diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,10819 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7167602436984829,
+  "eval_steps": 500,
+  "global_step": 9000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 4.6942,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1e-05,
+      "loss": 4.7828,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-05,
+      "loss": 4.6849,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-05,
+      "loss": 4.7359,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-05,
+      "loss": 4.6983,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-05,
+      "loss": 4.644,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5e-05,
+      "loss": 4.5187,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-05,
+      "loss": 4.5579,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5e-05,
+      "loss": 4.578,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-05,
+      "loss": 4.5831,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 4.6096,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 4.5414,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 4.5711,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-05,
+      "loss": 4.5152,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 4.4487,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-05,
+      "loss": 4.3695,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.5e-05,
+      "loss": 4.4659,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-05,
+      "loss": 4.3659,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.5e-05,
+      "loss": 4.3929,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001,
+      "loss": 4.4216,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999996024218523e-05,
+      "loss": 4.4018,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999984096880413e-05,
+      "loss": 4.3597,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99996421800464e-05,
+      "loss": 4.439,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999936387622818e-05,
+      "loss": 4.3521,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999900605779205e-05,
+      "loss": 4.3989,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999856872530703e-05,
+      "loss": 4.3338,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999805187946864e-05,
+      "loss": 4.2993,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999745552109885e-05,
+      "loss": 4.2473,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9996779651146e-05,
+      "loss": 4.2912,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999602427068499e-05,
+      "loss": 4.2601,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999518938091707e-05,
+      "loss": 4.267,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999427498317e-05,
+      "loss": 4.2148,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999328107889794e-05,
+      "loss": 4.2266,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999220766968153e-05,
+      "loss": 4.2288,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99910547572278e-05,
+      "loss": 4.2851,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998982234337026e-05,
+      "loss": 4.2129,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.998851043006885e-05,
+      "loss": 4.2106,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998711901940988e-05,
+      "loss": 4.15,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998564811360616e-05,
+      "loss": 4.1142,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998409771499686e-05,
+      "loss": 4.1932,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998246782604763e-05,
+      "loss": 4.1948,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99807584493505e-05,
+      "loss": 4.1887,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997896958762389e-05,
+      "loss": 4.2004,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997710124371267e-05,
+      "loss": 4.1001,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997515342058809e-05,
+      "loss": 4.2075,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997312612134778e-05,
+      "loss": 4.0638,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99710193492158e-05,
+      "loss": 4.1659,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996883310754256e-05,
+      "loss": 4.1513,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996656739980487e-05,
+      "loss": 4.0307,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996422222960593e-05,
+      "loss": 4.0633,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996179760067527e-05,
+      "loss": 4.1079,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995929351686885e-05,
+      "loss": 4.0861,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995670998216888e-05,
+      "loss": 4.0651,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995404700068405e-05,
+      "loss": 4.1165,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99513045766493e-05,
+      "loss": 4.1147,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994848271442594e-05,
+      "loss": 4.0725,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994558141850163e-05,
+      "loss": 4.0731,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.994260069349032e-05,
+      "loss": 4.1369,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99395405441323e-05,
+      "loss": 4.0649,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993640097529419e-05,
+      "loss": 4.0254,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.993318199196885e-05,
+      "loss": 4.1072,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.992988359927549e-05,
+      "loss": 4.0304,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992650580245957e-05,
+      "loss": 4.0497,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992304860689284e-05,
+      "loss": 4.0945,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991951201807334e-05,
+      "loss": 4.0605,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991589604162536e-05,
+      "loss": 4.0561,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991220068329941e-05,
+      "loss": 3.9424,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990842594897227e-05,
+      "loss": 4.0298,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990457184464695e-05,
+      "loss": 4.0149,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990063837645267e-05,
+      "loss": 3.9618,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98966255506449e-05,
+      "loss": 4.1018,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989253337360526e-05,
+      "loss": 3.9386,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.988836185184161e-05,
+      "loss": 4.0345,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.988411099198797e-05,
+      "loss": 3.9962,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987978080080453e-05,
+      "loss": 3.9985,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987537128517765e-05,
+      "loss": 3.9344,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987088245211984e-05,
+      "loss": 3.9529,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986631430876973e-05,
+      "loss": 4.0632,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986166686239213e-05,
+      "loss": 3.9907,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985694012037791e-05,
+      "loss": 4.0267,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.985213409024407e-05,
+      "loss": 4.0315,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98472487796337e-05,
+      "loss": 3.9565,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.984228419631596e-05,
+      "loss": 3.9627,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983724034818611e-05,
+      "loss": 4.0307,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983211724326545e-05,
+      "loss": 3.9366,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98269148897013e-05,
+      "loss": 3.9646,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.982163329576704e-05,
+      "loss": 3.9946,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.981627246986203e-05,
+      "loss": 4.0068,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.981083242051168e-05,
+      "loss": 3.9392,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.980531315636738e-05,
+      "loss": 3.8882,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.979971468620647e-05,
+      "loss": 3.9324,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.979403701893226e-05,
+      "loss": 3.9649,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.978828016357405e-05,
+      "loss": 3.831,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.978244412928699e-05,
+      "loss": 3.9977,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.977652892535224e-05,
+      "loss": 3.8904,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.977053456117679e-05,
+      "loss": 3.9068,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.976446104629358e-05,
+      "loss": 4.0171,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.975830839036137e-05,
+      "loss": 3.9263,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.975207660316485e-05,
+      "loss": 3.8902,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.974576569461449e-05,
+      "loss": 3.8908,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973937567474658e-05,
+      "loss": 3.9495,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973290655372327e-05,
+      "loss": 3.9359,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972635834183249e-05,
+      "loss": 3.9539,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971973104948793e-05,
+      "loss": 3.8985,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.971302468722907e-05,
+      "loss": 3.8712,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.970623926572111e-05,
+      "loss": 3.8512,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9699374795755e-05,
+      "loss": 3.8015,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.969243128824742e-05,
+      "loss": 3.8675,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.968540875424065e-05,
+      "loss": 3.8848,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967830720490277e-05,
+      "loss": 3.9619,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.967112665152743e-05,
+      "loss": 3.8321,
+      "step": 555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.966386710553399e-05,
+      "loss": 3.7751,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.965652857846737e-05,
+      "loss": 3.888,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.964911108199813e-05,
+      "loss": 3.8488,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.964161462792241e-05,
+      "loss": 3.8746,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.963403922816191e-05,
+      "loss": 3.8921,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.962638489476389e-05,
+      "loss": 3.8137,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.961865163990114e-05,
+      "loss": 3.9471,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.961083947587192e-05,
+      "loss": 3.8715,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.960294841510005e-05,
+      "loss": 3.8753,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.959497847013477e-05,
+      "loss": 3.7872,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.958692965365078e-05,
+      "loss": 3.8676,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95788019784482e-05,
+      "loss": 3.7823,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95705954574526e-05,
+      "loss": 3.8179,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95623101037149e-05,
+      "loss": 3.8291,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955394593041143e-05,
+      "loss": 3.9407,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95455029508438e-05,
+      "loss": 3.8398,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9536981178439e-05,
+      "loss": 3.8793,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.952838062674932e-05,
+      "loss": 3.7892,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951970130945232e-05,
+      "loss": 3.729,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.951094324035082e-05,
+      "loss": 3.8171,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.950210643337291e-05,
+      "loss": 3.8023,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.949319090257186e-05,
+      "loss": 3.8391,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.948419666212614e-05,
+      "loss": 3.8643,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.947512372633943e-05,
+      "loss": 3.8256,
+      "step": 675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.946597210964052e-05,
+      "loss": 3.8892,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.945674182658333e-05,
+      "loss": 3.7721,
+      "step": 685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.944743289184692e-05,
+      "loss": 3.8196,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.943804532023538e-05,
+      "loss": 3.7935,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.942857912667791e-05,
+      "loss": 3.8135,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.941903432622869e-05,
+      "loss": 3.8458,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.940941093406695e-05,
+      "loss": 3.9027,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93997089654969e-05,
+      "loss": 3.7112,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93899284359477e-05,
+      "loss": 3.7702,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.938006936097343e-05,
+      "loss": 3.7204,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.937013175625312e-05,
+      "loss": 3.7208,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.936011563759066e-05,
+      "loss": 3.8012,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93500210209148e-05,
+      "loss": 3.8407,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.933984792227918e-05,
+      "loss": 3.857,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.932959635786214e-05,
+      "loss": 3.793,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.931926634396692e-05,
+      "loss": 3.8489,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930885789702147e-05,
+      "loss": 3.7901,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929837103357844e-05,
+      "loss": 3.7984,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928780577031526e-05,
+      "loss": 3.7905,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.927716212403397e-05,
+      "loss": 3.7457,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926644011166133e-05,
+      "loss": 3.7893,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.925563975024865e-05,
+      "loss": 3.7334,
+      "step": 785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.924476105697192e-05,
+      "loss": 3.8116,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923380404913163e-05,
+      "loss": 3.7852,
+      "step": 795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.922276874415285e-05,
+      "loss": 3.8064,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921165515958519e-05,
+      "loss": 3.8788,
+      "step": 805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920046331310271e-05,
+      "loss": 3.8111,
+      "step": 810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.918919322250396e-05,
+      "loss": 3.7122,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.917784490571187e-05,
+      "loss": 3.7705,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.916641838077382e-05,
+      "loss": 3.8083,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.915491366586159e-05,
+      "loss": 3.733,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.914333077927125e-05,
+      "loss": 3.7442,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.91316697394232e-05,
+      "loss": 3.7237,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.911993056486216e-05,
+      "loss": 3.7868,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.910811327425707e-05,
+      "loss": 3.8094,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.909621788640113e-05,
+      "loss": 3.8008,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.908424442021172e-05,
+      "loss": 3.7155,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.907219289473037e-05,
+      "loss": 3.6726,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.906006332912282e-05,
+      "loss": 3.7196,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.904785574267883e-05,
+      "loss": 3.6389,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903557015481231e-05,
+      "loss": 3.8362,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.902320658506117e-05,
+      "loss": 3.7614,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.901076505308733e-05,
+      "loss": 3.7371,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899824557867675e-05,
+      "loss": 3.8077,
+      "step": 895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.898564818173928e-05,
+      "loss": 3.7623,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.897297288230876e-05,
+      "loss": 3.6897,
+      "step": 905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.896021970054282e-05,
+      "loss": 3.6569,
+      "step": 910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.894738865672306e-05,
+      "loss": 3.7735,
+      "step": 915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89344797712548e-05,
+      "loss": 3.7146,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.892149306466724e-05,
+      "loss": 3.7589,
+      "step": 925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.89084285576133e-05,
+      "loss": 3.6673,
+      "step": 930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.889528627086962e-05,
+      "loss": 3.6378,
+      "step": 935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.888206622533653e-05,
+      "loss": 3.7231,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.886876844203807e-05,
+      "loss": 3.7688,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.885539294212185e-05,
+      "loss": 3.6341,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.88419397468591e-05,
+      "loss": 3.7139,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.88284088776446e-05,
+      "loss": 3.6872,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.881480035599667e-05,
+      "loss": 3.7708,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.880111420355711e-05,
+      "loss": 3.6776,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.878735044209117e-05,
+      "loss": 3.7223,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.877350909348757e-05,
+      "loss": 3.7746,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.875959017975833e-05,
+      "loss": 3.7101,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87455937230389e-05,
+      "loss": 3.6918,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.873151974558804e-05,
+      "loss": 3.8019,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.871736826978775e-05,
+      "loss": 3.742,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.870313931814329e-05,
+      "loss": 3.6611,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.868883291328316e-05,
+      "loss": 3.6668,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.867444907795901e-05,
+      "loss": 3.7924,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.865998783504563e-05,
+      "loss": 3.7292,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.864544920754093e-05,
+      "loss": 3.6375,
+      "step": 1025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.863083321856585e-05,
+      "loss": 3.6995,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.861613989136442e-05,
+      "loss": 3.6144,
+      "step": 1035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.860136924930358e-05,
+      "loss": 3.7128,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.858652131587329e-05,
+      "loss": 3.7044,
+      "step": 1045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.857159611468641e-05,
+      "loss": 3.7889,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.855659366947864e-05,
+      "loss": 3.7131,
+      "step": 1055
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.854151400410862e-05,
+      "loss": 3.7154,
+      "step": 1060
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.852635714255767e-05,
+      "loss": 3.7178,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.851112310892998e-05,
+      "loss": 3.6711,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.84958119274524e-05,
+      "loss": 3.7292,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.848042362247452e-05,
+      "loss": 3.8056,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.846495821846853e-05,
+      "loss": 3.6307,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.844941574002925e-05,
+      "loss": 3.5814,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.843379621187411e-05,
+      "loss": 3.6567,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.841809965884303e-05,
+      "loss": 3.6431,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.840232610589844e-05,
+      "loss": 3.7094,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.838647557812521e-05,
+      "loss": 3.6881,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.837054810073064e-05,
+      "loss": 3.6095,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.835454369904438e-05,
+      "loss": 3.6257,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.833846239851845e-05,
+      "loss": 3.5595,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.832230422472714e-05,
+      "loss": 3.6242,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.830606920336703e-05,
+      "loss": 3.7295,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82897573602568e-05,
+      "loss": 3.6477,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827336872133745e-05,
+      "loss": 3.6793,
+      "step": 1145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.825690331267203e-05,
+      "loss": 3.6984,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.824036116044566e-05,
+      "loss": 3.569,
+      "step": 1155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.822374229096554e-05,
+      "loss": 3.6613,
+      "step": 1160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.820704673066088e-05,
+      "loss": 3.6229,
+      "step": 1165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.819027450608283e-05,
+      "loss": 3.6221,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.817342564390446e-05,
+      "loss": 3.6367,
+      "step": 1175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.815650017092077e-05,
+      "loss": 3.617,
+      "step": 1180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.81394981140485e-05,
+      "loss": 3.604,
+      "step": 1185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.812241950032627e-05,
+      "loss": 3.711,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.81052643569144e-05,
+      "loss": 3.6117,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.808803271109493e-05,
+      "loss": 3.6292,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.807072459027156e-05,
+      "loss": 3.7184,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.805334002196964e-05,
+      "loss": 3.5923,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.803587903383602e-05,
+      "loss": 3.6493,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.801834165363917e-05,
+      "loss": 3.5904,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.800072790926899e-05,
+      "loss": 3.6719,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.798303782873685e-05,
+      "loss": 3.5674,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.796527144017549e-05,
+      "loss": 3.5526,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.794742877183904e-05,
+      "loss": 3.6448,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.79295098521029e-05,
+      "loss": 3.5928,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.791151470946378e-05,
+      "loss": 3.6718,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.789344337253956e-05,
+      "loss": 3.6312,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.787529587006931e-05,
+      "loss": 3.6153,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.785707223091326e-05,
+      "loss": 3.7067,
+      "step": 1265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783877248405267e-05,
+      "loss": 3.6906,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782039665858987e-05,
+      "loss": 3.6093,
+      "step": 1275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780194478374815e-05,
+      "loss": 3.5865,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.778341688887178e-05,
+      "loss": 3.6759,
+      "step": 1285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.776481300342589e-05,
+      "loss": 3.6521,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.774613315699648e-05,
+      "loss": 3.6359,
+      "step": 1295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.772737737929033e-05,
+      "loss": 3.7053,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.770854570013502e-05,
