 1.969,
+      "step": 2640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.188440851659718e-05,
+      "loss": 1.837,
+      "step": 2650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.185258266764266e-05,
+      "loss": 1.9689,
+      "step": 2660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.182075681868815e-05,
+      "loss": 1.8176,
+      "step": 2670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.178893096973362e-05,
+      "loss": 2.0166,
+      "step": 2680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.17571051207791e-05,
+      "loss": 1.7734,
+      "step": 2690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.172527927182459e-05,
+      "loss": 1.6355,
+      "step": 2700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.169345342287006e-05,
+      "loss": 1.4228,
+      "step": 2710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.166162757391554e-05,
+      "loss": 1.9741,
+      "step": 2720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.162980172496102e-05,
+      "loss": 1.849,
+      "step": 2730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.159797587600649e-05,
+      "loss": 1.8072,
+      "step": 2740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.156615002705198e-05,
+      "loss": 1.8057,
+      "step": 2750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.153432417809745e-05,
+      "loss": 1.766,
+      "step": 2760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.150249832914294e-05,
+      "loss": 1.8951,
+      "step": 2770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.147067248018841e-05,
+      "loss": 1.5905,
+      "step": 2780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.143884663123388e-05,
+      "loss": 1.8121,
+      "step": 2790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.140702078227937e-05,
+      "loss": 2.4688,
+      "step": 2800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.137519493332485e-05,
+      "loss": 1.8452,
+      "step": 2810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.134336908437034e-05,
+      "loss": 2.1774,
+      "step": 2820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.13115432354158e-05,
+      "loss": 2.1027,
+      "step": 2830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.127971738646129e-05,
+      "loss": 1.7269,
+      "step": 2840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.124789153750677e-05,
+      "loss": 1.5584,
+      "step": 2850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.121606568855224e-05,
+      "loss": 1.9226,
+      "step": 2860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.118423983959773e-05,
+      "loss": 1.9731,
+      "step": 2870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.115241399064321e-05,
+      "loss": 1.9055,
+      "step": 2880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.112058814168868e-05,
+      "loss": 2.1257,
+      "step": 2890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.108876229273416e-05,
+      "loss": 2.1161,
+      "step": 2900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.105693644377963e-05,
+      "loss": 2.0938,
+      "step": 2910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.102511059482513e-05,
+      "loss": 1.5036,
+      "step": 2920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.09932847458706e-05,
+      "loss": 2.4857,
+      "step": 2930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.096145889691607e-05,
+      "loss": 1.5093,
+      "step": 2940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.092963304796155e-05,
+      "loss": 2.1482,
+      "step": 2950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.089780719900704e-05,
+      "loss": 2.0191,
+      "step": 2960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.086598135005252e-05,
+      "loss": 1.9049,
+      "step": 2970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.083415550109799e-05,
+      "loss": 1.6495,
+      "step": 2980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.080232965214348e-05,
+      "loss": 2.2134,
+      "step": 2990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.077050380318896e-05,
+      "loss": 2.6072,
+      "step": 3000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.073867795423443e-05,
+      "loss": 1.7293,
+      "step": 3010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.070685210527991e-05,
+      "loss": 1.7487,
+      "step": 3020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.06750262563254e-05,
+      "loss": 2.1817,
+      "step": 3030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.064320040737088e-05,
+      "loss": 2.0928,
+      "step": 3040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.061137455841635e-05,
+      "loss": 1.8595,
+      "step": 3050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.057954870946182e-05,
+      "loss": 2.2285,
+      "step": 3060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.054772286050732e-05,
+      "loss": 2.1055,
+      "step": 3070
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.051589701155279e-05,
+      "loss": 1.9423,
+      "step": 3080
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.048407116259827e-05,
+      "loss": 1.9643,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.045224531364374e-05,
+      "loss": 2.5573,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.042041946468923e-05,
+      "loss": 2.2847,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.038859361573471e-05,
+      "loss": 1.9266,
+      "step": 3120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.035676776678018e-05,
+      "loss": 1.7437,
+      "step": 3130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.032494191782566e-05,
+      "loss": 1.5789,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.029311606887115e-05,
+      "loss": 2.0244,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.026129021991662e-05,
+      "loss": 2.1154,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.02294643709621e-05,
+      "loss": 2.1877,
+      "step": 3170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.019763852200758e-05,
+      "loss": 2.0215,
+      "step": 3180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.016581267305307e-05,
+      "loss": 1.7053,
+      "step": 3190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.013398682409854e-05,
+      "loss": 2.1041,
+      "step": 3200
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.010216097514401e-05,
+      "loss": 1.8648,
+      "step": 3210
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.00703351261895e-05,
+      "loss": 1.8743,
+      "step": 3220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.003850927723497e-05,
+      "loss": 1.8577,
+      "step": 3230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.000668342828046e-05,
+      "loss": 2.1161,
+      "step": 3240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.997485757932593e-05,
+      "loss": 1.8529,
+      "step": 3250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.994303173037141e-05,
+      "loss": 1.8198,
+      "step": 3260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.99112058814169e-05,
+      "loss": 1.6714,
+      "step": 3270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.987938003246237e-05,
+      "loss": 1.9358,
+      "step": 3280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.984755418350785e-05,
+      "loss": 2.1932,
+      "step": 3290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.981572833455333e-05,
+      "loss": 1.5896,
+      "step": 3300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.97839024855988e-05,
+      "loss": 1.9534,
+      "step": 3310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.975207663664429e-05,
+      "loss": 1.8981,
+      "step": 3320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.972025078768977e-05,
+      "loss": 1.697,
+      "step": 3330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.968842493873525e-05,
+      "loss": 1.92,
+      "step": 3340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.965659908978072e-05,
+      "loss": 1.7039,
+      "step": 3350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.96247732408262e-05,
+      "loss": 1.7979,
+      "step": 3360
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.959294739187168e-05,
+      "loss": 2.0006,
+      "step": 3370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.956112154291716e-05,
+      "loss": 1.8557,
+      "step": 3380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.952929569396265e-05,
+      "loss": 1.6347,
+      "step": 3390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.949746984500812e-05,
+      "loss": 1.667,
+      "step": 3400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.94656439960536e-05,
+      "loss": 1.9593,
+      "step": 3410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.943381814709908e-05,
+      "loss": 1.504,
+      "step": 3420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.940199229814455e-05,
+      "loss": 2.2819,
+      "step": 3430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.937016644919004e-05,
+      "loss": 1.7575,
+      "step": 3440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.933834060023552e-05,
+      "loss": 1.8409,
+      "step": 3450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.930651475128099e-05,
+      "loss": 1.9553,
+      "step": 3460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.927468890232647e-05,
+      "loss": 2.1456,
+      "step": 3470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.924286305337194e-05,
+      "loss": 1.6622,
+      "step": 3480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.921103720441744e-05,
+      "loss": 2.3896,
+      "step": 3490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.917921135546291e-05,
+      "loss": 1.6848,
+      "step": 3500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.914738550650838e-05,
+      "loss": 1.8145,
+      "step": 3510
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.911555965755386e-05,
+      "loss": 1.8479,
+      "step": 3520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.908373380859935e-05,
+      "loss": 2.0713,
+      "step": 3530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.905190795964483e-05,
+      "loss": 1.6836,
+      "step": 3540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.90200821106903e-05,
+      "loss": 2.1469,
+      "step": 3550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.898825626173579e-05,
+      "loss": 1.9421,
