diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8200 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 6807,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.936857562408223e-07,
+      "loss": 8.0371,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4684287812041115e-06,
+      "loss": 7.6761,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.936857562408223e-06,
+      "loss": 8.1376,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4052863436123355e-06,
+      "loss": 7.9662,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.873715124816446e-06,
+      "loss": 7.441,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3421439060205585e-06,
+      "loss": 8.0036,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.810572687224671e-06,
+      "loss": 8.2087,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0279001468428782e-05,
+      "loss": 8.1128,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1747430249632892e-05,
+      "loss": 7.8649,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3215859030837005e-05,
+      "loss": 7.4465,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4684287812041117e-05,
+      "loss": 7.2594,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.615271659324523e-05,
+      "loss": 7.1477,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7621145374449342e-05,
+      "loss": 6.4907,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9089574155653454e-05,
+      "loss": 6.839,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0558002936857563e-05,
+      "loss": 6.4178,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2026431718061676e-05,
+      "loss": 5.8355,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.3494860499265785e-05,
+      "loss": 6.2787,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4963289280469897e-05,
+      "loss": 6.3641,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.643171806167401e-05,
+      "loss": 6.0727,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7900146842878122e-05,
+      "loss": 6.1137,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9368575624082234e-05,
+      "loss": 5.5486,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.0837004405286347e-05,
+      "loss": 5.8688,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.230543318649046e-05,
+      "loss": 5.6604,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.377386196769457e-05,
+      "loss": 5.1517,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.5242290748898684e-05,
+      "loss": 5.0356,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6710719530102796e-05,
+      "loss": 5.2555,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.817914831130691e-05,
+      "loss": 5.4059,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.9647577092511014e-05,
+      "loss": 5.24,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.1116005873715127e-05,
+      "loss": 4.8884,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.258443465491924e-05,
+      "loss": 5.0816,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.405286343612335e-05,
+      "loss": 4.985,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.5521292217327464e-05,
+      "loss": 4.8594,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.698972099853157e-05,
+      "loss": 5.1042,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.845814977973568e-05,
+      "loss": 4.6756,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9926578560939794e-05,
+      "loss": 4.7418,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.1395007342143906e-05,
+      "loss": 4.952,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.286343612334802e-05,
+      "loss": 4.8024,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.433186490455213e-05,
+      "loss": 4.6133,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.5800293685756244e-05,
+      "loss": 4.7399,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.7268722466960356e-05,
+      "loss": 4.5914,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.873715124816447e-05,
+      "loss": 4.7979,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.020558002936858e-05,
+      "loss": 4.4394,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.167400881057269e-05,
+      "loss": 4.4931,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.31424375917768e-05,
+      "loss": 4.0828,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.461086637298092e-05,
+      "loss": 4.6045,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.607929515418503e-05,
+      "loss": 4.7074,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.754772393538914e-05,
+      "loss": 4.5746,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.901615271659326e-05,
+      "loss": 4.6345,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.048458149779737e-05,
+      "loss": 4.6318,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.195301027900148e-05,
+      "loss": 4.6217,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.342143906020559e-05,
+      "loss": 4.304,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.48898678414097e-05,
+      "loss": 4.0523,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.635829662261382e-05,
+      "loss": 4.2805,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.782672540381793e-05,
+      "loss": 3.8923,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.929515418502203e-05,
+      "loss": 4.5916,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.076358296622614e-05,
+      "loss": 4.1592,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.223201174743025e-05,
+      "loss": 4.2993,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.370044052863437e-05,
+      "loss": 3.7691,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.516886930983848e-05,
+      "loss": 4.0821,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.663729809104259e-05,
+      "loss": 4.009,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.81057268722467e-05,
+      "loss": 4.1831,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.957415565345081e-05,
+      "loss": 4.1099,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.104258443465493e-05,
+      "loss": 4.0626,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.251101321585903e-05,
+      "loss": 3.9612,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.397944199706314e-05,
+      "loss": 3.9486,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.544787077826725e-05,
+      "loss": 3.6232,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.691629955947136e-05,
+      "loss": 3.8523,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.838472834067548e-05,
+      "loss": 3.6291,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.985315712187959e-05,
+      "loss": 3.6497,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00010132158590308371,
+      "loss": 4.0705,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00010279001468428781,
+      "loss": 3.5615,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00010425844346549194,
+      "loss": 3.5258,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00010572687224669604,
+      "loss": 3.4318,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00010719530102790014,
+      "loss": 3.7495,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00010866372980910426,
+      "loss": 3.6911,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011013215859030836,
+      "loss": 3.551,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011160058737151249,
+      "loss": 3.8947,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011306901615271659,
+      "loss": 3.4287,
+      "step": 385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011453744493392071,
+      "loss": 3.6862,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011600587371512481,
+      "loss": 3.6173,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011747430249632894,
+      "loss": 3.5599,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011894273127753304,
+      "loss": 3.2464,
+      "step": 405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012041116005873716,
+      "loss": 3.4171,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00012187958883994126,
+      "loss": 3.259,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012334801762114539,
+      "loss": 3.4354,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012481644640234947,
+      "loss": 3.518,
+      "step": 425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001262848751835536,
+      "loss": 3.5075,
+      "step": 430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001277533039647577,
+      "loss": 3.3865,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00012922173274596184,
+      "loss": 3.2455,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00013069016152716592,
+      "loss": 3.3011,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00013215859030837006,
+      "loss": 3.0195,
+      "step": 450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00013362701908957415,
+      "loss": 3.0395,
+      "step": 455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00013509544787077829,
+      "loss": 3.2593,
+      "step": 460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00013656387665198237,
+      "loss": 3.3803,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001380323054331865,
+      "loss": 2.7479,
+      "step": 470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001395007342143906,
+      "loss": 3.294,
+      "step": 475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00014096916299559473,
+      "loss": 3.0379,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00014243759177679882,
+      "loss": 3.1439,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00014390602055800296,
+      "loss": 3.0701,
+      "step": 490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00014537444933920705,
+      "loss": 3.1837,
+      "step": 495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00014684287812041118,
+      "loss": 3.0289,
+      "step": 500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00014831130690161527,
+      "loss": 3.519,
+      "step": 505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001497797356828194,
+      "loss": 3.0623,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001512481644640235,
+      "loss": 3.0159,
+      "step": 515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00015271659324522763,
+      "loss": 3.2545,
+      "step": 520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00015418502202643172,
+      "loss": 3.2286,
+      "step": 525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00015565345080763586,
+      "loss": 2.9999,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00015712187958883994,
+      "loss": 3.0678,
+      "step": 535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00015859030837004406,
+      "loss": 3.0911,
+      "step": 540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016005873715124817,
+      "loss": 3.3058,
+      "step": 545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00016152716593245228,
+      "loss": 3.1464,
+      "step": 550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001629955947136564,
+      "loss": 2.9742,
+      "step": 555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001644640234948605,
+      "loss": 3.1363,
+      "step": 560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016593245227606462,
+      "loss": 2.8022,
+      "step": 565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016740088105726873,
+      "loss": 2.6464,
+      "step": 570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00016886930983847284,
+      "loss": 3.0191,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017033773861967696,
+      "loss": 2.8873,
+      "step": 580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017180616740088107,
+      "loss": 2.8938,
+      "step": 585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017327459618208518,
+      "loss": 2.5943,
+      "step": 590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001747430249632893,
+      "loss": 3.0093,
+      "step": 595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001762114537444934,
+      "loss": 3.1039,
+      "step": 600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017767988252569752,
+      "loss": 2.7724,
+      "step": 605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017914831130690163,
+      "loss": 2.8095,
+      "step": 610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018061674008810574,
+      "loss": 2.9231,
+      "step": 615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00018208516886930985,
+      "loss": 2.7797,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018355359765051397,
+      "loss": 2.7899,
+      "step": 625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018502202643171805,
+      "loss": 2.5249,
+      "step": 630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001864904552129222,
+      "loss": 2.7455,
+      "step": 635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018795888399412628,
+      "loss": 2.8902,
+      "step": 640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00018942731277533042,
+      "loss": 2.9109,
+      "step": 645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001908957415565345,
+      "loss": 3.0111,
+      "step": 650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019236417033773864,
+      "loss": 2.8863,
+      "step": 655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019383259911894273,
+      "loss": 2.6726,
+      "step": 660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019530102790014687,
+      "loss": 2.6821,
+      "step": 665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019676945668135095,
+      "loss": 2.9596,
+      "step": 670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001982378854625551,
+      "loss": 2.7658,
+      "step": 675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019970631424375918,
+      "loss": 3.316,
+      "step": 680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019999978960491256,
+      "loss": 2.8106,
+      "step": 685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001999989348763872,
+      "loss": 2.9072,
+      "step": 690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001999974226703463,
+      "loss": 2.8696,
+      "step": 695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019999525299673244,
+      "loss": 2.4362,
+      "step": 700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001999924258698108,
+      "loss": 3.0676,
+      "step": 705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001999889413081694,
+      "loss": 2.7216,
+      "step": 710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019998479933471862,
+      "loss": 2.7508,
+      "step": 715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001999799999766913,
+      "loss": 2.827,
+      "step": 720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019997454326564252,
+      "loss": 2.8742,
+      "step": 725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001999684292374493,
