diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9024 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 7497,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.999997804999731e-05,
+      "loss": 2.5926,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9999912200085595e-05,
+      "loss": 2.726,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999980245055394e-05,
+      "loss": 2.7046,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999648801884152e-05,
+      "loss": 2.5412,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9999451254750736e-05,
+      "loss": 2.4895,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.999920981002093e-05,
+      "loss": 2.5147,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9998924468754672e-05,
+      "loss": 2.3979,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998595232204618e-05,
+      "loss": 2.3704,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9998222101816113e-05,
+      "loss": 2.3891,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99978050792272e-05,
+      "loss": 2.3385,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.999734416626861e-05,
+      "loss": 2.3395,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9996839364963744e-05,
+      "loss": 2.2868,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9996290677528688e-05,
+      "loss": 2.2824,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9995698106372176e-05,
+      "loss": 2.2458,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99950616540956e-05,
+      "loss": 2.2594,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999438132349298e-05,
+      "loss": 2.2887,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993657117550972e-05,
+      "loss": 2.2976,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992889039448838e-05,
+      "loss": 2.2611,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992077092558443e-05,
+      "loss": 2.2257,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991221280444237e-05,
+      "loss": 2.3287,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990321606863224e-05,
+      "loss": 2.3082,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989378075764988e-05,
+      "loss": 2.251,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998839069129162e-05,
+      "loss": 2.2239,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998735945777774e-05,
+      "loss": 2.2154,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986284379750462e-05,
+      "loss": 2.2939,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985165461929385e-05,
+      "loss": 2.2763,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984002709226558e-05,
+      "loss": 2.1672,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.998279612674646e-05,
+      "loss": 2.1404,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9981545719785994e-05,
+      "loss": 2.1851,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9980251493834447e-05,
+      "loss": 2.1481,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997891345457347e-05,
+      "loss": 2.1352,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9977531607877055e-05,
+      "loss": 2.2454,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9976105959811512e-05,
+      "loss": 2.3146,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9974636516635436e-05,
+      "loss": 2.208,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9973123284799683e-05,
+      "loss": 2.1468,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.997156627094734e-05,
+      "loss": 2.1792,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9969965481913704e-05,
+      "loss": 2.3303,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9968320924726236e-05,
+      "loss": 2.2114,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9966632606604543e-05,
+      "loss": 2.2289,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.996490053496034e-05,
+      "loss": 2.2365,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9963124717397425e-05,
+      "loss": 2.1721,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9961305161711643e-05,
+      "loss": 2.1837,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9959441875890834e-05,
+      "loss": 2.1694,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9957534868114835e-05,
+      "loss": 2.1377,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99555841467554e-05,
+      "loss": 2.1963,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9953589720376204e-05,
+      "loss": 2.2826,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.995155159773278e-05,
+      "loss": 2.2644,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9949469787772484e-05,
+      "loss": 2.1792,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9947344299634464e-05,
+      "loss": 2.2485,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9945175142649617e-05,
+      "loss": 2.2065,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.994296232634054e-05,
+      "loss": 2.2887,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9940705860421493e-05,
+      "loss": 2.1542,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9938405754798377e-05,
+      "loss": 2.1692,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9936062019568646e-05,
+      "loss": 2.1624,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9933674665021295e-05,
+      "loss": 2.19,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9931243701636826e-05,
+      "loss": 2.1613,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9928769140087157e-05,
+      "loss": 2.2509,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.992625099123562e-05,
+      "loss": 2.1907,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.992368926613689e-05,
+      "loss": 2.1528,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9921083976036937e-05,
+      "loss": 2.1551,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9918435132372992e-05,
+      "loss": 2.2466,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.991574274677348e-05,
+      "loss": 2.2388,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9913006831057967e-05,
+      "loss": 2.1348,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9910227397237136e-05,
+      "loss": 2.1659,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9907404457512693e-05,
+      "loss": 2.1995,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.990453802427735e-05,
+      "loss": 2.2362,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.990162811011475e-05,
+      "loss": 2.1695,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9898674727799418e-05,
+      "loss": 2.2246,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9895677890296704e-05,
+      "loss": 2.1766,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9892637610762723e-05,
+      "loss": 2.1765,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9889553902544307e-05,
+      "loss": 2.251,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9886426779178935e-05,
+      "loss": 2.1699,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988325625439468e-05,
+      "loss": 2.2406,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.988004234211015e-05,
+      "loss": 2.1664,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9876785056434422e-05,
+      "loss": 2.1711,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.987348441166698e-05,
+      "loss": 2.1874,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9870140422297655e-05,
+      "loss": 2.1686,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9866753103006564e-05,
+      "loss": 2.2158,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9863322468664038e-05,
+      "loss": 2.147,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.985984853433057e-05,
+      "loss": 2.1805,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9856331315256728e-05,
+      "loss": 2.2011,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9852770826883106e-05,
+      "loss": 2.0972,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984916708484025e-05,
+      "loss": 2.243,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.984552010494859e-05,
+      "loss": 2.2098,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9841829903218377e-05,
+      "loss": 2.1151,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9838096495849582e-05,
+      "loss": 2.2735,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9834319899231884e-05,
+      "loss": 2.1943,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.983050012994453e-05,
+      "loss": 2.2244,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9826637204756312e-05,
+      "loss": 2.178,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.982273114062548e-05,
+      "loss": 2.1652,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9818781954699656e-05,
+      "loss": 2.2102,
+      "step": 455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9814789664315762e-05,
+      "loss": 2.1286,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.981075428699996e-05,
+      "loss": 2.2028,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.980667584046756e-05,
+      "loss": 2.122,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.980255434262294e-05,
+      "loss": 2.0737,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979838981155948e-05,
+      "loss": 2.1987,
+      "step": 480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.979418226555947e-05,
+      "loss": 2.2296,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9789931723094046e-05,
+      "loss": 2.1883,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.978563820282309e-05,
+      "loss": 2.0761,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9781301723595154e-05,
+      "loss": 2.1523,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.977692230444739e-05,
+      "loss": 2.2492,
+      "step": 505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.977249996460544e-05,
+      "loss": 2.179,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.976803472348339e-05,
+      "loss": 2.1714,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.976352660068365e-05,
+      "loss": 2.1702,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9758975615996874e-05,
+      "loss": 2.0948,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.975438178940189e-05,
+      "loss": 2.1055,
+      "step": 530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.97497451410656e-05,
+      "loss": 2.1615,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.97450656913429e-05,
+      "loss": 2.1442,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9740343460776564e-05,
+      "loss": 2.1537,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9735578470097195e-05,
+      "loss": 2.2454,
+      "step": 550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9730770740223098e-05,
+      "loss": 2.2078,
+      "step": 555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.972592029226022e-05,
+      "loss": 2.1602,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.972102714750202e-05,
+      "loss": 2.229,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.971609132742941e-05,
+      "loss": 2.1704,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9711112853710645e-05,
+      "loss": 2.2131,
+      "step": 575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9706091748201227e-05,
+      "loss": 2.2287,
+      "step": 580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9701028032943805e-05,
+      "loss": 2.2162,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.96959217301681e-05,
+      "loss": 2.2365,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969077286229078e-05,
+      "loss": 2.1714,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9685581451915382e-05,
+      "loss": 2.1122,
+      "step": 600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968034752183219e-05,
+      "loss": 2.1883,
+      "step": 605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9675071095018175e-05,
+      "loss": 2.1136,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.966975219463684e-05,
+      "loss": 2.2358,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9664390844038168e-05,
+      "loss": 2.1575,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.965898706675849e-05,
+      "loss": 2.176,
+      "step": 625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9653540886520387e-05,
+      "loss": 2.0719,
+      "step": 630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.96480523272326e-05,
+      "loss": 2.1178,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9642521412989902e-05,
+      "loss": 2.1896,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.963694816807301e-05,
+      "loss": 2.1458,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9631332616948472e-05,
+      "loss": 2.1789,
+      "step": 650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9625674784268564e-05,
+      "loss": 2.1564,
+      "step": 655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9619974694871173e-05,
+      "loss": 2.1995,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9614232373779692e-05,
+      "loss": 2.1413,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9608447846202918e-05,
+      "loss": 2.1179,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9602621137534923e-05,
+      "loss": 2.1635,
+      "step": 675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.959675227335497e-05,
+      "loss": 2.187,
+      "step": 680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9590841279427368e-05,
+      "loss": 2.2399,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9584888181701388e-05,
+      "loss": 2.1589,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9578893006311135e-05,
+      "loss": 2.1393,
+      "step": 695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9572855779575427e-05,
+      "loss": 2.1048,
+      "step": 700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9566776527997694e-05,
+      "loss": 2.1851,
+      "step": 705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9560655278265852e-05,
+      "loss": 2.1909,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9554492057252196e-05,
+      "loss": 2.1493,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9548286892013262e-05,
+      "loss": 2.1845,
+      "step": 720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9542039809789736e-05,
+      "loss": 2.2492,
+      "step": 725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9535750838006307e-05,
+      "loss": 2.1798,
+      "step": 730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9529420004271568e-05,
+      "loss": 2.1765,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.952304733637788e-05,
+      "loss": 2.1119,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9516632862301258e-05,
+      "loss": 2.0904,
+      "step": 745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9510176610201254e-05,
+      "loss": 2.1828,
+      "step": 750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.950367860842081e-05,
+      "loss": 2.1519,
+      "step": 755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9497138885486157e-05,
+      "loss": 2.0773,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.949055747010669e-05,
+      "loss": 2.1506,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9483934391174816e-05,
+      "loss": 2.2158,
+      "step": 770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9477269677765863e-05,
+      "loss": 2.1949,
+      "step": 775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9470563359137923e-05,
+      "loss": 2.1424,
+      "step": 780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9463815464731737e-05,
+      "loss": 2.2106,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.945702602417057e-05,
+      "loss": 2.1789,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.945019506726007e-05,
+      "loss": 2.0718,
+      "step": 795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9443322623988133e-05,
+      "loss": 2.1993,
+      "step": 800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9436408724524794e-05,
+      "loss": 2.1222,
+      "step": 805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9429453399222077e-05,
+      "loss": 2.2176,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.942245667861386e-05,
+      "loss": 2.1398,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9415418593415755e-05,
+      "loss": 2.1975,
+      "step": 820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9408339174524957e-05,
+      "loss": 2.0749,
+      "step": 825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9401218453020118e-05,
+      "loss": 2.1371,
+      "step": 830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.939405646016121e-05,
+      "loss": 2.162,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.938685322738939e-05,
+      "loss": 2.1034,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9379608786326843e-05,
+      "loss": 2.1339,
+      "step": 845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.937232316877668e-05,
+      "loss": 2.1118,
+      "step": 850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.936499640672276e-05,
+      "loss": 2.1367,
+      "step": 855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9357628532329574e-05,
+      "loss": 2.1633,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9350219577942095e-05,
+      "loss": 2.0912,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9342769576085633e-05,
+      "loss": 2.1311,
+      "step": 870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9335278559465703e-05,
+      "loss": 2.1007,
+      "step": 875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9327746560967878e-05,
+      "loss": 2.1426,
+      "step": 880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9320173613657628e-05,
+      "loss": 2.1446,
+      "step": 885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9312559750780196e-05,
+      "loss": 2.1508,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9304905005760447e-05,
+      "loss": 2.1822,
+      "step": 895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9297209412202712e-05,
+      "loss": 2.0216,
+      "step": 900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9289473003890658e-05,
+      "loss": 2.2102,
+      "step": 905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9281695814787113e-05,
+      "loss": 2.0846,
+      "step": 910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9273877879033946e-05,
+      "loss": 2.1596,
+      "step": 915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.92660192309519e-05,
+      "loss": 2.2179,
+      "step": 920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9258119905040443e-05,
+      "loss": 2.1029,
+      "step": 925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9250179935977617e-05,
+      "loss": 2.1343,
+      "step": 930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9242199358619897e-05,
+      "loss": 2.0859,
+      "step": 935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9234178208002014e-05,
+      "loss": 2.1138,
+      "step": 940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.922611651933683e-05,
+      "loss": 2.0537,
+      "step": 945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.921801432801516e-05,
+      "loss": 2.1042,
+      "step": 950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.920987166960563e-05,
+      "loss": 2.1944,
+      "step": 955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.920168857985451e-05,
+      "loss": 2.0814,
+      "step": 960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9193465094685577e-05,
+      "loss": 2.1275,
+      "step": 965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9185201250199926e-05,
+      "loss": 2.0843,
+      "step": 970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9176897082675845e-05,
+      "loss": 2.119,
+      "step": 975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9168552628568632e-05,
+      "loss": 2.1449,
+      "step": 980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.916016792451044e-05,
+      "loss": 2.1665,
+      "step": 985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9151743007310134e-05,
+      "loss": 2.0731,
+      "step": 990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.91432779139531e-05,
+      "loss": 2.1782,
+      "step": 995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9134772681601102e-05,
+      "loss": 2.2128,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9126227347592113e-05,
+      "loss": 2.1159,
+      "step": 1005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9117641949440156e-05,
+      "loss": 2.1405,
+      "step": 1010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9109016524835136e-05,
+      "loss": 2.1809,
+      "step": 1015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9100351111642666e-05,
+      "loss": 2.092,
+      "step": 1020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.909164574790392e-05,
+      "loss": 2.1187,
+      "step": 1025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9082900471835448e-05,
+      "loss": 2.1296,
+      "step": 1030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9074115321829015e-05,
+      "loss": 2.0347,
+      "step": 1035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9065290336451435e-05,
+      "loss": 2.1343,
+      "step": 1040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.90564255544444e-05,
+      "loss": 2.1648,