+      "loss": 3.6537,
+      "step": 1305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.768963814947877e-05,
+      "loss": 3.7156,
+      "step": 1310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.767065475739052e-05,
+      "loss": 3.6197,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.76515955540598e-05,
+      "loss": 3.7069,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.763246056979669e-05,
+      "loss": 3.7176,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.76132498350318e-05,
+      "loss": 3.6145,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.75939633803162e-05,
+      "loss": 3.6255,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.757460123632136e-05,
+      "loss": 3.5853,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.755516343383919e-05,
+      "loss": 3.5463,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.753565000378184e-05,
+      "loss": 3.6211,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.751606097718178e-05,
+      "loss": 3.5817,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.749639638519168e-05,
+      "loss": 3.6224,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.747665625908437e-05,
+      "loss": 3.5653,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.745684063025285e-05,
+      "loss": 3.6668,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.743694953021014e-05,
+      "loss": 3.6129,
+      "step": 1375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.741698299058932e-05,
+      "loss": 3.6293,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.739694104314343e-05,
+      "loss": 3.6062,
+      "step": 1385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.737682371974542e-05,
+      "loss": 3.5647,
+      "step": 1390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735663105238814e-05,
+      "loss": 3.6538,
+      "step": 1395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733636307318422e-05,
+      "loss": 3.6094,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73160198143661e-05,
+      "loss": 3.587,
+      "step": 1405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.729560130828593e-05,
+      "loss": 3.6404,
+      "step": 1410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.727510758741549e-05,
+      "loss": 3.6076,
+      "step": 1415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.725453868434622e-05,
+      "loss": 3.6094,
+      "step": 1420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.72338946317891e-05,
+      "loss": 3.6192,
+      "step": 1425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.721317546257464e-05,
+      "loss": 3.5425,
+      "step": 1430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.719238120965277e-05,
+      "loss": 3.4998,
+      "step": 1435
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.717151190609287e-05,
+      "loss": 3.6023,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.715056758508365e-05,
+      "loss": 3.6063,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.712954827993314e-05,
+      "loss": 3.5854,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.710845402406859e-05,
+      "loss": 3.5709,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.708728485103646e-05,
+      "loss": 3.6337,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.706604079450236e-05,
+      "loss": 3.6097,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7044721888251e-05,
+      "loss": 3.5907,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.702332816618606e-05,
+      "loss": 3.5769,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.700185966233028e-05,
+      "loss": 3.7142,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.698031641082527e-05,
+      "loss": 3.6429,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.695869844593157e-05,
+      "loss": 3.6309,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.693700580202845e-05,
+      "loss": 3.6244,
+      "step": 1495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.691523851361405e-05,
+      "loss": 3.5841,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.689339661530511e-05,
+      "loss": 3.6318,
+      "step": 1505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.687148014183712e-05,
+      "loss": 3.5772,
+      "step": 1510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.684948912806411e-05,
+      "loss": 3.5391,
+      "step": 1515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682742360895865e-05,
+      "loss": 3.601,
+      "step": 1520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.680528361961181e-05,
+      "loss": 3.5806,
+      "step": 1525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.678306919523312e-05,
+      "loss": 3.5674,
+      "step": 1530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.676078037115045e-05,
+      "loss": 3.5536,
+      "step": 1535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.673841718280999e-05,
+      "loss": 3.6266,
+      "step": 1540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.671597966577621e-05,
+      "loss": 3.6625,
+      "step": 1545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.669346785573177e-05,
+      "loss": 3.668,
+      "step": 1550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.667088178847748e-05,
+      "loss": 3.5286,
+      "step": 1555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.664822149993225e-05,
+      "loss": 3.6136,
+      "step": 1560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.662548702613304e-05,
+      "loss": 3.5837,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.660267840323472e-05,
+      "loss": 3.594,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.657979566751019e-05,
+      "loss": 3.5919,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.655683885535012e-05,
+      "loss": 3.6834,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.653380800326303e-05,
+      "loss": 3.5753,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.651070314787515e-05,
+      "loss": 3.5206,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.648752432593046e-05,
+      "loss": 3.4816,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.646427157429052e-05,
+      "loss": 3.599,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.644094492993446e-05,
+      "loss": 3.5214,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.641754442995894e-05,
+      "loss": 3.5812,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.639407011157807e-05,
+      "loss": 3.5747,
+      "step": 1615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.637052201212336e-05,
+      "loss": 3.6232,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.634690016904365e-05,
+      "loss": 3.5959,
+      "step": 1625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.632320461990505e-05,
+      "loss": 3.6579,
+      "step": 1630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.629943540239088e-05,
+      "loss": 3.5989,
+      "step": 1635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.627559255430166e-05,
+      "loss": 3.6357,
+      "step": 1640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.625167611355492e-05,
+      "loss": 3.5521,
+      "step": 1645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.622768611818532e-05,
+      "loss": 3.5342,
+      "step": 1650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.620362260634444e-05,
+      "loss": 3.6339,
+      "step": 1655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.617948561630077e-05,
+      "loss": 3.4787,
+      "step": 1660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.615527518643969e-05,
+      "loss": 3.5853,
+      "step": 1665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.613099135526336e-05,
+      "loss": 3.5324,
+      "step": 1670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.610663416139061e-05,
+      "loss": 3.4186,
+      "step": 1675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.608220364355705e-05,
+      "loss": 3.4482,
+      "step": 1680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.605769984061481e-05,
+      "loss": 3.6165,
+      "step": 1685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.603312279153262e-05,
+      "loss": 3.5361,
+      "step": 1690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.600847253539564e-05,
+      "loss": 3.6431,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.59837491114055e-05,
+      "loss": 3.5211,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.595895255888017e-05,
+      "loss": 3.5357,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.593408291725392e-05,
+      "loss": 3.4829,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.590914022607726e-05,
+      "loss": 3.5244,
+      "step": 1715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.588412452501685e-05,
+      "loss": 3.5994,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.585903585385548e-05,
+      "loss": 3.6754,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.583387425249199e-05,
+      "loss": 3.4901,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.580863976094119e-05,
+      "loss": 3.5048,
+      "step": 1735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.578333241933381e-05,
+      "loss": 3.6038,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.575795226791642e-05,
+      "loss": 3.6498,
+      "step": 1745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.573249934705139e-05,
+      "loss": 3.5392,
+      "step": 1750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.570697369721686e-05,
+      "loss": 3.5651,
+      "step": 1755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.568137535900655e-05,
+      "loss": 3.4781,
+      "step": 1760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.565570437312986e-05,
+      "loss": 3.504,
+      "step": 1765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.562996078041163e-05,
+      "loss": 3.5478,
+      "step": 1770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.560414462179225e-05,
+      "loss": 3.5587,
+      "step": 1775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.557825593832748e-05,
+      "loss": 3.5953,
+      "step": 1780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.555229477118842e-05,
+      "loss": 3.5326,
+      "step": 1785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.552626116166145e-05,
+      "loss": 3.5102,
+      "step": 1790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.550015515114812e-05,
+      "loss": 3.5282,
+      "step": 1795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.547397678116519e-05,
+      "loss": 3.4984,
+      "step": 1800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.544772609334441e-05,
+      "loss": 3.5408,
+      "step": 1805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.542140312943258e-05,
+      "loss": 3.571,
+      "step": 1810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.539500793129146e-05,
+      "loss": 3.5226,
+      "step": 1815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.536854054089767e-05,
+      "loss": 3.5068,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.534200100034263e-05,
+      "loss": 3.5671,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.53153893518325e-05,
+      "loss": 3.6019,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.528870563768812e-05,
+      "loss": 3.4649,
+      "step": 1835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.526194990034494e-05,
+      "loss": 3.5709,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.523512218235293e-05,
+      "loss": 3.4742,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.520822252637658e-05,
+      "loss": 3.5305,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.518125097519472e-05,
+      "loss": 3.5924,
+      "step": 1855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.515420757170057e-05,
+      "loss": 3.4474,
+      "step": 1860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.512709235890159e-05,
+      "loss": 3.4977,
+      "step": 1865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.509990537991942e-05,
+      "loss": 3.6109,
+      "step": 1870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.507264667798987e-05,
+      "loss": 3.5768,
+      "step": 1875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.504531629646283e-05,
+      "loss": 3.5078,
+      "step": 1880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.501791427880208e-05,
+      "loss": 3.5609,
+      "step": 1885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.499044066858548e-05,
+      "loss": 3.5146,
+      "step": 1890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.49628955095046e-05,
+      "loss": 3.4628,
+      "step": 1895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.493527884536486e-05,
+      "loss": 3.3574,
+      "step": 1900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.490759072008539e-05,
+      "loss": 3.5114,
+      "step": 1905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.487983117769897e-05,
+      "loss": 3.5263,
+      "step": 1910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.485200026235197e-05,
+      "loss": 3.487,
+      "step": 1915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.48240980183042e-05,
+      "loss": 3.4806,
+      "step": 1920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.479612448992898e-05,
+      "loss": 3.5092,
+      "step": 1925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.476807972171298e-05,
+      "loss": 3.5249,
+      "step": 1930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.47399637582561e-05,
+      "loss": 3.4768,
+      "step": 1935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.471177664427155e-05,
+      "loss": 3.4919,
+      "step": 1940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.468351842458564e-05,
+      "loss": 3.5028,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.465518914413775e-05,
+      "loss": 3.5483,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.462678884798036e-05,
+      "loss": 3.5497,
+      "step": 1955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.459831758127873e-05,
+      "loss": 3.5099,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.456977538931113e-05,
+      "loss": 3.5606,
+      "step": 1965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.454116231746856e-05,
+      "loss": 3.5443,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.451247841125475e-05,
+      "loss": 3.5503,
+      "step": 1975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.448372371628605e-05,
+      "loss": 3.6481,
+      "step": 1980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.445489827829143e-05,
+      "loss": 3.4327,
+      "step": 1985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.442600214311237e-05,
+      "loss": 3.5457,
+      "step": 1990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.439703535670272e-05,
+      "loss": 3.5742,
+      "step": 1995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.436799796512875e-05,
+      "loss": 3.5603,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.433889001456898e-05,
+      "loss": 3.5279,
+      "step": 2005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.430971155131416e-05,
+      "loss": 3.4488,
+      "step": 2010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.428046262176715e-05,
+      "loss": 3.3552,
+      "step": 2015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.42511432724429e-05,
+      "loss": 3.5797,
+      "step": 2020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.422175354996833e-05,
+      "loss": 3.462,
+      "step": 2025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.419229350108232e-05,
+      "loss": 3.4895,
+      "step": 2030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.416276317263553e-05,
+      "loss": 3.5937,
+      "step": 2035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41331626115904e-05,
+      "loss": 3.495,
+      "step": 2040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.41034918650211e-05,
+      "loss": 3.5199,
+      "step": 2045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.40737509801134e-05,
+      "loss": 3.5694,
+      "step": 2050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.404394000416456e-05,
+      "loss": 3.456,
+      "step": 2055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.401405898458338e-05,
+      "loss": 3.4895,
+      "step": 2060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.398410796889004e-05,
+      "loss": 3.5074,
+      "step": 2065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.395408700471597e-05,
+      "loss": 3.5423,
+      "step": 2070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.392399613980393e-05,
+      "loss": 3.4718,
+      "step": 2075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.389383542200778e-05,
+      "loss": 3.4885,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.38636048992925e-05,
+      "loss": 3.5272,
+      "step": 2085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.383330461973407e-05,
+      "loss": 3.4273,
+      "step": 2090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.380293463151939e-05,
+      "loss": 3.5103,
+      "step": 2095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.377249498294624e-05,
+      "loss": 3.5177,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.374198572242318e-05,
+      "loss": 3.4721,
+      "step": 2105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.37114068984695e-05,
+      "loss": 3.5344,
+      "step": 2110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.368075855971504e-05,
+      "loss": 3.4826,
+      "step": 2115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.365004075490027e-05,
+      "loss": 3.5711,
+      "step": 2120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.361925353287609e-05,
+      "loss": 3.5316,
+      "step": 2125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.358839694260381e-05,
+      "loss": 3.4581,
+      "step": 2130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.355747103315507e-05,
+      "loss": 3.5513,
+      "step": 2135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.352647585371168e-05,
+      "loss": 3.4465,
+      "step": 2140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.349541145356573e-05,
+      "loss": 3.5729,
+      "step": 2145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.346427788211929e-05,
+      "loss": 3.5033,
+      "step": 2150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.343307518888448e-05,
+      "loss": 3.5428,
+      "step": 2155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.340180342348332e-05,
+      "loss": 3.5128,
+      "step": 2160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.337046263564772e-05,
+      "loss": 3.4937,
+      "step": 2165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.333905287521932e-05,
+      "loss": 3.5154,
+      "step": 2170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.330757419214945e-05,
+      "loss": 3.6503,
+      "step": 2175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.327602663649906e-05,
+      "loss": 3.4777,
+      "step": 2180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.324441025843862e-05,
+      "loss": 3.5135,
+      "step": 2185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.321272510824806e-05,
+      "loss": 3.607,
+      "step": 2190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.318097123631668e-05,
+      "loss": 3.377,
+      "step": 2195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.314914869314304e-05,
+      "loss": 3.567,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.311725752933496e-05,
+      "loss": 3.4721,
+      "step": 2205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.308529779560933e-05,
+      "loss": 3.4765,
+      "step": 2210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.305326954279214e-05,
+      "loss": 3.5354,
+      "step": 2215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.302117282181833e-05,
+      "loss": 3.4196,
+      "step": 2220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.298900768373169e-05,
+      "loss": 3.4827,
+      "step": 2225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.295677417968486e-05,
+      "loss": 3.5881,
+      "step": 2230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.292447236093919e-05,
+      "loss": 3.5657,
+      "step": 2235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.289210227886466e-05,
+      "loss": 3.4733,
+      "step": 2240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.285966398493981e-05,
+      "loss": 3.4613,
+      "step": 2245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.28271575307517e-05,
+      "loss": 3.469,
+      "step": 2250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.279458296799574e-05,
+      "loss": 3.5411,
+      "step": 2255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.276194034847566e-05,
+      "loss": 3.6552,
+      "step": 2260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.272922972410343e-05,
+      "loss": 3.49,
+      "step": 2265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.269645114689916e-05,
+      "loss": 3.4776,
+      "step": 2270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.266360466899104e-05,
+      "loss": 3.5281,
+      "step": 2275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.263069034261525e-05,
+      "loss": 3.4744,
+      "step": 2280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.259770822011586e-05,
+      "loss": 3.4711,
+      "step": 2285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.256465835394473e-05,
+      "loss": 3.5001,
+      "step": 2290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.253154079666149e-05,
+      "loss": 3.4914,
+      "step": 2295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.24983556009334e-05,
+      "loss": 3.449,
+      "step": 2300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.246510281953532e-05,
+      "loss": 3.3332,
+      "step": 2305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.243178250534955e-05,
+      "loss": 3.5098,
+      "step": 2310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.23983947113658e-05,
+      "loss": 3.4639,
+      "step": 2315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.23649394906811e-05,
+      "loss": 3.4336,
+      "step": 2320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.233141689649974e-05,
+      "loss": 3.334,
+      "step": 2325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.229782698213307e-05,
+      "loss": 3.5559,
+      "step": 2330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.226416980099959e-05,
+      "loss": 3.5456,
+      "step": 2335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.223044540662473e-05,
+      "loss": 3.4469,
+      "step": 2340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.219665385264082e-05,
+      "loss": 3.5366,
+      "step": 2345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.216279519278699e-05,
+      "loss": 3.44,
+      "step": 2350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.21288694809091e-05,
+      "loss": 3.5158,
+      "step": 2355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.209487677095964e-05,
+      "loss": 3.5065,
+      "step": 2360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.206081711699764e-05,
+      "loss": 3.477,
+      "step": 2365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.202669057318857e-05,
+      "loss": 3.3703,
+      "step": 2370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.199249719380434e-05,
+      "loss": 3.5143,
+      "step": 2375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.19582370332231e-05,
+      "loss": 3.4243,
+      "step": 2380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.192391014592921e-05,
+      "loss": 3.4971,
+      "step": 2385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.188951658651316e-05,
+      "loss": 3.4576,
+      "step": 2390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.185505640967144e-05,
+      "loss": 3.449,
+      "step": 2395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.182052967020654e-05,
+      "loss": 3.4834,
+      "step": 2400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.178593642302673e-05,
+      "loss": 3.4718,
+      "step": 2405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.17512767231461e-05,
+      "loss": 3.4929,
+      "step": 2410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.171655062568443e-05,
+      "loss": 3.4378,
+      "step": 2415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.168175818586703e-05,
+      "loss": 3.5424,
+      "step": 2420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.164689945902478e-05,
+      "loss": 3.5078,
+      "step": 2425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.161197450059395e-05,
+      "loss": 3.5152,
+      "step": 2430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.157698336611615e-05,
+      "loss": 3.3418,
+      "step": 2435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.154192611123819e-05,
+      "loss": 3.4643,
+      "step": 2440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.150680279171208e-05,
+      "loss": 3.4382,