+      "step": 3560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.895643041278127e-05,
+      "loss": 1.424,
+      "step": 3570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.892460456382674e-05,
+      "loss": 2.1441,
+      "step": 3580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.889277871487222e-05,
+      "loss": 1.9829,
+      "step": 3590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.88609528659177e-05,
+      "loss": 1.6689,
+      "step": 3600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.882912701696319e-05,
+      "loss": 1.9551,
+      "step": 3610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.879730116800866e-05,
+      "loss": 1.5118,
+      "step": 3620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.876547531905413e-05,
+      "loss": 1.8151,
+      "step": 3630
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.873364947009963e-05,
+      "loss": 2.0585,
+      "step": 3640
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.87018236211451e-05,
+      "loss": 1.9604,
+      "step": 3650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.866999777219058e-05,
+      "loss": 1.6825,
+      "step": 3660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.863817192323605e-05,
+      "loss": 2.1185,
+      "step": 3670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.860634607428154e-05,
+      "loss": 1.7995,
+      "step": 3680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.857452022532702e-05,
+      "loss": 1.7444,
+      "step": 3690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.854269437637249e-05,
+      "loss": 1.7131,
+      "step": 3700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.851086852741797e-05,
+      "loss": 1.5694,
+      "step": 3710
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.847904267846346e-05,
+      "loss": 1.8997,
+      "step": 3720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.844721682950893e-05,
+      "loss": 1.7479,
+      "step": 3730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.841539098055441e-05,
+      "loss": 2.0054,
+      "step": 3740
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.83835651315999e-05,
+      "loss": 1.783,
+      "step": 3750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.835173928264538e-05,
+      "loss": 1.6751,
+      "step": 3760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.831991343369085e-05,
+      "loss": 1.9195,
+      "step": 3770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.828808758473632e-05,
+      "loss": 1.6263,
+      "step": 3780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.825626173578181e-05,
+      "loss": 1.8235,
+      "step": 3790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.822443588682728e-05,
+      "loss": 1.8654,
+      "step": 3800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.819261003787277e-05,
+      "loss": 2.0507,
+      "step": 3810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.816078418891824e-05,
+      "loss": 1.7236,
+      "step": 3820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.812895833996372e-05,
+      "loss": 2.1645,
+      "step": 3830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.80971324910092e-05,
+      "loss": 2.1542,
+      "step": 3840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.806530664205468e-05,
+      "loss": 1.6501,
+      "step": 3850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.803348079310016e-05,
+      "loss": 1.4847,
+      "step": 3860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.800165494414564e-05,
+      "loss": 1.9785,
+      "step": 3870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.796982909519111e-05,
+      "loss": 1.8721,
+      "step": 3880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.79380032462366e-05,
+      "loss": 1.9387,
+      "step": 3890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.790617739728208e-05,
+      "loss": 1.6376,
+      "step": 3900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.787435154832756e-05,
+      "loss": 1.7417,
+      "step": 3910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.784252569937303e-05,
+      "loss": 1.4899,
+      "step": 3920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.78106998504185e-05,
+      "loss": 1.8071,
+      "step": 3930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.7778874001464e-05,
+      "loss": 1.9723,
+      "step": 3940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.774704815250947e-05,
+      "loss": 1.7802,
+      "step": 3950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.771522230355495e-05,
+      "loss": 1.7954,
+      "step": 3960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.768339645460043e-05,
+      "loss": 1.825,
+      "step": 3970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.765157060564591e-05,
+      "loss": 2.056,
+      "step": 3980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.761974475669139e-05,
+      "loss": 2.1008,
+      "step": 3990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.758791890773686e-05,
+      "loss": 1.9251,
+      "step": 4000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.755609305878235e-05,
+      "loss": 1.7959,
+      "step": 4010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.752426720982783e-05,
+      "loss": 1.9192,
+      "step": 4020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.74924413608733e-05,
+      "loss": 2.0,
+      "step": 4030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.746061551191878e-05,
+      "loss": 1.8604,
+      "step": 4040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.742878966296427e-05,
+      "loss": 1.7686,
+      "step": 4050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.739696381400975e-05,
+      "loss": 1.7677,
+      "step": 4060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.736513796505522e-05,
+      "loss": 1.7588,
+      "step": 4070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.733331211610069e-05,
+      "loss": 1.8295,
+      "step": 4080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.730148626714619e-05,
+      "loss": 1.9025,
+      "step": 4090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.726966041819166e-05,
+      "loss": 1.7978,
+      "step": 4100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.723783456923714e-05,
+      "loss": 1.7281,
+      "step": 4110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.720600872028261e-05,
+      "loss": 1.5019,
+      "step": 4120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.71741828713281e-05,
+      "loss": 1.604,
+      "step": 4130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.714235702237358e-05,
+      "loss": 1.6965,
+      "step": 4140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.711053117341905e-05,
+      "loss": 1.7006,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.707870532446453e-05,
+      "loss": 1.8867,
+      "step": 4160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.704687947551002e-05,
+      "loss": 1.9758,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.701505362655549e-05,
+      "loss": 1.7924,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.698322777760097e-05,
+      "loss": 1.4875,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.695140192864645e-05,
+      "loss": 1.7991,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.691957607969194e-05,
+      "loss": 2.2652,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.688775023073741e-05,
+      "loss": 1.8837,
+      "step": 4220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.685592438178288e-05,
+      "loss": 1.7792,
+      "step": 4230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.682409853282837e-05,
+      "loss": 2.4074,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.679227268387384e-05,
+      "loss": 1.8309,
+      "step": 4250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.676044683491933e-05,
+      "loss": 1.6532,
+      "step": 4260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.67286209859648e-05,
+      "loss": 1.6922,
+      "step": 4270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.66967951370103e-05,
+      "loss": 1.5487,
+      "step": 4280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.666496928805577e-05,
+      "loss": 1.9437,
+      "step": 4290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.663314343910124e-05,
+      "loss": 1.8552,
+      "step": 4300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.660131759014672e-05,
+      "loss": 2.2202,
+      "step": 4310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.65694917411922e-05,
+      "loss": 1.7241,
+      "step": 4320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.653766589223769e-05,
+      "loss": 2.0968,
+      "step": 4330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.650584004328316e-05,
+      "loss": 2.2153,
+      "step": 4340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.647401419432864e-05,
+      "loss": 1.7501,
+      "step": 4350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.644218834537412e-05,
+      "loss": 2.2423,
+      "step": 4360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.64103624964196e-05,
+      "loss": 1.6641,
+      "step": 4370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.637853664746508e-05,
+      "loss": 2.1719,
+      "step": 4380
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.634671079851056e-05,
+      "loss": 1.7141,
+      "step": 4390
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.631488494955603e-05,
+      "loss": 1.8436,
+      "step": 4400
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.628305910060152e-05,
+      "loss": 1.9603,
+      "step": 4410
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.625123325164699e-05,
+      "loss": 1.7869,
+      "step": 4420
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.621940740269247e-05,
+      "loss": 1.9844,
+      "step": 4430
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.618758155373795e-05,
+      "loss": 1.9699,
+      "step": 4440
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.615575570478342e-05,
+      "loss": 1.879,
+      "step": 4450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.61239298558289e-05,
+      "loss": 1.8398,
+      "step": 4460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.609210400687439e-05,
+      "loss": 2.1011,
+      "step": 4470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.606027815791987e-05,
+      "loss": 1.9542,
+      "step": 4480
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.602845230896534e-05,