+      "step": 1045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9047521014724303e-05,
+      "loss": 2.0968,
+      "step": 1050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9038576756382084e-05,
+      "loss": 2.1439,
+      "step": 1055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9029592818683042e-05,
+      "loss": 2.1372,
+      "step": 1060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9020569241066665e-05,
+      "loss": 2.0636,
+      "step": 1065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.901150606314646e-05,
+      "loss": 2.1161,
+      "step": 1070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9002403324709793e-05,
+      "loss": 2.1448,
+      "step": 1075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8993261065717685e-05,
+      "loss": 2.1542,
+      "step": 1080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8984079326304653e-05,
+      "loss": 2.1861,
+      "step": 1085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8974858146778547e-05,
+      "loss": 2.1522,
+      "step": 1090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8965597567620344e-05,
+      "loss": 2.1143,
+      "step": 1095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8956297629483995e-05,
+      "loss": 2.1022,
+      "step": 1100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.894695837319623e-05,
+      "loss": 2.1759,
+      "step": 1105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.893757983975639e-05,
+      "loss": 2.1132,
+      "step": 1110
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.892816207033625e-05,
+      "loss": 2.1244,
+      "step": 1115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8918705106279812e-05,
+      "loss": 2.1174,
+      "step": 1120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8909208989103155e-05,
+      "loss": 2.1678,
+      "step": 1125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.889967376049424e-05,
+      "loss": 2.1666,
+      "step": 1130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8890099462312732e-05,
+      "loss": 2.0998,
+      "step": 1135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8880486136589794e-05,
+      "loss": 2.1457,
+      "step": 1140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.887083382552794e-05,
+      "loss": 2.1633,
+      "step": 1145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8861142571500813e-05,
+      "loss": 2.1287,
+      "step": 1150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8851412417053033e-05,
+      "loss": 2.0356,
+      "step": 1155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8841643404899974e-05,
+      "loss": 2.1444,
+      "step": 1160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.883183557792761e-05,
+      "loss": 2.1141,
+      "step": 1165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8821988979192307e-05,
+      "loss": 2.1039,
+      "step": 1170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8812103651920633e-05,
+      "loss": 2.1084,
+      "step": 1175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8802179639509188e-05,
+      "loss": 2.1054,
+      "step": 1180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.879221698552439e-05,
+      "loss": 2.0563,
+      "step": 1185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8782215733702286e-05,
+      "loss": 2.2365,
+      "step": 1190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.877217592794839e-05,
+      "loss": 2.177,
+      "step": 1195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8762097612337444e-05,
+      "loss": 2.1924,
+      "step": 1200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.875198083111327e-05,
+      "loss": 2.1563,
+      "step": 1205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8741825628688534e-05,
+      "loss": 2.0484,
+      "step": 1210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.873163204964458e-05,
+      "loss": 2.0747,
+      "step": 1215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.872140013873123e-05,
+      "loss": 2.0507,
+      "step": 1220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8711129940866577e-05,
+      "loss": 2.1103,
+      "step": 1225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8700821501136797e-05,
+      "loss": 2.1322,
+      "step": 1230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8690474864795942e-05,
+      "loss": 2.0776,
+      "step": 1235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8680090077265753e-05,
+      "loss": 2.1604,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8669667184135454e-05,
+      "loss": 2.1689,
+      "step": 1245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8659206231161548e-05,
+      "loss": 2.0437,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8648707264267628e-05,
+      "loss": 2.1611,
+      "step": 1255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8638170329544164e-05,
+      "loss": 2.1317,
+      "step": 1260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8627595473248303e-05,
+      "loss": 2.0664,
+      "step": 1265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.861698274180367e-05,
+      "loss": 2.0979,
+      "step": 1270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8606332181800165e-05,
+      "loss": 2.0975,
+      "step": 1275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8595643839993747e-05,
+      "loss": 2.0339,
+      "step": 1280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8584917763306246e-05,
+      "loss": 2.1365,
+      "step": 1285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8574153998825145e-05,
+      "loss": 2.1832,
+      "step": 1290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.856335259380337e-05,
+      "loss": 2.1306,
+      "step": 1295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8552513595659103e-05,
+      "loss": 2.1845,
+      "step": 1300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8541637051975544e-05,
+      "loss": 2.1158,
+      "step": 1305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8530723010500733e-05,
+      "loss": 2.1405,
+      "step": 1310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8519771519147312e-05,
+      "loss": 2.0784,
+      "step": 1315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.850878262599234e-05,
+      "loss": 2.1533,
+      "step": 1320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8497756379277058e-05,
+      "loss": 2.1382,
+      "step": 1325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8486692827406694e-05,
+      "loss": 2.129,
+      "step": 1330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8475592018950256e-05,
+      "loss": 2.1339,
+      "step": 1335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.846445400264029e-05,
+      "loss": 2.161,
+      "step": 1340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8453278827372697e-05,
+      "loss": 2.0769,
+      "step": 1345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.84420665422065e-05,
+      "loss": 2.07,
+      "step": 1350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.843081719636364e-05,
+      "loss": 2.2284,
+      "step": 1355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.841953083922875e-05,
+      "loss": 2.0727,
+      "step": 1360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8408207520348945e-05,
+      "loss": 2.1778,
+      "step": 1365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.83968472894336e-05,
+      "loss": 2.0996,
+      "step": 1370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.838545019635413e-05,
+      "loss": 2.0714,
+      "step": 1375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8374016291143792e-05,
+      "loss": 2.1285,
+      "step": 1380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8362545623997428e-05,
+      "loss": 2.0806,
+      "step": 1385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.835103824527127e-05,
+      "loss": 2.079,
+      "step": 1390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8339494205482722e-05,
+      "loss": 2.1132,
+      "step": 1395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8327913555310125e-05,
+      "loss": 2.0798,
+      "step": 1400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8316296345592534e-05,
+      "loss": 2.1899,
+      "step": 1405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8304642627329513e-05,
+      "loss": 2.0772,
+      "step": 1410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8292952451680886e-05,
+      "loss": 2.1043,
+      "step": 1415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8281225869966532e-05,
+      "loss": 2.1246,
+      "step": 1420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8269462933666157e-05,
+      "loss": 2.026,
+      "step": 1425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8257663694419045e-05,
+      "loss": 2.1274,
+      "step": 1430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8245828204023875e-05,
+      "loss": 2.0544,
+      "step": 1435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.823395651443845e-05,
+      "loss": 2.0656,
+      "step": 1440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8222048677779495e-05,
+      "loss": 2.164,
+      "step": 1445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.821010474632242e-05,
+      "loss": 2.0952,
+      "step": 1450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8198124772501085e-05,
+      "loss": 2.1039,
+      "step": 1455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.818610880890759e-05,
+      "loss": 2.0748,
+      "step": 1460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.817405690829202e-05,
+      "loss": 2.0959,
+      "step": 1465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.816196912356222e-05,
+      "loss": 2.1587,
+      "step": 1470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8149845507783574e-05,
+      "loss": 2.0942,
+      "step": 1475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8137686114178763e-05,
+      "loss": 2.1103,
+      "step": 1480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.812549099612753e-05,
+      "loss": 2.2153,
+      "step": 1485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8113260207166453e-05,
+      "loss": 2.0573,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8100993800988696e-05,
+      "loss": 2.0931,
+      "step": 1495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8088691831443798e-05,
+      "loss": 2.0398,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8076354352537402e-05,
+      "loss": 2.0999,
+      "step": 1505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.806398141843105e-05,
+      "loss": 2.1188,
+      "step": 1510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8051573083441935e-05,
+      "loss": 2.1364,
+      "step": 1515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8039129402042648e-05,
+      "loss": 2.05,
+      "step": 1520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8026650428860958e-05,
+      "loss": 2.0816,
+      "step": 1525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8014136218679566e-05,
+      "loss": 2.1499,
+      "step": 1530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8001586826435863e-05,
+      "loss": 2.2182,
+      "step": 1535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7989002307221685e-05,
+      "loss": 2.0472,
+      "step": 1540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7976382716283075e-05,
+      "loss": 2.1268,
+      "step": 1545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7963728109020047e-05,
+      "loss": 2.1228,
+      "step": 1550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.795103854098634e-05,
+      "loss": 2.0701,
+      "step": 1555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7938314067889154e-05,
+      "loss": 2.1223,
+      "step": 1560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7925554745588938e-05,
+      "loss": 2.0372,
+      "step": 1565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.791276063009912e-05,
+      "loss": 2.0967,
+      "step": 1570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.789993177758588e-05,
+      "loss": 2.0381,
+      "step": 1575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7887068244367887e-05,
+      "loss": 2.1974,
+      "step": 1580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.787417008691605e-05,
+      "loss": 2.1828,
+      "step": 1585
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7861237361853297e-05,
+      "loss": 2.1431,
+      "step": 1590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7848270125954295e-05,
+      "loss": 2.0985,
+      "step": 1595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7835268436145217e-05,
+      "loss": 2.1519,
+      "step": 1600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7822232349503485e-05,
+      "loss": 2.1818,
+      "step": 1605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.780916192325753e-05,
+      "loss": 2.1099,
+      "step": 1610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.779605721478652e-05,
+      "loss": 2.1819,
+      "step": 1615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.778291828162015e-05,
+      "loss": 2.1122,
+      "step": 1620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7769745181438333e-05,
+      "loss": 2.1088,
+      "step": 1625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.775653797207099e-05,
+      "loss": 2.1283,
+      "step": 1630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.774329671149777e-05,
+      "loss": 2.1591,
+      "step": 1635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7730021457847822e-05,
+      "loss": 2.0889,
+      "step": 1640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7716712269399512e-05,
+      "loss": 2.1279,
+      "step": 1645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7703369204580187e-05,
+      "loss": 2.1395,
+      "step": 1650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7689992321965907e-05,
+      "loss": 2.1582,
+      "step": 1655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7676581680281194e-05,
+      "loss": 2.0378,
+      "step": 1660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.766313733839877e-05,
+      "loss": 2.0885,
+      "step": 1665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7649659355339313e-05,
+      "loss": 2.1027,
+      "step": 1670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7636147790271163e-05,
+      "loss": 2.1033,
+      "step": 1675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7622602702510104e-05,
+      "loss": 2.1079,
+      "step": 1680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7609024151519082e-05,
+      "loss": 1.9935,
+      "step": 1685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7595412196907937e-05,
+      "loss": 2.1249,
+      "step": 1690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.758176689843316e-05,
+      "loss": 2.1409,
+      "step": 1695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.756808831599762e-05,
+      "loss": 2.0591,
+      "step": 1700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.75543765096503e-05,
+      "loss": 2.0628,
+      "step": 1705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.754063153958603e-05,
+      "loss": 2.1537,
+      "step": 1710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7526853466145248e-05,
+      "loss": 2.1149,
+      "step": 1715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7513042349813692e-05,
+      "loss": 2.1868,
+      "step": 1720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7499198251222184e-05,
+      "loss": 2.1345,
+      "step": 1725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7485321231146314e-05,
+      "loss": 2.1213,
+      "step": 1730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.747141135050621e-05,
+      "loss": 2.0826,
+      "step": 1735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7457468670366258e-05,
+      "loss": 2.1221,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7443493251934826e-05,
+      "loss": 2.1497,
+      "step": 1745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7429485156564014e-05,
+      "loss": 1.9985,
+      "step": 1750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7415444445749367e-05,
+      "loss": 2.1146,
+      "step": 1755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.740137118112961e-05,
+      "loss": 2.0794,
+      "step": 1760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7387265424486384e-05,
+      "loss": 2.1051,
+      "step": 1765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7373127237743966e-05,
+      "loss": 2.0684,
+      "step": 1770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7358956682969012e-05,
+      "loss": 2.1082,
+      "step": 1775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7344753822370253e-05,
+      "loss": 2.1632,
+      "step": 1780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 2.1144,
+      "step": 1785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7316251433245157e-05,
+      "loss": 2.1013,
+      "step": 1790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7301952029844318e-05,
+      "loss": 2.0976,
+      "step": 1795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.728762057087014e-05,
+      "loss": 2.1355,
+      "step": 1800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.727325711923773e-05,
+      "loss": 2.1501,
+      "step": 1805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.725886173800265e-05,
+      "loss": 2.1784,
+      "step": 1810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.724443449036064e-05,
+      "loss": 2.1633,
+      "step": 1815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7229975439647316e-05,
+      "loss": 2.0991,
+      "step": 1820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7215484649337918e-05,
+      "loss": 2.1432,
+      "step": 1825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7200962183047028e-05,
+      "loss": 2.0887,
+      "step": 1830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7186408104528278e-05,
+      "loss": 2.168,
+      "step": 1835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.717182247767408e-05,
+      "loss": 2.1152,
+      "step": 1840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7157205366515347e-05,
+      "loss": 2.0208,
+      "step": 1845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.71425568352212e-05,
+      "loss": 2.0639,
+      "step": 1850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7127876948098706e-05,
+      "loss": 2.1513,
+      "step": 1855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.711316576959257e-05,
+      "loss": 2.0711,
+      "step": 1860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7098423364284878e-05,
+      "loss": 2.0285,
+      "step": 1865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7083649796894798e-05,
+      "loss": 2.1167,
+      "step": 1870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7068845132278294e-05,
+      "loss": 2.1325,
+      "step": 1875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7054009435427855e-05,
+      "loss": 2.1438,
+      "step": 1880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7039142771472197e-05,
+      "loss": 2.1716,
+      "step": 1885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7024245205675986e-05,
+      "loss": 2.0852,
+      "step": 1890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.700931680343954e-05,
+      "loss": 2.0742,
+      "step": 1895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6994357630298558e-05,
+      "loss": 2.1573,
+      "step": 1900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.697936775192381e-05,
+      "loss": 2.1373,
+      "step": 1905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6964347234120874e-05,
+      "loss": 2.1254,
+      "step": 1910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6949296142829833e-05,
+      "loss": 2.1806,
+      "step": 1915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6934214544124986e-05,
+      "loss": 2.0781,
+      "step": 1920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6919102504214557e-05,
+      "loss": 2.1924,
+      "step": 1925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.690396008944041e-05,
+      "loss": 2.0889,
+      "step": 1930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6888787366277752e-05,
+      "loss": 2.0604,
+      "step": 1935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6873584401334856e-05,
+      "loss": 2.102,
+      "step": 1940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.685835126135273e-05,
+      "loss": 2.0902,
+      "step": 1945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.684308801320488e-05,
+      "loss": 2.1354,
+      "step": 1950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6827794723896968e-05,
+      "loss": 2.0281,
+      "step": 1955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.681247146056654e-05,
+      "loss": 2.1993,
+      "step": 1960
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6797118290482737e-05,
+      "loss": 2.0583,
+      "step": 1965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.678173528104598e-05,
+      "loss": 2.1452,
+      "step": 1970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.676632249978769e-05,
+      "loss": 2.0835,
+      "step": 1975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6750880014369983e-05,