+      "step": 2445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.14716134633949e-05,
+      "loss": 3.49,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.143635818224866e-05,
+      "loss": 3.5082,
+      "step": 2455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.140103700434027e-05,
+      "loss": 3.5109,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.136564998584147e-05,
+      "loss": 3.4094,
+      "step": 2465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.133019718302865e-05,
+      "loss": 3.4167,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.129467865228287e-05,
+      "loss": 3.4241,
+      "step": 2475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.125909445008969e-05,
+      "loss": 3.5457,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.122344463303912e-05,
+      "loss": 3.4157,
+      "step": 2485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.11877292578255e-05,
+      "loss": 3.4833,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.115194838124746e-05,
+      "loss": 3.4268,
+      "step": 2495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.111610206020775e-05,
+      "loss": 3.3836,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.108019035171326e-05,
+      "loss": 3.486,
+      "step": 2505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.104421331287481e-05,
+      "loss": 3.4619,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.100817100090715e-05,
+      "loss": 3.4898,
+      "step": 2515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.097206347312881e-05,
+      "loss": 3.3892,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.093589078696207e-05,
+      "loss": 3.38,
+      "step": 2525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.089965299993277e-05,
+      "loss": 3.5966,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.086335016967033e-05,
+      "loss": 3.4651,
+      "step": 2535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.082698235390762e-05,
+      "loss": 3.4198,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.079054961048084e-05,
+      "loss": 3.4805,
+      "step": 2545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.075405199732941e-05,
+      "loss": 3.4511,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.071748957249595e-05,
+      "loss": 3.3768,
+      "step": 2555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.068086239412617e-05,
+      "loss": 3.4762,
+      "step": 2560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.06441705204687e-05,
+      "loss": 3.4741,
+      "step": 2565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.060741400987511e-05,
+      "loss": 3.4517,
+      "step": 2570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.057059292079972e-05,
+      "loss": 3.4101,
+      "step": 2575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.05337073117996e-05,
+      "loss": 3.3871,
+      "step": 2580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.049675724153439e-05,
+      "loss": 3.4241,
+      "step": 2585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.045974276876625e-05,
+      "loss": 3.5433,
+      "step": 2590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.042266395235975e-05,
+      "loss": 3.4374,
+      "step": 2595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.03855208512818e-05,
+      "loss": 3.4917,
+      "step": 2600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.034831352460155e-05,
+      "loss": 3.489,
+      "step": 2605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.031104203149029e-05,
+      "loss": 3.4671,
+      "step": 2610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.027370643122132e-05,
+      "loss": 3.4513,
+      "step": 2615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.023630678316995e-05,
+      "loss": 3.39,
+      "step": 2620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.019884314681328e-05,
+      "loss": 3.4402,
+      "step": 2625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.01613155817302e-05,
+      "loss": 3.4681,
+      "step": 2630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.01237241476013e-05,
+      "loss": 3.3896,
+      "step": 2635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.008606890420869e-05,
+      "loss": 3.3966,
+      "step": 2640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.004834991143597e-05,
+      "loss": 3.4971,
+      "step": 2645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.001056722926815e-05,
+      "loss": 3.3684,
+      "step": 2650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.99727209177915e-05,
+      "loss": 3.5489,
+      "step": 2655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.993481103719347e-05,
+      "loss": 3.4837,
+      "step": 2660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.989683764776263e-05,
+      "loss": 3.4455,
+      "step": 2665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.985880080988854e-05,
+      "loss": 3.5187,
+      "step": 2670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.982070058406167e-05,
+      "loss": 3.5234,
+      "step": 2675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.978253703087328e-05,
+      "loss": 3.5362,
+      "step": 2680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.974431021101535e-05,
+      "loss": 3.5598,
+      "step": 2685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.970602018528047e-05,
+      "loss": 3.4363,
+      "step": 2690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.966766701456177e-05,
+      "loss": 3.3995,
+      "step": 2695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.962925075985275e-05,
+      "loss": 3.4133,
+      "step": 2700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.959077148224727e-05,
+      "loss": 3.4896,
+      "step": 2705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.955222924293942e-05,
+      "loss": 3.4082,
+      "step": 2710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.951362410322341e-05,
+      "loss": 3.4358,
+      "step": 2715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.947495612449348e-05,
+      "loss": 3.4844,
+      "step": 2720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.94362253682438e-05,
+      "loss": 3.4643,
+      "step": 2725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.939743189606837e-05,
+      "loss": 3.4519,
+      "step": 2730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.935857576966096e-05,
+      "loss": 3.303,
+      "step": 2735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.931965705081494e-05,
+      "loss": 3.4388,
+      "step": 2740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.928067580142323e-05,
+      "loss": 3.3959,
+      "step": 2745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.92416320834782e-05,
+      "loss": 3.446,
+      "step": 2750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.92025259590716e-05,
+      "loss": 3.4604,
+      "step": 2755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.916335749039437e-05,
+      "loss": 3.5391,
+      "step": 2760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.912412673973662e-05,
+      "loss": 3.3397,
+      "step": 2765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.908483376948751e-05,
+      "loss": 3.4077,
+      "step": 2770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.904547864213512e-05,
+      "loss": 3.3902,
+      "step": 2775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.900606142026644e-05,
+      "loss": 3.3682,
+      "step": 2780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.896658216656717e-05,
+      "loss": 3.5104,
+      "step": 2785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.892704094382165e-05,
+      "loss": 3.4919,
+      "step": 2790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.888743781491277e-05,
+      "loss": 3.4907,
+      "step": 2795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.884777284282194e-05,
+      "loss": 3.3789,
+      "step": 2800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.88080460906288e-05,
+      "loss": 3.2901,
+      "step": 2805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.876825762151135e-05,
+      "loss": 3.3936,
+      "step": 2810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.872840749874567e-05,
+      "loss": 3.555,
+      "step": 2815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.868849578570591e-05,
+      "loss": 3.4477,
+      "step": 2820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.864852254586419e-05,
+      "loss": 3.3553,
+      "step": 2825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.860848784279043e-05,
+      "loss": 3.4809,
+      "step": 2830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.856839174015234e-05,
+      "loss": 3.4873,
+      "step": 2835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.852823430171527e-05,
+      "loss": 3.4045,
+      "step": 2840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.848801559134207e-05,
+      "loss": 3.3637,
+      "step": 2845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.844773567299308e-05,
+      "loss": 3.4424,
+      "step": 2850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.840739461072595e-05,
+      "loss": 3.4,
+      "step": 2855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.836699246869559e-05,
+      "loss": 3.4152,
+      "step": 2860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.832652931115404e-05,
+      "loss": 3.5277,
+      "step": 2865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.828600520245036e-05,
+      "loss": 3.4834,
+      "step": 2870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.824542020703053e-05,
+      "loss": 3.3749,
+      "step": 2875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.820477438943742e-05,
+      "loss": 3.3547,
+      "step": 2880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.816406781431054e-05,
+      "loss": 3.3373,
+      "step": 2885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.812330054638611e-05,
+      "loss": 3.474,
+      "step": 2890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.808247265049683e-05,
+      "loss": 3.4258,
+      "step": 2895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.804158419157178e-05,
+      "loss": 3.3889,
+      "step": 2900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.80006352346364e-05,
+      "loss": 3.3624,
+      "step": 2905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.795962584481236e-05,
+      "loss": 3.3606,
+      "step": 2910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.791855608731741e-05,
+      "loss": 3.4122,
+      "step": 2915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.787742602746524e-05,
+      "loss": 3.524,
+      "step": 2920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.783623573066558e-05,
+      "loss": 3.5582,
+      "step": 2925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.779498526242385e-05,
+      "loss": 3.2856,
+      "step": 2930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.775367468834116e-05,
+      "loss": 3.4641,
+      "step": 2935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.771230407411427e-05,
+      "loss": 3.4481,
+      "step": 2940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.767087348553537e-05,
+      "loss": 3.3438,
+      "step": 2945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.762938298849206e-05,
+      "loss": 3.3799,
+      "step": 2950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.758783264896719e-05,
+      "loss": 3.4216,
+      "step": 2955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.754622253303881e-05,
+      "loss": 3.3931,
+      "step": 2960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.750455270687999e-05,
+      "loss": 3.3399,
+      "step": 2965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.746282323675876e-05,
+      "loss": 3.3986,
+      "step": 2970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.742103418903806e-05,
+      "loss": 3.4072,
+      "step": 2975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.737918563017553e-05,
+      "loss": 3.4715,
+      "step": 2980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.733727762672343e-05,
+      "loss": 3.4706,
+      "step": 2985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.729531024532864e-05,
+      "loss": 3.4497,
+      "step": 2990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.725328355273237e-05,
+      "loss": 3.4577,
+      "step": 2995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.721119761577021e-05,
+      "loss": 3.4584,
+      "step": 3000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.716905250137196e-05,
+      "loss": 3.367,
+      "step": 3005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.712684827656152e-05,
+      "loss": 3.4083,
+      "step": 3010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.708458500845682e-05,
+      "loss": 3.5261,
+      "step": 3015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.704226276426964e-05,
+      "loss": 3.4557,
+      "step": 3020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.699988161130559e-05,
+      "loss": 3.4302,
+      "step": 3025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.695744161696395e-05,
+      "loss": 3.3727,
+      "step": 3030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.691494284873758e-05,
+      "loss": 3.4272,
+      "step": 3035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.687238537421281e-05,
+      "loss": 3.429,
+      "step": 3040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.68297692610693e-05,
+      "loss": 3.4621,
+      "step": 3045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.678709457708003e-05,
+      "loss": 3.4113,
+      "step": 3050
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.674436139011107e-05,
+      "loss": 3.4395,
+      "step": 3055
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.670156976812155e-05,
+      "loss": 3.4487,
+      "step": 3060
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.665871977916351e-05,
+      "loss": 3.4243,
+      "step": 3065
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.661581149138185e-05,
+      "loss": 3.3994,
+      "step": 3070
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.657284497301415e-05,
+      "loss": 3.4511,
+      "step": 3075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.652982029239058e-05,
+      "loss": 3.4993,
+      "step": 3080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.648673751793388e-05,
+      "loss": 3.4241,
+      "step": 3085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.64435967181591e-05,
+      "loss": 3.4658,
+      "step": 3090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.640039796167358e-05,
+      "loss": 3.4843,
+      "step": 3095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.635714131717687e-05,
+      "loss": 3.4162,
+      "step": 3100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.631382685346056e-05,
+      "loss": 3.411,
+      "step": 3105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.627045463940819e-05,
+      "loss": 3.4282,
+      "step": 3110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.622702474399513e-05,
+      "loss": 3.3778,
+      "step": 3115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.618353723628847e-05,
+      "loss": 3.4491,
+      "step": 3120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.613999218544697e-05,
+      "loss": 3.4359,
+      "step": 3125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.609638966072087e-05,
+      "loss": 3.3209,
+      "step": 3130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.60527297314518e-05,
+      "loss": 3.4207,
+      "step": 3135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.600901246707269e-05,
+      "loss": 3.4317,
+      "step": 3140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.596523793710767e-05,
+      "loss": 3.3601,
+      "step": 3145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.592140621117192e-05,
+      "loss": 3.508,
+      "step": 3150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.58775173589716e-05,
+      "loss": 3.3683,
+      "step": 3155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.583357145030368e-05,
+      "loss": 3.3702,
+      "step": 3160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.578956855505591e-05,
+      "loss": 3.4031,
+      "step": 3165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.574550874320665e-05,
+      "loss": 3.3817,
+      "step": 3170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.570139208482476e-05,
+      "loss": 3.38,
+      "step": 3175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.565721865006952e-05,
+      "loss": 3.4495,
+      "step": 3180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.561298850919051e-05,
+      "loss": 3.4216,
+      "step": 3185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.556870173252746e-05,
+      "loss": 3.3887,
+      "step": 3190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.552435839051022e-05,
+      "loss": 3.4144,
+      "step": 3195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.547995855365854e-05,
+      "loss": 3.3708,
+      "step": 3200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.543550229258206e-05,
+      "loss": 3.4078,
+      "step": 3205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.539098967798009e-05,
+      "loss": 3.446,
+      "step": 3210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.534642078064166e-05,
+      "loss": 3.3723,
+      "step": 3215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.53017956714452e-05,
+      "loss": 3.3989,
+      "step": 3220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.525711442135859e-05,
+      "loss": 3.445,
+      "step": 3225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.521237710143901e-05,
+      "loss": 3.5061,
+      "step": 3230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.516758378283278e-05,
+      "loss": 3.4727,
+      "step": 3235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.512273453677527e-05,
+      "loss": 3.347,
+      "step": 3240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.507782943459078e-05,
+      "loss": 3.4397,
+      "step": 3245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.503286854769246e-05,
+      "loss": 3.4332,
+      "step": 3250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.498785194758223e-05,
+      "loss": 3.3747,
+      "step": 3255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.494277970585049e-05,
+      "loss": 3.3486,
+      "step": 3260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.489765189417625e-05,
+      "loss": 3.3998,
+      "step": 3265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.485246858432678e-05,
+      "loss": 3.3772,
+      "step": 3270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.480722984815772e-05,
+      "loss": 3.435,
+      "step": 3275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.476193575761276e-05,
+      "loss": 3.3637,
+      "step": 3280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.471658638472369e-05,
+      "loss": 3.4799,
+      "step": 3285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.467118180161017e-05,
+      "loss": 3.3062,
+      "step": 3290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.46257220804797e-05,
+      "loss": 3.3486,
+      "step": 3295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.458020729362742e-05,
+      "loss": 3.366,
+      "step": 3300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.453463751343609e-05,
+      "loss": 3.3472,
+      "step": 3305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.448901281237591e-05,
+      "loss": 3.4045,
+      "step": 3310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.44433332630044e-05,
+      "loss": 3.4768,
+      "step": 3315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.439759893796633e-05,
+      "loss": 3.3336,
+      "step": 3320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.435180990999356e-05,
+      "loss": 3.4473,
+      "step": 3325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.430596625190497e-05,
+      "loss": 3.42,
+      "step": 3330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.42600680366063e-05,
+      "loss": 3.372,
+      "step": 3335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.421411533709009e-05,
+      "loss": 3.4726,
+      "step": 3340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.416810822643546e-05,
+      "loss": 3.4135,
+      "step": 3345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.41220467778081e-05,
+      "loss": 3.422,
+      "step": 3350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.407593106446013e-05,
+      "loss": 3.393,
+      "step": 3355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.402976115972994e-05,
+      "loss": 3.3534,
+      "step": 3360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.39835371370421e-05,
+      "loss": 3.3964,
+      "step": 3365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.39372590699073e-05,
+      "loss": 3.397,
+      "step": 3370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.389092703192208e-05,
+      "loss": 3.4361,
+      "step": 3375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.384454109676886e-05,
+      "loss": 3.3532,
+      "step": 3380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.379810133821583e-05,
+      "loss": 3.4604,
+      "step": 3385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.375160783011669e-05,
+      "loss": 3.3546,
+      "step": 3390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.370506064641064e-05,
+      "loss": 3.459,
+      "step": 3395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.365845986112226e-05,
+      "loss": 3.467,
+      "step": 3400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.361180554836138e-05,
+      "loss": 3.3447,
+      "step": 3405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.356509778232295e-05,
+      "loss": 3.3328,
+      "step": 3410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.351833663728688e-05,
+      "loss": 3.3733,
+      "step": 3415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.347152218761803e-05,
+      "loss": 3.3222,
+      "step": 3420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.342465450776602e-05,
+      "loss": 3.4111,
+      "step": 3425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.33777336722651e-05,
+      "loss": 3.4702,
+      "step": 3430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.333075975573405e-05,
+      "loss": 3.5019,
+      "step": 3435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.32837328328761e-05,
+      "loss": 3.3594,
+      "step": 3440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.323665297847876e-05,
+      "loss": 3.4535,
+      "step": 3445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.318952026741368e-05,
+      "loss": 3.4227,
+      "step": 3450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.314233477463667e-05,
+      "loss": 3.4089,
+      "step": 3455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.309509657518736e-05,
+      "loss": 3.3265,
+      "step": 3460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.304780574418927e-05,
+      "loss": 3.4319,
+      "step": 3465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.30004623568496e-05,
+      "loss": 3.3498,
+      "step": 3470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.295306648845912e-05,
+      "loss": 3.3462,
+      "step": 3475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.290561821439212e-05,
+      "loss": 3.4399,
+      "step": 3480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.285811761010615e-05,
+      "loss": 3.4775,
+      "step": 3485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.281056475114204e-05,
+      "loss": 3.4148,
+      "step": 3490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.276295971312369e-05,
+      "loss": 3.5441,
+      "step": 3495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.271530257175797e-05,
+      "loss": 3.5119,
+      "step": 3500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.266759340283468e-05,
+      "loss": 3.4918,
+      "step": 3505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.261983228222627e-05,
+      "loss": 3.3821,
+      "step": 3510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.257201928588786e-05,
+      "loss": 3.3782,
+      "step": 3515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.25241544898571e-05,
+      "loss": 3.3227,
+      "step": 3520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.247623797025392e-05,
+      "loss": 3.4454,
+      "step": 3525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.24282698032806e-05,
+      "loss": 3.3875,
+      "step": 3530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.238025006522151e-05,
+      "loss": 3.4055,
+      "step": 3535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.233217883244305e-05,
+      "loss": 3.3418,
+      "step": 3540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.228405618139348e-05,
+      "loss": 3.4406,