+      "loss": 2.2341,
+      "step": 4490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.599662646001081e-05,
+      "loss": 1.7047,
+      "step": 4500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.596480061105631e-05,
+      "loss": 1.8369,
+      "step": 4510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.593297476210178e-05,
+      "loss": 1.7466,
+      "step": 4520
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.590114891314726e-05,
+      "loss": 1.736,
+      "step": 4530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.586932306419274e-05,
+      "loss": 1.6473,
+      "step": 4540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.583749721523822e-05,
+      "loss": 1.8275,
+      "step": 4550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.58056713662837e-05,
+      "loss": 1.8659,
+      "step": 4560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.577384551732917e-05,
+      "loss": 1.6171,
+      "step": 4570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.574201966837466e-05,
+      "loss": 1.5898,
+      "step": 4580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.571019381942014e-05,
+      "loss": 1.7374,
+      "step": 4590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.567836797046561e-05,
+      "loss": 1.7308,
+      "step": 4600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.56465421215111e-05,
+      "loss": 1.8294,
+      "step": 4610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.561471627255658e-05,
+      "loss": 1.5242,
+      "step": 4620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.558289042360206e-05,
+      "loss": 2.284,
+      "step": 4630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.555106457464753e-05,
+      "loss": 1.7441,
+      "step": 4640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.5519238725693e-05,
+      "loss": 2.1215,
+      "step": 4650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.54874128767385e-05,
+      "loss": 1.9204,
+      "step": 4660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.545558702778397e-05,
+      "loss": 1.8143,
+      "step": 4670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.542376117882945e-05,
+      "loss": 1.8353,
+      "step": 4680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.539193532987492e-05,
+      "loss": 1.9182,
+      "step": 4690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.53601094809204e-05,
+      "loss": 1.7577,
+      "step": 4700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.532828363196589e-05,
+      "loss": 1.5695,
+      "step": 4710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.529645778301136e-05,
+      "loss": 1.2831,
+      "step": 4720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.526463193405684e-05,
+      "loss": 1.7315,
+      "step": 4730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.523280608510233e-05,
+      "loss": 1.7665,
+      "step": 4740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.52009802361478e-05,
+      "loss": 1.8807,
+      "step": 4750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.516915438719328e-05,
+      "loss": 1.7689,
+      "step": 4760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.513732853823876e-05,
+      "loss": 1.6982,
+      "step": 4770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.510550268928425e-05,
+      "loss": 2.0575,
+      "step": 4780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.507367684032972e-05,
+      "loss": 1.8133,
+      "step": 4790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.504185099137519e-05,
+      "loss": 1.949,
+      "step": 4800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.501002514242068e-05,
+      "loss": 1.5134,
+      "step": 4810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.497819929346615e-05,
+      "loss": 1.9392,
+      "step": 4820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.494637344451164e-05,
+      "loss": 1.7536,
+      "step": 4830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.491454759555711e-05,
+      "loss": 2.1834,
+      "step": 4840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.488272174660259e-05,
+      "loss": 1.672,
+      "step": 4850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.485089589764808e-05,
+      "loss": 1.7865,
+      "step": 4860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.481907004869355e-05,
+      "loss": 1.6641,
+      "step": 4870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.478724419973903e-05,
+      "loss": 1.8628,
+      "step": 4880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.475541835078451e-05,
+      "loss": 1.7441,
+      "step": 4890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.472359250183e-05,
+      "loss": 1.4225,
+      "step": 4900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.469176665287547e-05,
+      "loss": 1.8587,
+      "step": 4910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.465994080392095e-05,
+      "loss": 1.7314,
+      "step": 4920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.462811495496643e-05,
+      "loss": 2.0339,
+      "step": 4930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.45962891060119e-05,
+      "loss": 1.9146,
+      "step": 4940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.456446325705739e-05,
+      "loss": 1.9494,
+      "step": 4950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.453263740810287e-05,
+      "loss": 1.5817,
+      "step": 4960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.450081155914834e-05,
+      "loss": 1.8283,
+      "step": 4970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.446898571019383e-05,
+      "loss": 1.5496,
+      "step": 4980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.44371598612393e-05,
+      "loss": 1.6279,
+      "step": 4990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.440533401228479e-05,
+      "loss": 2.1064,
+      "step": 5000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.437350816333026e-05,
+      "loss": 1.8572,
+      "step": 5010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.434168231437573e-05,
+      "loss": 1.9051,
+      "step": 5020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.430985646542122e-05,
+      "loss": 1.8268,
+      "step": 5030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.42780306164667e-05,
+      "loss": 2.0814,
+      "step": 5040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.424620476751218e-05,
+      "loss": 1.5997,
+      "step": 5050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.421437891855765e-05,
+      "loss": 2.0973,
+      "step": 5060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.418255306960314e-05,
+      "loss": 2.0325,
+      "step": 5070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.415072722064862e-05,
+      "loss": 1.9608,
+      "step": 5080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.411890137169409e-05,
+      "loss": 2.1482,
+      "step": 5090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.408707552273957e-05,
+      "loss": 1.7059,
+      "step": 5100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.405524967378506e-05,
+      "loss": 1.6616,
+      "step": 5110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.402342382483053e-05,
+      "loss": 1.331,
+      "step": 5120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.399159797587601e-05,
+      "loss": 1.9625,
+      "step": 5130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.395977212692148e-05,
+      "loss": 1.7034,
+      "step": 5140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.392794627796698e-05,
+      "loss": 2.0897,
+      "step": 5150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.389612042901245e-05,
+      "loss": 1.681,
+      "step": 5160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.386429458005792e-05,
+      "loss": 1.7284,
+      "step": 5170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.38324687311034e-05,
+      "loss": 1.7513,
+      "step": 5180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.380064288214889e-05,
+      "loss": 1.3583,
+      "step": 5190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.376881703319437e-05,
+      "loss": 1.7959,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.373699118423984e-05,
+      "loss": 2.121,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.370516533528532e-05,
+      "loss": 2.2103,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.367333948633081e-05,
+      "loss": 1.9372,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.364151363737628e-05,
+      "loss": 1.7694,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.360968778842176e-05,
+      "loss": 1.8922,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.357786193946725e-05,
+      "loss": 1.8685,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.354603609051272e-05,
+      "loss": 1.7645,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.35142102415582e-05,
+      "loss": 2.0918,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.348238439260367e-05,
+      "loss": 1.6694,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.345055854364917e-05,
+      "loss": 1.6349,
+      "step": 5300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.341873269469464e-05,
+      "loss": 1.7016,
+      "step": 5310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.33869068457401e-05,
+      "loss": 1.9029,
+      "step": 5320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.335508099678559e-05,
+      "loss": 1.7197,
+      "step": 5330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.332325514783107e-05,
+      "loss": 2.0524,
+      "step": 5340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.329142929887656e-05,
+      "loss": 1.7017,
+      "step": 5350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.325960344992203e-05,
+      "loss": 2.2195,
+      "step": 5360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.322777760096751e-05,
+      "loss": 1.7724,
+      "step": 5370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.3195951752013e-05,
+      "loss": 1.9593,
+      "step": 5380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.316412590305846e-05,
+      "loss": 1.8544,
+      "step": 5390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.313230005410395e-05,
+      "loss": 2.0837,
+      "step": 5400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.310047420514943e-05,
+      "loss": 1.97,
+      "step": 5410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.30686483561949e-05,