+      "loss": 2.6331,
+      "step": 730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019996165793231038,
+      "loss": 2.7033,
+      "step": 735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001999542293947461,
+      "loss": 2.7013,
+      "step": 740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019994614367359792,
+      "loss": 2.7281,
+      "step": 745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019993740082202818,
+      "loss": 2.7481,
+      "step": 750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019992800089751984,
+      "loss": 2.5903,
+      "step": 755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001999179439618759,
+      "loss": 2.7431,
+      "step": 760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019990723008121917,
+      "loss": 2.9438,
+      "step": 765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019989585932599172,
+      "loss": 2.7703,
+      "step": 770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019988383177095459,
+      "loss": 2.6782,
+      "step": 775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000199871147495187,
+      "loss": 2.6494,
+      "step": 780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019985780658208618,
+      "loss": 2.9518,
+      "step": 785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019984380911936648,
+      "loss": 2.668,
+      "step": 790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019982915519905912,
+      "loss": 2.6909,
+      "step": 795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019981384491751133,
+      "loss": 2.8454,
+      "step": 800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019979787837538587,
+      "loss": 2.6661,
+      "step": 805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019978125567766023,
+      "loss": 2.4861,
+      "step": 810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019976397693362614,
+      "loss": 2.9039,
+      "step": 815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001997460422568886,
+      "loss": 2.6432,
+      "step": 820
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019972745176536537,
+      "loss": 2.5623,
+      "step": 825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019970820558128604,
+      "loss": 2.4012,
+      "step": 830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001996883038311913,
+      "loss": 2.7303,
+      "step": 835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019966774664593206,
+      "loss": 3.0838,
+      "step": 840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019964653416066868,
+      "loss": 2.6049,
+      "step": 845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001996246665148699,
+      "loss": 2.4024,
+      "step": 850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019960214385231217,
+      "loss": 2.6611,
+      "step": 855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019957896632107845,
+      "loss": 2.5144,
+      "step": 860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019955513407355743,
+      "loss": 2.6198,
+      "step": 865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001995306472664425,
+      "loss": 2.4248,
+      "step": 870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019950550606073056,
+      "loss": 2.6576,
+      "step": 875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019947971062172118,
+      "loss": 2.6264,
+      "step": 880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019945326111901542,
+      "loss": 2.7561,
+      "step": 885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019942615772651455,
+      "loss": 2.6115,
+      "step": 890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001993984006224193,
+      "loss": 2.4612,
+      "step": 895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019936998998922826,
+      "loss": 2.7587,
+      "step": 900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019934092601373694,
+      "loss": 2.6534,
+      "step": 905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019931120888703652,
+      "loss": 2.4858,
+      "step": 910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001992808388045125,
+      "loss": 2.6839,
+      "step": 915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019924981596584345,
+      "loss": 2.5574,
+      "step": 920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019921814057499978,
+      "loss": 2.634,
+      "step": 925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001991858128402422,
+      "loss": 2.5158,
+      "step": 930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001991528329741206,
+      "loss": 2.5677,
+      "step": 935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019911920119347254,
+      "loss": 2.3971,
+      "step": 940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001990849177194217,
+      "loss": 2.5978,
+      "step": 945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019904998277737668,
+      "loss": 2.6475,
+      "step": 950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019901439659702924,
+      "loss": 2.4652,
+      "step": 955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019897815941235307,
+      "loss": 2.526,
+      "step": 960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019894127146160204,
+      "loss": 2.5277,
+      "step": 965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019890373298730868,
+      "loss": 2.6482,
+      "step": 970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001988655442362827,
+      "loss": 2.6791,
+      "step": 975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019882670545960914,
+      "loss": 2.5003,
+      "step": 980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019878721691264704,
+      "loss": 2.668,
+      "step": 985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019874707885502745,
+      "loss": 2.6535,
+      "step": 990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019870629155065186,
+      "loss": 2.4613,
+      "step": 995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001986648552676905,
+      "loss": 2.4555,
+      "step": 1000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001986227702785805,
+      "loss": 2.6678,
+      "step": 1005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001985800368600242,
+      "loss": 2.3696,
+      "step": 1010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001985366552929871,
+      "loss": 2.4033,
+      "step": 1015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019849262586269642,
+      "loss": 2.5128,
+      "step": 1020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019844794885863877,
+      "loss": 2.6752,
+      "step": 1025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019840262457455855,
+      "loss": 2.6137,
+      "step": 1030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019835665330845595,
+      "loss": 2.4843,
+      "step": 1035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019831003536258487,
+      "loss": 2.5648,
+      "step": 1040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019826277104345109,
+      "loss": 2.368,
+      "step": 1045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001982148606618102,
+      "loss": 2.6524,
+      "step": 1050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019816630453266555,
+      "loss": 2.5964,
+      "step": 1055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001981171029752662,
+      "loss": 2.5593,
+      "step": 1060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019806725631310476,
+      "loss": 2.357,
+      "step": 1065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019801676487391529,
+      "loss": 2.482,
+      "step": 1070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001979656289896712,
+      "loss": 2.5125,
+      "step": 1075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000197913848996583,
+      "loss": 2.7377,
+      "step": 1080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019786142523509615,
+      "loss": 2.4826,
+      "step": 1085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019780835804988876,
+      "loss": 2.4345,
+      "step": 1090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019775464778986934,
+      "loss": 2.2278,
+      "step": 1095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019770029480817454,
+      "loss": 2.5896,
+      "step": 1100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019764529946216682,
+      "loss": 2.4121,
+      "step": 1105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019758966211343206,
+      "loss": 2.4519,
+      "step": 1110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019753338312777718,
+      "loss": 2.5232,
+      "step": 1115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019747646287522784,
+      "loss": 2.3545,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001974189017300259,
+      "loss": 2.3761,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019736070007062692,
+      "loss": 2.6134,
+      "step": 1130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019730185827969784,
+      "loss": 2.6982,
+      "step": 1135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019724237674411432,
+      "loss": 2.4818,
+      "step": 1140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019718225585495824,
+      "loss": 2.4621,
+      "step": 1145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019712149600751517,
+      "loss": 2.3575,
+      "step": 1150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019706009760127164,
+      "loss": 2.7274,
+      "step": 1155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019699806103991272,
+      "loss": 2.3379,
+      "step": 1160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019693538673131917,
+      "loss": 2.2026,
+      "step": 1165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019687207508756486,
+      "loss": 2.3081,
+      "step": 1170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019680812652491408,
+      "loss": 2.411,
+      "step": 1175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001967435414638187,
+      "loss": 2.4966,
+      "step": 1180
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019667832032891554,
+      "loss": 2.6727,
+      "step": 1185
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00019661246354902342,
+      "loss": 2.673,
+      "step": 1190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019654597155714044,
+      "loss": 2.2485,
+      "step": 1195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019647884479044123,
+      "loss": 2.4359,
+      "step": 1200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019641108369027385,
+      "loss": 2.5372,
+      "step": 1205
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00019634268870215703,
+      "loss": 2.1397,
+      "step": 1210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019627366027577726,
+      "loss": 2.1738,
+      "step": 1215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019620399886498578,
+      "loss": 2.5208,
+      "step": 1220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001961337049277955,
+      "loss": 2.589,
+      "step": 1225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019606277892637823,
+      "loss": 2.4162,
+      "step": 1230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019599122132706146,
+      "loss": 2.5716,
+      "step": 1235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0001959190326003253,
+      "loss": 2.4076,
+      "step": 1240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019584621322079942,
+      "loss": 2.371,
+      "step": 1245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019577276366726003,
+      "loss": 2.4083,
+      "step": 1250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019569868442262655,
+      "loss": 2.4514,
+      "step": 1255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019562397597395857,
+      "loss": 2.3607,
+      "step": 1260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001955486388124525,
+      "loss": 2.6568,
+      "step": 1265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019547267343343857,
+      "loss": 2.2921,
+      "step": 1270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001953960803363774,
+      "loss": 2.502,
+      "step": 1275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00019531886002485674,
+      "loss": 2.2983,
+      "step": 1280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019524101300658813,
+      "loss": 2.5903,
+      "step": 1285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001951625397934037,
+      "loss": 2.3507,
+      "step": 1290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001950834409012527,
+      "loss": 2.3861,
+      "step": 1295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00019500371685019806,
+      "loss": 2.3619,
+      "step": 1300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001949233681644131,
+      "loss": 2.4827,
+      "step": 1305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019484239537217798,
+      "loss": 2.7222,
+      "step": 1310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019476079900587626,
+      "loss": 2.3926,
+      "step": 1315
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019467857960199142,
+      "loss": 2.4906,
+      "step": 1320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019459573770110335,
+      "loss": 2.5517,
+      "step": 1325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001945122738478847,
+      "loss": 2.2752,
+      "step": 1330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019442818859109737,
+      "loss": 2.5604,
+      "step": 1335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019434348248358892,
+      "loss": 2.3671,
+      "step": 1340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019425815608228888,
+      "loss": 2.4842,
+      "step": 1345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019417220994820514,
+      "loss": 2.6403,
+      "step": 1350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019408564464642024,
+      "loss": 2.3457,
+      "step": 1355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019399846074608757,
+      "loss": 2.2025,
+      "step": 1360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019391065882042786,
+      "loss": 2.3618,
+      "step": 1365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00019382223944672516,
+      "loss": 2.453,
+      "step": 1370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019373320320632313,
+      "loss": 2.4178,
+      "step": 1375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019364355068462126,
+      "loss": 2.5761,
+      "step": 1380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019355328247107106,
+      "loss": 2.4834,
+      "step": 1385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019346239915917204,
+      "loss": 2.5458,
+      "step": 1390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00019337090134646787,
+      "loss": 2.4164,