+      "step": 3545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.223588218860289e-05,
+      "loss": 3.4968,
+      "step": 3550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.218765693068298e-05,
+      "loss": 3.484,
+      "step": 3555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.213938048432697e-05,
+      "loss": 3.397,
+      "step": 3560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.209105292630952e-05,
+      "loss": 3.366,
+      "step": 3565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.204267433348654e-05,
+      "loss": 3.31,
+      "step": 3570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.199424478279512e-05,
+      "loss": 3.4176,
+      "step": 3575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.194576435125338e-05,
+      "loss": 3.4284,
+      "step": 3580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.189723311596038e-05,
+      "loss": 3.4067,
+      "step": 3585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.184865115409594e-05,
+      "loss": 3.3759,
+      "step": 3590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.180001854292056e-05,
+      "loss": 3.3052,
+      "step": 3595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.175133535977528e-05,
+      "loss": 3.3955,
+      "step": 3600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.170260168208162e-05,
+      "loss": 3.3783,
+      "step": 3605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.165381758734135e-05,
+      "loss": 3.4579,
+      "step": 3610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.16049831531364e-05,
+      "loss": 3.4481,
+      "step": 3615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.155609845712882e-05,
+      "loss": 3.3457,
+      "step": 3620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.150716357706052e-05,
+      "loss": 3.3214,
+      "step": 3625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.14581785907533e-05,
+      "loss": 3.4277,
+      "step": 3630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.140914357610857e-05,
+      "loss": 3.3314,
+      "step": 3635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.136005861110735e-05,
+      "loss": 3.3504,
+      "step": 3640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.131092377381007e-05,
+      "loss": 3.3851,
+      "step": 3645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.126173914235646e-05,
+      "loss": 3.4,
+      "step": 3650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.121250479496548e-05,
+      "loss": 3.421,
+      "step": 3655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.116322080993515e-05,
+      "loss": 3.3598,
+      "step": 3660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.111388726564238e-05,
+      "loss": 3.3592,
+      "step": 3665
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.106450424054294e-05,
+      "loss": 3.3246,
+      "step": 3670
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.101507181317128e-05,
+      "loss": 3.3881,
+      "step": 3675
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.096559006214039e-05,
+      "loss": 3.3278,
+      "step": 3680
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.091605906614174e-05,
+      "loss": 3.4759,
+      "step": 3685
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.086647890394511e-05,
+      "loss": 3.4235,
+      "step": 3690
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.081684965439844e-05,
+      "loss": 3.3102,
+      "step": 3695
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.076717139642775e-05,
+      "loss": 3.4678,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.071744420903699e-05,
+      "loss": 3.2851,
+      "step": 3705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.066766817130796e-05,
+      "loss": 3.345,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.06178433624001e-05,
+      "loss": 3.3317,
+      "step": 3715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.056796986155043e-05,
+      "loss": 3.3437,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.051804774807342e-05,
+      "loss": 3.3405,
+      "step": 3725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.046807710136081e-05,
+      "loss": 3.3683,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.041805800088156e-05,
+      "loss": 3.4042,
+      "step": 3735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.036799052618169e-05,
+      "loss": 3.3136,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.031787475688413e-05,
+      "loss": 3.3916,
+      "step": 3745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.02677107726886e-05,
+      "loss": 3.3481,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.021749865337153e-05,
+      "loss": 3.4524,
+      "step": 3755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.01672384787859e-05,
+      "loss": 3.4389,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.011693032886108e-05,
+      "loss": 3.4936,
+      "step": 3765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.006657428360276e-05,
+      "loss": 3.278,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.001617042309278e-05,
+      "loss": 3.3959,
+      "step": 3775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.996571882748905e-05,
+      "loss": 3.4236,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.991521957702537e-05,
+      "loss": 3.391,
+      "step": 3785
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.986467275201135e-05,
+      "loss": 3.341,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.98140784328322e-05,
+      "loss": 3.3802,
+      "step": 3795
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.976343669994876e-05,
+      "loss": 3.402,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.971274763389717e-05,
+      "loss": 3.4288,
+      "step": 3805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.966201131528893e-05,
+      "loss": 3.4788,
+      "step": 3810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.961122782481059e-05,
+      "loss": 3.3784,
+      "step": 3815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.956039724322382e-05,
+      "loss": 3.3513,
+      "step": 3820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.950951965136512e-05,
+      "loss": 3.3883,
+      "step": 3825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.945859513014578e-05,
+      "loss": 3.3876,
+      "step": 3830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.940762376055168e-05,
+      "loss": 3.4297,
+      "step": 3835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.935660562364324e-05,
+      "loss": 3.364,
+      "step": 3840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.930554080055526e-05,
+      "loss": 3.3797,
+      "step": 3845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.925442937249677e-05,
+      "loss": 3.2127,
+      "step": 3850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.920327142075089e-05,
+      "loss": 3.2899,
+      "step": 3855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.915206702667477e-05,
+      "loss": 3.2997,
+      "step": 3860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.910081627169943e-05,
+      "loss": 3.3108,
+      "step": 3865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.904951923732955e-05,
+      "loss": 3.3589,
+      "step": 3870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.899817600514346e-05,
+      "loss": 3.3179,
+      "step": 3875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.894678665679297e-05,
+      "loss": 3.3564,
+      "step": 3880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.889535127400318e-05,
+      "loss": 3.3114,
+      "step": 3885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.884386993857245e-05,
+      "loss": 3.3034,
+      "step": 3890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.879234273237219e-05,
+      "loss": 3.3161,
+      "step": 3895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.874076973734675e-05,
+      "loss": 3.2877,
+      "step": 3900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.868915103551332e-05,
+      "loss": 3.4732,
+      "step": 3905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.863748670896179e-05,
+      "loss": 3.3421,
+      "step": 3910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.858577683985455e-05,
+      "loss": 3.3957,
+      "step": 3915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.85340215104265e-05,
+      "loss": 3.3129,
+      "step": 3920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.848222080298477e-05,
+      "loss": 3.4425,
+      "step": 3925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.843037479990869e-05,
+      "loss": 3.4322,
+      "step": 3930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.837848358364959e-05,
+      "loss": 3.327,
+      "step": 3935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.832654723673072e-05,
+      "loss": 3.4104,
+      "step": 3940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.827456584174713e-05,
+      "loss": 3.2463,
+      "step": 3945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.822253948136549e-05,
+      "loss": 3.4628,
+      "step": 3950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.817046823832396e-05,
+      "loss": 3.3873,
+      "step": 3955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.811835219543209e-05,
+      "loss": 3.3031,
+      "step": 3960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.80661914355707e-05,
+      "loss": 3.3844,
+      "step": 3965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.801398604169168e-05,
+      "loss": 3.2495,
+      "step": 3970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.796173609681795e-05,
+      "loss": 3.4649,
+      "step": 3975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.790944168404324e-05,
+      "loss": 3.4311,
+      "step": 3980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.7857102886532e-05,
+      "loss": 3.3376,
+      "step": 3985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.78047197875193e-05,
+      "loss": 3.4527,
+      "step": 3990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.775229247031062e-05,
+      "loss": 3.3318,
+      "step": 3995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.769982101828182e-05,
+      "loss": 3.386,
+      "step": 4000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.764730551487888e-05,
+      "loss": 3.4073,
+      "step": 4005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.759474604361788e-05,
+      "loss": 3.3507,
+      "step": 4010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.75421426880848e-05,
+      "loss": 3.3164,
+      "step": 4015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.748949553193541e-05,
+      "loss": 3.3153,
+      "step": 4020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.743680465889517e-05,
+      "loss": 3.3825,
+      "step": 4025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.738407015275901e-05,
+      "loss": 3.2484,
+      "step": 4030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.733129209739131e-05,
+      "loss": 3.438,
+      "step": 4035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.727847057672566e-05,
+      "loss": 3.4041,
+      "step": 4040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.722560567476478e-05,
+      "loss": 3.3529,
+      "step": 4045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.717269747558042e-05,
+      "loss": 3.2576,
+      "step": 4050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.711974606331313e-05,
+      "loss": 3.3842,
+      "step": 4055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.70667515221722e-05,
+      "loss": 3.2844,
+      "step": 4060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.701371393643554e-05,
+      "loss": 3.4692,
+      "step": 4065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.696063339044947e-05,
+      "loss": 3.4288,
+      "step": 4070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.690750996862867e-05,
+      "loss": 3.417,
+      "step": 4075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.685434375545596e-05,
+      "loss": 3.3601,
+      "step": 4080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.680113483548228e-05,
+      "loss": 3.2945,
+      "step": 4085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.674788329332641e-05,
+      "loss": 3.3622,
+      "step": 4090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.669458921367494e-05,
+      "loss": 3.3491,
+      "step": 4095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.664125268128215e-05,
+      "loss": 3.3619,
+      "step": 4100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.658787378096978e-05,
+      "loss": 3.3342,
+      "step": 4105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.653445259762698e-05,
+      "loss": 3.4673,
+      "step": 4110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.648098921621011e-05,
+      "loss": 3.3552,
+      "step": 4115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.642748372174267e-05,
+      "loss": 3.318,
+      "step": 4120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.637393619931512e-05,
+      "loss": 3.2206,
+      "step": 4125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.632034673408475e-05,
+      "loss": 3.2604,
+      "step": 4130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.62667154112756e-05,
+      "loss": 3.4387,
+      "step": 4135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.621304231617819e-05,
+      "loss": 3.3651,
+      "step": 4140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.615932753414952e-05,
+      "loss": 3.3741,
+      "step": 4145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.610557115061292e-05,
+      "loss": 3.4008,
+      "step": 4150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.605177325105784e-05,
+      "loss": 3.4082,
+      "step": 4155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.599793392103972e-05,
+      "loss": 3.487,
+      "step": 4160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.594405324617996e-05,
+      "loss": 3.4425,
+      "step": 4165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.589013131216567e-05,
+      "loss": 3.3718,
+      "step": 4170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.583616820474955e-05,
+      "loss": 3.2864,
+      "step": 4175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.578216400974986e-05,
+      "loss": 3.3821,
+      "step": 4180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.572811881305011e-05,
+      "loss": 3.4328,
+      "step": 4185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.567403270059907e-05,
+      "loss": 3.2414,
+      "step": 4190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.561990575841054e-05,
+      "loss": 3.2998,
+      "step": 4195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.556573807256335e-05,
+      "loss": 3.3135,
+      "step": 4200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.551152972920098e-05,
+      "loss": 3.2735,
+      "step": 4205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.545728081453167e-05,
+      "loss": 3.37,
+      "step": 4210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.540299141482817e-05,
+      "loss": 3.3561,
+      "step": 4215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.534866161642755e-05,
+      "loss": 3.3509,
+      "step": 4220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.529429150573122e-05,
+      "loss": 3.2985,
+      "step": 4225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.523988116920462e-05,
+      "loss": 3.3036,
+      "step": 4230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.51854306933772e-05,
+      "loss": 3.3956,
+      "step": 4235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.513094016484224e-05,
+      "loss": 3.3001,
+      "step": 4240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.507640967025671e-05,
+      "loss": 3.4428,
+      "step": 4245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.502183929634115e-05,
+      "loss": 3.4523,
+      "step": 4250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.496722912987953e-05,
+      "loss": 3.3106,
+      "step": 4255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.491257925771904e-05,
+      "loss": 3.3829,
+      "step": 4260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.485788976677009e-05,
+      "loss": 3.2739,
+      "step": 4265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.480316074400606e-05,
+      "loss": 3.4135,
+      "step": 4270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.47483922764632e-05,
+      "loss": 3.4687,
+      "step": 4275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.469358445124051e-05,
+      "loss": 3.2421,
+      "step": 4280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.463873735549953e-05,
+      "loss": 3.2255,
+      "step": 4285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.458385107646432e-05,
+      "loss": 3.4038,
+      "step": 4290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.452892570142121e-05,
+      "loss": 3.3514,
+      "step": 4295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.447396131771872e-05,
+      "loss": 3.3292,
+      "step": 4300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.441895801276739e-05,
+      "loss": 3.3903,
+      "step": 4305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.436391587403967e-05,
+      "loss": 3.4055,
+      "step": 4310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.43088349890698e-05,
+      "loss": 3.2701,
+      "step": 4315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.425371544545355e-05,
+      "loss": 3.3066,
+      "step": 4320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.419855733084826e-05,
+      "loss": 3.3826,
+      "step": 4325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.414336073297256e-05,
+      "loss": 3.4043,
+      "step": 4330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.408812573960631e-05,
+      "loss": 3.4217,
+      "step": 4335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.403285243859038e-05,
+      "loss": 3.3543,
+      "step": 4340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.397754091782663e-05,
+      "loss": 3.3894,
+      "step": 4345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.392219126527767e-05,
+      "loss": 3.2913,
+      "step": 4350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.386680356896674e-05,
+      "loss": 3.4379,
+      "step": 4355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.381137791697758e-05,
+      "loss": 3.3444,
+      "step": 4360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.375591439745431e-05,
+      "loss": 3.3316,
+      "step": 4365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.370041309860127e-05,
+      "loss": 3.4055,
+      "step": 4370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.364487410868286e-05,
+      "loss": 3.3507,
+      "step": 4375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.358929751602346e-05,
+      "loss": 3.2947,
+      "step": 4380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.353368340900719e-05,
+      "loss": 3.298,
+      "step": 4385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.347803187607788e-05,
+      "loss": 3.3415,
+      "step": 4390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.342234300573889e-05,
+      "loss": 3.3755,
+      "step": 4395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.33666168865529e-05,
+      "loss": 3.3226,
+      "step": 4400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.331085360714187e-05,
+      "loss": 3.4361,
+      "step": 4405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.325505325618683e-05,
+      "loss": 3.3453,
+      "step": 4410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.319921592242781e-05,
+      "loss": 3.2739,
+      "step": 4415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.31433416946636e-05,
+      "loss": 3.3647,
+      "step": 4420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.308743066175172e-05,
+      "loss": 3.2863,
+      "step": 4425
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.303148291260814e-05,
+      "loss": 3.3461,
+      "step": 4430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.29754985362073e-05,
+      "loss": 3.3515,
+      "step": 4435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.291947762158185e-05,
+      "loss": 3.3732,
+      "step": 4440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.286342025782257e-05,
+      "loss": 3.266,
+      "step": 4445
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.280732653407817e-05,
+      "loss": 3.3425,
+      "step": 4450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.275119653955523e-05,
+      "loss": 3.4323,
+      "step": 4455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.269503036351798e-05,
+      "loss": 3.2311,
+      "step": 4460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.263882809528817e-05,
+      "loss": 3.2801,
+      "step": 4465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.258258982424503e-05,
+      "loss": 3.3488,
+      "step": 4470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.252631563982492e-05,
+      "loss": 3.3676,
+      "step": 4475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.247000563152146e-05,
+      "loss": 3.3486,
+      "step": 4480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.241365988888508e-05,
+      "loss": 3.2963,
+      "step": 4485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.23572785015232e-05,
+      "loss": 3.3499,
+      "step": 4490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.23008615590998e-05,
+      "loss": 3.2497,
+      "step": 4495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.22444091513355e-05,
+      "loss": 3.4645,
+      "step": 4500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.218792136800724e-05,
+      "loss": 3.4265,
+      "step": 4505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.213139829894826e-05,
+      "loss": 3.3707,
+      "step": 4510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.207484003404793e-05,
+      "loss": 3.2959,
+      "step": 4515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.201824666325151e-05,
+      "loss": 3.2474,
+      "step": 4520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.196161827656022e-05,
+      "loss": 3.4023,
+      "step": 4525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.190495496403085e-05,
+      "loss": 3.3381,
+      "step": 4530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.184825681577582e-05,
+      "loss": 3.361,
+      "step": 4535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.179152392196284e-05,
+      "loss": 3.3721,
+      "step": 4540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.173475637281504e-05,
+      "loss": 3.2727,
+      "step": 4545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.16779542586105e-05,
+      "loss": 3.4205,
+      "step": 4550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.162111766968236e-05,
+      "loss": 3.348,
+      "step": 4555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.156424669641856e-05,
+      "loss": 3.426,
+      "step": 4560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.150734142926173e-05,
+      "loss": 3.3722,
+      "step": 4565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.145040195870902e-05,
+      "loss": 3.3655,
+      "step": 4570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1393428375312e-05,
+      "loss": 3.3602,
+      "step": 4575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.133642076967648e-05,
+      "loss": 3.3379,
+      "step": 4580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.127937923246233e-05,
+      "loss": 3.4339,
+      "step": 4585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.122230385438348e-05,
+      "loss": 3.2894,
+      "step": 4590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.116519472620761e-05,
+      "loss": 3.3456,
+      "step": 4595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.110805193875607e-05,
+      "loss": 3.342,
+      "step": 4600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.105087558290378e-05,
+      "loss": 3.3876,
+      "step": 4605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.099366574957898e-05,
+      "loss": 3.285,
+      "step": 4610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.093642252976321e-05,
+      "loss": 3.3045,
+      "step": 4615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.087914601449108e-05,
+      "loss": 3.3856,
+      "step": 4620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.082183629485018e-05,
+      "loss": 3.3605,
+      "step": 4625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.076449346198082e-05,
+      "loss": 3.3172,
+      "step": 4630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.070711760707608e-05,
+      "loss": 3.2294,
+      "step": 4635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.064970882138147e-05,
+      "loss": 3.3812,
+      "step": 4640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.059226719619494e-05,
+      "loss": 3.2673,