+      "loss": 1.4903,
+      "step": 5420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.303682250724039e-05,
+      "loss": 1.6233,
+      "step": 5430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.300499665828586e-05,
+      "loss": 1.5539,
+      "step": 5440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.297317080933135e-05,
+      "loss": 1.8161,
+      "step": 5450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.294134496037682e-05,
+      "loss": 1.752,
+      "step": 5460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.29095191114223e-05,
+      "loss": 2.069,
+      "step": 5470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.287769326246778e-05,
+      "loss": 2.061,
+      "step": 5480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.284586741351326e-05,
+      "loss": 1.6476,
+      "step": 5490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.281404156455874e-05,
+      "loss": 1.6349,
+      "step": 5500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.278221571560421e-05,
+      "loss": 1.4046,
+      "step": 5510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.27503898666497e-05,
+      "loss": 1.7181,
+      "step": 5520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.271856401769518e-05,
+      "loss": 1.8994,
+      "step": 5530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.268673816874065e-05,
+      "loss": 2.2467,
+      "step": 5540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.265491231978614e-05,
+      "loss": 1.7663,
+      "step": 5550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.26230864708316e-05,
+      "loss": 2.232,
+      "step": 5560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.25912606218771e-05,
+      "loss": 2.0191,
+      "step": 5570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.255943477292257e-05,
+      "loss": 2.1281,
+      "step": 5580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.252760892396804e-05,
+      "loss": 1.7599,
+      "step": 5590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.249578307501353e-05,
+      "loss": 1.9067,
+      "step": 5600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.246395722605901e-05,
+      "loss": 1.8876,
+      "step": 5610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.24321313771045e-05,
+      "loss": 2.0441,
+      "step": 5620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.240030552814996e-05,
+      "loss": 2.1636,
+      "step": 5630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.236847967919545e-05,
+      "loss": 1.6258,
+      "step": 5640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.233665383024093e-05,
+      "loss": 1.761,
+      "step": 5650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.23048279812864e-05,
+      "loss": 1.6266,
+      "step": 5660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.227300213233188e-05,
+      "loss": 1.6244,
+      "step": 5670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.224117628337737e-05,
+      "loss": 1.9169,
+      "step": 5680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.220935043442284e-05,
+      "loss": 1.3947,
+      "step": 5690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.217752458546832e-05,
+      "loss": 1.6006,
+      "step": 5700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.214569873651379e-05,
+      "loss": 1.4239,
+      "step": 5710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.211387288755929e-05,
+      "loss": 1.5619,
+      "step": 5720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.208204703860476e-05,
+      "loss": 1.5051,
+      "step": 5730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.205022118965023e-05,
+      "loss": 1.8509,
+      "step": 5740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.201839534069571e-05,
+      "loss": 1.7132,
+      "step": 5750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.19865694917412e-05,
+      "loss": 1.9296,
+      "step": 5760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.195474364278668e-05,
+      "loss": 1.8191,
+      "step": 5770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.192291779383215e-05,
+      "loss": 1.6306,
+      "step": 5780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.189109194487763e-05,
+      "loss": 1.5563,
+      "step": 5790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.185926609592312e-05,
+      "loss": 1.8907,
+      "step": 5800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.182744024696859e-05,
+      "loss": 1.6601,
+      "step": 5810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.179561439801407e-05,
+      "loss": 1.8245,
+      "step": 5820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.176378854905956e-05,
+      "loss": 1.8156,
+      "step": 5830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.173196270010503e-05,
+      "loss": 1.9084,
+      "step": 5840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.170013685115051e-05,
+      "loss": 1.5468,
+      "step": 5850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.166831100219598e-05,
+      "loss": 1.9778,
+      "step": 5860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.163648515324148e-05,
+      "loss": 1.701,
+      "step": 5870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.160465930428695e-05,
+      "loss": 1.6719,
+      "step": 5880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.157283345533242e-05,
+      "loss": 1.6509,
+      "step": 5890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.15410076063779e-05,
+      "loss": 1.673,
+      "step": 5900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.150918175742338e-05,
+      "loss": 1.3831,
+      "step": 5910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.147735590846887e-05,
+      "loss": 1.7159,
+      "step": 5920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.144553005951434e-05,
+      "loss": 1.832,
+      "step": 5930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.141370421055982e-05,
+      "loss": 1.93,
+      "step": 5940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13818783616053e-05,
+      "loss": 1.3991,
+      "step": 5950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.135005251265077e-05,
+      "loss": 1.7181,
+      "step": 5960
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131822666369626e-05,
+      "loss": 1.7019,
+      "step": 5970
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.128640081474174e-05,
+      "loss": 1.7486,
+      "step": 5980
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.125457496578721e-05,
+      "loss": 1.8143,
+      "step": 5990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.12227491168327e-05,
+      "loss": 1.7242,
+      "step": 6000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119092326787817e-05,
+      "loss": 1.3983,
+      "step": 6010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115909741892366e-05,
+      "loss": 1.5508,
+      "step": 6020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112727156996913e-05,
+      "loss": 2.059,
+      "step": 6030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10954457210146e-05,
+      "loss": 2.0893,
+      "step": 6040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.106361987206009e-05,
+      "loss": 2.0184,
+      "step": 6050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.103179402310557e-05,
+      "loss": 1.8744,
+      "step": 6060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.099996817415105e-05,
+      "loss": 1.4619,
+      "step": 6070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.096814232519652e-05,
+      "loss": 1.7368,
+      "step": 6080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.093631647624201e-05,
+      "loss": 1.7933,
+      "step": 6090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.090449062728749e-05,
+      "loss": 1.5629,
+      "step": 6100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.087266477833296e-05,
+      "loss": 2.0451,
+      "step": 6110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.084083892937845e-05,
+      "loss": 1.8756,
+      "step": 6120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.080901308042393e-05,
+      "loss": 1.7944,
+      "step": 6130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.07771872314694e-05,
+      "loss": 1.6811,
+      "step": 6140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.074536138251488e-05,
+      "loss": 1.4877,
+      "step": 6150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.071353553356035e-05,
+      "loss": 2.0818,
+      "step": 6160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.068170968460585e-05,
+      "loss": 1.4355,
+      "step": 6170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.064988383565132e-05,
+      "loss": 1.6414,
+      "step": 6180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.06180579866968e-05,
+      "loss": 2.0349,
+      "step": 6190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.058623213774227e-05,
+      "loss": 1.6727,
+      "step": 6200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.055440628878776e-05,
+      "loss": 1.8855,
+      "step": 6210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.052258043983324e-05,
+      "loss": 1.6737,
+      "step": 6220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.049075459087871e-05,
+      "loss": 1.485,
+      "step": 6230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.04589287419242e-05,
+      "loss": 1.277,
+      "step": 6240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.042710289296968e-05,
+      "loss": 1.778,
+      "step": 6250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.039527704401515e-05,
+      "loss": 2.0159,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.036345119506063e-05,
+      "loss": 1.7595,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.033162534610612e-05,
+      "loss": 1.7878,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.02997994971516e-05,
+      "loss": 2.0789,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.026797364819707e-05,
+      "loss": 1.613,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.023614779924254e-05,
+      "loss": 1.9158,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.020432195028804e-05,
+      "loss": 1.7132,
+      "step": 6320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.017249610133351e-05,
+      "loss": 1.8105,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.014067025237899e-05,
+      "loss": 1.4884,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.010884440342446e-05,