+      "loss": 2.0589,
+      "step": 1980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6735407892585378e-05,
+      "loss": 2.0813,
+      "step": 1985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6719906202356502e-05,
+      "loss": 2.0503,
+      "step": 1990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.670437501173578e-05,
+      "loss": 2.0837,
+      "step": 1995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6688814388905146e-05,
+      "loss": 2.081,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6673224402175743e-05,
+      "loss": 2.2051,
+      "step": 2005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6657605119987626e-05,
+      "loss": 2.137,
+      "step": 2010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6641956610909453e-05,
+      "loss": 2.0797,
+      "step": 2015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6626278943638177e-05,
+      "loss": 2.1248,
+      "step": 2020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6610572186998772e-05,
+      "loss": 2.075,
+      "step": 2025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.659483640994391e-05,
+      "loss": 2.1157,
+      "step": 2030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.657907168155366e-05,
+      "loss": 2.0207,
+      "step": 2035
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6563278071035182e-05,
+      "loss": 2.1333,
+      "step": 2040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6547455647722443e-05,
+      "loss": 2.1026,
+      "step": 2045
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6531604481075886e-05,
+      "loss": 2.0887,
+      "step": 2050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6515724640682146e-05,
+      "loss": 2.1504,
+      "step": 2055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.649981619625372e-05,
+      "loss": 2.0906,
+      "step": 2060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6483879217628696e-05,
+      "loss": 2.0802,
+      "step": 2065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6467913774770416e-05,
+      "loss": 2.093,
+      "step": 2070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6451919937767184e-05,
+      "loss": 2.0915,
+      "step": 2075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6435897776831956e-05,
+      "loss": 2.1309,
+      "step": 2080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.641984736230202e-05,
+      "loss": 2.1177,
+      "step": 2085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6403768764638708e-05,
+      "loss": 2.0697,
+      "step": 2090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6387662054427073e-05,
+      "loss": 2.115,
+      "step": 2095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.637152730237558e-05,
+      "loss": 2.1686,
+      "step": 2100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.63553645793158e-05,
+      "loss": 2.1476,
+      "step": 2105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6339173956202098e-05,
+      "loss": 2.0028,
+      "step": 2110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.632295550411132e-05,
+      "loss": 2.15,
+      "step": 2115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.630670929424247e-05,
+      "loss": 2.0369,
+      "step": 2120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6290435397916426e-05,
+      "loss": 2.0505,
+      "step": 2125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6274133886575597e-05,
+      "loss": 2.1644,
+      "step": 2130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6257804831783633e-05,
+      "loss": 2.1103,
+      "step": 2135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6241448305225085e-05,
+      "loss": 2.073,
+      "step": 2140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6225064378705124e-05,
+      "loss": 2.0924,
+      "step": 2145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6208653124149183e-05,
+      "loss": 2.1405,
+      "step": 2150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6192214613602687e-05,
+      "loss": 2.0792,
+      "step": 2155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6175748919230708e-05,
+      "loss": 2.1245,
+      "step": 2160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6159256113317646e-05,
+      "loss": 2.0774,
+      "step": 2165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.614273626826693e-05,
+      "loss": 2.0218,
+      "step": 2170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6126189456600694e-05,
+      "loss": 2.189,
+      "step": 2175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6109615750959445e-05,
+      "loss": 2.0766,
+      "step": 2180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.609301522410176e-05,
+      "loss": 2.0824,
+      "step": 2185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6076387948903965e-05,
+      "loss": 2.0341,
+      "step": 2190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.60597339983598e-05,
+      "loss": 2.1769,
+      "step": 2195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.604305344558012e-05,
+      "loss": 2.0102,
+      "step": 2200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6026346363792565e-05,
+      "loss": 2.1207,
+      "step": 2205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6009612826341226e-05,
+      "loss": 2.1376,
+      "step": 2210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5992852906686346e-05,
+      "loss": 2.0659,
+      "step": 2215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.597606667840398e-05,
+      "loss": 2.1084,
+      "step": 2220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5959254215185683e-05,
+      "loss": 2.0434,
+      "step": 2225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.594241559083817e-05,
+      "loss": 2.0074,
+      "step": 2230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5925550879283017e-05,
+      "loss": 2.0913,
+      "step": 2235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5908660154556313e-05,
+      "loss": 2.0843,
+      "step": 2240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.589174349080835e-05,
+      "loss": 2.0662,
+      "step": 2245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.587480096230329e-05,
+      "loss": 2.0527,
+      "step": 2250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5857832643418844e-05,
+      "loss": 2.0494,
+      "step": 2255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.5840838608645943e-05,
+      "loss": 2.121,
+      "step": 2260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5823818932588406e-05,
+      "loss": 2.1162,
+      "step": 2265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.580677368996262e-05,
+      "loss": 2.1505,
+      "step": 2270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.578970295559721e-05,
+      "loss": 2.1857,
+      "step": 2275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5772606804432704e-05,
+      "loss": 2.159,
+      "step": 2280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.5755485311521223e-05,
+      "loss": 2.0198,
+      "step": 2285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5738338552026126e-05,
+      "loss": 2.0953,
+      "step": 2290
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5721166601221697e-05,
+      "loss": 2.0743,
+      "step": 2295
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5703969534492812e-05,
+      "loss": 2.0906,
+      "step": 2300
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.56867474273346e-05,
+      "loss": 2.0602,
+      "step": 2305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5669500355352117e-05,
+      "loss": 2.1485,
+      "step": 2310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5652228394260026e-05,
+      "loss": 2.1229,
+      "step": 2315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.563493161988224e-05,
+      "loss": 2.1125,
+      "step": 2320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.561761010815161e-05,
+      "loss": 2.2295,
+      "step": 2325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5600263935109584e-05,
+      "loss": 2.0598,
+      "step": 2330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5582893176905867e-05,
+      "loss": 2.1363,
+      "step": 2335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.55654979097981e-05,
+      "loss": 2.0272,
+      "step": 2340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.554807821015151e-05,
+      "loss": 2.0747,
+      "step": 2345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5530634154438592e-05,
+      "loss": 2.0505,
+      "step": 2350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5513165819238764e-05,
+      "loss": 2.0423,
+      "step": 2355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5495673281238016e-05,
+      "loss": 2.0228,
+      "step": 2360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.547815661722861e-05,
+      "loss": 2.0991,
+      "step": 2365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5460615904108704e-05,
+      "loss": 2.0544,
+      "step": 2370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5443051218882042e-05,
+      "loss": 2.0873,
+      "step": 2375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5425462638657597e-05,
+      "loss": 2.0415,
+      "step": 2380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.540785024064925e-05,
+      "loss": 2.1257,
+      "step": 2385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5390214102175432e-05,
+      "loss": 2.0762,
+      "step": 2390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5372554300658807e-05,
+      "loss": 2.0193,
+      "step": 2395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5354870913625907e-05,
+      "loss": 2.0364,
+      "step": 2400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5337164018706822e-05,
+      "loss": 2.1337,
+      "step": 2405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5319433693634814e-05,
+      "loss": 2.0862,
+      "step": 2410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.530168001624603e-05,
+      "loss": 2.1117,
+      "step": 2415
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5283903064479125e-05,
+      "loss": 2.1142,
+      "step": 2420
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5266102916374918e-05,
+      "loss": 2.07,
+      "step": 2425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.524827965007608e-05,
+      "loss": 2.0325,
+      "step": 2430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5230433343826748e-05,
+      "loss": 2.106,
+      "step": 2435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.521256407597222e-05,
+      "loss": 2.0486,
+      "step": 2440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5194671924958597e-05,
+      "loss": 2.0351,
+      "step": 2445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5176756969332428e-05,
+      "loss": 2.1353,
+      "step": 2450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5158819287740372e-05,
+      "loss": 2.097,
+      "step": 2455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5140858958928872e-05,
+      "loss": 2.0776,
+      "step": 2460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5122876061743772e-05,
+      "loss": 2.1002,
+      "step": 2465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5104870675130006e-05,
+      "loss": 2.0693,
+      "step": 2470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5086842878131224e-05,
+      "loss": 1.9931,
+      "step": 2475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5068792749889473e-05,
+      "loss": 2.0622,
+      "step": 2480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5050720369644822e-05,
+      "loss": 2.0894,
+      "step": 2485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5032625816735027e-05,
+      "loss": 1.9941,
+      "step": 2490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5014509170595193e-05,
+      "loss": 2.0138,
+      "step": 2495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4996370510757397e-05,
+      "loss": 2.1484,
+      "step": 2500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4978209916850368e-05,
+      "loss": 2.0602,
+      "step": 2505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4960027468599128e-05,
+      "loss": 2.1208,
+      "step": 2510
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.494182324582463e-05,
+      "loss": 2.0889,
+      "step": 2515
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4923597328443423e-05,
+      "loss": 2.0358,
+      "step": 2520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4905349796467295e-05,
+      "loss": 2.11,
+      "step": 2525
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4887080730002918e-05,
+      "loss": 2.0893,
+      "step": 2530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4868790209251509e-05,
+      "loss": 2.0928,
+      "step": 2535
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4850478314508459e-05,
+      "loss": 2.0905,
+      "step": 2540
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4832145126162996e-05,
+      "loss": 2.0338,
+      "step": 2545
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4813790724697832e-05,
+      "loss": 2.07,
+      "step": 2550
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.479541519068879e-05,
+      "loss": 2.0586,
+      "step": 2555
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4777018604804484e-05,
+      "loss": 2.064,
+      "step": 2560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4758601047805934e-05,
+      "loss": 2.0792,
+      "step": 2565
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4740162600546217e-05,
+      "loss": 2.0141,
+      "step": 2570
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4721703343970137e-05,
+      "loss": 2.1187,
+      "step": 2575
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4703223359113836e-05,
+      "loss": 2.1028,
+      "step": 2580
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4684722727104455e-05,
+      "loss": 1.9526,
+      "step": 2585
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4666201529159781e-05,
+      "loss": 2.1122,
+      "step": 2590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.464765984658788e-05,
+      "loss": 2.0869,
+      "step": 2595
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4629097760786755e-05,
+      "loss": 2.0335,
+      "step": 2600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4610515353243964e-05,
+      "loss": 2.1524,
+      "step": 2605
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4591912705536294e-05,
+      "loss": 2.0777,
+      "step": 2610
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4573289899329374e-05,
+      "loss": 2.1048,
+      "step": 2615
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4554647016377334e-05,
+      "loss": 2.1319,
+      "step": 2620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4535984138522442e-05,
+      "loss": 1.9869,
+      "step": 2625
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4517301347694743e-05,
+      "loss": 1.9912,
+      "step": 2630
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4498598725911693e-05,
+      "loss": 2.0666,
+      "step": 2635
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4479876355277815e-05,
+      "loss": 2.1449,
+      "step": 2640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4461134317984326e-05,
+      "loss": 2.1128,
+      "step": 2645
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4442372696308782e-05,
+      "loss": 2.1342,
+      "step": 2650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4423591572614705e-05,
+      "loss": 2.0594,
+      "step": 2655
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4404791029351249e-05,
+      "loss": 2.098,
+      "step": 2660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4385971149052803e-05,
+      "loss": 2.0177,
+      "step": 2665
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4367132014338648e-05,
+      "loss": 2.0658,
+      "step": 2670
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4348273707912601e-05,
+      "loss": 2.1465,
+      "step": 2675
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4329396312562638e-05,
+      "loss": 2.0353,
+      "step": 2680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.431049991116053e-05,
+      "loss": 2.1074,
+      "step": 2685
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4291584586661494e-05,
+      "loss": 2.0332,
+      "step": 2690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.427265042210381e-05,
+      "loss": 2.1348,
+      "step": 2695
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4253697500608475e-05,
+      "loss": 2.0878,
+      "step": 2700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4234725905378826e-05,
+      "loss": 2.0234,
+      "step": 2705
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4215735719700167e-05,
+      "loss": 2.2253,
+      "step": 2710
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4196727026939434e-05,
+      "loss": 2.0872,
+      "step": 2715
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4177699910544793e-05,
+      "loss": 2.0715,
+      "step": 2720
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4158654454045292e-05,
+      "loss": 2.0865,
+      "step": 2725
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4139590741050505e-05,
+      "loss": 2.1196,
+      "step": 2730
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4120508855250134e-05,
+      "loss": 2.0734,
+      "step": 2735
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4101408880413672e-05,
+      "loss": 2.1182,
+      "step": 2740
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4082290900390016e-05,
+      "loss": 2.0678,
+      "step": 2745
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.406315499910711e-05,
+      "loss": 2.1042,
+      "step": 2750
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4044001260571574e-05,
+      "loss": 2.0094,
+      "step": 2755
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4024829768868324e-05,
+      "loss": 1.9978,
+      "step": 2760
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4005640608160223e-05,
+      "loss": 2.1296,
+      "step": 2765
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3986433862687699e-05,
+      "loss": 2.0618,
+      "step": 2770
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3967209616768369e-05,
+      "loss": 2.1228,
+      "step": 2775
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.394796795479669e-05,
+      "loss": 2.0857,
+      "step": 2780
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.392870896124356e-05,
+      "loss": 2.0438,
+      "step": 2785
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3909432720655976e-05,
+      "loss": 2.137,
+      "step": 2790
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3890139317656645e-05,
+      "loss": 2.0582,
+      "step": 2795
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3870828836943617e-05,
+      "loss": 2.0735,
+      "step": 2800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3851501363289907e-05,
+      "loss": 2.1025,
+      "step": 2805
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3832156981543145e-05,
+      "loss": 2.0914,
+      "step": 2810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.381279577662517e-05,
+      "loss": 2.1104,
+      "step": 2815
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3793417833531683e-05,
+      "loss": 2.0286,
+      "step": 2820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.377402323733186e-05,
+      "loss": 2.0399,
+      "step": 2825
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3754612073168002e-05,
+      "loss": 2.0269,
+      "step": 2830
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3735184426255117e-05,
+      "loss": 2.0914,
+      "step": 2835
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3715740381880592e-05,
+      "loss": 2.0965,
+      "step": 2840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.369628002540379e-05,
+      "loss": 2.049,
+      "step": 2845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3676803442255688e-05,
+      "loss": 2.1164,
+      "step": 2850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3657310717938495e-05,
+      "loss": 2.1314,
+      "step": 2855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.3637801938025282e-05,
+      "loss": 2.0716,
+      "step": 2860
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3618277188159601e-05,
+      "loss": 2.1287,
+      "step": 2865
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3598736554055116e-05,
+      "loss": 2.0827,
+      "step": 2870
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3579180121495225e-05,
+      "loss": 2.1128,
+      "step": 2875
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.3559607976332675e-05,
+      "loss": 2.0894,
+      "step": 2880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.354002020448919e-05,
+      "loss": 2.0323,
+      "step": 2885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3520416891955101e-05,
+      "loss": 2.1541,
+      "step": 2890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3500798124788965e-05,
+      "loss": 2.1082,
+      "step": 2895
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3481163989117177e-05,
+      "loss": 2.1331,
+      "step": 2900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3461514571133598e-05,
+      "loss": 2.0552,