+      "step": 4645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.053479282286662e-05,
+      "loss": 3.319,
+      "step": 4650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.047728579279871e-05,
+      "loss": 3.2923,
+      "step": 4655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.041974619744538e-05,
+      "loss": 3.3043,
+      "step": 4660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.036217412831256e-05,
+      "loss": 3.3756,
+      "step": 4665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.030456967695785e-05,
+      "loss": 3.32,
+      "step": 4670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.024693293499033e-05,
+      "loss": 3.2614,
+      "step": 4675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.018926399407045e-05,
+      "loss": 3.3459,
+      "step": 4680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.013156294590981e-05,
+      "loss": 3.3456,
+      "step": 4685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.007382988227116e-05,
+      "loss": 3.3065,
+      "step": 4690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.001606489496808e-05,
+      "loss": 3.3681,
+      "step": 4695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.995826807586501e-05,
+      "loss": 3.2413,
+      "step": 4700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 6.99004395168769e-05,
+      "loss": 3.3469,
+      "step": 4705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.984257930996927e-05,
+      "loss": 3.2277,
+      "step": 4710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.978468754715793e-05,
+      "loss": 3.2716,
+      "step": 4715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.972676432050888e-05,
+      "loss": 3.2861,
+      "step": 4720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.966880972213813e-05,
+      "loss": 3.3404,
+      "step": 4725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.961082384421166e-05,
+      "loss": 3.4185,
+      "step": 4730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.95528067789451e-05,
+      "loss": 3.3534,
+      "step": 4735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.949475861860372e-05,
+      "loss": 3.3369,
+      "step": 4740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.943667945550227e-05,
+      "loss": 3.2669,
+      "step": 4745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.937856938200475e-05,
+      "loss": 3.2831,
+      "step": 4750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.932042849052433e-05,
+      "loss": 3.38,
+      "step": 4755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.926225687352324e-05,
+      "loss": 3.3631,
+      "step": 4760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.920405462351251e-05,
+      "loss": 3.2571,
+      "step": 4765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.914582183305192e-05,
+      "loss": 3.4537,
+      "step": 4770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.90875585947498e-05,
+      "loss": 3.2558,
+      "step": 4775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.902926500126292e-05,
+      "loss": 3.3742,
+      "step": 4780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.89709411452963e-05,
+      "loss": 3.3014,
+      "step": 4785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.891258711960312e-05,
+      "loss": 3.3252,
+      "step": 4790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.885420301698453e-05,
+      "loss": 3.4309,
+      "step": 4795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.879578893028947e-05,
+      "loss": 3.3445,
+      "step": 4800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.873734495241464e-05,
+      "loss": 3.3136,
+      "step": 4805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.86788711763042e-05,
+      "loss": 3.2953,
+      "step": 4810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.862036769494974e-05,
+      "loss": 3.4181,
+      "step": 4815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.856183460139008e-05,
+      "loss": 3.4764,
+      "step": 4820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.850327198871115e-05,
+      "loss": 3.3456,
+      "step": 4825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.844467995004582e-05,
+      "loss": 3.435,
+      "step": 4830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.838605857857372e-05,
+      "loss": 3.3972,
+      "step": 4835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.832740796752115e-05,
+      "loss": 3.3396,
+      "step": 4840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.826872821016093e-05,
+      "loss": 3.2642,
+      "step": 4845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.821001939981224e-05,
+      "loss": 3.3893,
+      "step": 4850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.81512816298404e-05,
+      "loss": 3.3482,
+      "step": 4855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.809251499365685e-05,
+      "loss": 3.3027,
+      "step": 4860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.803371958471891e-05,
+      "loss": 3.3825,
+      "step": 4865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.797489549652964e-05,
+      "loss": 3.3328,
+      "step": 4870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.791604282263775e-05,
+      "loss": 3.314,
+      "step": 4875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.785716165663737e-05,
+      "loss": 3.4199,
+      "step": 4880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.779825209216799e-05,
+      "loss": 3.3824,
+      "step": 4885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.773931422291419e-05,
+      "loss": 3.3915,
+      "step": 4890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.768034814260563e-05,
+      "loss": 3.3314,
+      "step": 4895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.76213539450168e-05,
+      "loss": 3.3272,
+      "step": 4900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.756233172396695e-05,
+      "loss": 3.3267,
+      "step": 4905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.750328157331982e-05,
+      "loss": 3.3655,
+      "step": 4910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.744420358698362e-05,
+      "loss": 3.2832,
+      "step": 4915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.73850978589108e-05,
+      "loss": 3.294,
+      "step": 4920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.732596448309798e-05,
+      "loss": 3.2728,
+      "step": 4925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.726680355358568e-05,
+      "loss": 3.2973,
+      "step": 4930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.72076151644583e-05,
+      "loss": 3.3613,
+      "step": 4935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.714839940984385e-05,
+      "loss": 3.2915,
+      "step": 4940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.708915638391391e-05,
+      "loss": 3.3351,
+      "step": 4945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.70298861808834e-05,
+      "loss": 3.3452,
+      "step": 4950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.697058889501048e-05,
+      "loss": 3.2676,
+      "step": 4955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.691126462059636e-05,
+      "loss": 3.2629,
+      "step": 4960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.685191345198518e-05,
+      "loss": 3.3358,
+      "step": 4965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.679253548356388e-05,
+      "loss": 3.3723,
+      "step": 4970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.673313080976195e-05,
+      "loss": 3.2911,
+      "step": 4975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.66736995250514e-05,
+      "loss": 3.4014,
+      "step": 4980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.661424172394656e-05,
+      "loss": 3.2229,
+      "step": 4985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.655475750100393e-05,
+      "loss": 3.2656,
+      "step": 4990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.649524695082201e-05,
+      "loss": 3.3531,
+      "step": 4995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.643571016804115e-05,
+      "loss": 3.3616,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.637614724734349e-05,
+      "loss": 3.363,
+      "step": 5005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.631655828345266e-05,
+      "loss": 3.3628,
+      "step": 5010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.625694337113376e-05,
+      "loss": 3.3642,
+      "step": 5015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.619730260519312e-05,
+      "loss": 3.3643,
+      "step": 5020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.613763608047823e-05,
+      "loss": 3.3568,
+      "step": 5025
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.607794389187746e-05,
+      "loss": 3.3362,
+      "step": 5030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.60182261343201e-05,
+      "loss": 3.449,
+      "step": 5035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.595848290277603e-05,
+      "loss": 3.3926,
+      "step": 5040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.589871429225566e-05,
+      "loss": 3.2945,
+      "step": 5045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.58389203978098e-05,
+      "loss": 3.3106,
+      "step": 5050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.577910131452937e-05,
+      "loss": 3.3874,
+      "step": 5055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.571925713754547e-05,
+      "loss": 3.3205,
+      "step": 5060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.5659387962029e-05,
+      "loss": 3.3353,
+      "step": 5065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.559949388319071e-05,
+      "loss": 3.4095,
+      "step": 5070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.553957499628088e-05,
+      "loss": 3.2138,
+      "step": 5075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.547963139658927e-05,
+      "loss": 3.316,
+      "step": 5080
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.541966317944494e-05,
+      "loss": 3.3345,
+      "step": 5085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.535967044021613e-05,
+      "loss": 3.2767,
+      "step": 5090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.529965327431001e-05,
+      "loss": 3.3363,
+      "step": 5095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.523961177717267e-05,
+      "loss": 3.313,
+      "step": 5100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.517954604428883e-05,
+      "loss": 3.2976,
+      "step": 5105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.511945617118177e-05,
+      "loss": 3.2353,
+      "step": 5110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.505934225341321e-05,
+      "loss": 3.4176,
+      "step": 5115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.499920438658305e-05,
+      "loss": 3.2884,
+      "step": 5120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.493904266632931e-05,
+      "loss": 3.2146,
+      "step": 5125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.487885718832791e-05,
+      "loss": 3.2208,
+      "step": 5130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.481864804829259e-05,
+      "loss": 3.4005,
+      "step": 5135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.47584153419747e-05,
+      "loss": 3.2824,
+      "step": 5140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.469815916516306e-05,
+      "loss": 3.2777,
+      "step": 5145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.463787961368385e-05,
+      "loss": 3.3291,
+      "step": 5150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.457757678340037e-05,
+      "loss": 3.328,
+      "step": 5155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.4517250770213e-05,
+      "loss": 3.3599,
+      "step": 5160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.445690167005895e-05,
+      "loss": 3.2707,
+      "step": 5165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.439652957891213e-05,
+      "loss": 3.2722,
+      "step": 5170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.433613459278307e-05,
+      "loss": 3.26,
+      "step": 5175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.427571680771865e-05,
+      "loss": 3.4188,
+      "step": 5180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.421527631980205e-05,
+      "loss": 3.3822,
+      "step": 5185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.415481322515254e-05,
+      "loss": 3.4285,
+      "step": 5190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.409432761992533e-05,
+      "loss": 3.2581,
+      "step": 5195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.403381960031144e-05,
+      "loss": 3.2518,
+      "step": 5200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.397328926253757e-05,
+      "loss": 3.3204,
+      "step": 5205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.39127367028658e-05,
+      "loss": 3.3743,
+      "step": 5210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.385216201759372e-05,
+      "loss": 3.3839,
+      "step": 5215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.379156530305395e-05,
+      "loss": 3.3252,
+      "step": 5220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.373094665561424e-05,
+      "loss": 3.3594,
+      "step": 5225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.367030617167717e-05,
+      "loss": 3.2811,
+      "step": 5230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.360964394768007e-05,
+      "loss": 3.235,
+      "step": 5235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.354896008009486e-05,
+      "loss": 3.2488,
+      "step": 5240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.348825466542782e-05,
+      "loss": 3.2945,
+      "step": 5245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.342752780021957e-05,
+      "loss": 3.455,
+      "step": 5250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.33667795810448e-05,
+      "loss": 3.254,
+      "step": 5255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.330601010451216e-05,
+      "loss": 3.426,
+      "step": 5260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.32452194672641e-05,
+      "loss": 3.3492,
+      "step": 5265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.318440776597676e-05,
+      "loss": 3.3538,
+      "step": 5270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.312357509735975e-05,
+      "loss": 3.3777,
+      "step": 5275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.306272155815603e-05,
+      "loss": 3.3392,
+      "step": 5280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.300184724514172e-05,
+      "loss": 3.3146,
+      "step": 5285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.294095225512603e-05,
+      "loss": 3.2744,
+      "step": 5290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.288003668495106e-05,
+      "loss": 3.2361,
+      "step": 5295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.281910063149155e-05,
+      "loss": 3.356,
+      "step": 5300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.275814419165494e-05,
+      "loss": 3.2496,
+      "step": 5305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.269716746238094e-05,
+      "loss": 3.2057,
+      "step": 5310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.26361705406417e-05,
+      "loss": 3.3747,
+      "step": 5315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.25751535234413e-05,
+      "loss": 3.3452,
+      "step": 5320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.251411650781596e-05,
+      "loss": 3.2836,
+      "step": 5325
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.245305959083355e-05,
+      "loss": 3.3387,
+      "step": 5330
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.239198286959368e-05,
+      "loss": 3.3543,
+      "step": 5335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.233088644122744e-05,
+      "loss": 3.392,
+      "step": 5340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.226977040289721e-05,
+      "loss": 3.3302,
+      "step": 5345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.220863485179665e-05,
+      "loss": 3.3755,
+      "step": 5350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.214747988515033e-05,
+      "loss": 3.3226,
+      "step": 5355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.208630560021384e-05,
+      "loss": 3.3094,
+      "step": 5360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.202511209427335e-05,
+      "loss": 3.3615,
+      "step": 5365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.19638994646457e-05,
+      "loss": 3.4158,
+      "step": 5370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.19026678086781e-05,
+      "loss": 3.3426,
+      "step": 5375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.184141722374802e-05,
+      "loss": 3.2848,
+      "step": 5380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1780147807263e-05,
+      "loss": 3.4333,
+      "step": 5385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.171885965666062e-05,
+      "loss": 3.3069,
+      "step": 5390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.16575528694082e-05,
+      "loss": 3.293,
+      "step": 5395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.159622754300264e-05,
+      "loss": 3.2178,
+      "step": 5400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.153488377497043e-05,
+      "loss": 3.3165,
+      "step": 5405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.147352166286731e-05,
+      "loss": 3.3178,
+      "step": 5410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.141214130427824e-05,
+      "loss": 3.3582,
+      "step": 5415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.135074279681714e-05,
+      "loss": 3.3528,
+      "step": 5420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.128932623812687e-05,
+      "loss": 3.3861,
+      "step": 5425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.122789172587894e-05,
+      "loss": 3.317,
+      "step": 5430
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.116643935777342e-05,
+      "loss": 3.3699,
+      "step": 5435
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.110496923153881e-05,
+      "loss": 3.3406,
+      "step": 5440
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.104348144493179e-05,
+      "loss": 3.3044,
+      "step": 5445
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.09819760957372e-05,
+      "loss": 3.3848,
+      "step": 5450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0920453281767745e-05,
+      "loss": 3.2752,
+      "step": 5455
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.0858913100863934e-05,
+      "loss": 3.2643,
+      "step": 5460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0797355650893895e-05,
+      "loss": 3.269,
+      "step": 5465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0735781029753215e-05,
+      "loss": 3.2522,
+      "step": 5470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0674189335364785e-05,
+      "loss": 3.3882,
+      "step": 5475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0612580665678675e-05,
+      "loss": 3.2941,
+      "step": 5480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.055095511867189e-05,
+      "loss": 3.2898,
+      "step": 5485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.048931279234833e-05,
+      "loss": 3.3287,
+      "step": 5490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.042765378473859e-05,
+      "loss": 3.289,
+      "step": 5495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.036597819389972e-05,
+      "loss": 3.2634,
+      "step": 5500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.030428611791523e-05,
+      "loss": 3.2734,
+      "step": 5505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0242577654894774e-05,
+      "loss": 3.3155,
+      "step": 5510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0180852902974114e-05,
+      "loss": 3.2568,
+      "step": 5515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.011911196031489e-05,
+      "loss": 3.2582,
+      "step": 5520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0057354925104516e-05,
+      "loss": 3.2986,
+      "step": 5525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.999558189555596e-05,
+      "loss": 3.1828,
+      "step": 5530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.993379296990766e-05,
+      "loss": 3.301,
+      "step": 5535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.987198824642335e-05,
+      "loss": 3.3812,
+      "step": 5540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.981016782339182e-05,
+      "loss": 3.1721,
+      "step": 5545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9748331799126875e-05,
+      "loss": 3.2441,
+      "step": 5550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.968648027196714e-05,
+      "loss": 3.2217,
+      "step": 5555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9624613340275856e-05,
+      "loss": 3.2836,
+      "step": 5560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.956273110244079e-05,
+      "loss": 3.3632,
+      "step": 5565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.950083365687408e-05,
+      "loss": 3.3503,
+      "step": 5570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.943892110201195e-05,
+      "loss": 3.3494,
+      "step": 5575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.937699353631475e-05,
+      "loss": 3.329,
+      "step": 5580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.931505105826667e-05,
+      "loss": 3.243,
+      "step": 5585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.92530937663756e-05,
+      "loss": 3.2884,
+      "step": 5590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.919112175917303e-05,
+      "loss": 3.312,
+      "step": 5595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.912913513521378e-05,
+      "loss": 3.2823,
+      "step": 5600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.906713399307598e-05,
+      "loss": 3.2392,
+      "step": 5605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.900511843136084e-05,
+      "loss": 3.3705,
+      "step": 5610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.894308854869247e-05,
+      "loss": 3.3423,
+      "step": 5615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.888104444371777e-05,
+      "loss": 3.2347,
+      "step": 5620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.881898621510629e-05,
+      "loss": 3.3084,
+      "step": 5625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8756913961549974e-05,
+      "loss": 3.2612,
+      "step": 5630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.869482778176313e-05,
+      "loss": 3.2981,
+      "step": 5635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8632727774482196e-05,
+      "loss": 3.2577,
+      "step": 5640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8570614038465585e-05,
+      "loss": 3.3228,
+      "step": 5645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.850848667249357e-05,
+      "loss": 3.3351,
+      "step": 5650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.844634577536805e-05,
+      "loss": 3.3238,
+      "step": 5655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8384191445912504e-05,
+      "loss": 3.239,
+      "step": 5660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8322023782971735e-05,
+      "loss": 3.3894,
+      "step": 5665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8259842885411775e-05,
+      "loss": 3.4084,
+      "step": 5670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8197648852119656e-05,
+      "loss": 3.2715,
+      "step": 5675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813544178200335e-05,
+      "loss": 3.2753,
+      "step": 5680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.807322177399156e-05,
+      "loss": 3.2748,
+      "step": 5685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8010988927033526e-05,
+      "loss": 3.3901,
+      "step": 5690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.794874334009894e-05,
+      "loss": 3.3295,
+      "step": 5695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.788648511217774e-05,
+      "loss": 3.3265,
+      "step": 5700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.7824214342279955e-05,
+      "loss": 3.332,
+      "step": 5705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.77619311294356e-05,
+      "loss": 3.1945,
+      "step": 5710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.769963557269444e-05,
+      "loss": 3.312,
+      "step": 5715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.763732777112587e-05,
+      "loss": 3.3854,
+      "step": 5720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7575007823818806e-05,
+      "loss": 3.3435,
+      "step": 5725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7512675829881405e-05,
+      "loss": 3.3568,
+      "step": 5730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.745033188844105e-05,
+      "loss": 3.2786,
+      "step": 5735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.738797609864408e-05,