+      "step": 1395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019327878963454253,
+      "loss": 2.4525,
+      "step": 1400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019318606462901625,
+      "loss": 2.4169,
+      "step": 1405
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001930927269395416,
+      "loss": 2.5682,
+      "step": 1410
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019299877717979944,
+      "loss": 2.395,
+      "step": 1415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019290421596749487,
+      "loss": 2.3261,
+      "step": 1420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019280904392435328,
+      "loss": 2.2114,
+      "step": 1425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019271326167611606,
+      "loss": 2.3535,
+      "step": 1430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019261686985253668,
+      "loss": 2.4581,
+      "step": 1435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019251986908737646,
+      "loss": 2.3825,
+      "step": 1440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019242226001840043,
+      "loss": 2.1258,
+      "step": 1445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001923240432873731,
+      "loss": 2.3138,
+      "step": 1450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019222521954005424,
+      "loss": 2.5257,
+      "step": 1455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019212578942619474,
+      "loss": 2.4313,
+      "step": 1460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019202575359953213,
+      "loss": 2.3226,
+      "step": 1465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019192511271778656,
+      "loss": 2.1064,
+      "step": 1470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019182386744265623,
+      "loss": 2.574,
+      "step": 1475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019172201843981314,
+      "loss": 2.3097,
+      "step": 1480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00019161956637889872,
+      "loss": 2.4382,
+      "step": 1485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001915165119335194,
+      "loss": 2.1461,
+      "step": 1490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001914128557812422,
+      "loss": 2.5698,
+      "step": 1495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019130859860359026,
+      "loss": 2.5678,
+      "step": 1500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00019120374108603843,
+      "loss": 1.9771,
+      "step": 1505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001910982839180086,
+      "loss": 2.3394,
+      "step": 1510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001909922277928654,
+      "loss": 2.1381,
+      "step": 1515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019088557340791136,
+      "loss": 2.4063,
+      "step": 1520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019077832146438257,
+      "loss": 2.4501,
+      "step": 1525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019067047266744396,
+      "loss": 2.4332,
+      "step": 1530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001905620277261847,
+      "loss": 2.2714,
+      "step": 1535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019045298735361345,
+      "loss": 2.476,
+      "step": 1540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001903433522666538,
+      "loss": 2.435,
+      "step": 1545
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019023312318613945,
+      "loss": 2.3723,
+      "step": 1550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00019012230083680954,
+      "loss": 2.3283,
+      "step": 1555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001900108859473039,
+      "loss": 2.339,
+      "step": 1560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018989887925015814,
+      "loss": 2.4173,
+      "step": 1565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018978628148179897,
+      "loss": 2.1088,
+      "step": 1570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001896730933825393,
+      "loss": 2.2159,
+      "step": 1575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018955931569657333,
+      "loss": 2.4426,
+      "step": 1580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018944494917197172,
+      "loss": 2.2461,
+      "step": 1585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018932999456067675,
+      "loss": 2.3546,
+      "step": 1590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001892144526184971,
+      "loss": 2.4394,
+      "step": 1595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018909832410510315,
+      "loss": 2.1256,
+      "step": 1600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018898160978402198,
+      "loss": 2.2209,
+      "step": 1605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018886431042263208,
+      "loss": 2.3256,
+      "step": 1610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001887464267921587,
+      "loss": 2.608,
+      "step": 1615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018862795966766833,
+      "loss": 2.3155,
+      "step": 1620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001885089098280641,
+      "loss": 2.3065,
+      "step": 1625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001883892780560802,
+      "loss": 2.4474,
+      "step": 1630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018826906513827704,
+      "loss": 2.3907,
+      "step": 1635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018814827186503595,
+      "loss": 2.33,
+      "step": 1640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018802689903055396,
+      "loss": 2.0636,
+      "step": 1645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001879049474328387,
+      "loss": 2.4195,
+      "step": 1650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018778241787370303,
+      "loss": 2.4082,
+      "step": 1655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018765931115875985,
+      "loss": 2.1858,
+      "step": 1660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018753562809741673,
+      "loss": 2.234,
+      "step": 1665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018741136950287067,
+      "loss": 2.3279,
+      "step": 1670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001872865361921027,
+      "loss": 2.4532,
+      "step": 1675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018716112898587247,
+      "loss": 2.4711,
+      "step": 1680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000187035148708713,
+      "loss": 2.2462,
+      "step": 1685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018690859618892506,
+      "loss": 2.2064,
+      "step": 1690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001867814722585719,
+      "loss": 2.3891,
+      "step": 1695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001866537777534737,
+      "loss": 2.3496,
+      "step": 1700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018652551351320198,
+      "loss": 2.1906,
+      "step": 1705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018639668038107437,
+      "loss": 2.5242,
+      "step": 1710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001862672792041487,
+      "loss": 2.5202,
+      "step": 1715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001861373108332177,
+      "loss": 2.3353,
+      "step": 1720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001860067761228033,
+      "loss": 2.0448,
+      "step": 1725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018587567593115098,
+      "loss": 2.3772,
+      "step": 1730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001857440111202242,
+      "loss": 2.4137,
+      "step": 1735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018561178255569879,
+      "loss": 2.5623,
+      "step": 1740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000185478991106957,
+      "loss": 1.93,
+      "step": 1745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018534563764708206,
+      "loss": 2.282,
+      "step": 1750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018521172305285236,
+      "loss": 2.3154,
+      "step": 1755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018507724820473556,
+      "loss": 2.3849,
+      "step": 1760
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018494221398688307,
+      "loss": 2.2915,
+      "step": 1765
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018480662128712389,
+      "loss": 2.4104,
+      "step": 1770
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018467047099695905,
+      "loss": 2.1556,
+      "step": 1775
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018453376401155562,
+      "loss": 2.3384,
+      "step": 1780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018439650122974087,
+      "loss": 2.4587,
+      "step": 1785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001842586835539964,
+      "loss": 2.3147,
+      "step": 1790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018412031189045196,
+      "loss": 2.3839,
+      "step": 1795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018398138714887993,
+      "loss": 2.208,
+      "step": 1800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018384191024268894,
+      "loss": 2.1344,
+      "step": 1805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018370188208891803,
+      "loss": 2.1627,
+      "step": 1810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018356130360823068,
+      "loss": 2.3003,
+      "step": 1815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018342017572490858,
+      "loss": 2.2818,
+      "step": 1820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001832784993668458,
+      "loss": 2.4602,
+      "step": 1825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001831362754655424,
+      "loss": 2.4241,
+      "step": 1830
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001829935049560985,
+      "loss": 2.2407,
+      "step": 1835
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.0001828501887772081,
+      "loss": 2.4529,
+      "step": 1840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018270632787115295,
+      "loss": 2.3067,
+      "step": 1845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001825619231837962,
+      "loss": 2.1068,
+      "step": 1850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.0001824169756645763,
+      "loss": 2.3169,
+      "step": 1855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018227148626650072,
+      "loss": 2.3431,
+      "step": 1860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018212545594613978,
+      "loss": 2.1555,
+      "step": 1865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018197888566362023,
+      "loss": 2.4435,
+      "step": 1870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018183177638261895,
+      "loss": 2.3,
+      "step": 1875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018168412907035672,
+      "loss": 2.455,
+      "step": 1880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018153594469759175,
+      "loss": 2.2663,
+      "step": 1885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018138722423861333,
+      "loss": 2.4622,
+      "step": 1890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018123796867123548,
+      "loss": 2.3199,
+      "step": 1895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018108817897679043,
+      "loss": 2.5241,
+      "step": 1900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018093785614012228,
+      "loss": 2.2236,
+      "step": 1905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001807870011495803,
+      "loss": 2.4421,
+      "step": 1910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018063561499701282,
+      "loss": 2.2333,
+      "step": 1915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018048369867776029,
+      "loss": 2.2838,
+      "step": 1920
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018033125319064902,
+      "loss": 2.2361,
+      "step": 1925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018017827953798444,
+      "loss": 2.2901,
+      "step": 1930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001800247787255447,
+      "loss": 2.4939,
+      "step": 1935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017987075176257382,
+      "loss": 2.2162,
+      "step": 1940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017971619966177524,
+      "loss": 2.5374,
+      "step": 1945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017956112343930512,
+      "loss": 2.2674,
+      "step": 1950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017940552411476566,
+      "loss": 2.2996,
+      "step": 1955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00017924940271119827,
+      "loss": 2.3013,
+      "step": 1960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017909276025507696,
+      "loss": 2.3724,
+      "step": 1965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017893559777630173,
+      "loss": 2.4073,
+      "step": 1970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017877791630819149,
+      "loss": 2.4049,
+      "step": 1975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017861971688747747,
+      "loss": 1.8259,
+      "step": 1980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00017846100055429642,
+      "loss": 2.3023,
+      "step": 1985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017830176835218368,
+      "loss": 2.2579,
+      "step": 1990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001781420213280662,
+      "loss": 2.366,
+      "step": 1995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00017798176053225606,
+      "loss": 2.1387,
+      "step": 2000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001778209870184431,
+      "loss": 2.334,
+      "step": 2005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00017765970184368835,
+      "loss": 2.2572,
+      "step": 2010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001774979060684168,
+      "loss": 2.1728,
+      "step": 2015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001773356007564107,
+      "loss": 2.3457,
+      "step": 2020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001771727869748023,
+      "loss": 2.2485,
+      "step": 2025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.000177009465794067,
+      "loss": 2.4565,
+      "step": 2030
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001768456382880163,
+      "loss": 2.2527,
+      "step": 2035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017668130553379063,
+      "loss": 2.1906,
+      "step": 2040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00017651646861185252,
+      "loss": 1.9663,
+      "step": 2045
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001763511286059791,
+      "loss": 2.1846,
+      "step": 2050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001761852866032554,
+      "loss": 2.4796,
+      "step": 2055