+      "loss": 1.2167,
+      "step": 6350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.007701855446994e-05,
+      "loss": 1.8082,
+      "step": 6360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.004519270551543e-05,
+      "loss": 1.6277,
+      "step": 6370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.00133668565609e-05,
+      "loss": 1.7452,
+      "step": 6380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.998154100760638e-05,
+      "loss": 1.7174,
+      "step": 6390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.994971515865187e-05,
+      "loss": 2.2034,
+      "step": 6400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.991788930969734e-05,
+      "loss": 1.6318,
+      "step": 6410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.988606346074282e-05,
+      "loss": 1.6383,
+      "step": 6420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.98542376117883e-05,
+      "loss": 1.4466,
+      "step": 6430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.982241176283379e-05,
+      "loss": 1.9118,
+      "step": 6440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.979058591387926e-05,
+      "loss": 1.8459,
+      "step": 6450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.975876006492473e-05,
+      "loss": 1.7772,
+      "step": 6460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.972693421597022e-05,
+      "loss": 1.6248,
+      "step": 6470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.96951083670157e-05,
+      "loss": 1.5957,
+      "step": 6480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.966328251806118e-05,
+      "loss": 1.4671,
+      "step": 6490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.963145666910665e-05,
+      "loss": 1.569,
+      "step": 6500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.959963082015213e-05,
+      "loss": 1.9128,
+      "step": 6510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.956780497119761e-05,
+      "loss": 1.7862,
+      "step": 6520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.953597912224308e-05,
+      "loss": 1.7062,
+      "step": 6530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.950415327328857e-05,
+      "loss": 1.4484,
+      "step": 6540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.947232742433405e-05,
+      "loss": 1.9372,
+      "step": 6550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.944050157537952e-05,
+      "loss": 2.0335,
+      "step": 6560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.9408675726425e-05,
+      "loss": 1.8989,
+      "step": 6570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.937684987747049e-05,
+      "loss": 1.4778,
+      "step": 6580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.934502402851597e-05,
+      "loss": 1.5545,
+      "step": 6590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.931319817956144e-05,
+      "loss": 1.7245,
+      "step": 6600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.928137233060691e-05,
+      "loss": 1.7073,
+      "step": 6610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.92495464816524e-05,
+      "loss": 1.5795,
+      "step": 6620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.921772063269788e-05,
+      "loss": 1.7212,
+      "step": 6630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.918589478374336e-05,
+      "loss": 1.3638,
+      "step": 6640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.915406893478883e-05,
+      "loss": 1.482,
+      "step": 6650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.912224308583432e-05,
+      "loss": 1.6127,
+      "step": 6660
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.90904172368798e-05,
+      "loss": 1.8949,
+      "step": 6670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.905859138792527e-05,
+      "loss": 1.6316,
+      "step": 6680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.902676553897076e-05,
+      "loss": 1.7575,
+      "step": 6690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.899493969001624e-05,
+      "loss": 1.6866,
+      "step": 6700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.896311384106171e-05,
+      "loss": 1.6949,
+      "step": 6710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.893128799210719e-05,
+      "loss": 2.1837,
+      "step": 6720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.889946214315266e-05,
+      "loss": 1.754,
+      "step": 6730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.886763629419816e-05,
+      "loss": 1.6275,
+      "step": 6740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.883581044524363e-05,
+      "loss": 1.6981,
+      "step": 6750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.88039845962891e-05,
+      "loss": 1.6664,
+      "step": 6760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.877215874733458e-05,
+      "loss": 1.7061,
+      "step": 6770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.874033289838007e-05,
+      "loss": 1.6414,
+      "step": 6780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.870850704942555e-05,
+      "loss": 1.7035,
+      "step": 6790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.867668120047102e-05,
+      "loss": 1.5274,
+      "step": 6800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86448553515165e-05,
+      "loss": 2.1185,
+      "step": 6810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.861302950256199e-05,
+      "loss": 1.949,
+      "step": 6820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.858120365360746e-05,
+      "loss": 1.7355,
+      "step": 6830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.854937780465294e-05,
+      "loss": 2.224,
+      "step": 6840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.851755195569843e-05,
+      "loss": 1.7225,
+      "step": 6850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.848572610674391e-05,
+      "loss": 2.2467,
+      "step": 6860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.845390025778938e-05,
+      "loss": 1.4054,
+      "step": 6870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.842207440883485e-05,
+      "loss": 1.3582,
+      "step": 6880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.839024855988035e-05,
+      "loss": 1.356,
+      "step": 6890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.835842271092582e-05,
+      "loss": 1.4775,
+      "step": 6900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.83265968619713e-05,
+      "loss": 1.8954,
+      "step": 6910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.829477101301677e-05,
+      "loss": 1.7116,
+      "step": 6920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.826294516406225e-05,
+      "loss": 1.4697,
+      "step": 6930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.823111931510774e-05,
+      "loss": 1.5814,
+      "step": 6940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.819929346615321e-05,
+      "loss": 1.9475,
+      "step": 6950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.816746761719869e-05,
+      "loss": 1.9221,
+      "step": 6960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.813564176824418e-05,
+      "loss": 1.8046,
+      "step": 6970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.810381591928965e-05,
+      "loss": 1.4795,
+      "step": 6980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.807199007033513e-05,
+      "loss": 1.6698,
+      "step": 6990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.804016422138061e-05,
+      "loss": 1.4368,
+      "step": 7000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.80083383724261e-05,
+      "loss": 2.2295,
+      "step": 7010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.797651252347157e-05,
+      "loss": 1.7967,
+      "step": 7020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.794468667451704e-05,
+      "loss": 1.8404,
+      "step": 7030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.791286082556253e-05,
+      "loss": 1.7147,
+      "step": 7040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.7881034976608e-05,
+      "loss": 1.893,
+      "step": 7050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.784920912765349e-05,
+      "loss": 1.8299,
+      "step": 7060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.781738327869896e-05,
+      "loss": 1.9237,
+      "step": 7070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.778555742974444e-05,
+      "loss": 2.1464,
+      "step": 7080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.775373158078992e-05,
+      "loss": 1.4533,
+      "step": 7090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.77219057318354e-05,
+      "loss": 1.5716,
+      "step": 7100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.769007988288088e-05,
+      "loss": 1.9034,
+      "step": 7110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.765825403392636e-05,
+      "loss": 1.6687,
+      "step": 7120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.762642818497183e-05,
+      "loss": 1.5473,
+      "step": 7130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.759460233601732e-05,
+      "loss": 1.4395,
+      "step": 7140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.75627764870628e-05,
+      "loss": 1.9144,
+      "step": 7150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.753095063810828e-05,
+      "loss": 1.4831,
+      "step": 7160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.749912478915375e-05,
+      "loss": 1.9204,
+      "step": 7170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.746729894019922e-05,
+      "loss": 2.0492,
+      "step": 7180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.743547309124472e-05,
+      "loss": 1.5555,
+      "step": 7190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.740364724229019e-05,
+      "loss": 1.4942,
+      "step": 7200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.737182139333567e-05,
+      "loss": 1.6958,
+      "step": 7210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.733999554438114e-05,
+      "loss": 1.4307,
+      "step": 7220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.730816969542663e-05,
+      "loss": 1.9068,
+      "step": 7230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.727634384647211e-05,
+      "loss": 1.8229,
+      "step": 7240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.724451799751758e-05,
+      "loss": 1.6404,
+      "step": 7250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.721269214856307e-05,