+      "loss": 3.3574,
+      "step": 5740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7325608559655696e-05,
+      "loss": 3.3764,
+      "step": 5745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.72632293706598e-05,
+      "loss": 3.3233,
+      "step": 5750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.720083863085877e-05,
+      "loss": 3.2363,
+      "step": 5755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.713843643947341e-05,
+      "loss": 3.2708,
+      "step": 5760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.707602289574269e-05,
+      "loss": 3.3484,
+      "step": 5765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.701359809892367e-05,
+      "loss": 3.2699,
+      "step": 5770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.695116214829128e-05,
+      "loss": 3.3105,
+      "step": 5775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6888715143138204e-05,
+      "loss": 3.3329,
+      "step": 5780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.682625718277471e-05,
+      "loss": 3.3854,
+      "step": 5785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.676378836652846e-05,
+      "loss": 3.2137,
+      "step": 5790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6701308793744425e-05,
+      "loss": 3.281,
+      "step": 5795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6638818563784614e-05,
+      "loss": 3.3314,
+      "step": 5800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6576317776028075e-05,
+      "loss": 3.4054,
+      "step": 5805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.651380652987056e-05,
+      "loss": 3.3177,
+      "step": 5810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6451284924724526e-05,
+      "loss": 3.2198,
+      "step": 5815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6388753060018826e-05,
+      "loss": 3.267,
+      "step": 5820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6326211035198704e-05,
+      "loss": 3.2733,
+      "step": 5825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6263658949725514e-05,
+      "loss": 3.3151,
+      "step": 5830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6201096903076634e-05,
+      "loss": 3.3414,
+      "step": 5835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.613852499474528e-05,
+      "loss": 3.2863,
+      "step": 5840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.607594332424032e-05,
+      "loss": 3.2711,
+      "step": 5845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.60133519910862e-05,
+      "loss": 3.2744,
+      "step": 5850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5950751094822695e-05,
+      "loss": 3.3517,
+      "step": 5855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5888140735004804e-05,
+      "loss": 3.3615,
+      "step": 5860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.582552101120255e-05,
+      "loss": 3.296,
+      "step": 5865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.57628920230009e-05,
+      "loss": 3.2852,
+      "step": 5870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.57002538699995e-05,
+      "loss": 3.3754,
+      "step": 5875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.563760665181259e-05,
+      "loss": 3.3331,
+      "step": 5880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5574950468068854e-05,
+      "loss": 3.3363,
+      "step": 5885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.551228541841118e-05,
+      "loss": 3.3004,
+      "step": 5890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.544961160249661e-05,
+      "loss": 3.2633,
+      "step": 5895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.538692911999609e-05,
+      "loss": 3.3281,
+      "step": 5900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.532423807059437e-05,
+      "loss": 3.3404,
+      "step": 5905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5261538553989787e-05,
+      "loss": 3.3965,
+      "step": 5910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.519883066989422e-05,
+      "loss": 3.2314,
+      "step": 5915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.513611451803277e-05,
+      "loss": 3.208,
+      "step": 5920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.507339019814374e-05,
+      "loss": 3.3816,
+      "step": 5925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.501065780997838e-05,
+      "loss": 3.3146,
+      "step": 5930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.494791745330083e-05,
+      "loss": 3.2916,
+      "step": 5935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.488516922788787e-05,
+      "loss": 3.2998,
+      "step": 5940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4822413233528746e-05,
+      "loss": 3.3856,
+      "step": 5945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.4759649570025165e-05,
+      "loss": 3.4089,
+      "step": 5950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.469687833719093e-05,
+      "loss": 3.3318,
+      "step": 5955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.463409963485194e-05,
+      "loss": 3.213,
+      "step": 5960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.457131356284595e-05,
+      "loss": 3.3121,
+      "step": 5965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.450852022102246e-05,
+      "loss": 3.283,
+      "step": 5970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4445719709242484e-05,
+      "loss": 3.3776,
+      "step": 5975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.438291212737847e-05,
+      "loss": 3.3398,
+      "step": 5980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4320097575314136e-05,
+      "loss": 3.2639,
+      "step": 5985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.42572761529442e-05,
+      "loss": 3.3952,
+      "step": 5990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.419444796017443e-05,
+      "loss": 3.2606,
+      "step": 5995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.413161309692124e-05,
+      "loss": 3.2891,
+      "step": 6000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.406877166311173e-05,
+      "loss": 3.2596,
+      "step": 6005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4005923758683406e-05,
+      "loss": 3.4417,
+      "step": 6010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.39430694835841e-05,
+      "loss": 3.27,
+      "step": 6015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.388020893777174e-05,
+      "loss": 3.2345,
+      "step": 6020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3817342221214254e-05,
+      "loss": 3.3165,
+      "step": 6025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.375446943388935e-05,
+      "loss": 3.3617,
+      "step": 6030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.369159067578445e-05,
+      "loss": 3.2816,
+      "step": 6035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3628706046896426e-05,
+      "loss": 3.2987,
+      "step": 6040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.356581564723148e-05,
+      "loss": 3.2887,
+      "step": 6045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.350291957680501e-05,
+      "loss": 3.2825,
+      "step": 6050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.344001793564144e-05,
+      "loss": 3.3326,
+      "step": 6055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3377110823774034e-05,
+      "loss": 3.2861,
+      "step": 6060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3314198341244746e-05,
+      "loss": 3.3047,
+      "step": 6065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3251280588104125e-05,
+      "loss": 3.2728,
+      "step": 6070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3188357664411034e-05,
+      "loss": 3.3102,
+      "step": 6075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.312542967023262e-05,
+      "loss": 3.221,
+      "step": 6080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.306249670564404e-05,
+      "loss": 3.2518,
+      "step": 6085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2999558870728394e-05,
+      "loss": 3.2871,
+      "step": 6090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.293661626557651e-05,
+      "loss": 3.2977,
+      "step": 6095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.287366899028679e-05,
+      "loss": 3.3596,
+      "step": 6100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2810717144965114e-05,
+      "loss": 3.2744,
+      "step": 6105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2747760829724555e-05,
+      "loss": 3.1845,
+      "step": 6110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2684800144685364e-05,
+      "loss": 3.2781,
+      "step": 6115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.262183518997469e-05,
+      "loss": 3.3123,
+      "step": 6120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.25588660657265e-05,
+      "loss": 3.2958,
+      "step": 6125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.24958928720814e-05,
+      "loss": 3.3479,
+      "step": 6130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2432915709186424e-05,
+      "loss": 3.3515,
+      "step": 6135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.236993467719499e-05,
+      "loss": 3.3486,
+      "step": 6140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.230694987626656e-05,
+      "loss": 3.3138,
+      "step": 6145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.224396140656673e-05,
+      "loss": 3.1646,
+      "step": 6150
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.218096936826681e-05,
+      "loss": 3.2474,
+      "step": 6155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2117973861543846e-05,
+      "loss": 3.3066,
+      "step": 6160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.205497498658036e-05,
+      "loss": 3.264,
+      "step": 6165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.199197284356431e-05,
+      "loss": 3.2532,
+      "step": 6170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.192896753268874e-05,
+      "loss": 3.3082,
+      "step": 6175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.186595915415183e-05,
+      "loss": 3.2006,
+      "step": 6180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.180294780815659e-05,
+      "loss": 3.1348,
+      "step": 6185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.173993359491073e-05,
+      "loss": 3.3678,
+      "step": 6190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.167691661462658e-05,
+      "loss": 3.217,
+      "step": 6195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.161389696752082e-05,
+      "loss": 3.3111,
+      "step": 6200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.155087475381441e-05,
+      "loss": 3.264,
+      "step": 6205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.148785007373232e-05,
+      "loss": 3.3099,
+      "step": 6210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1424823027503545e-05,
+      "loss": 3.2771,
+      "step": 6215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.136179371536076e-05,
+      "loss": 3.3142,
+      "step": 6220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.129876223754029e-05,
+      "loss": 3.2615,
+      "step": 6225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.123572869428188e-05,
+      "loss": 3.3727,
+      "step": 6230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.117269318582856e-05,
+      "loss": 3.3317,
+      "step": 6235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1109655812426515e-05,
+      "loss": 3.272,
+      "step": 6240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1046616674324844e-05,
+      "loss": 3.2934,
+      "step": 6245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0983575871775514e-05,
+      "loss": 3.2374,
+      "step": 6250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.092053350503307e-05,
+      "loss": 3.264,
+      "step": 6255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.085748967435463e-05,
+      "loss": 3.3279,
+      "step": 6260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.079444447999955e-05,
+      "loss": 3.3059,
+      "step": 6265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.073139802222941e-05,
+      "loss": 3.1654,
+      "step": 6270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0668350401307786e-05,
+      "loss": 3.2516,
+      "step": 6275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.06053017175001e-05,
+      "loss": 3.2396,
+      "step": 6280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.054225207107348e-05,
+      "loss": 3.343,
+      "step": 6285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0479201562296566e-05,
+      "loss": 3.3333,
+      "step": 6290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.041615029143937e-05,
+      "loss": 3.2847,
+      "step": 6295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.035309835877312e-05,
+      "loss": 3.2608,
+      "step": 6300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.029004586457011e-05,
+      "loss": 3.3348,
+      "step": 6305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.022699290910351e-05,
+      "loss": 3.2119,
+      "step": 6310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0163939592647224e-05,
+      "loss": 3.3146,
+      "step": 6315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.010088601547575e-05,
+      "loss": 3.3513,
+      "step": 6320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.003783227786396e-05,
+      "loss": 3.2127,
+      "step": 6325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9974778480087034e-05,
+      "loss": 3.3516,
+      "step": 6330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.991172472242021e-05,
+      "loss": 3.4253,
+      "step": 6335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.984867110513866e-05,
+      "loss": 3.2655,
+      "step": 6340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.978561772851737e-05,
+      "loss": 3.2592,
+      "step": 6345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.972256469283089e-05,
+      "loss": 3.2952,
+      "step": 6350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.965951209835328e-05,
+      "loss": 3.2121,
+      "step": 6355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.959646004535787e-05,
+      "loss": 3.1262,
+      "step": 6360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9533408634117126e-05,
+      "loss": 3.4554,
+      "step": 6365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.947035796490252e-05,
+      "loss": 3.2466,
+      "step": 6370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.940730813798429e-05,
+      "loss": 3.1519,
+      "step": 6375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93442592536314e-05,
+      "loss": 3.2149,
+      "step": 6380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.928121141211127e-05,
+      "loss": 3.2975,
+      "step": 6385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.92181647136897e-05,
+      "loss": 3.3339,
+      "step": 6390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915511925863061e-05,
+      "loss": 3.103,
+      "step": 6395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9092075147196005e-05,
+      "loss": 3.4425,
+      "step": 6400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.902903247964573e-05,
+      "loss": 3.3147,
+      "step": 6405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8965991356237325e-05,
+      "loss": 3.3728,
+      "step": 6410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8902951877225886e-05,
+      "loss": 3.2625,
+      "step": 6415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.883991414286389e-05,
+      "loss": 3.254,
+      "step": 6420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.877687825340104e-05,
+      "loss": 3.3388,
+      "step": 6425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.87138443090841e-05,
+      "loss": 3.2628,
+      "step": 6430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.865081241015676e-05,
+      "loss": 3.3439,
+      "step": 6435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8587782656859414e-05,
+      "loss": 3.3653,
+      "step": 6440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8524755149429104e-05,
+      "loss": 3.3219,
+      "step": 6445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8461729988099254e-05,
+      "loss": 3.2079,
+      "step": 6450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.839870727309956e-05,
+      "loss": 3.2502,
+      "step": 6455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8335687104655864e-05,
+      "loss": 3.2648,
+      "step": 6460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.827266958298991e-05,
+      "loss": 3.1998,
+      "step": 6465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.820965480831928e-05,
+      "loss": 3.2575,
+      "step": 6470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.814664288085714e-05,
+      "loss": 3.2625,
+      "step": 6475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.808363390081217e-05,
+      "loss": 3.3456,
+      "step": 6480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8020627968388326e-05,
+      "loss": 3.3267,
+      "step": 6485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.795762518378476e-05,
+      "loss": 3.2276,
+      "step": 6490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.789462564719557e-05,
+      "loss": 3.3478,
+      "step": 6495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.783162945880972e-05,
+      "loss": 3.2905,
+      "step": 6500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.776863671881085e-05,
+      "loss": 3.2879,
+      "step": 6505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.770564752737709e-05,
+      "loss": 3.3863,
+      "step": 6510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.764266198468097e-05,
+      "loss": 3.3323,
+      "step": 6515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.757968019088917e-05,
+      "loss": 3.2962,
+      "step": 6520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7516702246162436e-05,
+      "loss": 3.237,
+      "step": 6525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7453728250655385e-05,
+      "loss": 3.3558,
+      "step": 6530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7390758304516366e-05,
+      "loss": 3.3581,
+      "step": 6535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.732779250788725e-05,
+      "loss": 3.3233,
+      "step": 6540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.726483096090338e-05,
+      "loss": 3.4287,
+      "step": 6545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.720187376369325e-05,
+      "loss": 3.2799,
+      "step": 6550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.713892101637851e-05,
+      "loss": 3.2997,
+      "step": 6555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.70759728190737e-05,
+      "loss": 3.2488,
+      "step": 6560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7013029271886136e-05,
+      "loss": 3.3185,
+      "step": 6565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.695009047491573e-05,
+      "loss": 3.2975,
+      "step": 6570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.6887156528254835e-05,
+      "loss": 3.2391,
+      "step": 6575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.682422753198812e-05,
+      "loss": 3.2682,
+      "step": 6580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.676130358619234e-05,
+      "loss": 3.3436,
+      "step": 6585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.669838479093625e-05,
+      "loss": 3.2187,
+      "step": 6590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.663547124628039e-05,
+      "loss": 3.3501,
+      "step": 6595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.657256305227696e-05,
+      "loss": 3.2933,
+      "step": 6600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6509660308969684e-05,
+      "loss": 3.1682,
+      "step": 6605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.644676311639356e-05,
+      "loss": 3.3349,
+      "step": 6610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.638387157457478e-05,
+      "loss": 3.2787,
+      "step": 6615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6320985783530574e-05,
+      "loss": 3.2679,
+      "step": 6620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6258105843269e-05,
+      "loss": 3.3825,
+      "step": 6625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6195231853788814e-05,
+      "loss": 3.2471,
+      "step": 6630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6132363915079316e-05,
+      "loss": 3.3569,
+      "step": 6635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.606950212712018e-05,
+      "loss": 3.3174,
+      "step": 6640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.60066465898813e-05,
+      "loss": 3.2425,
+      "step": 6645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.594379740332263e-05,
+      "loss": 3.2878,
+      "step": 6650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5880954667394014e-05,
+      "loss": 3.2971,
+      "step": 6655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.581811848203506e-05,
+      "loss": 3.2535,
+      "step": 6660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5755288947174935e-05,
+      "loss": 3.2619,
+      "step": 6665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.569246616273225e-05,
+      "loss": 3.3308,
+      "step": 6670
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5629650228614854e-05,
+      "loss": 3.2756,
+      "step": 6675
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5566841244719736e-05,
+      "loss": 3.3963,
+      "step": 6680
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.550403931093279e-05,
+      "loss": 3.1959,
+      "step": 6685
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.544124452712875e-05,
+      "loss": 3.337,
+      "step": 6690
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5378456993170924e-05,
+      "loss": 3.3612,
+      "step": 6695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.531567680891114e-05,
+      "loss": 3.2293,
+      "step": 6700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.525290407418951e-05,
+      "loss": 3.2185,
+      "step": 6705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5190138888834284e-05,
+      "loss": 3.2742,
+      "step": 6710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.512738135266176e-05,
+      "loss": 3.3035,
+      "step": 6715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.506463156547601e-05,
+      "loss": 3.251,
+      "step": 6720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.500188962706884e-05,
+      "loss": 3.343,
+      "step": 6725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.493915563721951e-05,
+      "loss": 3.2323,
+      "step": 6730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4876429695694704e-05,
+      "loss": 3.1565,
+      "step": 6735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.481371190224825e-05,
+      "loss": 3.182,
+      "step": 6740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4751002356621055e-05,
+      "loss": 3.3033,
+      "step": 6745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4688301158540925e-05,
+      "loss": 3.2375,
+      "step": 6750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.462560840772232e-05,
+      "loss": 3.33,
+      "step": 6755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.456292420386635e-05,
+      "loss": 3.3277,
+      "step": 6760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4500248646660455e-05,
+      "loss": 3.1499,
+      "step": 6765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.443758183577841e-05,
+      "loss": 3.2964,
+      "step": 6770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4374923870879985e-05,
+      "loss": 3.3005,
+      "step": 6775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.431227485161097e-05,
+      "loss": 3.2783,
+      "step": 6780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.424963487760286e-05,
+      "loss": 3.3259,
+      "step": 6785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4187004048472816e-05,
+      "loss": 3.3946,
+      "step": 6790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4124382463823415e-05,
+      "loss": 3.2855,
+      "step": 6795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.406177022324256e-05,
+      "loss": 3.3309,
+      "step": 6800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.39991674263033e-05,
+      "loss": 3.2591,
+      "step": 6805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.393657417256363e-05,
+      "loss": 3.2998,
+      "step": 6810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3873990561566405e-05,
+      "loss": 3.3103,
+      "step": 6815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.3811416692839106e-05,
+      "loss": 3.2764,
+      "step": 6820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.374885266589377e-05,
+      "loss": 3.2246,
+      "step": 6825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.368629858022675e-05,