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.0001760189436940669,
+      "loss": 2.245,
+      "step": 2060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017585210097209242,
+      "loss": 2.3413,
+      "step": 2065
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017568475953429706,
+      "loss": 2.2338,
+      "step": 2070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00017551692048092487,
+      "loss": 2.2242,
+      "step": 2075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017534858491549167,
+      "loss": 2.2587,
+      "step": 2080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017517975394477765,
+      "loss": 2.3805,
+      "step": 2085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00017501042867882043,
+      "loss": 2.0019,
+      "step": 2090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001748406102309073,
+      "loss": 2.2803,
+      "step": 2095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017467029971756837,
+      "loss": 2.0584,
+      "step": 2100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017449949825856881,
+      "loss": 2.2131,
+      "step": 2105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017432820697690183,
+      "loss": 2.1911,
+      "step": 2110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017415642699878108,
+      "loss": 2.1714,
+      "step": 2115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00017398415945363326,
+      "loss": 2.013,
+      "step": 2120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017381140547409091,
+      "loss": 2.2408,
+      "step": 2125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017363816619598462,
+      "loss": 2.5007,
+      "step": 2130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017346444275833587,
+      "loss": 2.1915,
+      "step": 2135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00017329023630334935,
+      "loss": 2.5022,
+      "step": 2140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017311554797640552,
+      "loss": 2.3124,
+      "step": 2145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001729403789260531,
+      "loss": 2.5562,
+      "step": 2150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001727647303040015,
+      "loss": 2.3645,
+      "step": 2155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017258860326511318,
+      "loss": 2.0959,
+      "step": 2160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00017241199896739614,
+      "loss": 2.4902,
+      "step": 2165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017223491857199636,
+      "loss": 2.2757,
+      "step": 2170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017205736324318999,
+      "loss": 2.2418,
+      "step": 2175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001718793341483758,
+      "loss": 2.269,
+      "step": 2180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00017170083245806757,
+      "loss": 2.209,
+      "step": 2185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017152185934588623,
+      "loss": 2.2342,
+      "step": 2190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017134241598855236,
+      "loss": 2.328,
+      "step": 2195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001711625035658782,
+      "loss": 2.1285,
+      "step": 2200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00017098212326076008,
+      "loss": 2.3896,
+      "step": 2205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001708012762591706,
+      "loss": 2.2122,
+      "step": 2210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017061996375015078,
+      "loss": 2.0428,
+      "step": 2215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017043818692580228,
+      "loss": 2.3604,
+      "step": 2220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017025594698127965,
+      "loss": 2.2105,
+      "step": 2225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00017007324511478223,
+      "loss": 2.1662,
+      "step": 2230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016989008252754655,
+      "loss": 2.2152,
+      "step": 2235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016970646042383826,
+      "loss": 2.2752,
+      "step": 2240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016952238001094428,
+      "loss": 2.28,
+      "step": 2245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016933784249916476,
+      "loss": 2.2531,
+      "step": 2250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00016915284910180533,
+      "loss": 2.2469,
+      "step": 2255
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016896740103516895,
+      "loss": 2.1296,
+      "step": 2260
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001687814995185479,
+      "loss": 1.9912,
+      "step": 2265
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016859514577421592,
+      "loss": 2.1739,
+      "step": 2270
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016840834102741997,
+      "loss": 2.4593,
+      "step": 2275
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00016822108650637238,
+      "loss": 2.0605,
+      "step": 2280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016803338344224266,
+      "loss": 2.3659,
+      "step": 2285
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016784523306914934,
+      "loss": 2.1919,
+      "step": 2290
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016765663662415204,
+      "loss": 2.1289,
+      "step": 2295
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00016746759534724316,
+      "loss": 2.2207,
+      "step": 2300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016727811048133985,
+      "loss": 2.1974,
+      "step": 2305
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016708818327227574,
+      "loss": 2.1457,
+      "step": 2310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016689781496879283,
+      "loss": 2.5133,
+      "step": 2315
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016670700682253328,
+      "loss": 2.0362,
+      "step": 2320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00016651576008803112,
+      "loss": 2.2621,
+      "step": 2325
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016632407602270398,
+      "loss": 2.1938,
+      "step": 2330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016613195588684488,
+      "loss": 2.266,
+      "step": 2335
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00016593940094361407,
+      "loss": 2.1861,
+      "step": 2340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.0001657464124590304,
+      "loss": 2.2115,
+      "step": 2345
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016555299170196332,
+      "loss": 2.2941,
+      "step": 2350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016535913994412436,
+      "loss": 2.1746,
+      "step": 2355
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016516485846005882,
+      "loss": 2.1835,
+      "step": 2360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00016497014852713738,
+      "loss": 1.9548,
+      "step": 2365
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.0001647750114255477,
+      "loss": 2.1774,
+      "step": 2370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000164579448438286,
+      "loss": 2.3656,
+      "step": 2375
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016438346085114865,
+      "loss": 2.3787,
+      "step": 2380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016418704995272373,
+      "loss": 2.3596,
+      "step": 2385
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00016399021703438247,
+      "loss": 2.214,
+      "step": 2390
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.0001637929633902708,
+      "loss": 2.2219,
+      "step": 2395
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016359529031730093,
+      "loss": 2.2045,
+      "step": 2400
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016339719911514272,
+      "loss": 2.1769,
+      "step": 2405
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016319869108621512,
+      "loss": 1.9191,
+      "step": 2410
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00016299976753567772,
+      "loss": 2.1758,
+      "step": 2415
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016280042977142204,
+      "loss": 2.2566,
+      "step": 2420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016260067910406304,
+      "loss": 2.1223,
+      "step": 2425
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016240051684693042,
+      "loss": 1.9765,
+      "step": 2430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00016219994431606005,
+      "loss": 2.1405,
+      "step": 2435
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016199896283018527,
+      "loss": 2.1078,
+      "step": 2440
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016179757371072824,
+      "loss": 2.124,
+      "step": 2445
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00016159577828179123,
+      "loss": 2.1611,
+      "step": 2450
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001613935778701479,
+      "loss": 2.1657,
+      "step": 2455
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001611909738052347,
+      "loss": 2.1706,
+      "step": 2460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.000160987967419142,
+      "loss": 1.9781,
+      "step": 2465
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016078456004660536,
+      "loss": 2.1244,
+      "step": 2470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016058075302499673,
+      "loss": 2.1409,
+      "step": 2475
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00016037654769431576,
+      "loss": 2.1768,
+      "step": 2480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00016017194539718086,
+      "loss": 2.1801,
+      "step": 2485
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001599669474788205,
+      "loss": 2.1986,
+      "step": 2490
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00015976155528706415,
+      "loss": 2.2472,
+      "step": 2495
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001595557701723338,
+      "loss": 2.18,
+      "step": 2500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00015934959348763467,
+      "loss": 2.2502,
+      "step": 2505
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015914302658854657,
+      "loss": 2.4171,
+      "step": 2510
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015893607083321477,
+      "loss": 2.1392,
+      "step": 2515
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015872872758234148,
+      "loss": 2.2273,
+      "step": 2520
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00015852099819917639,
+      "loss": 2.2318,
+      "step": 2525
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015831288404950802,
+      "loss": 2.4229,
+      "step": 2530
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001581043865016547,
+      "loss": 2.2818,
+      "step": 2535
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015789550692645556,
+      "loss": 2.2214,
+      "step": 2540
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015768624669726145,
+      "loss": 2.1146,
+      "step": 2545
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00015747660718992598,
+      "loss": 2.2155,
+      "step": 2550
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015726658978279642,
+      "loss": 2.2385,
+      "step": 2555
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015705619585670478,
+      "loss": 2.2751,
+      "step": 2560
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015684542679495847,
+      "loss": 2.1865,
+      "step": 2565
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00015663428398333157,
+      "loss": 1.9878,
+      "step": 2570
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001564227688100552,
+      "loss": 2.1527,
+      "step": 2575
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015621088266580904,
+      "loss": 1.9778,
+      "step": 2580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015599862694371157,
+      "loss": 2.2056,
+      "step": 2585
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015578600303931136,
+      "loss": 2.2311,
+      "step": 2590
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00015557301235057767,
+      "loss": 2.0443,
+      "step": 2595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015535965627789126,
+      "loss": 2.0348,
+      "step": 2600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015514593622403532,
+      "loss": 2.4104,
+      "step": 2605
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001549318535941861,
+      "loss": 2.2535,
+      "step": 2610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00015471740979590377,
+      "loss": 2.0757,
+      "step": 2615
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001545026062391231,
+      "loss": 2.1442,
+      "step": 2620
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015428744433614415,
+      "loss": 2.2531,
+      "step": 2625
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00015407192550162318,
+      "loss": 2.0542,
+      "step": 2630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001538560511525632,
+      "loss": 2.2617,
+      "step": 2635
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.0001536398227083046,
+      "loss": 2.3173,
+      "step": 2640
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015342324159051587,
+      "loss": 1.9859,
+      "step": 2645
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015320630922318444,
+      "loss": 2.1397,
+      "step": 2650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015298902703260692,
+      "loss": 2.1437,
+      "step": 2655
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.0001527713964473802,
+      "loss": 2.1295,
+      "step": 2660
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00015255341889839157,
+      "loss": 2.0041,
+      "step": 2665
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015233509581880973,
+      "loss": 2.5931,
+      "step": 2670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.0001521164286440751,
+      "loss": 2.1652,
+      "step": 2675
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015189741881189054,
+      "loss": 2.2568,
+      "step": 2680
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00015167806776221178,
+      "loss": 2.057,
+      "step": 2685
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000151458376937238,
+      "loss": 2.0953,
+      "step": 2690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015123834778140233,
+      "loss": 2.0007,
+      "step": 2695
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015101798174136247,
+      "loss": 2.0058,
+      "step": 2700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.000150797280265991,
+      "loss": 2.2804,
+      "step": 2705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00015057624480636594,
+      "loss": 2.1407,
+      "step": 2710
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.0001503548768157612,