+      "step": 2905
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3441849957099193e-05,
+      "loss": 2.1692,
+      "step": 2910
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3422170233341625e-05,
+      "loss": 2.1162,
+      "step": 2915
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3402475486254887e-05,
+      "loss": 2.109,
+      "step": 2920
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3382765802298934e-05,
+      "loss": 2.0635,
+      "step": 2925
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3363041267999291e-05,
+      "loss": 2.137,
+      "step": 2930
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.334330196994667e-05,
+      "loss": 2.0237,
+      "step": 2935
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3323547994796597e-05,
+      "loss": 2.0276,
+      "step": 2940
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.330377942926904e-05,
+      "loss": 2.0646,
+      "step": 2945
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3283996360148015e-05,
+      "loss": 2.1255,
+      "step": 2950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3264198874281198e-05,
+      "loss": 2.0739,
+      "step": 2955
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3244387058579566e-05,
+      "loss": 2.1738,
+      "step": 2960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3224561000017002e-05,
+      "loss": 2.1162,
+      "step": 2965
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.320472078562991e-05,
+      "loss": 2.0098,
+      "step": 2970
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3184866502516846e-05,
+      "loss": 2.0545,
+      "step": 2975
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3164998237838122e-05,
+      "loss": 2.0972,
+      "step": 2980
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3145116078815427e-05,
+      "loss": 2.0943,
+      "step": 2985
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3125220112731458e-05,
+      "loss": 2.0336,
+      "step": 2990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.310531042692951e-05,
+      "loss": 2.0876,
+      "step": 2995
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3085387108813116e-05,
+      "loss": 2.0413,
+      "step": 3000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3065450245845652e-05,
+      "loss": 2.1667,
+      "step": 3005
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3045499925549962e-05,
+      "loss": 2.082,
+      "step": 3010
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.302553623550796e-05,
+      "loss": 2.0349,
+      "step": 3015
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.300555926336025e-05,
+      "loss": 2.1066,
+      "step": 3020
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.298556909680576e-05,
+      "loss": 2.1254,
+      "step": 3025
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.296556582360133e-05,
+      "loss": 2.078,
+      "step": 3030
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.2945549531561334e-05,
+      "loss": 2.1126,
+      "step": 3035
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2925520308557316e-05,
+      "loss": 2.0396,
+      "step": 3040
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2905478242517563e-05,
+      "loss": 2.0167,
+      "step": 3045
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2885423421426762e-05,
+      "loss": 2.0488,
+      "step": 3050
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.286535593332559e-05,
+      "loss": 2.1691,
+      "step": 3055
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.2845275866310325e-05,
+      "loss": 2.1113,
+      "step": 3060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2825183308532478e-05,
+      "loss": 2.1532,
+      "step": 3065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2805078348198387e-05,
+      "loss": 2.0869,
+      "step": 3070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2784961073568835e-05,
+      "loss": 2.1402,
+      "step": 3075
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.2764831572958674e-05,
+      "loss": 2.045,
+      "step": 3080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.274468993473642e-05,
+      "loss": 2.1223,
+      "step": 3085
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2724536247323868e-05,
+      "loss": 2.085,
+      "step": 3090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.270437059919573e-05,
+      "loss": 2.1571,
+      "step": 3095
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2684193078879205e-05,
+      "loss": 2.1077,
+      "step": 3100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.266400377495362e-05,
+      "loss": 2.0425,
+      "step": 3105
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.2643802776050027e-05,
+      "loss": 2.1263,
+      "step": 3110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2623590170850826e-05,
+      "loss": 2.1301,
+      "step": 3115
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2603366048089363e-05,
+      "loss": 2.0984,
+      "step": 3120
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2583130496549545e-05,
+      "loss": 2.0693,
+      "step": 3125
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2562883605065456e-05,
+      "loss": 2.1245,
+      "step": 3130
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.2542625462520968e-05,
+      "loss": 2.1125,
+      "step": 3135
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2522356157849323e-05,
+      "loss": 2.1237,
+      "step": 3140
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2502075780032792e-05,
+      "loss": 2.0992,
+      "step": 3145
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.248178441810224e-05,
+      "loss": 2.0085,
+      "step": 3150
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2461482161136756e-05,
+      "loss": 2.1386,
+      "step": 3155
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.244116909826326e-05,
+      "loss": 2.0545,
+      "step": 3160
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2420845318656107e-05,
+      "loss": 2.0876,
+      "step": 3165
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2400510911536705e-05,
+      "loss": 2.0853,
+      "step": 3170
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2380165966173105e-05,
+      "loss": 2.1409,
+      "step": 3175
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2359810571879638e-05,
+      "loss": 2.071,
+      "step": 3180
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2339444818016488e-05,
+      "loss": 2.1455,
+      "step": 3185
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2319068793989326e-05,
+      "loss": 2.1625,
+      "step": 3190
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.229868258924891e-05,
+      "loss": 2.0579,
+      "step": 3195
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2278286293290693e-05,
+      "loss": 2.045,
+      "step": 3200
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.225787999565442e-05,
+      "loss": 2.0943,
+      "step": 3205
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2237463785923752e-05,
+      "loss": 2.0928,
+      "step": 3210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.221703775372586e-05,
+      "loss": 2.071,
+      "step": 3215
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2196601988731035e-05,
+      "loss": 2.21,
+      "step": 3220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2176156580652293e-05,
+      "loss": 2.0664,
+      "step": 3225
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2155701619244997e-05,
+      "loss": 1.9691,
+      "step": 3230
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.213523719430643e-05,
+      "loss": 2.0508,
+      "step": 3235
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.211476339567543e-05,
+      "loss": 1.9978,
+      "step": 3240
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2094280313231985e-05,
+      "loss": 2.1024,
+      "step": 3245
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2073788036896842e-05,
+      "loss": 2.0739,
+      "step": 3250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2053286656631093e-05,
+      "loss": 2.1285,
+      "step": 3255
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2032776262435824e-05,
+      "loss": 2.0807,
+      "step": 3260
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2012256944351664e-05,
+      "loss": 2.0765,
+      "step": 3265
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1991728792458432e-05,
+      "loss": 2.0483,
+      "step": 3270
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1971191896874734e-05,
+      "loss": 2.0142,
+      "step": 3275
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1950646347757548e-05,
+      "loss": 1.9939,
+      "step": 3280
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.1930092235301847e-05,
+      "loss": 2.17,
+      "step": 3285
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1909529649740188e-05,
+      "loss": 2.1462,
+      "step": 3290
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1888958681342345e-05,
+      "loss": 2.1365,
+      "step": 3295
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.186837942041487e-05,
+      "loss": 2.0604,
+      "step": 3300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1847791957300735e-05,
+      "loss": 2.1323,
+      "step": 3305
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.1827196382378914e-05,
+      "loss": 2.0574,
+      "step": 3310
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1806592786063991e-05,
+      "loss": 2.1837,
+      "step": 3315
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1785981258805764e-05,
+      "loss": 2.0659,
+      "step": 3320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1765361891088851e-05,
+      "loss": 2.0591,
+      "step": 3325
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.1744734773432282e-05,
+      "loss": 2.1358,
+      "step": 3330
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.172409999638912e-05,
+      "loss": 2.0352,
+      "step": 3335
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.170345765054605e-05,
+      "loss": 2.0648,
+      "step": 3340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1682807826522973e-05,
+      "loss": 1.956,
+      "step": 3345
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1662150614972633e-05,
+      "loss": 2.0617,
+      "step": 3350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.16414861065802e-05,
+      "loss": 2.1478,
+      "step": 3355
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.1620814392062873e-05,
+      "loss": 2.097,
+      "step": 3360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1600135562169494e-05,
+      "loss": 2.0129,
+      "step": 3365
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1579449707680137e-05,
+      "loss": 2.0587,
+      "step": 3370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1558756919405712e-05,
+      "loss": 2.1292,
+      "step": 3375
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.1538057288187573e-05,
+      "loss": 2.1235,
+      "step": 3380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.151735090489711e-05,
+      "loss": 2.0787,
+      "step": 3385
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1496637860435355e-05,
+      "loss": 2.0729,
+      "step": 3390
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1475918245732588e-05,
+      "loss": 2.0815,
+      "step": 3395
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1455192151747931e-05,
+      "loss": 2.1627,
+      "step": 3400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1434459669468944e-05,
+      "loss": 2.1101,
+      "step": 3405
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1413720889911232e-05,
+      "loss": 2.0376,
+      "step": 3410
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1392975904118058e-05,
+      "loss": 2.0389,
+      "step": 3415
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1372224803159912e-05,
+      "loss": 2.0645,
+      "step": 3420
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.135146767813414e-05,
+      "loss": 2.0577,
+      "step": 3425
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.133070462016454e-05,
+      "loss": 2.0651,
+      "step": 3430
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1309935720400934e-05,
+      "loss": 2.1368,
+      "step": 3435
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1289161070018813e-05,
+      "loss": 1.9803,
+      "step": 3440
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1268380760218901e-05,
+      "loss": 2.1219,
+      "step": 3445
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.124759488222677e-05,
+      "loss": 2.0363,
+      "step": 3450
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1226803527292435e-05,
+      "loss": 2.0921,
+      "step": 3455
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1206006786689951e-05,
+      "loss": 2.0454,
+      "step": 3460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.118520475171703e-05,
+      "loss": 2.0804,
+      "step": 3465
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1164397513694608e-05,
+      "loss": 2.0569,
+      "step": 3470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.114358516396647e-05,
+      "loss": 2.128,
+      "step": 3475
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1122767793898853e-05,
+      "loss": 2.0417,
+      "step": 3480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1101945494880013e-05,
+      "loss": 2.0421,
+      "step": 3485
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1081118358319857e-05,
+      "loss": 2.108,
+      "step": 3490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1060286475649529e-05,
+      "loss": 2.0822,
+      "step": 3495
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1039449938321e-05,
+      "loss": 2.0601,
+      "step": 3500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1018608837806685e-05,
+      "loss": 2.1049,
+      "step": 3505
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.0997763265599023e-05,
+      "loss": 2.1129,
+      "step": 3510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0976913313210084e-05,
+      "loss": 2.1099,
+      "step": 3515
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.095605907217118e-05,
+      "loss": 2.0894,
+      "step": 3520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.093520063403243e-05,
+      "loss": 2.1987,
+      "step": 3525
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0914338090362395e-05,
+      "loss": 2.0848,
+      "step": 3530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.0893471532747651e-05,
+      "loss": 2.1345,
+      "step": 3535
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0872601052792398e-05,
+      "loss": 2.1285,
+      "step": 3540
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0851726742118051e-05,
+      "loss": 2.1164,
+      "step": 3545
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0830848692362852e-05,
+      "loss": 2.1502,
+      "step": 3550
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.0809966995181443e-05,
+      "loss": 2.082,
+      "step": 3555
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.078908174224449e-05,
+      "loss": 2.0757,
+      "step": 3560
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0768193025238265e-05,
+      "loss": 2.1279,
+      "step": 3565
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 2.0593,
+      "step": 3570
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0726405565838713e-05,
+      "loss": 2.0968,
+      "step": 3575
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0705507006892356e-05,
+      "loss": 2.1628,
+      "step": 3580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.0684605350769862e-05,
+      "loss": 2.0794,
+      "step": 3585
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0663700689229506e-05,
+      "loss": 1.9637,
+      "step": 3590
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.064279311404277e-05,
+      "loss": 2.146,
+      "step": 3595
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0621882716993916e-05,
+      "loss": 2.1017,
+      "step": 3600
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.06009695898796e-05,
+      "loss": 2.1323,
+      "step": 3605
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.0580053824508463e-05,
+      "loss": 2.0876,
+      "step": 3610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.055913551270072e-05,
+      "loss": 2.1592,
+      "step": 3615
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0538214746287775e-05,
+      "loss": 2.0061,
+      "step": 3620
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0517291617111809e-05,
+      "loss": 2.0367,
+      "step": 3625
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0496366217025362e-05,
+      "loss": 2.0932,
+      "step": 3630
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0475438637890958e-05,
+      "loss": 2.061,
+      "step": 3635
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0454508971580675e-05,
+      "loss": 2.0244,
+      "step": 3640
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0433577309975763e-05,
+      "loss": 2.0368,
+      "step": 3645
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0412643744966226e-05,
+      "loss": 2.1809,
+      "step": 3650
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0391708368450429e-05,
+      "loss": 2.1164,
+      "step": 3655
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0370771272334679e-05,
+      "loss": 2.111,
+      "step": 3660
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0349832548532848e-05,
+      "loss": 2.1022,
+      "step": 3665
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0328892288965939e-05,
+      "loss": 2.1016,
+      "step": 3670
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0307950585561705e-05,
+      "loss": 2.1116,
+      "step": 3675
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0287007530254233e-05,
+      "loss": 2.0237,
+      "step": 3680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.026606321498355e-05,
+      "loss": 2.0666,
+      "step": 3685
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0245117731695213e-05,
+      "loss": 2.023,
+      "step": 3690
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0224171172339899e-05,
+      "loss": 2.1712,
+      "step": 3695
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0203223628873019e-05,
+      "loss": 2.1528,
+      "step": 3700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0182275193254296e-05,
+      "loss": 2.056,
+      "step": 3705
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0161325957447377e-05,
+      "loss": 2.092,
+      "step": 3710
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0140376013419419e-05,
+      "loss": 2.0285,
+      "step": 3715
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0119425453140687e-05,
+      "loss": 2.1339,
+      "step": 3720
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0098474368584153e-05,
+      "loss": 2.0977,
+      "step": 3725
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0077522851725086e-05,
+      "loss": 2.0364,
+      "step": 3730
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0056570994540658e-05,
+      "loss": 2.2121,
+      "step": 3735
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0035618889009535e-05,
+      "loss": 2.0931,
+      "step": 3740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0014666627111467e-05,
+      "loss": 2.0551,
+      "step": 3745
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.993714300826901e-06,
+      "loss": 2.0648,
+      "step": 3750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.972762002136554e-06,
+      "loss": 2.0979,
+      "step": 3755
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.951809823021037e-06,
+      "loss": 2.0781,
+      "step": 3760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.930857855460421e-06,
+      "loss": 2.1147,
+      "step": 3765
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.909906191433855e-06,
+      "loss": 2.0669,
+      "step": 3770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.88895492291916e-06,
+      "loss": 2.0526,
+      "step": 3775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.868004141892412e-06,
+      "loss": 2.145,
+      "step": 3780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.847053940327556e-06,
+      "loss": 2.0464,
+      "step": 3785
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.826104410195976e-06,
+      "loss": 2.1414,
+      "step": 3790
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.805155643466133e-06,
+      "loss": 2.0413,
+      "step": 3795
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.784207732103122e-06,
+      "loss": 2.0787,
+      "step": 3800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.76326076806828e-06,
+      "loss": 2.0104,
+      "step": 3805
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.742314843318796e-06,
+      "loss": 2.0562,
+      "step": 3810
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.721370049807287e-06,
+      "loss": 2.0541,
+      "step": 3815
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.700426479481408e-06,
+      "loss": 2.1302,
+      "step": 3820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.67948422428345e-06,
+      "loss": 2.0897,
+      "step": 3825
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.658543376149917e-06,