+      "loss": 3.3114,
+      "step": 6830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.36237545353186e-05,
+      "loss": 3.3416,
+      "step": 6835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.356122063063388e-05,
+      "loss": 3.2551,
+      "step": 6840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3498696965621056e-05,
+      "loss": 3.4208,
+      "step": 6845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3436183639712326e-05,
+      "loss": 3.2101,
+      "step": 6850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.337368075232338e-05,
+      "loss": 3.3047,
+      "step": 6855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.331118840285338e-05,
+      "loss": 3.1439,
+      "step": 6860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.324870669068467e-05,
+      "loss": 3.231,
+      "step": 6865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.318623571518273e-05,
+      "loss": 3.1587,
+      "step": 6870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3123775575695904e-05,
+      "loss": 3.3883,
+      "step": 6875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.306132637155537e-05,
+      "loss": 3.2532,
+      "step": 6880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.299888820207487e-05,
+      "loss": 3.2816,
+      "step": 6885
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.293646116655061e-05,
+      "loss": 3.2112,
+      "step": 6890
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.28740453642611e-05,
+      "loss": 3.2756,
+      "step": 6895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.281164089446695e-05,
+      "loss": 3.2755,
+      "step": 6900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.274924785641082e-05,
+      "loss": 3.3124,
+      "step": 6905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.26868663493171e-05,
+      "loss": 3.3087,
+      "step": 6910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2624496472391915e-05,
+      "loss": 3.2184,
+      "step": 6915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2562138324822845e-05,
+      "loss": 3.3022,
+      "step": 6920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.249979200577886e-05,
+      "loss": 3.1928,
+      "step": 6925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2437457614410064e-05,
+      "loss": 3.2555,
+      "step": 6930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.237513524984766e-05,
+      "loss": 3.3404,
+      "step": 6935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.231282501120366e-05,
+      "loss": 3.3323,
+      "step": 6940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.225052699757083e-05,
+      "loss": 3.3501,
+      "step": 6945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.218824130802251e-05,
+      "loss": 3.2505,
+      "step": 6950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.212596804161237e-05,
+      "loss": 3.3771,
+      "step": 6955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.206370729737442e-05,
+      "loss": 3.3208,
+      "step": 6960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.200145917432266e-05,
+      "loss": 3.1705,
+      "step": 6965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1939223771451096e-05,
+      "loss": 3.3011,
+      "step": 6970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.187700118773346e-05,
+      "loss": 3.2441,
+      "step": 6975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.181479152212311e-05,
+      "loss": 3.2963,
+      "step": 6980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.175259487355285e-05,
+      "loss": 3.2732,
+      "step": 6985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.169041134093482e-05,
+      "loss": 3.2555,
+      "step": 6990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1628241023160254e-05,
+      "loss": 3.2335,
+      "step": 6995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.156608401909939e-05,
+      "loss": 3.258,
+      "step": 7000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.150394042760132e-05,
+      "loss": 3.3028,
+      "step": 7005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.144181034749374e-05,
+      "loss": 3.3495,
+      "step": 7010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.137969387758294e-05,
+      "loss": 3.2122,
+      "step": 7015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.131759111665349e-05,
+      "loss": 3.2203,
+      "step": 7020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1255502163468215e-05,
+      "loss": 3.2809,
+      "step": 7025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.119342711676794e-05,
+      "loss": 3.2594,
+      "step": 7030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.113136607527142e-05,
+      "loss": 3.26,
+      "step": 7035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.106931913767508e-05,
+      "loss": 3.2743,
+      "step": 7040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.100728640265296e-05,
+      "loss": 3.268,
+      "step": 7045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.094526796885652e-05,
+      "loss": 3.266,
+      "step": 7050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.088326393491441e-05,
+      "loss": 3.2584,
+      "step": 7055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.082127439943247e-05,
+      "loss": 3.2751,
+      "step": 7060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.075929946099342e-05,
+      "loss": 3.3459,
+      "step": 7065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0697339218156786e-05,
+      "loss": 3.2955,
+      "step": 7070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063539376945872e-05,
+      "loss": 3.2314,
+      "step": 7075
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0573463213411854e-05,
+      "loss": 3.2562,
+      "step": 7080
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.051154764850512e-05,
+      "loss": 3.2312,
+      "step": 7085
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.044964717320364e-05,
+      "loss": 3.3068,
+      "step": 7090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.03877618859485e-05,
+      "loss": 3.4005,
+      "step": 7095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0325891885156655e-05,
+      "loss": 3.2406,
+      "step": 7100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.026403726922077e-05,
+      "loss": 3.3119,
+      "step": 7105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.020219813650898e-05,
+      "loss": 3.1184,
+      "step": 7110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0140374585364875e-05,
+      "loss": 3.3416,
+      "step": 7115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.00785667141072e-05,
+      "loss": 3.198,
+      "step": 7120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.001677462102981e-05,
+      "loss": 3.2264,
+      "step": 7125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9954998404401414e-05,
+      "loss": 3.3262,
+      "step": 7130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.989323816246556e-05,
+      "loss": 3.2663,
+      "step": 7135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.983149399344029e-05,
+      "loss": 3.2851,
+      "step": 7140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9769765995518146e-05,
+      "loss": 3.2586,
+      "step": 7145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.970805426686597e-05,
+      "loss": 3.3034,
+      "step": 7150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.964635890562466e-05,
+      "loss": 3.2202,
+      "step": 7155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.958468000990917e-05,
+      "loss": 3.2949,
+      "step": 7160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.952301767780817e-05,
+      "loss": 3.2887,
+      "step": 7165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.946137200738409e-05,
+      "loss": 3.2995,
+      "step": 7170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9399743096672784e-05,
+      "loss": 3.2535,
+      "step": 7175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9338131043683515e-05,
+      "loss": 3.2665,
+      "step": 7180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.927653594639867e-05,
+      "loss": 3.2938,
+      "step": 7185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.921495790277375e-05,
+      "loss": 3.3098,
+      "step": 7190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.915339701073706e-05,
+      "loss": 3.2817,
+      "step": 7195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.909185336818966e-05,
+      "loss": 3.181,
+      "step": 7200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.903032707300521e-05,
+      "loss": 3.3426,
+      "step": 7205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.896881822302973e-05,
+      "loss": 3.2384,
+      "step": 7210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.8907326916081525e-05,
+      "loss": 3.3,
+      "step": 7215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8845853249950985e-05,
+      "loss": 3.363,
+      "step": 7220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.878439732240047e-05,
+      "loss": 3.3352,
+      "step": 7225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8722959231164103e-05,
+      "loss": 3.2482,
+      "step": 7230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8661539073947666e-05,
+      "loss": 3.2452,
+      "step": 7235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.860013694842839e-05,
+      "loss": 3.3039,
+      "step": 7240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.853875295225486e-05,
+      "loss": 3.3155,
+      "step": 7245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.847738718304683e-05,
+      "loss": 3.2994,
+      "step": 7250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8416039738395035e-05,
+      "loss": 3.3456,
+      "step": 7255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.83547107158611e-05,
+      "loss": 3.2594,
+      "step": 7260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8293400212977336e-05,
+      "loss": 3.2979,
+      "step": 7265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.823210832724662e-05,
+      "loss": 3.3034,
+      "step": 7270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.817083515614219e-05,
+      "loss": 3.2788,
+      "step": 7275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8109580797107555e-05,
+      "loss": 3.352,
+      "step": 7280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.804834534755628e-05,
+      "loss": 3.2263,
+      "step": 7285
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7987128904871885e-05,
+      "loss": 3.244,
+      "step": 7290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.792593156640766e-05,
+      "loss": 3.2882,
+      "step": 7295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.786475342948647e-05,
+      "loss": 3.3186,
+      "step": 7300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.780359459140071e-05,
+      "loss": 3.295,
+      "step": 7305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.774245514941203e-05,
+      "loss": 3.2737,
+      "step": 7310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.768133520075126e-05,
+      "loss": 3.3136,
+      "step": 7315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.762023484261822e-05,
+      "loss": 3.2782,
+      "step": 7320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.755915417218159e-05,
+      "loss": 3.2433,
+      "step": 7325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7498093286578704e-05,
+      "loss": 3.3227,
+      "step": 7330
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.74370522829155e-05,
+      "loss": 3.2516,
+      "step": 7335
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.737603125826621e-05,
+      "loss": 3.3203,
+      "step": 7340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.731503030967336e-05,
+      "loss": 3.272,
+      "step": 7345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7254049534147535e-05,
+      "loss": 3.3364,
+      "step": 7350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7193089028667205e-05,
+      "loss": 3.2643,
+      "step": 7355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7132148890178665e-05,
+      "loss": 3.2189,
+      "step": 7360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7071229215595734e-05,
+      "loss": 3.197,
+      "step": 7365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.70103301017998e-05,
+      "loss": 3.2003,
+      "step": 7370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6949451645639434e-05,
+      "loss": 3.183,
+      "step": 7375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.688859394393044e-05,
+      "loss": 3.273,
+      "step": 7380
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.682775709345558e-05,
+      "loss": 3.3454,
+      "step": 7385
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.676694119096446e-05,
+      "loss": 3.2857,
+      "step": 7390
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.67061463331734e-05,
+      "loss": 3.2489,
+      "step": 7395
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.664537261676518e-05,
+      "loss": 3.3098,
+      "step": 7400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.658462013838906e-05,
+      "loss": 3.2934,
+      "step": 7405
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6523888994660434e-05,
+      "loss": 3.2328,
+      "step": 7410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.646317928216082e-05,
+      "loss": 3.2655,
+      "step": 7415
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6402491097437627e-05,
+      "loss": 3.3121,
+      "step": 7420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6341824537004066e-05,
+      "loss": 3.2981,
+      "step": 7425
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.62811796973389e-05,
+      "loss": 3.2749,
+      "step": 7430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.62205566748864e-05,
+      "loss": 3.2222,
+      "step": 7435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.61599555660561e-05,
+      "loss": 3.3513,
+      "step": 7440
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6099376467222726e-05,
+      "loss": 3.3243,
+      "step": 7445
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6038819474725995e-05,
+      "loss": 3.3609,
+      "step": 7450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.5978284684870424e-05,
+      "loss": 3.3018,
+      "step": 7455
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.591777219392527e-05,
+      "loss": 3.3091,
+      "step": 7460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.58572820981243e-05,
+      "loss": 3.2193,
+      "step": 7465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.579681449366569e-05,
+      "loss": 3.1175,
+      "step": 7470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5736369476711804e-05,
+      "loss": 3.3211,
+      "step": 7475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.567594714338914e-05,
+      "loss": 3.2371,
+      "step": 7480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.561554758978809e-05,
+      "loss": 3.2748,
+      "step": 7485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.555517091196281e-05,
+      "loss": 3.2694,
+      "step": 7490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.549481720593112e-05,
+      "loss": 3.3368,
+      "step": 7495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5434486567674254e-05,
+      "loss": 3.2745,
+      "step": 7500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5374179093136795e-05,
+      "loss": 3.2288,
+      "step": 7505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.531389487822647e-05,
+      "loss": 3.3287,
+      "step": 7510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.525363401881404e-05,
+      "loss": 3.3132,
+      "step": 7515
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.51933966107331e-05,
+      "loss": 3.2288,
+      "step": 7520
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.513318274977996e-05,
+      "loss": 3.2945,
+      "step": 7525
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5072992531713466e-05,
+      "loss": 3.1956,
+      "step": 7530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.501282605225491e-05,
+      "loss": 3.1796,
+      "step": 7535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.495268340708777e-05,
+      "loss": 3.2145,
+      "step": 7540
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.489256469185768e-05,
+      "loss": 3.2729,
+      "step": 7545
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.483247000217217e-05,
+      "loss": 3.2246,
+      "step": 7550
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.477239943360059e-05,
+      "loss": 3.3022,
+      "step": 7555
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4712353081673934e-05,
+      "loss": 3.2216,
+      "step": 7560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.465233104188463e-05,
+      "loss": 3.3541,
+      "step": 7565
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.459233340968654e-05,
+      "loss": 3.29,
+      "step": 7570
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4532360280494614e-05,
+      "loss": 3.2046,
+      "step": 7575
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.447241174968488e-05,
+      "loss": 3.269,
+      "step": 7580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4412487912594254e-05,
+      "loss": 3.2912,
+      "step": 7585
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.435258886452035e-05,
+      "loss": 3.2786,
+      "step": 7590
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.429271470072141e-05,
+      "loss": 3.255,
+      "step": 7595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.423286551641604e-05,
+      "loss": 3.3293,
+      "step": 7600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.417304140678317e-05,
+      "loss": 3.2822,
+      "step": 7605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4113242466961816e-05,
+      "loss": 3.2061,
+      "step": 7610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4053468792051e-05,
+      "loss": 3.2789,
+      "step": 7615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.399372047710954e-05,
+      "loss": 3.2582,
+      "step": 7620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.393399761715595e-05,
+      "loss": 3.2223,
+      "step": 7625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3874300307168226e-05,
+      "loss": 3.2598,
+      "step": 7630
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3814628642083774e-05,
+      "loss": 3.1955,
+      "step": 7635
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.375498271679917e-05,
+      "loss": 3.2248,
+      "step": 7640
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.369536262617009e-05,
+      "loss": 3.2472,
+      "step": 7645
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.363576846501112e-05,
+      "loss": 3.3702,
+      "step": 7650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3576200328095595e-05,
+      "loss": 3.2979,
+      "step": 7655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.351665831015549e-05,
+      "loss": 3.3465,
+      "step": 7660
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.345714250588121e-05,
+      "loss": 3.2558,
+      "step": 7665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.339765300992149e-05,
+      "loss": 3.2802,
+      "step": 7670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.333818991688324e-05,
+      "loss": 3.2417,
+      "step": 7675
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.327875332133134e-05,
+      "loss": 3.1038,
+      "step": 7680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.321934331778857e-05,
+      "loss": 3.2478,
+      "step": 7685
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.31599600007354e-05,
+      "loss": 3.254,
+      "step": 7690
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3100603464609885e-05,
+      "loss": 3.2242,
+      "step": 7695
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.304127380380745e-05,
+      "loss": 3.3464,
+      "step": 7700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.298197111268082e-05,
+      "loss": 3.2081,
+      "step": 7705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.292269548553979e-05,
+      "loss": 3.2334,
+      "step": 7710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2863447016651167e-05,
+      "loss": 3.2407,
+      "step": 7715
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.2804225800238506e-05,
+      "loss": 3.244,
+      "step": 7720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2745031930482075e-05,
+      "loss": 3.2873,
+      "step": 7725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.268586550151861e-05,
+      "loss": 3.3145,
+      "step": 7730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.262672660744124e-05,
+      "loss": 3.2304,
+      "step": 7735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.256761534229929e-05,
+      "loss": 3.2914,
+      "step": 7740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2508531800098154e-05,
+      "loss": 3.2681,
+      "step": 7745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.244947607479914e-05,
+      "loss": 3.308,
+      "step": 7750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2390448260319285e-05,
+      "loss": 3.2038,
+      "step": 7755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.23314484505313e-05,
+      "loss": 3.3103,
+      "step": 7760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.227247673926329e-05,
+      "loss": 3.3995,
+      "step": 7765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2213533220298744e-05,
+      "loss": 3.2252,
+      "step": 7770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.215461798737624e-05,
+      "loss": 3.3238,
+      "step": 7775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.209573113418947e-05,
+      "loss": 3.1943,
+      "step": 7780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.203687275438689e-05,
+      "loss": 3.2566,
+      "step": 7785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.197804294157175e-05,
+      "loss": 3.2204,
+      "step": 7790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1919241789301834e-05,
+      "loss": 3.2811,
+      "step": 7795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.186046939108935e-05,
+      "loss": 3.2251,
+      "step": 7800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.180172584040079e-05,
+      "loss": 3.2799,
+      "step": 7805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.174301123065675e-05,
+      "loss": 3.263,
+      "step": 7810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1684325655231836e-05,
+      "loss": 3.2729,
+      "step": 7815
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1625669207454435e-05,
+      "loss": 3.2625,
+      "step": 7820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.156704198060665e-05,
+      "loss": 3.2804,
+      "step": 7825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.150844406792408e-05,
+      "loss": 3.3449,
+      "step": 7830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.144987556259575e-05,
+      "loss": 3.2979,
+      "step": 7835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.139133655776386e-05,
+      "loss": 3.2375,
+      "step": 7840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1332827146523743e-05,
+      "loss": 3.3401,
+      "step": 7845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.127434742192367e-05,
+      "loss": 3.2289,
+      "step": 7850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.121589747696465e-05,
+      "loss": 3.3349,
+      "step": 7855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.115747740460039e-05,
+      "loss": 3.3123,
+      "step": 7860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.109908729773705e-05,
+      "loss": 3.2661,
+      "step": 7865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.104072724923317e-05,
+      "loss": 3.2237,
+      "step": 7870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0982397351899454e-05,
+      "loss": 3.2947,
+      "step": 7875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.092409769849868e-05,
+      "loss": 3.2348,
+      "step": 7880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.086582838174551e-05,
+      "loss": 3.2683,
+      "step": 7885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0807589494306395e-05,
+      "loss": 3.3022,
+      "step": 7890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.074938112879936e-05,
+      "loss": 3.2979,
+      "step": 7895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.069120337779389e-05,
+      "loss": 3.28,
+      "step": 7900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.063305633381081e-05,
+      "loss": 3.2153,
+      "step": 7905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.057494008932208e-05,
+      "loss": 3.2734,
+      "step": 7910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.051685473675071e-05,
+      "loss": 3.1986,
+      "step": 7915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0458800368470564e-05,