+      "loss": 2.1644,
+      "step": 2715
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00015013317774963708,
+      "loss": 2.2753,
+      "step": 2720
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00014991114906563055,
+      "loss": 2.0797,
+      "step": 2725
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00014968879222354597,
+      "loss": 2.3884,
+      "step": 2730
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00014946610868534502,
+      "loss": 2.1815,
+      "step": 2735
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00014924309991513757,
+      "loss": 2.0124,
+      "step": 2740
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001490197673791717,
+      "loss": 2.2273,
+      "step": 2745
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00014879611254582428,
+      "loss": 2.1957,
+      "step": 2750
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00014857213688559124,
+      "loss": 2.1517,
+      "step": 2755
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00014834784187107785,
+      "loss": 2.2585,
+      "step": 2760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00014812322897698912,
+      "loss": 2.0488,
+      "step": 2765
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00014789829968012,
+      "loss": 2.0742,
+      "step": 2770
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00014767305545934588,
+      "loss": 2.1186,
+      "step": 2775
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00014744749779561258,
+      "loss": 2.2138,
+      "step": 2780
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001472216281719269,
+      "loss": 2.2996,
+      "step": 2785
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001469954480733465,
+      "loss": 2.2487,
+      "step": 2790
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00014676895898697062,
+      "loss": 1.9628,
+      "step": 2795
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00014654216240192995,
+      "loss": 2.2419,
+      "step": 2800
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00014631505980937688,
+      "loss": 2.2693,
+      "step": 2805
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001460876527024758,
+      "loss": 2.1449,
+      "step": 2810
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00014585994257639324,
+      "loss": 2.2163,
+      "step": 2815
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00014563193092828803,
+      "loss": 2.314,
+      "step": 2820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00014540361925730147,
+      "loss": 2.3863,
+      "step": 2825
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00014517500906454742,
+      "loss": 2.0355,
+      "step": 2830
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00014494610185310252,
+      "loss": 1.8842,
+      "step": 2835
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00014471689912799626,
+      "loss": 2.2408,
+      "step": 2840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00014448740239620108,
+      "loss": 2.3598,
+      "step": 2845
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014425761316662241,
+      "loss": 2.2498,
+      "step": 2850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001440275329500889,
+      "loss": 2.2595,
+      "step": 2855
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014379716325934236,
+      "loss": 2.1201,
+      "step": 2860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001435665056090278,
+      "loss": 2.059,
+      "step": 2865
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00014333556151568364,
+      "loss": 2.0564,
+      "step": 2870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014310433249773146,
+      "loss": 2.071,
+      "step": 2875
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014287282007546627,
+      "loss": 2.2842,
+      "step": 2880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014264102577104645,
+      "loss": 2.0015,
+      "step": 2885
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00014240895110848365,
+      "loss": 2.0825,
+      "step": 2890
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001421765976136328,
+      "loss": 2.3226,
+      "step": 2895
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001419439668141822,
+      "loss": 2.1655,
+      "step": 2900
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.0001417110602396434,
+      "loss": 2.0336,
+      "step": 2905
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00014147787942134089,
+      "loss": 2.1605,
+      "step": 2910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00014124442589240265,
+      "loss": 2.2324,
+      "step": 2915
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014101070118774936,
+      "loss": 2.1054,
+      "step": 2920
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014077670684408485,
+      "loss": 1.9368,
+      "step": 2925
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00014054244439988566,
+      "loss": 2.2169,
+      "step": 2930
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.0001403079153953911,
+      "loss": 2.2096,
+      "step": 2935
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00014007312137259307,
+      "loss": 2.2111,
+      "step": 2940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00013983806387522592,
+      "loss": 2.2063,
+      "step": 2945
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00013960274444875628,
+      "loss": 2.2548,
+      "step": 2950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.000139367164640373,
+      "loss": 2.0605,
+      "step": 2955
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00013913132599897683,
+      "loss": 2.3223,
+      "step": 2960
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00013889523007517028,
+      "loss": 2.1303,
+      "step": 2965
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00013865887842124755,
+      "loss": 2.179,
+      "step": 2970
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.0001384222725911842,
+      "loss": 1.9177,
+      "step": 2975
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00013818541414062683,
+      "loss": 2.3301,
+      "step": 2980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001379483046268832,
+      "loss": 2.2388,
+      "step": 2985
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00013771094560891155,
+      "loss": 2.2499,
+      "step": 2990
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00013747333864731073,
+      "loss": 2.2831,
+      "step": 2995
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00013723548530430974,
+      "loss": 2.3621,
+      "step": 3000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00013699738714375748,
+      "loss": 1.9664,
+      "step": 3005
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00013675904573111247,
+      "loss": 2.2979,
+      "step": 3010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00013652046263343262,
+      "loss": 2.169,
+      "step": 3015
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00013628163941936485,
+      "loss": 2.1979,
+      "step": 3020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00013604257765913484,
+      "loss": 2.2215,
+      "step": 3025
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001358032789245366,
+      "loss": 2.0594,
+      "step": 3030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00013556374478892232,
+      "loss": 2.2148,
+      "step": 3035
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00013532397682719185,
+      "loss": 2.1494,
+      "step": 3040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00013508397661578242,
+      "loss": 2.2675,
+      "step": 3045
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001348437457326582,
+      "loss": 2.222,
+      "step": 3050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00013460328575730019,
+      "loss": 2.0916,
+      "step": 3055
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00013436259827069534,
+      "loss": 2.358,
+      "step": 3060
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00013412168485532676,
+      "loss": 2.2071,
+      "step": 3065
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00013388054709516272,
+      "loss": 2.3862,
+      "step": 3070
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001336391865756468,
+      "loss": 2.1506,
+      "step": 3075
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00013339760488368695,
+      "loss": 1.9016,
+      "step": 3080
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00013315580360764542,
+      "loss": 2.0225,
+      "step": 3085
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00013291378433732818,
+      "loss": 2.0375,
+      "step": 3090
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00013267154866397447,
+      "loss": 2.3393,
+      "step": 3095
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00013242909818024628,
+      "loss": 2.2597,
+      "step": 3100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001321864344802181,
+      "loss": 2.2359,
+      "step": 3105
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00013194355915936611,
+      "loss": 2.3076,
+      "step": 3110
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.000131700473814558,
+      "loss": 1.9924,
+      "step": 3115
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00013145718004404223,
+      "loss": 2.263,
+      "step": 3120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00013121367944743777,
+      "loss": 2.1274,
+      "step": 3125
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001309699736257232,
+      "loss": 2.2208,
+      "step": 3130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00013072606418122667,
+      "loss": 2.1972,
+      "step": 3135
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00013048195271761498,
+      "loss": 2.1493,
+      "step": 3140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00013023764083988323,
+      "loss": 2.4829,
+      "step": 3145
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001299931301543442,
+      "loss": 2.0439,
+      "step": 3150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00012974842226861773,
+      "loss": 2.3323,
+      "step": 3155
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001295035187916204,
+      "loss": 2.1791,
+      "step": 3160
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00012925842133355454,
+      "loss": 2.0196,
+      "step": 3165
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00012901313150589806,
+      "loss": 2.3059,
+      "step": 3170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001287676509213936,
+      "loss": 1.8997,
+      "step": 3175
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00012852198119403798,
+      "loss": 2.1809,
+      "step": 3180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00012827612393907163,
+      "loss": 2.2636,
+      "step": 3185
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001280300807729679,
+      "loss": 1.9401,
+      "step": 3190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001277838533134226,
+      "loss": 2.053,
+      "step": 3195
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00012753744317934307,
+      "loss": 2.3496,
+      "step": 3200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001272908519908379,
+      "loss": 2.3493,
+      "step": 3205
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00012704408136920585,
+      "loss": 2.1661,
+      "step": 3210
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.0001267971329369256,
+      "loss": 2.2261,
+      "step": 3215
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00012655000831764495,
+      "loss": 2.3115,
+      "step": 3220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00012630270913616985,
+      "loss": 2.0858,
+      "step": 3225
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00012605523701845431,
+      "loss": 2.2376,
+      "step": 3230
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00012580759359158905,
+      "loss": 2.3248,
+      "step": 3235
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00012555978048379133,
+      "loss": 2.2613,
+      "step": 3240
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00012531179932439397,
+      "loss": 1.9674,
+      "step": 3245
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00012506365174383467,
+      "loss": 2.0881,
+      "step": 3250
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.0001248153393736454,
+      "loss": 2.1558,
+      "step": 3255
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00012456686384644148,
+      "loss": 2.2714,
+      "step": 3260
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00012431822679591112,
+      "loss": 2.1985,
+      "step": 3265
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00012406942985680437,
+      "loss": 2.3348,
+      "step": 3270
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00012382047466492262,
+      "loss": 2.1145,
+      "step": 3275
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001235713628571077,
+      "loss": 1.9917,
+      "step": 3280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00012332209607123117,
+      "loss": 2.1811,
+      "step": 3285
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001230726759461836,
+      "loss": 2.2461,
+      "step": 3290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00012282310412186365,
+      "loss": 2.1153,
+      "step": 3295
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001225733822391675,
+      "loss": 1.8411,
+      "step": 3300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00012232351193997774,
+      "loss": 1.9911,
+      "step": 3305
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000122073494867153,
+      "loss": 2.091,
+      "step": 3310
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00012182333266451684,
+      "loss": 1.978,
+      "step": 3315
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00012157302697684695,
+      "loss": 2.0108,
+      "step": 3320
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00012132257944986454,
+      "loss": 2.1627,
+      "step": 3325
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00012107199173022327,
+      "loss": 2.1021,
+      "step": 3330
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00012082126546549864,
+      "loss": 2.0461,
+      "step": 3335
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000120570402304177,
+      "loss": 2.2276,
+      "step": 3340
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00012031940389564478,
+      "loss": 2.0143,
+      "step": 3345
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00012006827189017773,
+      "loss": 2.3599,
+      "step": 3350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00011981700793892982,
+      "loss": 1.8899,
+      "step": 3355
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00011956561369392274,
+      "loss": 2.1409,
+      "step": 3360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001193140908080346,
+      "loss": 2.0616,