+      "loss": 2.0729,
+      "step": 3830
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.637604027011145e-06,
+      "loss": 2.0358,
+      "step": 3835
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.61666626879089e-06,
+      "loss": 2.025,
+      "step": 3840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.595730193405924e-06,
+      "loss": 2.0552,
+      "step": 3845
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.574795892765628e-06,
+      "loss": 2.1235,
+      "step": 3850
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.553863458771585e-06,
+      "loss": 2.097,
+      "step": 3855
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.532932983317202e-06,
+      "loss": 2.1525,
+      "step": 3860
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.512004558287277e-06,
+      "loss": 2.0289,
+      "step": 3865
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.491078275557596e-06,
+      "loss": 2.0842,
+      "step": 3870
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.470154226994564e-06,
+      "loss": 1.9881,
+      "step": 3875
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.449232504454763e-06,
+      "loss": 2.0582,
+      "step": 3880
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.428313199784557e-06,
+      "loss": 2.1515,
+      "step": 3885
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.407396404819717e-06,
+      "loss": 2.0348,
+      "step": 3890
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.38648221138498e-06,
+      "loss": 2.1124,
+      "step": 3895
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.365570711293658e-06,
+      "loss": 2.1524,
+      "step": 3900
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.34466199634726e-06,
+      "loss": 2.0611,
+      "step": 3905
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.323756158335054e-06,
+      "loss": 2.0999,
+      "step": 3910
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.302853289033673e-06,
+      "loss": 2.102,
+      "step": 3915
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.281953480206725e-06,
+      "loss": 2.0639,
+      "step": 3920
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.26105682360439e-06,
+      "loss": 2.0317,
+      "step": 3925
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.240163410962999e-06,
+      "loss": 2.0991,
+      "step": 3930
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.219273334004635e-06,
+      "loss": 2.1076,
+      "step": 3935
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.19838668443676e-06,
+      "loss": 1.9736,
+      "step": 3940
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.177503553951775e-06,
+      "loss": 1.9861,
+      "step": 3945
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.156624034226623e-06,
+      "loss": 2.0982,
+      "step": 3950
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.13574821692242e-06,
+      "loss": 2.0654,
+      "step": 3955
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.114876193684011e-06,
+      "loss": 2.1508,
+      "step": 3960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.094008056139587e-06,
+      "loss": 2.0808,
+      "step": 3965
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.07314389590028e-06,
+      "loss": 2.1509,
+      "step": 3970
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.052283804559774e-06,
+      "loss": 2.0443,
+      "step": 3975
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.031427873693875e-06,
+      "loss": 2.0152,
+      "step": 3980
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.010576194860129e-06,
+      "loss": 2.0672,
+      "step": 3985
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.989728859597424e-06,
+      "loss": 2.1208,
+      "step": 3990
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.968885959425567e-06,
+      "loss": 2.1253,
+      "step": 3995
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.9480475858449e-06,
+      "loss": 2.0516,
+      "step": 4000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.927213830335904e-06,
+      "loss": 2.0799,
+      "step": 4005
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.906384784358767e-06,
+      "loss": 2.0959,
+      "step": 4010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.88556053935301e-06,
+      "loss": 2.0965,
+      "step": 4015
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.864741186737094e-06,
+      "loss": 2.0327,
+      "step": 4020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.843926817907976e-06,
+      "loss": 2.1073,
+      "step": 4025
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.823117524240751e-06,
+      "loss": 1.9879,
+      "step": 4030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 8.802313397088223e-06,
+      "loss": 2.0548,
+      "step": 4035
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.781514527780531e-06,
+      "loss": 2.1127,
+      "step": 4040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.760721007624718e-06,
+      "loss": 2.0538,
+      "step": 4045
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.739932927904345e-06,
+      "loss": 2.1022,
+      "step": 4050
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.719150379879099e-06,
+      "loss": 2.1688,
+      "step": 4055
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 8.698373454784374e-06,
+      "loss": 2.0854,
+      "step": 4060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.677602243830882e-06,
+      "loss": 2.1103,
+      "step": 4065
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.656836838204255e-06,
+      "loss": 2.0988,
+      "step": 4070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.63607732906463e-06,
+      "loss": 2.0324,
+      "step": 4075
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.615323807546258e-06,
+      "loss": 2.0913,
+      "step": 4080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 8.594576364757126e-06,
+      "loss": 2.0598,
+      "step": 4085
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.573835091778505e-06,
+      "loss": 2.0585,
+      "step": 4090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.553100079664598e-06,
+      "loss": 2.0281,
+      "step": 4095
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.53237141944212e-06,
+      "loss": 2.1304,
+      "step": 4100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.511649202109902e-06,
+      "loss": 2.1373,
+      "step": 4105
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.490933518638488e-06,
+      "loss": 2.0538,
+      "step": 4110
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.470224459969737e-06,
+      "loss": 2.1275,
+      "step": 4115
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.449522117016432e-06,
+      "loss": 2.088,
+      "step": 4120
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.428826580661867e-06,
+      "loss": 2.1329,
+      "step": 4125
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.408137941759459e-06,
+      "loss": 2.1218,
+      "step": 4130
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.387456291132343e-06,
+      "loss": 2.0591,
+      "step": 4135
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.366781719572978e-06,
+      "loss": 2.2194,
+      "step": 4140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.34611431784274e-06,
+      "loss": 2.0977,
+      "step": 4145
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.325454176671541e-06,
+      "loss": 2.1123,
+      "step": 4150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.304801386757407e-06,
+      "loss": 2.0763,
+      "step": 4155
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.284156038766095e-06,
+      "loss": 2.0852,
+      "step": 4160
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 2.0575,
+      "step": 4165
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.242888031051234e-06,
+      "loss": 2.075,
+      "step": 4170
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.222265552494262e-06,
+      "loss": 2.0556,
+      "step": 4175
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.201650878192468e-06,
+      "loss": 2.0116,
+      "step": 4180
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.18104409864429e-06,
+      "loss": 1.9719,
+      "step": 4185
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.160445304313497e-06,
+      "loss": 2.0438,
+      "step": 4190
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.139854585628806e-06,
+      "loss": 2.0374,
+      "step": 4195
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.119272032983487e-06,
+      "loss": 2.1279,
+      "step": 4200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.098697736734957e-06,
+      "loss": 2.0143,
+      "step": 4205
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.078131787204382e-06,
+      "loss": 2.1391,
+      "step": 4210
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.0575742746763e-06,
+      "loss": 2.0629,
+      "step": 4215
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.037025289398196e-06,
+      "loss": 2.1205,
+      "step": 4220
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.016484921580127e-06,
+      "loss": 2.0495,
+      "step": 4225
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.995953261394321e-06,
+      "loss": 2.0628,
+      "step": 4230
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.975430398974775e-06,
+      "loss": 2.0952,
+      "step": 4235
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.954916424416869e-06,
+      "loss": 2.0941,
+      "step": 4240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.934411427776958e-06,
+      "loss": 2.1818,
+      "step": 4245
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.913915499071994e-06,
+      "loss": 2.0785,
+      "step": 4250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.89342872827911e-06,
+      "loss": 2.1087,
+      "step": 4255
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.872951205335242e-06,
+      "loss": 2.0482,
+      "step": 4260
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.85248302013673e-06,
+      "loss": 2.2156,
+      "step": 4265
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.832024262538914e-06,
+      "loss": 2.04,
+      "step": 4270
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.811575022355749e-06,
+      "loss": 2.0792,
+      "step": 4275
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.791135389359417e-06,
+      "loss": 1.9923,
+      "step": 4280
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.770705453279916e-06,
+      "loss": 2.1058,
+      "step": 4285
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.750285303804674e-06,
+      "loss": 2.0794,
+      "step": 4290
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.729875030578157e-06,
+      "loss": 2.1222,
+      "step": 4295
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.70947472320148e-06,
+      "loss": 2.0841,
+      "step": 4300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.689084471232001e-06,
+      "loss": 2.0811,
+      "step": 4305
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.668704364182935e-06,
+      "loss": 2.0769,
+      "step": 4310
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.648334491522966e-06,
+      "loss": 2.1287,
+      "step": 4315
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.627974942675846e-06,
+      "loss": 2.0796,
+      "step": 4320
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.607625807020003e-06,
+      "loss": 2.013,
+      "step": 4325
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.587287173888156e-06,
+      "loss": 2.082,
+      "step": 4330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.566959132566914e-06,
+      "loss": 2.0406,
+      "step": 4335
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.546641772296388e-06,
+      "loss": 1.9931,
+      "step": 4340
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.526335182269804e-06,
+      "loss": 2.019,
+      "step": 4345
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.506039451633101e-06,
+      "loss": 2.0648,
+      "step": 4350
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.485754669484549e-06,
+      "loss": 1.9721,
+      "step": 4355
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.465480924874348e-06,
+      "loss": 2.0578,
+      "step": 4360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.445218306804253e-06,
+      "loss": 2.1054,
+      "step": 4365
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.424966904227167e-06,
+      "loss": 2.0934,
+      "step": 4370
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.404726806046753e-06,
+      "loss": 2.0193,
+      "step": 4375
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.384498101117062e-06,
+      "loss": 2.0823,
+      "step": 4380
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.364280878242114e-06,
+      "loss": 2.1134,
+      "step": 4385
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.344075226175527e-06,
+      "loss": 2.1058,
+      "step": 4390
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.323881233620129e-06,
+      "loss": 2.1015,
+      "step": 4395
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.303698989227554e-06,
+      "loss": 2.0341,
+      "step": 4400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.283528581597867e-06,
+      "loss": 2.1213,
+      "step": 4405
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.263370099279173e-06,
+      "loss": 2.0839,
+      "step": 4410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.2432236307672136e-06,
+      "loss": 2.0384,
+      "step": 4415
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.223089264505001e-06,
+      "loss": 2.0842,
+      "step": 4420
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.202967088882405e-06,
+      "loss": 2.087,
+      "step": 4425
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.1828571922358e-06,
+      "loss": 1.9524,
+      "step": 4430
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.162759662847638e-06,
+      "loss": 2.0555,
+      "step": 4435
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.142674588946079e-06,
+      "loss": 2.1134,
+      "step": 4440
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.122602058704615e-06,
+      "loss": 2.0735,
+      "step": 4445
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.102542160241664e-06,
+      "loss": 2.1126,
+      "step": 4450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.082494981620184e-06,
+      "loss": 2.0787,
+      "step": 4455
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.062460610847308e-06,
+      "loss": 1.9744,
+      "step": 4460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.042439135873932e-06,
+      "loss": 2.1265,
+      "step": 4465
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.022430644594339e-06,
+      "loss": 1.9675,
+      "step": 4470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.002435224845822e-06,
+      "loss": 2.0836,
+      "step": 4475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.98245296440828e-06,
+      "loss": 2.0921,
+      "step": 4480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.962483951003851e-06,
+      "loss": 2.0507,
+      "step": 4485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.942528272296505e-06,
+      "loss": 1.9563,
+      "step": 4490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.922586015891695e-06,
+      "loss": 2.1548,
+      "step": 4495
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.902657269335933e-06,
+      "loss": 2.1145,
+      "step": 4500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.882742120116419e-06,
+      "loss": 2.0528,
+      "step": 4505
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.862840655660679e-06,
+      "loss": 2.1135,
+      "step": 4510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.842952963336154e-06,
+      "loss": 2.0965,
+      "step": 4515
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.82307913044981e-06,
+      "loss": 2.042,
+      "step": 4520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.803219244247798e-06,
+      "loss": 2.1888,
+      "step": 4525
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.783373391915027e-06,
+      "loss": 2.0114,
+      "step": 4530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.763541660574795e-06,
+      "loss": 2.0276,
+      "step": 4535
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.74372413728841e-06,
+      "loss": 2.0472,
+      "step": 4540
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.723920909054822e-06,
+      "loss": 2.1113,
+      "step": 4545
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.704132062810204e-06,
+      "loss": 2.1455,
+      "step": 4550
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.6843576854276005e-06,
+      "loss": 2.0414,
+      "step": 4555
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.6645978637165485e-06,
+      "loss": 2.0696,
+      "step": 4560
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.644852684422669e-06,
+      "loss": 2.0612,
+      "step": 4565
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.625122234227306e-06,
+      "loss": 1.9787,
+      "step": 4570
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.605406599747156e-06,
+      "loss": 2.115,
+      "step": 4575
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.585705867533859e-06,
+      "loss": 2.1017,
+      "step": 4580
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.566020124073638e-06,
+      "loss": 2.0864,
+      "step": 4585
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.546349455786926e-06,
+      "loss": 2.1133,
+      "step": 4590
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.52669394902796e-06,
+      "loss": 2.0318,
+      "step": 4595
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.507053690084426e-06,
+      "loss": 2.0634,
+      "step": 4600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.487428765177073e-06,
+      "loss": 2.0525,
+      "step": 4605
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.467819260459332e-06,
+      "loss": 2.13,
+      "step": 4610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.448225262016938e-06,
+      "loss": 2.044,
+      "step": 4615
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.428646855867553e-06,
+      "loss": 2.1884,
+      "step": 4620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.409084127960398e-06,
+      "loss": 2.1139,
+      "step": 4625
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.389537164175852e-06,
+      "loss": 2.1151,
+      "step": 4630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.370006050325095e-06,
+      "loss": 2.0269,
+      "step": 4635
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.350490872149737e-06,
+      "loss": 2.1715,
+      "step": 4640
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.3309917153214115e-06,
+      "loss": 2.048,
+      "step": 4645
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.311508665441428e-06,
+      "loss": 2.1836,
+      "step": 4650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.292041808040393e-06,
+      "loss": 1.9866,
+      "step": 4655
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.272591228577816e-06,
+      "loss": 2.1363,
+      "step": 4660
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.253157012441751e-06,
+      "loss": 2.0852,
+      "step": 4665
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.233739244948416e-06,
+      "loss": 2.0689,
+      "step": 4670
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.214338011341825e-06,
+      "loss": 2.0112,
+      "step": 4675
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.194953396793401e-06,
+      "loss": 2.0693,
+      "step": 4680
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.175585486401612e-06,
+      "loss": 2.0233,
+      "step": 4685
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.156234365191598e-06,
+      "loss": 2.0183,
+      "step": 4690
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.136900118114789e-06,
+      "loss": 1.9348,
+      "step": 4695
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.117582830048539e-06,
+      "loss": 2.0473,
+      "step": 4700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.0982825857957565e-06,
+      "loss": 2.0919,
+      "step": 4705
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.0789994700845234e-06,
+      "loss": 2.0806,
+      "step": 4710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.059733567567725e-06,
+      "loss": 2.0353,
+      "step": 4715
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.040484962822686e-06,
+      "loss": 2.0148,
+      "step": 4720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.021253740350793e-06,
+      "loss": 2.0244,
+      "step": 4725
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.002039984577121e-06,
+      "loss": 2.1046,
+      "step": 4730
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.982843779850067e-06,
+      "loss": 2.0956,
+      "step": 4735
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.963665210440983e-06,
+      "loss": 2.0678,
+      "step": 4740
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.944504360543797e-06,
+      "loss": 2.0827,
+      "step": 4745
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.9253613142746505e-06,
+      "loss": 2.0919,