+      "loss": 3.2247,
+      "step": 7920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0400777076806243e-05,
+      "loss": 3.2155,
+      "step": 7925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0342784954032893e-05,
+      "loss": 3.2537,
+      "step": 7930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0284824092376137e-05,
+      "loss": 3.1874,
+      "step": 7935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0226894584011867e-05,
+      "loss": 3.1394,
+      "step": 7940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0168996521066095e-05,
+      "loss": 3.243,
+      "step": 7945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0111129995614852e-05,
+      "loss": 3.3244,
+      "step": 7950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0053295099683993e-05,
+      "loss": 3.2402,
+      "step": 7955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9995491925249087e-05,
+      "loss": 3.354,
+      "step": 7960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9937720564235244e-05,
+      "loss": 3.3085,
+      "step": 7965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9879981108516996e-05,
+      "loss": 3.3156,
+      "step": 7970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9822273649918115e-05,
+      "loss": 3.254,
+      "step": 7975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.976459828021151e-05,
+      "loss": 3.2322,
+      "step": 7980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9706955091119036e-05,
+      "loss": 3.2825,
+      "step": 7985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.964934417431139e-05,
+      "loss": 3.2707,
+      "step": 7990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9591765621407942e-05,
+      "loss": 3.2421,
+      "step": 7995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.953421952397658e-05,
+      "loss": 3.2354,
+      "step": 8000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9476705973533597e-05,
+      "loss": 3.3088,
+      "step": 8005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9419225061543503e-05,
+      "loss": 3.2401,
+      "step": 8010
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.936177687941894e-05,
+      "loss": 3.3493,
+      "step": 8015
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9304361518520445e-05,
+      "loss": 3.1792,
+      "step": 8020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9246979070156418e-05,
+      "loss": 3.1796,
+      "step": 8025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.918962962558286e-05,
+      "loss": 3.2588,
+      "step": 8030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9132313276003325e-05,
+      "loss": 3.2418,
+      "step": 8035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.907503011256877e-05,
+      "loss": 3.1275,
+      "step": 8040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.901778022637725e-05,
+      "loss": 3.2055,
+      "step": 8045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.896056370847404e-05,
+      "loss": 3.1804,
+      "step": 8050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8903380649851253e-05,
+      "loss": 3.225,
+      "step": 8055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.884623114144789e-05,
+      "loss": 3.3495,
+      "step": 8060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8789115274149443e-05,
+      "loss": 3.2203,
+      "step": 8065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8732033138788062e-05,
+      "loss": 3.1976,
+      "step": 8070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.867498482614214e-05,
+      "loss": 3.2655,
+      "step": 8075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.861797042693639e-05,
+      "loss": 3.268,
+      "step": 8080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.856099003184145e-05,
+      "loss": 3.3071,
+      "step": 8085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8504043731474005e-05,
+      "loss": 3.1338,
+      "step": 8090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8447131616396507e-05,
+      "loss": 3.2542,
+      "step": 8095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8390253777116926e-05,
+      "loss": 3.2667,
+      "step": 8100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.833341030408887e-05,
+      "loss": 3.2914,
+      "step": 8105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8276601287711192e-05,
+      "loss": 3.3584,
+      "step": 8110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8219826818328042e-05,
+      "loss": 3.3342,
+      "step": 8115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8163086986228504e-05,
+      "loss": 3.2234,
+      "step": 8120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.810638188164669e-05,
+      "loss": 3.31,
+      "step": 8125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.804971159476143e-05,
+      "loss": 3.2126,
+      "step": 8130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.799307621569618e-05,
+      "loss": 3.2805,
+      "step": 8135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7936475834518927e-05,
+      "loss": 3.2375,
+      "step": 8140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.787991054124194e-05,
+      "loss": 3.3354,
+      "step": 8145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7823380425821766e-05,
+      "loss": 3.3071,
+      "step": 8150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.776688557815889e-05,
+      "loss": 3.366,
+      "step": 8155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7710426088097836e-05,
+      "loss": 3.243,
+      "step": 8160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7654002045426796e-05,
+      "loss": 3.198,
+      "step": 8165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.75976135398777e-05,
+      "loss": 3.2828,
+      "step": 8170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.754126066112582e-05,
+      "loss": 3.2556,
+      "step": 8175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7484943498789906e-05,
+      "loss": 3.1849,
+      "step": 8180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7428662142431832e-05,
+      "loss": 3.2592,
+      "step": 8185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.737241668155653e-05,
+      "loss": 3.2097,
+      "step": 8190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.73162072056119e-05,
+      "loss": 3.2312,
+      "step": 8195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.726003380398854e-05,
+      "loss": 3.2091,
+      "step": 8200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.720389656601977e-05,
+      "loss": 3.3075,
+      "step": 8205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.714779558098129e-05,
+      "loss": 3.2165,
+      "step": 8210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.709173093809124e-05,
+      "loss": 3.1489,
+      "step": 8215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.703570272650989e-05,
+      "loss": 3.2931,
+      "step": 8220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6979711035339687e-05,
+      "loss": 3.2001,
+      "step": 8225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.692375595362483e-05,
+      "loss": 3.3254,
+      "step": 8230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.686783757035143e-05,
+      "loss": 3.2589,
+      "step": 8235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6811955974447235e-05,
+      "loss": 3.3136,
+      "step": 8240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6756111254781374e-05,
+      "loss": 3.1986,
+      "step": 8245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6700303500164465e-05,
+      "loss": 3.2442,
+      "step": 8250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6644532799348245e-05,
+      "loss": 3.2926,
+      "step": 8255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.658879924102562e-05,
+      "loss": 3.2593,
+      "step": 8260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6533102913830298e-05,
+      "loss": 3.2493,
+      "step": 8265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6477443906336898e-05,
+      "loss": 3.2348,
+      "step": 8270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6421822307060616e-05,
+      "loss": 3.2144,
+      "step": 8275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6366238204457205e-05,
+      "loss": 3.4113,
+      "step": 8280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6310691686922718e-05,
+      "loss": 3.2079,
+      "step": 8285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6255182842793513e-05,
+      "loss": 3.2325,
+      "step": 8290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6199711760346046e-05,
+      "loss": 3.2765,
+      "step": 8295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6144278527796597e-05,
+      "loss": 3.2023,
+      "step": 8300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6088883233301382e-05,
+      "loss": 3.1182,
+      "step": 8305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6033525964956206e-05,
+      "loss": 3.2889,
+      "step": 8310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.597820681079649e-05,
+      "loss": 3.3027,
+      "step": 8315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.5922925858796904e-05,
+      "loss": 3.3009,
+      "step": 8320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.5867683196871494e-05,
+      "loss": 3.2621,
+      "step": 8325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.5812478912873344e-05,
+      "loss": 3.2905,
+      "step": 8330
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.5757313094594505e-05,
+      "loss": 3.2894,
+      "step": 8335
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.57021858297659e-05,
+      "loss": 3.1675,
+      "step": 8340
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.564709720605709e-05,
+      "loss": 3.319,
+      "step": 8345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.5592047311076262e-05,
+      "loss": 3.1822,
+      "step": 8350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.553703623236987e-05,
+      "loss": 3.3185,
+      "step": 8355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5482064057422782e-05,
+      "loss": 3.271,
+      "step": 8360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.542713087365792e-05,
+      "loss": 3.2957,
+      "step": 8365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5372236768436258e-05,
+      "loss": 3.2764,
+      "step": 8370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5317381829056507e-05,
+      "loss": 3.3536,
+      "step": 8375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.526256614275524e-05,
+      "loss": 3.2458,
+      "step": 8380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5207789796706494e-05,
+      "loss": 3.3154,
+      "step": 8385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5153052878021777e-05,
+      "loss": 3.2589,
+      "step": 8390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.509835547374993e-05,
+      "loss": 3.2501,
+      "step": 8395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5043697670876888e-05,
+      "loss": 3.2972,
+      "step": 8400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4989079556325707e-05,
+      "loss": 3.2978,
+      "step": 8405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4934501216956176e-05,
+      "loss": 3.293,
+      "step": 8410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4879962739564973e-05,
+      "loss": 3.2676,
+      "step": 8415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.482546421088531e-05,
+      "loss": 3.2514,
+      "step": 8420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4771005717586893e-05,
+      "loss": 3.2453,
+      "step": 8425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4716587346275716e-05,
+      "loss": 3.3243,
+      "step": 8430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4662209183494022e-05,
+      "loss": 3.2065,
+      "step": 8435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4607871315720126e-05,
+      "loss": 3.2155,
+      "step": 8440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4553573829368154e-05,
+      "loss": 3.3724,
+      "step": 8445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.449931681078813e-05,
+      "loss": 3.1993,
+      "step": 8450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4445100346265643e-05,
+      "loss": 3.348,
+      "step": 8455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4390924522021875e-05,
+      "loss": 3.3548,
+      "step": 8460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4336789424213237e-05,
+      "loss": 3.268,
+      "step": 8465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4282695138931527e-05,
+      "loss": 3.1857,
+      "step": 8470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.422864175220354e-05,
+      "loss": 3.2996,
+      "step": 8475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4174629349991056e-05,
+      "loss": 3.1865,
+      "step": 8480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4120658018190666e-05,
+      "loss": 3.3497,
+      "step": 8485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4066727842633673e-05,
+      "loss": 3.3586,
+      "step": 8490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.401283890908596e-05,
+      "loss": 3.2139,
+      "step": 8495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.395899130324769e-05,
+      "loss": 3.2816,
+      "step": 8500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3905185110753446e-05,
+      "loss": 3.3715,
+      "step": 8505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3851420417171866e-05,
+      "loss": 3.2889,
+      "step": 8510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.379769730800567e-05,
+      "loss": 3.2694,
+      "step": 8515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3744015868691315e-05,
+      "loss": 3.3561,
+      "step": 8520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.369037618459913e-05,
+      "loss": 3.2566,
+      "step": 8525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.363677834103296e-05,
+      "loss": 3.2947,
+      "step": 8530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3583222423230117e-05,
+      "loss": 3.1786,
+      "step": 8535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.352970851636128e-05,
+      "loss": 3.2438,
+      "step": 8540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3476236705530258e-05,
+      "loss": 3.3684,
+      "step": 8545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3422807075774e-05,
+      "loss": 3.2836,
+      "step": 8550
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3369419712062247e-05,
+      "loss": 3.238,
+      "step": 8555
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3316074699297646e-05,
+      "loss": 3.2924,
+      "step": 8560
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3262772122315424e-05,
+      "loss": 3.235,
+      "step": 8565
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3209512065883342e-05,
+      "loss": 3.1788,
+      "step": 8570
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3156294614701524e-05,
+      "loss": 3.2783,
+      "step": 8575
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3103119853402365e-05,
+      "loss": 3.2919,
+      "step": 8580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3049987866550398e-05,
+      "loss": 3.2025,
+      "step": 8585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2996898738642015e-05,
+      "loss": 3.2997,
+      "step": 8590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2943852554105577e-05,
+      "loss": 3.2503,
+      "step": 8595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.2890849397301067e-05,
+      "loss": 3.2176,
+      "step": 8600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.283788935252013e-05,
+      "loss": 3.2273,
+      "step": 8605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2784972503985715e-05,
+      "loss": 3.1356,
+      "step": 8610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.273209893585221e-05,
+      "loss": 3.141,
+      "step": 8615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.267926873220509e-05,
+      "loss": 3.2623,
+      "step": 8620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.26264819770609e-05,
+      "loss": 3.2261,
+      "step": 8625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2573738754367063e-05,
+      "loss": 3.1421,
+      "step": 8630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2521039148001806e-05,
+      "loss": 3.3893,
+      "step": 8635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.246838324177401e-05,
+      "loss": 3.1579,
+      "step": 8640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2415771119422962e-05,
+      "loss": 3.233,
+      "step": 8645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2363202864618434e-05,
+      "loss": 3.2509,
+      "step": 8650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2310678560960342e-05,
+      "loss": 3.226,
+      "step": 8655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2258198291978804e-05,
+      "loss": 3.2335,
+      "step": 8660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2205762141133784e-05,
+      "loss": 3.1508,
+      "step": 8665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2153370191815197e-05,
+      "loss": 3.2353,
+      "step": 8670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.210102252734262e-05,
+      "loss": 3.2781,
+      "step": 8675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2048719230965166e-05,
+      "loss": 3.242,
+      "step": 8680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1996460385861474e-05,
+      "loss": 3.1507,
+      "step": 8685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1944246075139396e-05,
+      "loss": 3.2609,
+      "step": 8690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1892076381836074e-05,
+      "loss": 3.2656,
+      "step": 8695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.183995138891755e-05,
+      "loss": 3.293,
+      "step": 8700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1787871179278914e-05,
+      "loss": 3.3399,
+      "step": 8705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1735835835743955e-05,
+      "loss": 3.2327,
+      "step": 8710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1683845441065143e-05,
+      "loss": 3.3104,
+      "step": 8715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1631900077923428e-05,
+      "loss": 3.2212,
+      "step": 8720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1579999828928222e-05,
+      "loss": 3.2679,
+      "step": 8725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.152814477661712e-05,
+      "loss": 3.2256,
+      "step": 8730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.147633500345585e-05,
+      "loss": 3.2326,
+      "step": 8735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1424570591838183e-05,
+      "loss": 3.2779,
+      "step": 8740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1372851624085676e-05,
+      "loss": 3.2291,
+      "step": 8745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.132117818244771e-05,
+      "loss": 3.2613,
+      "step": 8750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1269550349101136e-05,
+      "loss": 3.263,
+      "step": 8755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1217968206150403e-05,
+      "loss": 3.2904,
+      "step": 8760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1166431835627228e-05,
+      "loss": 3.2569,
+      "step": 8765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.111494131949054e-05,
+      "loss": 3.241,
+      "step": 8770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.106349673962635e-05,
+      "loss": 3.2981,
+      "step": 8775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1012098177847628e-05,
+      "loss": 3.2934,
+      "step": 8780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0960745715894192e-05,
+      "loss": 3.3768,
+      "step": 8785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0909439435432442e-05,
+      "loss": 3.2989,
+      "step": 8790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0858179418055458e-05,
+      "loss": 3.2736,
+      "step": 8795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.080696574528264e-05,
+      "loss": 3.2335,
+      "step": 8800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0755798498559808e-05,
+      "loss": 3.263,
+      "step": 8805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.07046777592588e-05,
+      "loss": 3.3396,
+      "step": 8810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0653603608677618e-05,
+      "loss": 3.2272,
+      "step": 8815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0602576128040125e-05,
+      "loss": 3.217,
+      "step": 8820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0551595398495948e-05,
+      "loss": 3.3635,
+      "step": 8825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0500661501120377e-05,
+      "loss": 3.2659,
+      "step": 8830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0449774516914244e-05,
+      "loss": 3.2613,
+      "step": 8835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.03989345268038e-05,
+      "loss": 3.2626,
+      "step": 8840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0348141611640443e-05,
+      "loss": 3.2705,
+      "step": 8845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0297395852200847e-05,
+      "loss": 3.3683,
+      "step": 8850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.024669732918659e-05,
+      "loss": 3.1658,
+      "step": 8855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0196046123224226e-05,
+      "loss": 3.2262,
+      "step": 8860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0145442314864942e-05,
+      "loss": 3.2795,
+      "step": 8865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0094885984584656e-05,
+      "loss": 3.2798,
+      "step": 8870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.004437721278372e-05,
+      "loss": 3.2294,
+      "step": 8875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9993916079786857e-05,
+      "loss": 3.2355,
+      "step": 8880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.994350266584307e-05,
+      "loss": 3.3603,
+      "step": 8885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989313705112542e-05,
+      "loss": 3.2876,
+      "step": 8890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9842819315731027e-05,
+      "loss": 3.2763,
+      "step": 8895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9792549539680745e-05,
+      "loss": 3.3059,
+      "step": 8900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9742327802919285e-05,
+      "loss": 3.2085,
+      "step": 8905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.96921541853149e-05,
+      "loss": 3.2454,
+      "step": 8910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9642028766659308e-05,
+      "loss": 3.2503,
+      "step": 8915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9591951626667593e-05,
+      "loss": 3.2008,
+      "step": 8920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.954192284497809e-05,
+      "loss": 3.1598,
+      "step": 8925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9491942501152177e-05,
+      "loss": 3.214,
+      "step": 8930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9442010674674216e-05,
+      "loss": 3.3195,
+      "step": 8935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9392127444951446e-05,
+      "loss": 3.2227,
+      "step": 8940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9342292891313767e-05,
+      "loss": 3.2498,
+      "step": 8945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9292507093013746e-05,
+      "loss": 3.2752,
+      "step": 8950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9242770129226294e-05,
+      "loss": 3.247,
+      "step": 8955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9193082079048792e-05,
+      "loss": 3.1523,
+      "step": 8960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.914344302150074e-05,
+      "loss": 3.2029,
+      "step": 8965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9093853035523767e-05,
+      "loss": 3.2675,
+      "step": 8970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.904431219998143e-05,
+      "loss": 3.2487,
+      "step": 8975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8994820593659163e-05,
+      "loss": 3.3566,
+      "step": 8980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8945378295264128e-05,
+      "loss": 3.2971,
+      "step": 8985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8895985383424953e-05,
+      "loss": 3.2845,
+      "step": 8990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.884664193669187e-05,
+      "loss": 3.2422,
+      "step": 8995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8797348033536354e-05,
+      "loss": 3.2397,
+      "step": 9000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 12556,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 1.211524240637952e+18,
+  "trial_name": null,
+  "trial_params": null
+}