+      "step": 3365
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00011906244093498955,
+      "loss": 2.1514,
+      "step": 3370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00011881066572934644,
+      "loss": 2.1447,
+      "step": 3375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00011855876684648837,
+      "loss": 1.8911,
+      "step": 3380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00011830674594261145,
+      "loss": 2.25,
+      "step": 3385
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001180546046747141,
+      "loss": 2.157,
+      "step": 3390
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00011780234470058613,
+      "loss": 2.2546,
+      "step": 3395
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001175499676787978,
+      "loss": 2.0396,
+      "step": 3400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000117297475268689,
+      "loss": 2.1004,
+      "step": 3405
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00011704486913035819,
+      "loss": 1.9472,
+      "step": 3410
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00011679215092465163,
+      "loss": 2.1529,
+      "step": 3415
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00011653932231315245,
+      "loss": 2.2245,
+      "step": 3420
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00011628638495816955,
+      "loss": 2.1762,
+      "step": 3425
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00011603334052272696,
+      "loss": 2.0016,
+      "step": 3430
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001157801906705526,
+      "loss": 2.1391,
+      "step": 3435
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00011552693706606758,
+      "loss": 1.9266,
+      "step": 3440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00011527358137437516,
+      "loss": 2.2236,
+      "step": 3445
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00011502012526124978,
+      "loss": 1.9869,
+      "step": 3450
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00011476657039312613,
+      "loss": 2.021,
+      "step": 3455
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00011451291843708824,
+      "loss": 2.2469,
+      "step": 3460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00011425917106085844,
+      "loss": 2.0205,
+      "step": 3465
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00011400532993278643,
+      "loss": 2.0265,
+      "step": 3470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00011375139672183834,
+      "loss": 1.9317,
+      "step": 3475
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00011349737309758572,
+      "loss": 2.2193,
+      "step": 3480
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00011324326073019458,
+      "loss": 2.0085,
+      "step": 3485
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001129890612904144,
+      "loss": 2.1403,
+      "step": 3490
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001127347764495671,
+      "loss": 2.2252,
+      "step": 3495
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00011248040787953622,
+      "loss": 2.2423,
+      "step": 3500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00011222595725275562,
+      "loss": 2.0528,
+      "step": 3505
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00011197142624219887,
+      "loss": 1.9843,
+      "step": 3510
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00011171681652136793,
+      "loss": 2.131,
+      "step": 3515
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00011146212976428232,
+      "loss": 2.1704,
+      "step": 3520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00011120736764546799,
+      "loss": 2.1937,
+      "step": 3525
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00011095253183994645,
+      "loss": 2.3802,
+      "step": 3530
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001106976240232237,
+      "loss": 2.0748,
+      "step": 3535
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.0001104426458712791,
+      "loss": 2.0799,
+      "step": 3540
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00011018759906055463,
+      "loss": 2.0325,
+      "step": 3545
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00010993248526794347,
+      "loss": 2.2288,
+      "step": 3550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00010967730617077938,
+      "loss": 2.111,
+      "step": 3555
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00010942206344682541,
+      "loss": 2.1403,
+      "step": 3560
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00010916675877426296,
+      "loss": 2.1997,
+      "step": 3565
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00010891139383168072,
+      "loss": 1.9816,
+      "step": 3570
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010865597029806365,
+      "loss": 1.9475,
+      "step": 3575
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010840048985278195,
+      "loss": 1.9974,
+      "step": 3580
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010814495417557997,
+      "loss": 2.1288,
+      "step": 3585
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010788936494656523,
+      "loss": 2.1858,
+      "step": 3590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00010763372384619738,
+      "loss": 2.3401,
+      "step": 3595
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010737803255527702,
+      "loss": 2.0626,
+      "step": 3600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010712229275493489,
+      "loss": 1.9928,
+      "step": 3605
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010686650612662048,
+      "loss": 2.2724,
+      "step": 3610
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00010661067435209135,
+      "loss": 2.1122,
+      "step": 3615
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010635479911340176,
+      "loss": 2.2409,
+      "step": 3620
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010609888209289183,
+      "loss": 2.0935,
+      "step": 3625
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010584292497317633,
+      "loss": 2.0518,
+      "step": 3630
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00010558692943713373,
+      "loss": 1.9779,
+      "step": 3635
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.000105330897167895,
+      "loss": 2.0863,
+      "step": 3640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00010507482984883268,
+      "loss": 2.2319,
+      "step": 3645
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00010481872916354978,
+      "loss": 2.0148,
+      "step": 3650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00010456259679586862,
+      "loss": 2.1143,
+      "step": 3655
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00010430643442981986,
+      "loss": 2.1747,
+      "step": 3660
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001040502437496315,
+      "loss": 2.298,
+      "step": 3665
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00010379402643971746,
+      "loss": 2.0557,
+      "step": 3670
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00010353778418466697,
+      "loss": 2.0455,
+      "step": 3675
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00010328151866923316,
+      "loss": 2.2097,
+      "step": 3680
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00010302523157832216,
+      "loss": 2.2056,
+      "step": 3685
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00010276892459698182,
+      "loss": 2.272,
+      "step": 3690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00010251259941039098,
+      "loss": 2.0397,
+      "step": 3695
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00010225625770384797,
+      "loss": 2.1304,
+      "step": 3700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00010199990116275988,
+      "loss": 2.2824,
+      "step": 3705
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00010174353147263125,
+      "loss": 2.0028,
+      "step": 3710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00010148715031905312,
+      "loss": 1.9134,
+      "step": 3715
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00010123075938769187,
+      "loss": 2.1108,
+      "step": 3720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00010097436036427816,
+      "loss": 1.9122,
+      "step": 3725
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00010071795493459591,
+      "loss": 2.2849,
+      "step": 3730
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00010046154478447114,
+      "loss": 2.1937,
+      "step": 3735
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00010020513159976084,
+      "loss": 1.9927,
+      "step": 3740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.994871706634204e-05,
+      "loss": 2.1293,
+      "step": 3745
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.96923028701006e-05,
+      "loss": 2.1859,
+      "step": 3750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.943589069692014e-05,
+      "loss": 2.1459,
+      "step": 3755
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.917948223267105e-05,
+      "loss": 2.1753,
+      "step": 3760
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.892307916319919e-05,
+      "loss": 2.0521,
+      "step": 3765
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.866668317431514e-05,
+      "loss": 2.1496,
+      "step": 3770
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.841029595178282e-05,
+      "loss": 2.1725,
+      "step": 3775
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.815391918130848e-05,
+      "loss": 2.0178,
+      "step": 3780
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.789755454852971e-05,
+      "loss": 2.1582,
+      "step": 3785
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.764120373900436e-05,
+      "loss": 2.1756,
+      "step": 3790
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.738486843819919e-05,
+      "loss": 2.2592,
+      "step": 3795
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.712855033147921e-05,
+      "loss": 2.2422,
+      "step": 3800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.68722511040962e-05,
+      "loss": 2.1047,
+      "step": 3805
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.661597244117802e-05,
+      "loss": 2.0808,
+      "step": 3810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.635971602771716e-05,
+      "loss": 2.0067,
+      "step": 3815
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.61034835485598e-05,
+      "loss": 2.1168,
+      "step": 3820
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.584727668839487e-05,
+      "loss": 2.2862,
+      "step": 3825
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.559109713174282e-05,
+      "loss": 2.0351,
+      "step": 3830
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.533494656294458e-05,
+      "loss": 2.1812,
+      "step": 3835
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.507882666615049e-05,
+      "loss": 1.9858,
+      "step": 3840
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.482273912530913e-05,
+      "loss": 2.0261,
+      "step": 3845
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.456668562415657e-05,
+      "loss": 2.0648,
+      "step": 3850
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.431066784620486e-05,
+      "loss": 2.0273,
+      "step": 3855
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.405468747473127e-05,
+      "loss": 2.0114,
+      "step": 3860
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.379874619276707e-05,
+      "loss": 2.2219,
+      "step": 3865
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.354284568308665e-05,
+      "loss": 2.314,
+      "step": 3870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.328698762819623e-05,
+      "loss": 2.1478,
+      "step": 3875
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.303117371032284e-05,
+      "loss": 2.081,
+      "step": 3880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.277540561140342e-05,
+      "loss": 2.1293,
+      "step": 3885
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.251968501307365e-05,
+      "loss": 1.9751,
+      "step": 3890
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.226401359665686e-05,
+      "loss": 2.0561,
+      "step": 3895
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.2008393043153e-05,
+      "loss": 2.1842,
+      "step": 3900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.17528250332277e-05,
+      "loss": 2.0628,
+      "step": 3905
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.149731124720104e-05,
+      "loss": 2.0727,
+      "step": 3910
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.124185336503656e-05,
+      "loss": 2.1095,
+      "step": 3915
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.098645306633029e-05,
+      "loss": 2.0786,
+      "step": 3920
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.073111203029972e-05,
+      "loss": 1.992,
+      "step": 3925
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.04758319357726e-05,
+      "loss": 1.9248,
+      "step": 3930
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.0220614461176e-05,
+      "loss": 2.0359,
+      "step": 3935
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.99654612845253e-05,
+      "loss": 1.7823,
+      "step": 3940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.971037408341319e-05,
+      "loss": 2.0699,
+      "step": 3945
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.94553545349985e-05,
+      "loss": 2.0313,
+      "step": 3950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 8.92004043159953e-05,
+      "loss": 2.1288,
+      "step": 3955
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.894552510266172e-05,
+      "loss": 2.236,
+      "step": 3960
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.869071857078926e-05,
+      "loss": 2.0301,
+      "step": 3965
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.843598639569134e-05,
+      "loss": 2.0237,
+      "step": 3970
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.818133025219258e-05,
+      "loss": 2.2037,
+      "step": 3975
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 8.79267518146177e-05,
+      "loss": 2.0459,
+      "step": 3980
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.767225275678054e-05,
+      "loss": 2.4051,
+      "step": 3985
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.741783475197301e-05,
+      "loss": 2.1853,
+      "step": 3990
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.716349947295406e-05,
+      "loss": 1.9654,
+      "step": 3995
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 8.690924859193877e-05,
+      "loss": 1.9935,
+      "step": 4000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.665508378058737e-05,
+      "loss": 1.9519,
+      "step": 4005
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.640100670999413e-05,
+      "loss": 2.1032,
+      "step": 4010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.614701905067648e-05,