+      "loss": 1.7036,
+      "step": 7260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.718086629960855e-05,
+      "loss": 1.7163,
+      "step": 7270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.714904045065402e-05,
+      "loss": 1.6215,
+      "step": 7280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.71172146016995e-05,
+      "loss": 2.0836,
+      "step": 7290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.708538875274499e-05,
+      "loss": 1.9861,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.705356290379047e-05,
+      "loss": 1.8064,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.702173705483594e-05,
+      "loss": 1.616,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.698991120588141e-05,
+      "loss": 1.1965,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.695808535692691e-05,
+      "loss": 1.4957,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.692625950797238e-05,
+      "loss": 1.8038,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.689443365901786e-05,
+      "loss": 1.8001,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.686260781006333e-05,
+      "loss": 1.9065,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.683078196110881e-05,
+      "loss": 1.8561,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.67989561121543e-05,
+      "loss": 1.7497,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.676713026319977e-05,
+      "loss": 1.8024,
+      "step": 7400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.673530441424525e-05,
+      "loss": 1.666,
+      "step": 7410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.670347856529074e-05,
+      "loss": 1.4647,
+      "step": 7420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.667165271633622e-05,
+      "loss": 1.522,
+      "step": 7430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.663982686738169e-05,
+      "loss": 1.5436,
+      "step": 7440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.660800101842717e-05,
+      "loss": 1.6896,
+      "step": 7450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.657617516947266e-05,
+      "loss": 1.4487,
+      "step": 7460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.654434932051813e-05,
+      "loss": 1.5451,
+      "step": 7470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.651252347156361e-05,
+      "loss": 1.5389,
+      "step": 7480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.64806976226091e-05,
+      "loss": 1.856,
+      "step": 7490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.644887177365456e-05,
+      "loss": 1.6362,
+      "step": 7500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.641704592470005e-05,
+      "loss": 1.7324,
+      "step": 7510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.638522007574552e-05,
+      "loss": 1.6882,
+      "step": 7520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.635339422679102e-05,
+      "loss": 1.7187,
+      "step": 7530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.632156837783649e-05,
+      "loss": 1.4399,
+      "step": 7540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.628974252888196e-05,
+      "loss": 1.4401,
+      "step": 7550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.625791667992744e-05,
+      "loss": 1.5833,
+      "step": 7560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.622609083097292e-05,
+      "loss": 1.4363,
+      "step": 7570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.61942649820184e-05,
+      "loss": 1.9323,
+      "step": 7580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.616243913306388e-05,
+      "loss": 1.4852,
+      "step": 7590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.613061328410936e-05,
+      "loss": 1.7514,
+      "step": 7600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.609878743515484e-05,
+      "loss": 1.4138,
+      "step": 7610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.606696158620031e-05,
+      "loss": 1.7007,
+      "step": 7620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.60351357372458e-05,
+      "loss": 1.505,
+      "step": 7630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.600330988829128e-05,
+      "loss": 1.5586,
+      "step": 7640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.597148403933675e-05,
+      "loss": 1.777,
+      "step": 7650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.593965819038223e-05,
+      "loss": 1.7,
+      "step": 7660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.59078323414277e-05,
+      "loss": 1.467,
+      "step": 7670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.587600649247319e-05,
+      "loss": 1.9871,
+      "step": 7680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.584418064351867e-05,
+      "loss": 1.7776,
+      "step": 7690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.581235479456414e-05,
+      "loss": 1.4482,
+      "step": 7700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.578052894560963e-05,
+      "loss": 1.6553,
+      "step": 7710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.574870309665511e-05,
+      "loss": 1.6254,
+      "step": 7720
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.571687724770059e-05,
+      "loss": 1.5823,
+      "step": 7730
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.568505139874606e-05,
+      "loss": 1.9779,
+      "step": 7740
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.565322554979153e-05,
+      "loss": 1.6201,
+      "step": 7750
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.562139970083703e-05,
+      "loss": 1.5673,
+      "step": 7760
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.55895738518825e-05,
+      "loss": 1.4576,
+      "step": 7770
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.555774800292798e-05,
+      "loss": 1.8544,
+      "step": 7780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.552592215397345e-05,
+      "loss": 1.6615,
+      "step": 7790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.549409630501894e-05,
+      "loss": 1.7856,
+      "step": 7800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.546227045606442e-05,
+      "loss": 1.7265,
+      "step": 7810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.543044460710989e-05,
+      "loss": 1.9182,
+      "step": 7820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.539861875815538e-05,
+      "loss": 2.053,
+      "step": 7830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.536679290920086e-05,
+      "loss": 1.6568,
+      "step": 7840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.533496706024633e-05,
+      "loss": 1.7705,
+      "step": 7850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.530314121129181e-05,
+      "loss": 1.8134,
+      "step": 7860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.52713153623373e-05,
+      "loss": 1.4194,
+      "step": 7870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.523948951338278e-05,
+      "loss": 2.0076,
+      "step": 7880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.520766366442825e-05,
+      "loss": 1.7716,
+      "step": 7890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.517583781547372e-05,
+      "loss": 1.7052,
+      "step": 7900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.514401196651922e-05,
+      "loss": 1.6623,
+      "step": 7910
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.511218611756469e-05,
+      "loss": 1.6529,
+      "step": 7920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.508036026861017e-05,
+      "loss": 1.6233,
+      "step": 7930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.504853441965564e-05,
+      "loss": 1.7521,
+      "step": 7940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.501670857070112e-05,
+      "loss": 2.0434,
+      "step": 7950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.498488272174661e-05,
+      "loss": 1.4489,
+      "step": 7960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.495305687279208e-05,
+      "loss": 1.475,
+      "step": 7970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.492123102383756e-05,
+      "loss": 1.3923,
+      "step": 7980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.488940517488305e-05,
+      "loss": 1.6717,
+      "step": 7990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.485757932592852e-05,
+      "loss": 1.7205,
+      "step": 8000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.4825753476974e-05,
+      "loss": 1.5514,
+      "step": 8010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.479392762801948e-05,
+      "loss": 1.9066,
+      "step": 8020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.476210177906497e-05,
+      "loss": 1.8837,
+      "step": 8030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.473027593011044e-05,
+      "loss": 2.0779,
+      "step": 8040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.469845008115591e-05,
+      "loss": 1.6063,
+      "step": 8050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.46666242322014e-05,
+      "loss": 1.6868,
+      "step": 8060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.463479838324687e-05,
+      "loss": 1.6176,
+      "step": 8070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.460297253429236e-05,
+      "loss": 1.8274,
+      "step": 8080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.457114668533783e-05,
+      "loss": 1.7037,
+      "step": 8090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.453932083638333e-05,
+      "loss": 1.9955,
+      "step": 8100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.45074949874288e-05,
+      "loss": 1.7274,
+      "step": 8110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.447566913847427e-05,
+      "loss": 1.543,
+      "step": 8120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.444384328951975e-05,
+      "loss": 1.4292,
+      "step": 8130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.441201744056523e-05,
+      "loss": 1.6324,
+      "step": 8140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.438019159161072e-05,
+      "loss": 1.6815,
+      "step": 8150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.434836574265619e-05,
+      "loss": 1.6445,
+      "step": 8160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.431653989370167e-05,
+      "loss": 1.3703,
+      "step": 8170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.428471404474715e-05,
+      "loss": 1.7691,
+      "step": 8180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.425288819579262e-05,