+      "step": 4750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.906236155671529e-06,
+      "loss": 2.1765,
+      "step": 4755
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.887128968693887e-06,
+      "loss": 2.0592,
+      "step": 4760
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.8680398372222834e-06,
+      "loss": 2.015,
+      "step": 4765
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.8489688450580215e-06,
+      "loss": 2.0643,
+      "step": 4770
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.829916075922766e-06,
+      "loss": 2.1475,
+      "step": 4775
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.810881613458176e-06,
+      "loss": 2.0388,
+      "step": 4780
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.791865541225561e-06,
+      "loss": 2.0519,
+      "step": 4785
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.772867942705485e-06,
+      "loss": 2.0604,
+      "step": 4790
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.753888901297416e-06,
+      "loss": 2.0753,
+      "step": 4795
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.734928500319352e-06,
+      "loss": 1.9892,
+      "step": 4800
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.71598682300747e-06,
+      "loss": 2.124,
+      "step": 4805
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.697063952515745e-06,
+      "loss": 2.0879,
+      "step": 4810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.678159971915578e-06,
+      "loss": 2.0136,
+      "step": 4815
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.659274964195463e-06,
+      "loss": 2.1105,
+      "step": 4820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.6404090122605925e-06,
+      "loss": 2.0453,
+      "step": 4825
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.621562198932499e-06,
+      "loss": 2.0919,
+      "step": 4830
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.602734606948714e-06,
+      "loss": 1.9485,
+      "step": 4835
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.583926318962376e-06,
+      "loss": 2.124,
+      "step": 4840
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.565137417541866e-06,
+      "loss": 2.139,
+      "step": 4845
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.546367985170483e-06,
+      "loss": 2.0599,
+      "step": 4850
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.527618104246044e-06,
+      "loss": 2.0426,
+      "step": 4855
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.5088878570805336e-06,
+      "loss": 2.0937,
+      "step": 4860
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.490177325899744e-06,
+      "loss": 2.1013,
+      "step": 4865
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.471486592842927e-06,
+      "loss": 2.056,
+      "step": 4870
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.452815739962404e-06,
+      "loss": 2.0075,
+      "step": 4875
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.434164849223225e-06,
+      "loss": 2.061,
+      "step": 4880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.415534002502818e-06,
+      "loss": 2.1069,
+      "step": 4885
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.396923281590609e-06,
+      "loss": 2.1454,
+      "step": 4890
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.378332768187667e-06,
+      "loss": 2.0682,
+      "step": 4895
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.3597625439063685e-06,
+      "loss": 2.0854,
+      "step": 4900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.341212690269994e-06,
+      "loss": 2.0926,
+      "step": 4905
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.322683288712416e-06,
+      "loss": 2.0883,
+      "step": 4910
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.304174420577719e-06,
+      "loss": 2.082,
+      "step": 4915
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.285686167119845e-06,
+      "loss": 2.0835,
+      "step": 4920
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.267218609502234e-06,
+      "loss": 2.1362,
+      "step": 4925
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.248771828797474e-06,
+      "loss": 2.0765,
+      "step": 4930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.230345905986945e-06,
+      "loss": 2.1331,
+      "step": 4935
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.211940921960455e-06,
+      "loss": 2.131,
+      "step": 4940
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.193556957515888e-06,
+      "loss": 2.002,
+      "step": 4945
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.1751940933588714e-06,
+      "loss": 2.1444,
+      "step": 4950
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.156852410102382e-06,
+      "loss": 2.0563,
+      "step": 4955
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.1385319882664135e-06,
+      "loss": 2.068,
+      "step": 4960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.120232908277642e-06,
+      "loss": 2.0141,
+      "step": 4965
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.101955250469025e-06,
+      "loss": 2.0439,
+      "step": 4970
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.0836990950794906e-06,
+      "loss": 2.0652,
+      "step": 4975
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.06546452225358e-06,
+      "loss": 2.067,
+      "step": 4980
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.047251612041072e-06,
+      "loss": 2.049,
+      "step": 4985
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.029060444396653e-06,
+      "loss": 2.0319,
+      "step": 4990
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.010891099179558e-06,
+      "loss": 2.1232,
+      "step": 4995
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.992743656153224e-06,
+      "loss": 2.1031,
+      "step": 5000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.974618194984934e-06,
+      "loss": 2.1366,
+      "step": 5005
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.956514795245469e-06,
+      "loss": 2.0794,
+      "step": 5010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.938433536408771e-06,
+      "loss": 1.9702,
+      "step": 5015
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9203744978515735e-06,
+      "loss": 2.1197,
+      "step": 5020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.9023377588530626e-06,
+      "loss": 2.0746,
+      "step": 5025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.884323398594536e-06,
+      "loss": 2.0547,
+      "step": 5030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.866331496159042e-06,
+      "loss": 2.1689,
+      "step": 5035
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.848362130531039e-06,
+      "loss": 2.0292,
+      "step": 5040
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.8304153805960595e-06,
+      "loss": 1.977,
+      "step": 5045
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.812491325140343e-06,
+      "loss": 2.0431,
+      "step": 5050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.7945900428505e-06,
+      "loss": 2.0355,
+      "step": 5055
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 4.776711612313171e-06,
+      "loss": 2.1682,
+      "step": 5060
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.758856112014675e-06,
+      "loss": 1.9793,
+      "step": 5065
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.741023620340668e-06,
+      "loss": 2.0138,
+      "step": 5070
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.723214215575796e-06,
+      "loss": 1.9691,
+      "step": 5075
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.705427975903359e-06,
+      "loss": 2.0467,
+      "step": 5080
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.687664979404959e-06,
+      "loss": 2.0741,
+      "step": 5085
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.669925304060159e-06,
+      "loss": 2.0443,
+      "step": 5090
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.652209027746143e-06,
+      "loss": 2.0996,
+      "step": 5095
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.634516228237372e-06,
+      "loss": 2.0144,
+      "step": 5100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.616846983205246e-06,
+      "loss": 2.0942,
+      "step": 5105
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.599201370217763e-06,
+      "loss": 2.063,
+      "step": 5110
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.581579466739171e-06,
+      "loss": 2.0336,
+      "step": 5115
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.563981350129637e-06,
+      "loss": 2.117,
+      "step": 5120
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.546407097644903e-06,
+      "loss": 2.0373,
+      "step": 5125
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.528856786435945e-06,
+      "loss": 2.0733,
+      "step": 5130
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.511330493548638e-06,
+      "loss": 2.0492,
+      "step": 5135
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.493828295923416e-06,
+      "loss": 2.1111,
+      "step": 5140
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.476350270394942e-06,
+      "loss": 2.037,
+      "step": 5145
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.458896493691754e-06,
+      "loss": 2.0392,
+      "step": 5150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.441467042435941e-06,
+      "loss": 2.0423,
+      "step": 5155
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.424061993142803e-06,
+      "loss": 2.0407,
+      "step": 5160
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.406681422220518e-06,
+      "loss": 2.0928,
+      "step": 5165
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.389325405969799e-06,
+      "loss": 2.1026,
+      "step": 5170
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.371994020583563e-06,
+      "loss": 2.077,
+      "step": 5175
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.3546873421466104e-06,
+      "loss": 2.1275,
+      "step": 5180
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.337405446635264e-06,
+      "loss": 2.0782,
+      "step": 5185
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.320148409917055e-06,
+      "loss": 1.9864,
+      "step": 5190
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.302916307750383e-06,
+      "loss": 2.007,
+      "step": 5195
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.285709215784187e-06,
+      "loss": 2.0399,
+      "step": 5200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.268527209557605e-06,
+      "loss": 2.01,
+      "step": 5205
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.2513703644996615e-06,
+      "loss": 2.0768,
+      "step": 5210
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.234238755928913e-06,
+      "loss": 2.0703,
+      "step": 5215
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.217132459053129e-06,
+      "loss": 2.1422,
+      "step": 5220
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.200051548968962e-06,
+      "loss": 2.1622,
+      "step": 5225
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.182996100661616e-06,
+      "loss": 2.1308,
+      "step": 5230
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.16596618900452e-06,
+      "loss": 2.1187,
+      "step": 5235
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.148961888758991e-06,
+      "loss": 2.0699,
+      "step": 5240
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.131983274573922e-06,
+      "loss": 2.0442,
+      "step": 5245
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.1150304209854375e-06,
+      "loss": 2.1129,
+      "step": 5250
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.098103402416571e-06,
+      "loss": 2.0754,
+      "step": 5255
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.081202293176945e-06,
+      "loss": 2.037,
+      "step": 5260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.064327167462437e-06,
+      "loss": 2.087,
+      "step": 5265
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.047478099354857e-06,
+      "loss": 2.0902,
+      "step": 5270
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.030655162821626e-06,
+      "loss": 2.0168,
+      "step": 5275
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.013858431715445e-06,
+      "loss": 2.0462,
+      "step": 5280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.9970879797739695e-06,
+      "loss": 2.077,
+      "step": 5285
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.980343880619494e-06,
+      "loss": 2.0406,
+      "step": 5290
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9636262077586225e-06,
+      "loss": 2.1191,
+      "step": 5295
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.946935034581948e-06,
+      "loss": 1.9797,
+      "step": 5300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9302704343637265e-06,
+      "loss": 2.0947,
+      "step": 5305
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.913632480261567e-06,
+      "loss": 2.1328,
+      "step": 5310
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.897021245316101e-06,
+      "loss": 2.1311,
+      "step": 5315
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.880436802450645e-06,
+      "loss": 2.0713,
+      "step": 5320
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.863879224470922e-06,
+      "loss": 2.0167,
+      "step": 5325
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.847348584064709e-06,
+      "loss": 2.1206,
+      "step": 5330
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.830844953801518e-06,
+      "loss": 2.0814,
+      "step": 5335
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.814368406132304e-06,
+      "loss": 2.0911,
+      "step": 5340
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7979190133891163e-06,
+      "loss": 2.1073,
+      "step": 5345
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.781496847784798e-06,
+      "loss": 2.0458,
+      "step": 5350
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 2.0481,
+      "step": 5355
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.7487344862461907e-06,
+      "loss": 2.0473,
+      "step": 5360
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.732394434138685e-06,
+      "loss": 2.0432,
+      "step": 5365
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.716081896822984e-06,
+      "loss": 2.0772,
+      "step": 5370
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6997969459111414e-06,
+      "loss": 2.0831,
+      "step": 5375
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.6835396528941003e-06,
+      "loss": 2.1639,
+      "step": 5380
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.667310089141376e-06,
+      "loss": 1.9198,
+      "step": 5385
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.651108325900773e-06,
+      "loss": 2.1591,
+      "step": 5390
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6349344342980365e-06,
+      "loss": 2.0509,
+      "step": 5395
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.618788485336555e-06,
+      "loss": 2.0814,
+      "step": 5400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.6026705498970617e-06,
+      "loss": 2.1067,
+      "step": 5405
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.5865806987372996e-06,
+      "loss": 2.1184,
+      "step": 5410
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.570519002491717e-06,
+      "loss": 1.9818,
+      "step": 5415
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5544855316711768e-06,
+      "loss": 2.0269,
+      "step": 5420
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.538480356662621e-06,
+      "loss": 2.0397,
+      "step": 5425
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.5225035477287783e-06,
+      "loss": 2.0323,
+      "step": 5430
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.506555175007843e-06,
+      "loss": 2.1048,
+      "step": 5435
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.4906353085131917e-06,
+      "loss": 2.0833,
+      "step": 5440
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.474744018133036e-06,
+      "loss": 2.0813,
+      "step": 5445
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.4588813736301507e-06,
+      "loss": 2.137,
+      "step": 5450
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.443047444641557e-06,
+      "loss": 2.0274,
+      "step": 5455
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.427242300678213e-06,
+      "loss": 2.1431,
+      "step": 5460
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.411466011124701e-06,
+      "loss": 2.0952,
+      "step": 5465
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.3957186452389545e-06,
+      "loss": 2.0546,
+      "step": 5470
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.3800002721519064e-06,
+      "loss": 2.0325,
+      "step": 5475
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.364310960867223e-06,
+      "loss": 2.1614,
+      "step": 5480
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.348650780260997e-06,
+      "loss": 2.0274,
+      "step": 5485
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.333019799081425e-06,
+      "loss": 2.0534,
+      "step": 5490
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.317418085948524e-06,
+      "loss": 2.0118,
+      "step": 5495
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.301845709353818e-06,
+      "loss": 2.0819,
+      "step": 5500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.286302737660061e-06,
+      "loss": 1.9773,
+      "step": 5505
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.2707892391008945e-06,
+      "loss": 2.1252,
+      "step": 5510
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.2553052817805887e-06,
+      "loss": 2.0567,
+      "step": 5515
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.239850933673728e-06,
+      "loss": 2.0363,
+      "step": 5520
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.224426262624908e-06,
+      "loss": 2.0662,
+      "step": 5525
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.2090313363484383e-06,
+      "loss": 2.1223,
+      "step": 5530
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.1936662224280634e-06,
+      "loss": 2.1169,
+      "step": 5535
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.1783309883166327e-06,
+      "loss": 2.1307,
+      "step": 5540
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.16302570133583e-06,
+      "loss": 1.9699,
+      "step": 5545
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.147750428675883e-06,
+      "loss": 2.1088,
+      "step": 5550
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.1325052373952424e-06,
+      "loss": 2.1042,
+      "step": 5555
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.117290194420306e-06,
+      "loss": 2.1419,
+      "step": 5560
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.1021053665451204e-06,
+      "loss": 2.045,
+      "step": 5565
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0869508204310883e-06,
+      "loss": 2.1147,
+      "step": 5570
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0718266226066753e-06,
+      "loss": 2.0688,
+      "step": 5575
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0567328394671146e-06,
+      "loss": 2.1113,
+      "step": 5580
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.041669537274129e-06,
+      "loss": 2.0537,
+      "step": 5585
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0266367821556196e-06,
+      "loss": 2.0977,
+      "step": 5590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0116346401053908e-06,
+      "loss": 2.0842,
+      "step": 5595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.9966631769828524e-06,
+      "loss": 2.0854,
+      "step": 5600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.981722458512738e-06,
+      "loss": 1.9946,
+      "step": 5605
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.9668125502848035e-06,
+      "loss": 2.0781,
+      "step": 5610
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.9519335177535624e-06,
+      "loss": 2.0189,
+      "step": 5615
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.9370854262379723e-06,
+      "loss": 2.0735,
+      "step": 5620
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.9222683409211618e-06,
+      "loss": 2.1514,
+      "step": 5625
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.907482326850144e-06,
+      "loss": 2.0128,
+      "step": 5630
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.8927274489355296e-06,
+      "loss": 2.1505,
+      "step": 5635
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.878003771951239e-06,
+      "loss": 2.1523,
+      "step": 5640
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.86331136053422e-06,
+      "loss": 2.0934,
+      "step": 5645
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.848650279184173e-06,
+      "loss": 2.0922,
+      "step": 5650
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.834020592263249e-06,
+      "loss": 2.0501,
+      "step": 5655
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.8194223639957816e-06,
+      "loss": 2.0705,
+      "step": 5660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.804855658468002e-06,
+      "loss": 2.1016,
+      "step": 5665
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7903205396277546e-06,
+      "loss": 2.0355,
+      "step": 5670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7758170712842147e-06,