+      "loss": 2.1319,
+      "step": 4015
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.589312247256385e-05,
+      "loss": 1.9318,
+      "step": 4020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 8.563931864498709e-05,
+      "loss": 1.9461,
+      "step": 4025
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.538560923666697e-05,
+      "loss": 2.0611,
+      "step": 4030
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.51319959157036e-05,
+      "loss": 1.9359,
+      "step": 4035
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.487848034956527e-05,
+      "loss": 2.109,
+      "step": 4040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.462506420507764e-05,
+      "loss": 2.1609,
+      "step": 4045
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 8.437174914841261e-05,
+      "loss": 2.3065,
+      "step": 4050
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.411853684507744e-05,
+      "loss": 2.2701,
+      "step": 4055
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.38654289599038e-05,
+      "loss": 1.8895,
+      "step": 4060
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.36124271570369e-05,
+      "loss": 2.0993,
+      "step": 4065
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 8.335953309992442e-05,
+      "loss": 2.2109,
+      "step": 4070
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.310674845130563e-05,
+      "loss": 2.3028,
+      "step": 4075
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.285407487320042e-05,
+      "loss": 1.9683,
+      "step": 4080
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.260151402689848e-05,
+      "loss": 2.0915,
+      "step": 4085
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.234906757294829e-05,
+      "loss": 1.8933,
+      "step": 4090
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 8.209673717114618e-05,
+      "loss": 2.148,
+      "step": 4095
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.184452448052547e-05,
+      "loss": 2.1884,
+      "step": 4100
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.15924311593456e-05,
+      "loss": 1.8173,
+      "step": 4105
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.134045886508108e-05,
+      "loss": 2.274,
+      "step": 4110
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 8.108860925441076e-05,
+      "loss": 1.9474,
+      "step": 4115
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.083688398320681e-05,
+      "loss": 2.047,
+      "step": 4120
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.058528470652396e-05,
+      "loss": 1.8999,
+      "step": 4125
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.03338130785885e-05,
+      "loss": 1.9803,
+      "step": 4130
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 8.008247075278742e-05,
+      "loss": 2.0983,
+      "step": 4135
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.983125938165758e-05,
+      "loss": 1.9923,
+      "step": 4140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.958018061687494e-05,
+      "loss": 1.9653,
+      "step": 4145
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.932923610924343e-05,
+      "loss": 2.1081,
+      "step": 4150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.907842750868441e-05,
+      "loss": 1.8204,
+      "step": 4155
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.882775646422547e-05,
+      "loss": 1.8835,
+      "step": 4160
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.857722462399009e-05,
+      "loss": 2.0082,
+      "step": 4165
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.832683363518621e-05,
+      "loss": 2.0749,
+      "step": 4170
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.807658514409587e-05,
+      "loss": 2.0953,
+      "step": 4175
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.782648079606412e-05,
+      "loss": 2.0271,
+      "step": 4180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.757652223548836e-05,
+      "loss": 1.9882,
+      "step": 4185
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.732671110580746e-05,
+      "loss": 2.1378,
+      "step": 4190
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.707704904949085e-05,
+      "loss": 1.9259,
+      "step": 4195
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.682753770802791e-05,
+      "loss": 2.1103,
+      "step": 4200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.657817872191713e-05,
+      "loss": 2.2728,
+      "step": 4205
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.632897373065522e-05,
+      "loss": 2.0853,
+      "step": 4210
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.607992437272642e-05,
+      "loss": 2.177,
+      "step": 4215
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.583103228559164e-05,
+      "loss": 1.8806,
+      "step": 4220
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.558229910567794e-05,
+      "loss": 2.0561,
+      "step": 4225
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.533372646836736e-05,
+      "loss": 2.0183,
+      "step": 4230
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.508531600798657e-05,
+      "loss": 2.0099,
+      "step": 4235
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.483706935779584e-05,
+      "loss": 2.0262,
+      "step": 4240
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.458898814997852e-05,
+      "loss": 1.9678,
+      "step": 4245
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.434107401563016e-05,
+      "loss": 2.1944,
+      "step": 4250
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.409332858474772e-05,
+      "loss": 1.8389,
+      "step": 4255
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.384575348621909e-05,
+      "loss": 1.9925,
+      "step": 4260
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.359835034781227e-05,
+      "loss": 2.0963,
+      "step": 4265
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.335112079616456e-05,
+      "loss": 2.1287,
+      "step": 4270
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.31040664567719e-05,
+      "loss": 2.0697,
+      "step": 4275
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.285718895397848e-05,
+      "loss": 2.0554,
+      "step": 4280
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.261048991096558e-05,
+      "loss": 2.1972,
+      "step": 4285
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.236397094974119e-05,
+      "loss": 2.091,
+      "step": 4290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.211763369112934e-05,
+      "loss": 1.9977,
+      "step": 4295
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.18714797547594e-05,
+      "loss": 1.9408,
+      "step": 4300
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.162551075905538e-05,
+      "loss": 1.993,
+      "step": 4305
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.137972832122532e-05,
+      "loss": 2.1485,
+      "step": 4310
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.113413405725069e-05,
+      "loss": 2.253,
+      "step": 4315
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.088872958187578e-05,
+      "loss": 2.0418,
+      "step": 4320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.064351650859704e-05,
+      "loss": 1.7988,
+      "step": 4325
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.039849644965246e-05,
+      "loss": 2.2719,
+      "step": 4330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.015367101601091e-05,
+      "loss": 2.1789,
+      "step": 4335
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.990904181736187e-05,
+      "loss": 2.1065,
+      "step": 4340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.96646104621043e-05,
+      "loss": 1.947,
+      "step": 4345
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.942037855733661e-05,
+      "loss": 2.0803,
+      "step": 4350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.917634770884571e-05,
+      "loss": 2.0772,
+      "step": 4355
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.893251952109668e-05,
+      "loss": 1.9202,
+      "step": 4360
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.868889559722213e-05,
+      "loss": 2.1797,
+      "step": 4365
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.84454775390116e-05,
+      "loss": 2.1386,
+      "step": 4370
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.820226694690112e-05,
+      "loss": 2.0789,
+      "step": 4375
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.795926541996273e-05,
+      "loss": 2.0847,
+      "step": 4380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.771647455589384e-05,
+      "loss": 1.8474,
+      "step": 4385
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.74738959510068e-05,
+      "loss": 2.1413,
+      "step": 4390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.723153120021833e-05,
+      "loss": 1.9739,
+      "step": 4395
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.698938189703918e-05,
+      "loss": 1.9685,
+      "step": 4400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.674744963356357e-05,
+      "loss": 2.0361,
+      "step": 4405
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.65057360004586e-05,
+      "loss": 1.9542,
+      "step": 4410
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.626424258695403e-05,
+      "loss": 2.023,
+      "step": 4415
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.60229709808317e-05,
+      "loss": 1.9255,
+      "step": 4420
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.578192276841501e-05,
+      "loss": 2.1247,
+      "step": 4425
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.554109953455864e-05,
+      "loss": 2.1402,
+      "step": 4430
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.53005028626381e-05,
+      "loss": 1.9641,
+      "step": 4435
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.506013433453926e-05,
+      "loss": 1.9861,
+      "step": 4440
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.4819995530648e-05,
+      "loss": 2.1035,
+      "step": 4445
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.45800880298397e-05,
+      "loss": 2.0726,
+      "step": 4450
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.434041340946909e-05,
+      "loss": 1.9865,
+      "step": 4455
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.41009732453597e-05,
+      "loss": 2.1554,
+      "step": 4460
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.386176911179353e-05,
+      "loss": 2.265,
+      "step": 4465
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.362280258150074e-05,
+      "loss": 2.1058,
+      "step": 4470
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.33840752256492e-05,
+      "loss": 1.9767,
+      "step": 4475
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.314558861383442e-05,
+      "loss": 2.0712,
+      "step": 4480
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.29073443140689e-05,
+      "loss": 2.0934,
+      "step": 4485
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.266934389277204e-05,
+      "loss": 2.0419,
+      "step": 4490
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.24315889147597e-05,
+      "loss": 2.0022,
+      "step": 4495
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.219408094323415e-05,
+      "loss": 2.0074,
+      "step": 4500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.195682153977351e-05,
+      "loss": 2.1639,
+      "step": 4505
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.17198122643216e-05,
+      "loss": 2.1314,
+      "step": 4510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.148305467517768e-05,
+      "loss": 2.1673,
+      "step": 4515
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.124655032898631e-05,
+      "loss": 2.109,
+      "step": 4520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.1010300780726925e-05,
+      "loss": 2.1037,
+      "step": 4525
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.077430758370376e-05,
+      "loss": 1.8841,
+      "step": 4530
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.053857228953546e-05,
+      "loss": 2.105,
+      "step": 4535
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.03030964481452e-05,
+      "loss": 2.1448,
+      "step": 4540
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.0067881607750134e-05,
+      "loss": 1.927,
+      "step": 4545
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.983292931485142e-05,
+      "loss": 2.0516,
+      "step": 4550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.9598241114223986e-05,
+      "loss": 2.1061,
+      "step": 4555
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.936381854890646e-05,
+      "loss": 2.0763,
+      "step": 4560
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.912966316019093e-05,
+      "loss": 1.9717,
+      "step": 4565
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.8895776487612765e-05,
+      "loss": 2.211,
+      "step": 4570
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.8662160068940655e-05,
+      "loss": 2.0123,
+      "step": 4575
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.84288154401664e-05,
+      "loss": 2.0151,
+      "step": 4580
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.81957441354948e-05,
+      "loss": 1.9295,
+      "step": 4585
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.796294768733362e-05,
+      "loss": 1.8365,
+      "step": 4590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.773042762628342e-05,
+      "loss": 2.1724,
+      "step": 4595
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.749818548112762e-05,
+      "loss": 2.1015,
+      "step": 4600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.726622277882243e-05,
+      "loss": 2.0416,
+      "step": 4605
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.703454104448665e-05,
+      "loss": 2.0669,
+      "step": 4610
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.680314180139178e-05,
+      "loss": 2.0983,
+      "step": 4615
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.657202657095206e-05,
+      "loss": 1.9449,
+      "step": 4620
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.6341196872714394e-05,
+      "loss": 1.8182,
+      "step": 4625
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.611065422434828e-05,
+      "loss": 2.0388,
+      "step": 4630
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.588040014163585e-05,
+      "loss": 1.9935,
+      "step": 4635
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.565043613846219e-05,
+      "loss": 2.0616,
+      "step": 4640
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.542076372680498e-05,
+      "loss": 1.9655,
+      "step": 4645
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.519138441672471e-05,
+      "loss": 2.0487,
+      "step": 4650
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.496229971635487e-05,
+      "loss": 2.0029,
+      "step": 4655
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.473351113189194e-05,
+      "loss": 1.9061,
+      "step": 4660