+      "loss": 1.4671,
+      "step": 8190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.422106234683811e-05,
+      "loss": 1.7908,
+      "step": 8200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.418923649788359e-05,
+      "loss": 1.5641,
+      "step": 8210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.415741064892906e-05,
+      "loss": 1.4575,
+      "step": 8220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.412558479997454e-05,
+      "loss": 1.6646,
+      "step": 8230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.409375895102001e-05,
+      "loss": 1.7336,
+      "step": 8240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.406193310206551e-05,
+      "loss": 1.6668,
+      "step": 8250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.403010725311098e-05,
+      "loss": 1.589,
+      "step": 8260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.399828140415645e-05,
+      "loss": 1.8278,
+      "step": 8270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.396645555520194e-05,
+      "loss": 1.4182,
+      "step": 8280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.393462970624742e-05,
+      "loss": 1.4516,
+      "step": 8290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.39028038572929e-05,
+      "loss": 1.6442,
+      "step": 8300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.387097800833837e-05,
+      "loss": 1.6038,
+      "step": 8310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.383915215938386e-05,
+      "loss": 1.7188,
+      "step": 8320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.380732631042934e-05,
+      "loss": 1.856,
+      "step": 8330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.377550046147481e-05,
+      "loss": 1.4309,
+      "step": 8340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.37436746125203e-05,
+      "loss": 1.7427,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.371184876356578e-05,
+      "loss": 1.523,
+      "step": 8360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.368002291461125e-05,
+      "loss": 1.7628,
+      "step": 8370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.364819706565673e-05,
+      "loss": 1.5589,
+      "step": 8380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.36163712167022e-05,
+      "loss": 1.9238,
+      "step": 8390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.35845453677477e-05,
+      "loss": 1.5053,
+      "step": 8400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.355271951879317e-05,
+      "loss": 1.6986,
+      "step": 8410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.352089366983864e-05,
+      "loss": 1.7976,
+      "step": 8420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.348906782088412e-05,
+      "loss": 2.1376,
+      "step": 8430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.34572419719296e-05,
+      "loss": 1.6317,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.342541612297509e-05,
+      "loss": 1.4605,
+      "step": 8450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.339359027402056e-05,
+      "loss": 1.6641,
+      "step": 8460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.336176442506604e-05,
+      "loss": 1.4108,
+      "step": 8470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.332993857611153e-05,
+      "loss": 1.4088,
+      "step": 8480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.3298112727157e-05,
+      "loss": 1.7873,
+      "step": 8490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.326628687820248e-05,
+      "loss": 1.8185,
+      "step": 8500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.323446102924796e-05,
+      "loss": 1.5374,
+      "step": 8510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.320263518029343e-05,
+      "loss": 1.2006,
+      "step": 8520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.317080933133892e-05,
+      "loss": 1.452,
+      "step": 8530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.313898348238439e-05,
+      "loss": 1.5237,
+      "step": 8540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.310715763342989e-05,
+      "loss": 1.5874,
+      "step": 8550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.307533178447536e-05,
+      "loss": 1.5766,
+      "step": 8560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.304350593552083e-05,
+      "loss": 1.7485,
+      "step": 8570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.301168008656631e-05,
+      "loss": 1.9478,
+      "step": 8580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.297985423761179e-05,
+      "loss": 1.8084,
+      "step": 8590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.294802838865728e-05,
+      "loss": 2.0169,
+      "step": 8600
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.291620253970275e-05,
+      "loss": 1.7916,
+      "step": 8610
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.288437669074823e-05,
+      "loss": 1.6838,
+      "step": 8620
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.285255084179371e-05,
+      "loss": 1.7549,
+      "step": 8630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.282072499283918e-05,
+      "loss": 1.8129,
+      "step": 8640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.278889914388467e-05,
+      "loss": 1.4067,
+      "step": 8650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.275707329493015e-05,
+      "loss": 1.4214,
+      "step": 8660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.272524744597562e-05,
+      "loss": 1.6211,
+      "step": 8670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.26934215970211e-05,
+      "loss": 1.5487,
+      "step": 8680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.266159574806658e-05,
+      "loss": 1.6237,
+      "step": 8690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.262976989911207e-05,
+      "loss": 1.7543,
+      "step": 8700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.259794405015754e-05,
+      "loss": 1.9502,
+      "step": 8710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.256611820120303e-05,
+      "loss": 1.3376,
+      "step": 8720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.25342923522485e-05,
+      "loss": 1.6009,
+      "step": 8730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.250246650329398e-05,
+      "loss": 2.0273,
+      "step": 8740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.247064065433946e-05,
+      "loss": 1.8337,
+      "step": 8750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.243881480538493e-05,
+      "loss": 1.7087,
+      "step": 8760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.240698895643042e-05,
+      "loss": 1.4726,
+      "step": 8770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.23751631074759e-05,
+      "loss": 1.3815,
+      "step": 8780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.234333725852137e-05,
+      "loss": 2.0258,
+      "step": 8790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.231151140956685e-05,
+      "loss": 1.6538,
+      "step": 8800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.227968556061232e-05,
+      "loss": 1.6996,
+      "step": 8810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.224785971165782e-05,
+      "loss": 1.6224,
+      "step": 8820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.221603386270329e-05,
+      "loss": 1.4107,
+      "step": 8830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.218420801374876e-05,
+      "loss": 1.6466,
+      "step": 8840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.215238216479425e-05,
+      "loss": 1.8746,
+      "step": 8850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.212055631583973e-05,
+      "loss": 1.8413,
+      "step": 8860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.208873046688521e-05,
+      "loss": 1.722,
+      "step": 8870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.205690461793068e-05,
+      "loss": 1.9583,
+      "step": 8880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.202507876897617e-05,
+      "loss": 1.7814,
+      "step": 8890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.199325292002165e-05,
+      "loss": 1.5631,
+      "step": 8900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.196142707106712e-05,
+      "loss": 1.2928,
+      "step": 8910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.19296012221126e-05,
+      "loss": 1.6888,
+      "step": 8920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.189777537315809e-05,
+      "loss": 1.415,
+      "step": 8930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.186594952420356e-05,
+      "loss": 1.8556,
+      "step": 8940
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.183412367524904e-05,
+      "loss": 1.4755,
+      "step": 8950
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.180229782629451e-05,
+      "loss": 1.8606,
+      "step": 8960
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.177047197734001e-05,
+      "loss": 1.5918,
+      "step": 8970
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.173864612838548e-05,
+      "loss": 1.6491,
+      "step": 8980
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.170682027943095e-05,
+      "loss": 1.1818,
+      "step": 8990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.167499443047643e-05,
+      "loss": 1.6124,
+      "step": 9000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.164316858152192e-05,
+      "loss": 1.5302,
+      "step": 9010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.16113427325674e-05,
+      "loss": 1.6171,
+      "step": 9020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.157951688361287e-05,
+      "loss": 1.5863,
+      "step": 9030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.154769103465835e-05,
+      "loss": 1.6996,
+      "step": 9040
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.151586518570384e-05,
+      "loss": 1.6777,
+      "step": 9050
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.148403933674931e-05,
+      "loss": 1.2677,
+      "step": 9060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.145221348779479e-05,
+      "loss": 1.704,
+      "step": 9070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.142038763884027e-05,
+      "loss": 1.5171,
+      "step": 9080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.138856178988574e-05,
+      "loss": 1.4865,
+      "step": 9090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.135673594093123e-05,
+      "loss": 1.397,