+      "loss": 2.0765,
+      "step": 5675
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7613453171076233e-06,
+      "loss": 2.0409,
+      "step": 5680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.7469053406289857e-06,
+      "loss": 2.0348,
+      "step": 5685
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.732497205239807e-06,
+      "loss": 1.9971,
+      "step": 5690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.7181209741918093e-06,
+      "loss": 2.0887,
+      "step": 5695
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.703776710596654e-06,
+      "loss": 2.091,
+      "step": 5700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6894644774256663e-06,
+      "loss": 2.097,
+      "step": 5705
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.6751843375095543e-06,
+      "loss": 1.9934,
+      "step": 5710
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.660936353538146e-06,
+      "loss": 2.0891,
+      "step": 5715
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6467205880600954e-06,
+      "loss": 2.102,
+      "step": 5720
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.632537103482622e-06,
+      "loss": 2.1345,
+      "step": 5725
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6183859620712294e-06,
+      "loss": 2.1523,
+      "step": 5730
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.6042672259494352e-06,
+      "loss": 2.1234,
+      "step": 5735
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.590180957098498e-06,
+      "loss": 2.0256,
+      "step": 5740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5761272173571507e-06,
+      "loss": 2.0513,
+      "step": 5745
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5621060684213138e-06,
+      "loss": 2.0805,
+      "step": 5750
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.54811757184384e-06,
+      "loss": 2.001,
+      "step": 5755
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.5341617890342375e-06,
+      "loss": 2.0726,
+      "step": 5760
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.5202387812583984e-06,
+      "loss": 2.1048,
+      "step": 5765
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.506348609638336e-06,
+      "loss": 2.0817,
+      "step": 5770
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.492491335151909e-06,
+      "loss": 2.0392,
+      "step": 5775
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.478667018632562e-06,
+      "loss": 1.9889,
+      "step": 5780
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.4648757207690533e-06,
+      "loss": 2.0698,
+      "step": 5785
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.451117502105185e-06,
+      "loss": 2.0412,
+      "step": 5790
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.437392423039546e-06,
+      "loss": 2.0682,
+      "step": 5795
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.42370054382524e-06,
+      "loss": 2.0883,
+      "step": 5800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.410041924569625e-06,
+      "loss": 2.0525,
+      "step": 5805
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.3964166252340438e-06,
+      "loss": 2.0261,
+      "step": 5810
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3828247056335718e-06,
+      "loss": 2.0894,
+      "step": 5815
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3692662254367426e-06,
+      "loss": 2.0458,
+      "step": 5820
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3557412441652903e-06,
+      "loss": 2.0402,
+      "step": 5825
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.3422498211938918e-06,
+      "loss": 2.1306,
+      "step": 5830
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.328792015749899e-06,
+      "loss": 2.082,
+      "step": 5835
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3153678869130834e-06,
+      "loss": 2.1265,
+      "step": 5840
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3019774936153827e-06,
+      "loss": 2.128,
+      "step": 5845
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.28862089464063e-06,
+      "loss": 2.0362,
+      "step": 5850
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2752981486242985e-06,
+      "loss": 2.1417,
+      "step": 5855
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2620093140532528e-06,
+      "loss": 2.0554,
+      "step": 5860
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.2487544492654832e-06,
+      "loss": 2.0155,
+      "step": 5865
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.235533612449854e-06,
+      "loss": 2.1008,
+      "step": 5870
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.222346861645841e-06,
+      "loss": 2.105,
+      "step": 5875
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.209194254743295e-06,
+      "loss": 2.0809,
+      "step": 5880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.196075849482164e-06,
+      "loss": 2.1432,
+      "step": 5885
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1829917034522552e-06,
+      "loss": 2.1098,
+      "step": 5890
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1699418740929744e-06,
+      "loss": 2.0564,
+      "step": 5895
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1569264186930817e-06,
+      "loss": 2.1031,
+      "step": 5900
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.143945394390429e-06,
+      "loss": 2.1048,
+      "step": 5905
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1309988581717257e-06,
+      "loss": 2.1031,
+      "step": 5910
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.118086866872271e-06,
+      "loss": 2.1187,
+      "step": 5915
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1052094771757125e-06,
+      "loss": 2.0922,
+      "step": 5920
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0923667456137987e-06,
+      "loss": 2.056,
+      "step": 5925
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0795587285661278e-06,
+      "loss": 2.1164,
+      "step": 5930
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0667854822599022e-06,
+      "loss": 2.0951,
+      "step": 5935
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.054047062769675e-06,
+      "loss": 2.0682,
+      "step": 5940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.041343526017123e-06,
+      "loss": 1.9337,
+      "step": 5945
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0286749277707783e-06,
+      "loss": 2.0883,
+      "step": 5950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0160413236457855e-06,
+      "loss": 2.1396,
+      "step": 5955
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0034427691036817e-06,
+      "loss": 2.0591,
+      "step": 5960
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.990879319452125e-06,
+      "loss": 2.0639,
+      "step": 5965
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.978351029844665e-06,
+      "loss": 2.1451,
+      "step": 5970
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9658579552805036e-06,
+      "loss": 2.0165,
+      "step": 5975
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9534001506042455e-06,
+      "loss": 2.0768,
+      "step": 5980
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.9409776705056514e-06,
+      "loss": 2.0771,
+      "step": 5985
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.928590569519425e-06,
+      "loss": 2.0962,
+      "step": 5990
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9162389020249416e-06,
+      "loss": 2.007,
+      "step": 5995
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.90392272224603e-06,
+      "loss": 2.0936,
+      "step": 6000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8916420842507232e-06,
+      "loss": 2.0272,
+      "step": 6005
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.8793970419510333e-06,
+      "loss": 2.1202,
+      "step": 6010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.867187649102702e-06,
+      "loss": 2.1284,
+      "step": 6015
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8550139593049631e-06,
+      "loss": 1.9971,
+      "step": 6020
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8428760260003297e-06,
+      "loss": 2.0249,
+      "step": 6025
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.830773902474332e-06,
+      "loss": 2.0139,
+      "step": 6030
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.8187076418552974e-06,
+      "loss": 2.1007,
+      "step": 6035
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.806677297114121e-06,
+      "loss": 2.0515,
+      "step": 6040
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7946829210640238e-06,
+      "loss": 2.0003,
+      "step": 6045
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7827245663603133e-06,
+      "loss": 2.115,
+      "step": 6050
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7708022855001816e-06,
+      "loss": 2.0915,
+      "step": 6055
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7589161308224466e-06,
+      "loss": 1.9925,
+      "step": 6060
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7470661545073319e-06,
+      "loss": 2.059,
+      "step": 6065
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.735252408576239e-06,
+      "loss": 2.0153,
+      "step": 6070
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7234749448915255e-06,
+      "loss": 2.1008,
+      "step": 6075
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7117338151562568e-06,
+      "loss": 2.1203,
+      "step": 6080
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7000290709139977e-06,
+      "loss": 2.0918,
+      "step": 6085
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6883607635485877e-06,
+      "loss": 2.1489,
+      "step": 6090
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6767289442838985e-06,
+      "loss": 2.0792,
+      "step": 6095
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.6651336641836214e-06,
+      "loss": 2.0856,
+      "step": 6100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.65357497415105e-06,
+      "loss": 2.0075,
+      "step": 6105
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.642052924928832e-06,
+      "loss": 1.9995,
+      "step": 6110
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6305675670987686e-06,
+      "loss": 2.0732,
+      "step": 6115
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6191189510815942e-06,
+      "loss": 2.1005,
+      "step": 6120
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.607707127136734e-06,
+      "loss": 2.1333,
+      "step": 6125
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.596332145362104e-06,
+      "loss": 2.1098,
+      "step": 6130
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.5849940556938782e-06,
+      "loss": 2.0098,
+      "step": 6135
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5736929079062768e-06,
+      "loss": 2.1078,
+      "step": 6140
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5624287516113457e-06,
+      "loss": 1.9704,
+      "step": 6145
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5512016362587346e-06,
+      "loss": 2.023,
+      "step": 6150
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.540011611135489e-06,
+      "loss": 2.1033,
+      "step": 6155
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.5288587253658248e-06,
+      "loss": 2.0905,
+      "step": 6160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5177430279109128e-06,
+      "loss": 2.1233,
+      "step": 6165
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5066645675686787e-06,
+      "loss": 2.1036,
+      "step": 6170
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.495623392973562e-06,
+      "loss": 2.0099,
+      "step": 6175
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.4846195525963247e-06,
+      "loss": 1.9788,
+      "step": 6180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.4736530947438377e-06,
+      "loss": 2.0596,
+      "step": 6185
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4627240675588539e-06,
+      "loss": 2.0375,
+      "step": 6190
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.451832519019808e-06,
+      "loss": 2.0944,
+      "step": 6195
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4409784969406049e-06,
+      "loss": 2.0623,
+      "step": 6200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4301620489704072e-06,
+      "loss": 2.0963,
+      "step": 6205
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.4193832225934267e-06,
+      "loss": 2.1514,
+      "step": 6210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4086420651287157e-06,
+      "loss": 2.0243,
+      "step": 6215
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3979386237299641e-06,
+      "loss": 2.0516,
+      "step": 6220
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.387272945385285e-06,
+      "loss": 2.067,
+      "step": 6225
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3766450769170115e-06,
+      "loss": 2.038,
+      "step": 6230
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.3660550649814918e-06,
+      "loss": 2.1,
+      "step": 6235
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3555029560688838e-06,
+      "loss": 1.9969,
+      "step": 6240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3449887965029485e-06,
+      "loss": 2.0904,
+      "step": 6245
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3345126324408575e-06,
+      "loss": 2.0902,
+      "step": 6250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3240745098729745e-06,
+      "loss": 2.0967,
+      "step": 6255
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3136744746226626e-06,
+      "loss": 1.9768,
+      "step": 6260
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3033125723460826e-06,
+      "loss": 2.0477,
+      "step": 6265
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.2929888485319908e-06,
+      "loss": 2.1532,
+      "step": 6270
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.2827033485015406e-06,
+      "loss": 1.9443,
+      "step": 6275
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.2724561174080796e-06,
+      "loss": 2.1782,
+      "step": 6280
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2622472002369635e-06,
+      "loss": 2.1286,
+      "step": 6285
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2520766418053408e-06,
+      "loss": 2.0444,
+      "step": 6290
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.241944486761969e-06,
+      "loss": 2.0266,
+      "step": 6295
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2318507795870138e-06,
+      "loss": 2.1015,
+      "step": 6300
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2217955645918568e-06,
+      "loss": 2.0997,
+      "step": 6305
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2117788859188928e-06,
+      "loss": 2.0902,
+      "step": 6310
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2018007875413518e-06,
+      "loss": 2.163,
+      "step": 6315
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.1918613132630896e-06,
+      "loss": 1.9922,
+      "step": 6320
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.1819605067184025e-06,
+      "loss": 2.0706,
+      "step": 6325
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.1720984113718382e-06,
+      "loss": 2.0752,
+      "step": 6330
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1622750705179986e-06,
+      "loss": 2.0683,
+      "step": 6335
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1524905272813558e-06,
+      "loss": 2.1766,
+      "step": 6340
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1427448246160588e-06,
+      "loss": 2.064,
+      "step": 6345
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.133038005305751e-06,
+      "loss": 2.1203,
+      "step": 6350
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1233701119633723e-06,
+      "loss": 2.0746,
+      "step": 6355
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1137411870309788e-06,
+      "loss": 1.995,
+      "step": 6360
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.104151272779559e-06,
+      "loss": 1.9195,
+      "step": 6365
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0946004113088381e-06,
+      "loss": 2.0753,
+      "step": 6370
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0850886445471055e-06,
+      "loss": 2.087,
+      "step": 6375
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0756160142510197e-06,
+      "loss": 2.1446,
+      "step": 6380
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0661825620054366e-06,
+      "loss": 2.1177,
+      "step": 6385
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.056788329223215e-06,
+      "loss": 2.13,
+      "step": 6390
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0474333571450412e-06,
+      "loss": 2.0877,
+      "step": 6395
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0381176868392483e-06,
+      "loss": 2.1238,
+      "step": 6400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0288413592016345e-06,
+      "loss": 2.1037,
+      "step": 6405
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0196044149552798e-06,
+      "loss": 2.0599,
+      "step": 6410
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0104068946503787e-06,
+      "loss": 2.1005,
+      "step": 6415
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0012488386640484e-06,
+      "loss": 2.0853,
+      "step": 6420
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.921302872001603e-07,
+      "loss": 2.1091,
+      "step": 6425
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.830512802891601e-07,
+      "loss": 2.0541,
+      "step": 6430
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.74011857787892e-07,
+      "loss": 2.0986,
+      "step": 6435
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.650120593794266e-07,
+      "loss": 2.1166,
+      "step": 6440
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.560519245728816e-07,
+      "loss": 2.0038,
+      "step": 6445
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.471314927032571e-07,
+      "loss": 2.0662,
+      "step": 6450
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.382508029312521e-07,
+      "loss": 2.0667,
+      "step": 6455
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.294098942430996e-07,
+      "loss": 2.132,
+      "step": 6460
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.206088054503937e-07,
+      "loss": 2.1023,
+      "step": 6465
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.118475751899192e-07,
+      "loss": 2.0209,
+      "step": 6470
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.031262419234787e-07,
+      "loss": 2.04,
+      "step": 6475
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 8.94444843937734e-07,
+      "loss": 1.9607,
+      "step": 6480
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.858034193440257e-07,
+      "loss": 2.0577,
+      "step": 6485
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.772020060782127e-07,
+      "loss": 2.1076,
+      "step": 6490
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.686406419005022e-07,
+      "loss": 2.101,
+      "step": 6495
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.601193643952888e-07,
+      "loss": 2.0948,
+      "step": 6500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.516382109709854e-07,
+      "loss": 2.0716,
+      "step": 6505
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.431972188598581e-07,
+      "loss": 1.972,
+      "step": 6510
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.347964251178697e-07,
+      "loss": 1.9502,
+      "step": 6515
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.264358666245098e-07,
+      "loss": 1.9629,
+      "step": 6520
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.181155800826301e-07,
+      "loss": 2.0311,
+      "step": 6525
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.098356020182973e-07,
+      "loss": 1.9672,
+      "step": 6530
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.015959687806174e-07,
+      "loss": 2.1489,
+      "step": 6535
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.933967165415846e-07,
+      "loss": 2.14,
+      "step": 6540
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.852378812959227e-07,
+      "loss": 2.0682,
+      "step": 6545
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.771194988609221e-07,
+      "loss": 2.124,
+      "step": 6550
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 7.690416048762861e-07,
+      "loss": 2.1139,
+      "step": 6555
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.610042348039715e-07,
+      "loss": 2.0541,
+      "step": 6560
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.530074239280405e-07,
+      "loss": 2.0691,
+      "step": 6565
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.450512073544946e-07,
+      "loss": 2.0822,
+      "step": 6570
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.371356200111279e-07,
+      "loss": 2.0511,
+      "step": 6575
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.29260696647377e-07,
+      "loss": 2.1127,
+      "step": 6580
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.214264718341591e-07,
+      "loss": 2.0741,
+      "step": 6585
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.13632979963721e-07,
+      "loss": 2.0819,
+      "step": 6590
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.058802552495004e-07,