+      "step": 9100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.13249100919767e-05,
+      "loss": 1.3799,
+      "step": 9110
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.12930842430222e-05,
+      "loss": 1.7878,
+      "step": 9120
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.126125839406767e-05,
+      "loss": 1.7783,
+      "step": 9130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.122943254511314e-05,
+      "loss": 1.7521,
+      "step": 9140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.119760669615862e-05,
+      "loss": 1.7474,
+      "step": 9150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.11657808472041e-05,
+      "loss": 1.7479,
+      "step": 9160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.113395499824959e-05,
+      "loss": 1.7436,
+      "step": 9170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.110212914929506e-05,
+      "loss": 1.9751,
+      "step": 9180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.107030330034054e-05,
+      "loss": 1.5624,
+      "step": 9190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.103847745138602e-05,
+      "loss": 1.6065,
+      "step": 9200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.10066516024315e-05,
+      "loss": 1.553,
+      "step": 9210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.097482575347698e-05,
+      "loss": 1.8031,
+      "step": 9220
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.094299990452246e-05,
+      "loss": 1.385,
+      "step": 9230
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.091117405556793e-05,
+      "loss": 1.7275,
+      "step": 9240
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.087934820661342e-05,
+      "loss": 1.288,
+      "step": 9250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.084752235765889e-05,
+      "loss": 1.8881,
+      "step": 9260
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.081569650870438e-05,
+      "loss": 1.4875,
+      "step": 9270
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.078387065974985e-05,
+      "loss": 1.9456,
+      "step": 9280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.075204481079532e-05,
+      "loss": 1.7929,
+      "step": 9290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.07202189618408e-05,
+      "loss": 1.6181,
+      "step": 9300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.068839311288629e-05,
+      "loss": 1.6676,
+      "step": 9310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.065656726393177e-05,
+      "loss": 1.4524,
+      "step": 9320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.062474141497724e-05,
+      "loss": 1.5569,
+      "step": 9330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.059291556602273e-05,
+      "loss": 1.542,
+      "step": 9340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.056108971706821e-05,
+      "loss": 1.4374,
+      "step": 9350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.052926386811368e-05,
+      "loss": 1.6697,
+      "step": 9360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.049743801915916e-05,
+      "loss": 1.5604,
+      "step": 9370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.046561217020465e-05,
+      "loss": 1.8572,
+      "step": 9380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.043378632125013e-05,
+      "loss": 2.0608,
+      "step": 9390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.04019604722956e-05,
+      "loss": 1.6197,
+      "step": 9400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.037013462334107e-05,
+      "loss": 1.5616,
+      "step": 9410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.033830877438657e-05,
+      "loss": 1.1327,
+      "step": 9420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.030648292543204e-05,
+      "loss": 1.7424,
+      "step": 9430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.027465707647752e-05,
+      "loss": 1.4176,
+      "step": 9440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.024283122752299e-05,
+      "loss": 1.6296,
+      "step": 9450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.021100537856848e-05,
+      "loss": 1.7058,
+      "step": 9460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.017917952961396e-05,
+      "loss": 1.743,
+      "step": 9470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.014735368065943e-05,
+      "loss": 1.896,
+      "step": 9480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.011552783170491e-05,
+      "loss": 1.3877,
+      "step": 9490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.00837019827504e-05,
+      "loss": 1.7929,
+      "step": 9500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.005187613379587e-05,
+      "loss": 1.887,
+      "step": 9510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 7.002005028484135e-05,
+      "loss": 1.6872,
+      "step": 9520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.998822443588684e-05,
+      "loss": 1.4152,
+      "step": 9530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.995639858693232e-05,
+      "loss": 1.4784,
+      "step": 9540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.992457273797779e-05,
+      "loss": 1.8467,
+      "step": 9550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.989274688902326e-05,
+      "loss": 1.7269,
+      "step": 9560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.986092104006876e-05,
+      "loss": 1.7993,
+      "step": 9570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.982909519111423e-05,
+      "loss": 1.6533,
+      "step": 9580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.979726934215971e-05,
+      "loss": 1.8264,
+      "step": 9590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.976544349320518e-05,
+      "loss": 1.4738,
+      "step": 9600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.973361764425066e-05,
+      "loss": 1.515,
+      "step": 9610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.970179179529615e-05,
+      "loss": 1.9079,
+      "step": 9620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.966996594634162e-05,
+      "loss": 1.8671,
+      "step": 9630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.96381400973871e-05,
+      "loss": 1.6265,
+      "step": 9640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.960631424843258e-05,
+      "loss": 1.568,
+      "step": 9650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.957448839947805e-05,
+      "loss": 1.8229,
+      "step": 9660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.954266255052354e-05,
+      "loss": 1.5253,
+      "step": 9670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.951083670156902e-05,
+      "loss": 1.9236,
+      "step": 9680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.94790108526145e-05,
+      "loss": 2.1358,
+      "step": 9690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.944718500365998e-05,
+      "loss": 1.8796,
+      "step": 9700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.941535915470545e-05,
+      "loss": 1.5388,
+      "step": 9710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.938353330575094e-05,
+      "loss": 1.6827,
+      "step": 9720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.935170745679641e-05,
+      "loss": 1.7603,
+      "step": 9730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.93198816078419e-05,
+      "loss": 1.3614,
+      "step": 9740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.928805575888737e-05,
+      "loss": 1.7216,
+      "step": 9750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.925622990993285e-05,
+      "loss": 1.3782,
+      "step": 9760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.922440406097833e-05,
+      "loss": 1.6877,
+      "step": 9770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.91925782120238e-05,
+      "loss": 1.5273,
+      "step": 9780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.916075236306929e-05,
+      "loss": 1.4268,
+      "step": 9790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.912892651411477e-05,
+      "loss": 1.6359,
+      "step": 9800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.909710066516024e-05,
+      "loss": 1.6763,
+      "step": 9810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.906527481620573e-05,
+      "loss": 1.4216,
+      "step": 9820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.903344896725121e-05,
+      "loss": 1.4369,
+      "step": 9830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.900162311829669e-05,
+      "loss": 1.4794,
+      "step": 9840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.896979726934216e-05,
+      "loss": 1.7834,
+      "step": 9850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.893797142038763e-05,
+      "loss": 1.6101,
+      "step": 9860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.890614557143312e-05,
+      "loss": 1.9098,
+      "step": 9870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.88743197224786e-05,
+      "loss": 1.5039,
+      "step": 9880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.884249387352408e-05,
+      "loss": 1.6931,
+      "step": 9890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.881066802456955e-05,
+      "loss": 1.9016,
+      "step": 9900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.877884217561504e-05,
+      "loss": 1.6264,
+      "step": 9910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.874701632666052e-05,
+      "loss": 2.0439,
+      "step": 9920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.871519047770599e-05,
+      "loss": 1.8369,
+      "step": 9930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.868336462875147e-05,
+      "loss": 1.52,
+      "step": 9940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.865153877979696e-05,
+      "loss": 1.3288,
+      "step": 9950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.861971293084243e-05,
+      "loss": 1.3007,
+      "step": 9960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.858788708188791e-05,
+      "loss": 1.712,
+      "step": 9970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.855606123293338e-05,
+      "loss": 1.6743,
+      "step": 9980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.852423538397888e-05,
+      "loss": 1.7357,
+      "step": 9990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.849240953502435e-05,
+      "loss": 1.3981,
+      "step": 10000
+    }
+  ],
+  "max_steps": 31521,
+  "num_train_epochs": 3,
+  "total_flos": 9.480975121477796e+17,
+  "trial_name": null,
+  "trial_params": null
+}