+      "loss": 2.0707,
+      "step": 6595
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.981683317259613e-07,
+      "loss": 2.0275,
+      "step": 6600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.904972432484513e-07,
+      "loss": 2.0088,
+      "step": 6605
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.828670234930556e-07,
+      "loss": 2.1961,
+      "step": 6610
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.752777059564431e-07,
+      "loss": 2.0664,
+      "step": 6615
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.677293239557181e-07,
+      "loss": 2.0626,
+      "step": 6620
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.602219106282836e-07,
+      "loss": 2.0819,
+      "step": 6625
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.527554989316898e-07,
+      "loss": 2.0876,
+      "step": 6630
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.453301216434871e-07,
+      "loss": 2.0798,
+      "step": 6635
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.379458113610825e-07,
+      "loss": 2.0956,
+      "step": 6640
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.306026005016086e-07,
+      "loss": 2.1568,
+      "step": 6645
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.233005213017607e-07,
+      "loss": 2.0234,
+      "step": 6650
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.160396058176687e-07,
+      "loss": 1.9598,
+      "step": 6655
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.088198859247596e-07,
+      "loss": 2.0365,
+      "step": 6660
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.016413933176057e-07,
+      "loss": 2.0681,
+      "step": 6665
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.945041595097922e-07,
+      "loss": 2.1563,
+      "step": 6670
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.874082158337835e-07,
+      "loss": 2.0175,
+      "step": 6675
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.803535934407734e-07,
+      "loss": 2.0223,
+      "step": 6680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.733403233005563e-07,
+      "loss": 2.0616,
+      "step": 6685
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.663684362013944e-07,
+      "loss": 2.0521,
+      "step": 6690
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.594379627498747e-07,
+      "loss": 2.0198,
+      "step": 6695
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.525489333707812e-07,
+      "loss": 2.077,
+      "step": 6700
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.45701378306952e-07,
+      "loss": 2.1654,
+      "step": 6705
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.388953276191644e-07,
+      "loss": 2.0618,
+      "step": 6710
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.321308111859791e-07,
+      "loss": 2.0267,
+      "step": 6715
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.254078587036282e-07,
+      "loss": 2.1178,
+      "step": 6720
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.187264996858776e-07,
+      "loss": 2.1309,
+      "step": 6725
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.120867634638971e-07,
+      "loss": 2.0864,
+      "step": 6730
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.054886791861291e-07,
+      "loss": 2.0438,
+      "step": 6735
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.989322758181736e-07,
+      "loss": 2.0632,
+      "step": 6740
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.924175821426392e-07,
+      "loss": 2.0679,
+      "step": 6745
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.859446267590351e-07,
+      "loss": 2.0548,
+      "step": 6750
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.795134380836419e-07,
+      "loss": 1.9926,
+      "step": 6755
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7312404434938007e-07,
+      "loss": 2.09,
+      "step": 6760
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.667764736056912e-07,
+      "loss": 2.0877,
+      "step": 6765
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.604707537184139e-07,
+      "loss": 2.1266,
+      "step": 6770
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.542069123696624e-07,
+      "loss": 2.1788,
+      "step": 6775
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.479849770577038e-07,
+      "loss": 2.0806,
+      "step": 6780
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.418049750968356e-07,
+      "loss": 2.0333,
+      "step": 6785
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3566693361727254e-07,
+      "loss": 2.0392,
+      "step": 6790
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.2957087956501885e-07,
+      "loss": 2.0723,
+      "step": 6795
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.235168397017542e-07,
+      "loss": 2.1018,
+      "step": 6800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.1750484060471885e-07,
+      "loss": 2.0483,
+      "step": 6805
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1153490866658983e-07,
+      "loss": 2.0559,
+      "step": 6810
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0560707009537184e-07,
+      "loss": 2.0806,
+      "step": 6815
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.9972135091428165e-07,
+      "loss": 2.0804,
+      "step": 6820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.9387777696162757e-07,
+      "loss": 2.0594,
+      "step": 6825
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.880763738907034e-07,
+      "loss": 2.0933,
+      "step": 6830
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.823171671696713e-07,
+      "loss": 1.9843,
+      "step": 6835
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.766001820814513e-07,
+      "loss": 2.1302,
+      "step": 6840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7092544372361314e-07,
+      "loss": 2.0555,
+      "step": 6845
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.6529297700825694e-07,
+      "loss": 2.0146,
+      "step": 6850
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.59702806661919e-07,
+      "loss": 2.138,
+      "step": 6855
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.541549572254488e-07,
+      "loss": 2.1426,
+      "step": 6860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4864945305390863e-07,
+      "loss": 2.0187,
+      "step": 6865
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4318631831646297e-07,
+      "loss": 2.0497,
+      "step": 6870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.3776557699627844e-07,
+      "loss": 1.9925,
+      "step": 6875
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.323872528904104e-07,
+      "loss": 2.0441,
+      "step": 6880
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2705136960970554e-07,
+      "loss": 2.0205,
+      "step": 6885
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2175795057869494e-07,
+      "loss": 2.0661,
+      "step": 6890
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.165070190354913e-07,
+      "loss": 2.0878,
+      "step": 6895
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1129859803168516e-07,
+      "loss": 2.0933,
+      "step": 6900
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.061327104322487e-07,
+      "loss": 2.0879,
+      "step": 6905
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0100937891543116e-07,
+      "loss": 2.0346,
+      "step": 6910
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9592862597266013e-07,
+      "loss": 2.0495,
+      "step": 6915
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9089047390844395e-07,
+      "loss": 1.9751,
+      "step": 6920
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.858949448402737e-07,
+      "loss": 2.0457,
+      "step": 6925
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.809420606985236e-07,
+      "loss": 2.0559,
+      "step": 6930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.760318432263587e-07,
+      "loss": 2.0475,
+      "step": 6935
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7116431397963604e-07,
+      "loss": 2.0471,
+      "step": 6940
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.663394943268094e-07,
+      "loss": 2.0074,
+      "step": 6945
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.615574054488434e-07,
+      "loss": 1.9663,
+      "step": 6950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.568180683391097e-07,
+      "loss": 2.1215,
+      "step": 6955
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.521215038032987e-07,
+      "loss": 2.0805,
+      "step": 6960
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.474677324593333e-07,
+      "loss": 2.0014,
+      "step": 6965
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4285677473727123e-07,
+      "loss": 2.0507,
+      "step": 6970
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3828865087922038e-07,
+      "loss": 2.1024,
+      "step": 6975
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3376338093924477e-07,
+      "loss": 2.1282,
+      "step": 6980
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2928098478328442e-07,
+      "loss": 2.0834,
+      "step": 6985
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2484148208905987e-07,
+      "loss": 2.0063,
+      "step": 6990
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2044489234599008e-07,
+      "loss": 1.9253,
+      "step": 6995
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1609123485510697e-07,
+      "loss": 2.0383,
+      "step": 7000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.117805287289687e-07,
+      "loss": 2.0034,
+      "step": 7005
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0751279289157656e-07,
+      "loss": 2.0936,
+      "step": 7010
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0328804607829488e-07,
+      "loss": 2.0141,
+      "step": 7015
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9910630683576458e-07,
+      "loss": 2.1539,
+      "step": 7020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9496759352182204e-07,
+      "loss": 2.0881,
+      "step": 7025
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9087192430542135e-07,
+      "loss": 2.1645,
+      "step": 7030
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.868193171665522e-07,
+      "loss": 2.0837,
+      "step": 7035
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8280978989616338e-07,
+      "loss": 2.0468,
+      "step": 7040
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7884336009607928e-07,
+      "loss": 2.1323,
+      "step": 7045
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7492004517893125e-07,
+      "loss": 2.0991,
+      "step": 7050
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7103986236807312e-07,
+      "loss": 2.072,
+      "step": 7055
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6720282869751025e-07,
+      "loss": 2.0069,
+      "step": 7060
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6340896101182058e-07,
+      "loss": 2.0095,
+      "step": 7065
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5965827596608695e-07,
+      "loss": 2.0803,
+      "step": 7070
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5595079002581836e-07,
+      "loss": 2.0156,
+      "step": 7075
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.522865194668799e-07,
+      "loss": 2.131,
+      "step": 7080
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.486654803754206e-07,
+      "loss": 2.1106,
+      "step": 7085
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4508768864780586e-07,
+      "loss": 2.0862,
+      "step": 7090
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4155315999054176e-07,
+      "loss": 2.0843,
+      "step": 7095
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3806190992021185e-07,
+      "loss": 2.0172,
+      "step": 7100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3461395376340502e-07,
+      "loss": 2.1321,
+      "step": 7105
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3120930665665e-07,
+      "loss": 2.1629,
+      "step": 7110
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2784798354635196e-07,
+      "loss": 2.1107,
+      "step": 7115
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2452999918871943e-07,
+      "loss": 2.0813,
+      "step": 7120
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.212553681497064e-07,
+      "loss": 2.0589,
+      "step": 7125
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.1802410480494353e-07,
+      "loss": 2.035,
+      "step": 7130
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1483622333967936e-07,
+      "loss": 2.0216,
+      "step": 7135
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 1.978,
+      "step": 7140
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0859066183634414e-07,
+      "loss": 2.0482,
+      "step": 7145
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0553300921629206e-07,
+      "loss": 2.1282,
+      "step": 7150
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0251879331165559e-07,
+      "loss": 2.0081,
+      "step": 7155
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.954802735484214e-08,
+      "loss": 2.1178,
+      "step": 7160
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.662072438751835e-08,
+      "loss": 2.0898,
+      "step": 7165
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.373689726054458e-08,
+      "loss": 2.032,
+      "step": 7170
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.089655863392389e-08,
+      "loss": 1.996,
+      "step": 7175
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.809972097674424e-08,
+      "loss": 2.1137,
+      "step": 7180
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.53463965671253e-08,
+      "loss": 1.9833,
+      "step": 7185
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.263659749215946e-08,
+      "loss": 2.1127,
+      "step": 7190
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.997033564787093e-08,
+      "loss": 2.0218,
+      "step": 7195
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.734762273914675e-08,
+      "loss": 2.1112,
+      "step": 7200
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.476847027970024e-08,
+      "loss": 2.0171,
+      "step": 7205
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.223288959200991e-08,
+      "loss": 2.0654,
+      "step": 7210
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.97408918072795e-08,
+      "loss": 2.012,
+      "step": 7215
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.729248786537912e-08,
+      "loss": 2.1098,
+      "step": 7220
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.488768851480087e-08,
+      "loss": 2.0205,
+      "step": 7225
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.252650431261886e-08,
+      "loss": 2.0984,
+      "step": 7230
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.020894562443146e-08,
+      "loss": 2.0464,
+      "step": 7235
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.793502262432249e-08,
+      "loss": 2.103,
+      "step": 7240
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5704745294815624e-08,
+      "loss": 2.0956,
+      "step": 7245
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3518123426830046e-08,
+      "loss": 2.0641,
+      "step": 7250
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.137516661963493e-08,
+      "loss": 2.1449,
+      "step": 7255
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.927588428081387e-08,
+      "loss": 2.0376,
+      "step": 7260
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7220285626216054e-08,
+      "loss": 2.0383,
+      "step": 7265
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5208379679921865e-08,
+      "loss": 2.1728,
+      "step": 7270
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3240175274197325e-08,
+      "loss": 2.093,
+      "step": 7275
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.131568104946415e-08,
+      "loss": 2.0098,
+      "step": 7280
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.943490545425088e-08,
+      "loss": 1.998,
+      "step": 7285
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.759785674516292e-08,
+      "loss": 2.099,
+      "step": 7290
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.580454298684699e-08,
+      "loss": 2.041,
+      "step": 7295
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.405497205195007e-08,
+      "loss": 2.1463,
+      "step": 7300
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.234915162108942e-08,
+      "loss": 2.1113,
+      "step": 7305
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0687089182819264e-08,
+      "loss": 2.0994,
+      "step": 7310
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.906879203359303e-08,
+      "loss": 1.9752,
+      "step": 7315
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7494267277735632e-08,
+      "loss": 1.9797,
+      "step": 7320
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5963521827412352e-08,
+      "loss": 2.0584,
+      "step": 7325
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.447656240259777e-08,
+      "loss": 2.0813,
+      "step": 7330
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3033395531042448e-08,
+      "loss": 2.0428,
+      "step": 7335
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1634027548250723e-08,
+      "loss": 2.0691,
+      "step": 7340
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0278464597449644e-08,
+      "loss": 2.0987,
+      "step": 7345
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.896671262955896e-08,
+      "loss": 2.1581,
+      "step": 7350
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7698777403173383e-08,
+      "loss": 2.0984,
+      "step": 7355
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6474664484527058e-08,
+      "loss": 2.0336,
+      "step": 7360
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5294379247476898e-08,
+      "loss": 2.1252,
+      "step": 7365
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4157926873475947e-08,
+      "loss": 2.0692,
+      "step": 7370
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3065312351552283e-08,
+      "loss": 2.1003,
+      "step": 7375
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.201654047828127e-08,
+      "loss": 2.1239,
+      "step": 7380
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1011615857773328e-08,
+      "loss": 2.1577,
+      "step": 7385
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0050542901648419e-08,
+      "loss": 2.1533,
+      "step": 7390
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.13332582901716e-09,
+      "loss": 2.0182,
+      "step": 7395
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.259968666463058e-09,
+      "loss": 2.1697,
+      "step": 7400
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.43047524802476e-09,
+      "loss": 2.0327,
+      "step": 7405
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.644849215178273e-09,
+      "loss": 2.0703,
+      "step": 7410
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.903094016822542e-09,
+      "loss": 2.0087,
+      "step": 7415
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.205212909262791e-09,
+      "loss": 2.0258,
+      "step": 7420
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.551208956198317e-09,
+      "loss": 1.9995,
+      "step": 7425
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.941085028705827e-09,
+      "loss": 2.054,
+      "step": 7430
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3748438052305656e-09,
+      "loss": 2.1216,
+      "step": 7435
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.852487771571877e-09,
+      "loss": 2.0869,
+      "step": 7440
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3740192208721034e-09,
+      "loss": 2.0625,
+      "step": 7445
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9394402536099255e-09,
+      "loss": 2.0759,
+      "step": 7450
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5487527775848165e-09,
+      "loss": 2.0811,
+      "step": 7455
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.201958507918155e-09,
+      "loss": 2.1014,
+      "step": 7460
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.990589670343497e-10,
+      "loss": 2.0307,
+      "step": 7465
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.400554846641705e-10,
+      "loss": 2.0364,
+      "step": 7470
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.2494919783142574e-10,
+      "loss": 2.0335,
+      "step": 7475
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5374105085518297e-10,
+      "loss": 2.0692,
+      "step": 7480
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.264317953364458e-10,
+      "loss": 2.0747,
+      "step": 7485
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.3021990164815455e-11,
+      "loss": 2.0857,
+      "step": 7490
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5120015096090644e-12,
+      "loss": 2.1121,
+      "step": 7495
+    },
+    {
+      "epoch": 3.0,
+      "step": 7497,
+      "total_flos": 4.954742374840074e+17,
+      "train_loss": 2.0977668136346335,
+      "train_runtime": 13801.1295,
+      "train_samples_per_second": 8.691,
+      "train_steps_per_second": 0.543
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 7497,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 4.954742374840074e+17,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}