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.4505020167585396e-05,
+      "loss": 1.9066,
+      "step": 4665
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.4276828325727934e-05,
+      "loss": 2.1638,
+      "step": 4670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.4048937106645613e-05,
+      "loss": 2.0569,
+      "step": 4675
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.3821348008687967e-05,
+      "loss": 2.0398,
+      "step": 4680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.3594062528218025e-05,
+      "loss": 2.0051,
+      "step": 4685
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.336708215960258e-05,
+      "loss": 2.0979,
+      "step": 4690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.314040839520253e-05,
+      "loss": 1.8456,
+      "step": 4695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.291404272536275e-05,
+      "loss": 1.9396,
+      "step": 4700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.268798663840243e-05,
+      "loss": 1.9458,
+      "step": 4705
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.2462241620605366e-05,
+      "loss": 2.2062,
+      "step": 4710
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.223680915621014e-05,
+      "loss": 1.8613,
+      "step": 4715
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.2011690727400285e-05,
+      "loss": 1.8888,
+      "step": 4720
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.178688781429455e-05,
+      "loss": 2.0201,
+      "step": 4725
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.1562401894937365e-05,
+      "loss": 2.1154,
+      "step": 4730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.133823444528889e-05,
+      "loss": 2.0774,
+      "step": 4735
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.111438693921536e-05,
+      "loss": 1.7942,
+      "step": 4740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.089086084847954e-05,
+      "loss": 2.0274,
+      "step": 4745
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.066765764273078e-05,
+      "loss": 2.1798,
+      "step": 4750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.044477878949571e-05,
+      "loss": 1.9668,
+      "step": 4755
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.0222225754168175e-05,
+      "loss": 1.8329,
+      "step": 4760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 1.8233,
+      "step": 4765
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.97781029880911e-05,
+      "loss": 2.063,
+      "step": 4770
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.955653617737995e-05,
+      "loss": 2.1226,
+      "step": 4775
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.9335301024634094e-05,
+      "loss": 1.8759,
+      "step": 4780
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.911439898444036e-05,
+      "loss": 1.9641,
+      "step": 4785
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.889383150919543e-05,
+      "loss": 2.0135,
+      "step": 4790
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.867360004909635e-05,
+      "loss": 2.0777,
+      "step": 4795
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.845370605213091e-05,
+      "loss": 2.0989,
+      "step": 4800
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.823415096406806e-05,
+      "loss": 2.054,
+      "step": 4805
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.801493622844847e-05,
+      "loss": 1.9905,
+      "step": 4810
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.779606328657513e-05,
+      "loss": 2.1543,
+      "step": 4815
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.75775335775038e-05,
+      "loss": 1.8741,
+      "step": 4820
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.735934853803339e-05,
+      "loss": 1.8314,
+      "step": 4825
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.71415096026968e-05,
+      "loss": 2.0789,
+      "step": 4830
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.692401820375134e-05,
+      "loss": 1.9886,
+      "step": 4835
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.6706875771169265e-05,
+      "loss": 1.7691,
+      "step": 4840
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.64900837326284e-05,
+      "loss": 2.0365,
+      "step": 4845
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.627364351350288e-05,
+      "loss": 2.0404,
+      "step": 4850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.605755653685366e-05,
+      "loss": 2.0534,
+      "step": 4855
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.584182422341915e-05,
+      "loss": 2.2723,
+      "step": 4860
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.562644799160585e-05,
+      "loss": 2.1152,
+      "step": 4865
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.541142925747919e-05,
+      "loss": 1.9986,
+      "step": 4870
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.519676943475408e-05,
+      "loss": 2.1505,
+      "step": 4875
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.4982469934785574e-05,
+      "loss": 2.0285,
+      "step": 4880
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.4768532166559763e-05,
+      "loss": 2.2752,
+      "step": 4885
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.455495753668428e-05,
+      "loss": 2.055,
+      "step": 4890
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.4341747449379335e-05,
+      "loss": 1.9358,
+      "step": 4895
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.412890330646815e-05,
+      "loss": 2.0508,
+      "step": 4900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.391642650736811e-05,
+      "loss": 2.1712,
+      "step": 4905
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.370431844908119e-05,
+      "loss": 2.1583,
+      "step": 4910
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.349258052618509e-05,
+      "loss": 1.9494,
+      "step": 4915
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.328121413082388e-05,
+      "loss": 2.0583,
+      "step": 4920
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.307022065269887e-05,
+      "loss": 2.2009,
+      "step": 4925
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.285960147905946e-05,
+      "loss": 2.057,
+      "step": 4930
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.264935799469417e-05,
+      "loss": 2.0095,
+      "step": 4935
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.2439491581921373e-05,
+      "loss": 2.0035,
+      "step": 4940
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.223000362058023e-05,
+      "loss": 1.9214,
+      "step": 4945
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.202089548802157e-05,
+      "loss": 2.065,
+      "step": 4950
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.181216855909913e-05,
+      "loss": 2.0591,
+      "step": 4955
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.16038242061601e-05,
+      "loss": 1.9348,
+      "step": 4960
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.139586379903629e-05,
+      "loss": 1.8648,
+      "step": 4965
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.1188288705035226e-05,
+      "loss": 2.2961,
+      "step": 4970
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.098110028893105e-05,
+      "loss": 1.8785,
+      "step": 4975
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.077429991295549e-05,
+      "loss": 1.9871,
+      "step": 4980
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.056788893678898e-05,
+      "loss": 2.0064,
+      "step": 4985
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.036186871755173e-05,
+      "loss": 1.9484,
+      "step": 4990
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.015624060979486e-05,
+      "loss": 2.1859,
+      "step": 4995
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.995100596549128e-05,
+      "loss": 1.991,
+      "step": 5000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9746166134026995e-05,
+      "loss": 2.2204,
+      "step": 5005
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9541722462192196e-05,
+      "loss": 2.1229,
+      "step": 5010
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.9337676294172424e-05,
+      "loss": 1.8825,
+      "step": 5015
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.913402897153957e-05,
+      "loss": 1.9276,
+      "step": 5020
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.893078183324329e-05,
+      "loss": 1.9681,
+      "step": 5025
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.8727936215602077e-05,
+      "loss": 1.9694,
+      "step": 5030
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.852549345229445e-05,
+      "loss": 2.0049,
+      "step": 5035
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.832345487435019e-05,
+      "loss": 1.9067,
+      "step": 5040
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.812182181014169e-05,
+      "loss": 1.9862,
+      "step": 5045
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.792059558537518e-05,
+      "loss": 2.0121,
+      "step": 5050
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.7719777523081864e-05,
+      "loss": 2.0614,
+      "step": 5055
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.751936894360949e-05,
+      "loss": 2.0172,
+      "step": 5060
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.731937116461336e-05,
+      "loss": 2.0109,
+      "step": 5065
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.7119785501047977e-05,
+      "loss": 1.8886,
+      "step": 5070
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6920613265158124e-05,
+      "loss": 1.9128,
+      "step": 5075
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.672185576647047e-05,
+      "loss": 2.2601,
+      "step": 5080
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.652351431178473e-05,
+      "loss": 2.0205,
+      "step": 5085
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.6325590205165314e-05,
+      "loss": 2.0568,
+      "step": 5090
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.612808474793261e-05,
+      "loss": 1.9732,
+      "step": 5095
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.593099923865438e-05,
+      "loss": 2.1866,
+      "step": 5100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.573433497313731e-05,
+      "loss": 2.0571,
+      "step": 5105
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5538093244418525e-05,
+      "loss": 2.1196,
+      "step": 5110
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.5342275342757046e-05,
+      "loss": 1.9842,
+      "step": 5115
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.5146882555625226e-05,
+      "loss": 1.8952,
+      "step": 5120
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.495191616770034e-05,
+      "loss": 2.0592,
+      "step": 5125
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.475737746085631e-05,
+      "loss": 2.0654,
+      "step": 5130
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.456326771415498e-05,
+      "loss": 2.0577,
+      "step": 5135
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.436958820383783e-05,
+      "loss": 1.9978,
+      "step": 5140
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.417634020331769e-05,
+      "loss": 2.2291,
+      "step": 5145
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.398352498317029e-05,
+      "loss": 2.0303,
+      "step": 5150
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.379114381112581e-05,
+      "loss": 2.2062,
+      "step": 5155
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.359919795206065e-05,
+      "loss": 1.8579,
+      "step": 5160
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.3407688667989124e-05,
+      "loss": 2.2312,
+      "step": 5165
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.321661721805519e-05,
+      "loss": 1.9622,
+      "step": 5170
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.302598485852401e-05,
+      "loss": 2.0582,
+      "step": 5175
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.283579284277378e-05,
+      "loss": 2.0206,
+      "step": 5180
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.2646042421287625e-05,
+      "loss": 2.1524,
+      "step": 5185
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.245673484164521e-05,
+      "loss": 1.817,
+      "step": 5190
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.2267871348514475e-05,
+      "loss": 1.825,
+      "step": 5195
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.207945318364376e-05,
+      "loss": 2.0851,
+      "step": 5200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.1891481585853224e-05,
+      "loss": 1.9926,
+      "step": 5205
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.1703957791027104e-05,
+      "loss": 2.0991,
+      "step": 5210
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.151688303210525e-05,
+      "loss": 2.3285,
+      "step": 5215
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.133025853907531e-05,
+      "loss": 2.1269,
+      "step": 5220
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.114408553896437e-05,
+      "loss": 2.1899,
+      "step": 5225
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.09583652558311e-05,
+      "loss": 2.1094,
+      "step": 5230
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.077309891075766e-05,
+      "loss": 2.0178,
+      "step": 5235
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.058828772184155e-05,
+      "loss": 2.0036,
+      "step": 5240
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0403932904187694e-05,
+      "loss": 1.934,
+      "step": 5245
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.0220035669900493e-05,
+      "loss": 1.8375,
+      "step": 5250
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.0036597228075847e-05,
+      "loss": 2.0524,
+      "step": 5255
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.985361878479307e-05,
+      "loss": 2.232,
+      "step": 5260
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9671101543107037e-05,
+      "loss": 1.7783,
+      "step": 5265
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9489046703040478e-05,
+      "loss": 2.0166,
+      "step": 5270
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9307455461575728e-05,
+      "loss": 2.014,
+      "step": 5275
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9126329012647048e-05,
+      "loss": 1.8389,
+      "step": 5280
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.894566854713283e-05,
+      "loss": 2.0252,
+      "step": 5285
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8765475252847696e-05,
+      "loss": 2.0237,
+      "step": 5290
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8585750314534633e-05,
+      "loss": 2.193,
+      "step": 5295
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8406494913857264e-05,
+      "loss": 1.8726,
+      "step": 5300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8227710229392102e-05,
+      "loss": 1.8849,
+      "step": 5305
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8049397436620817e-05,
+      "loss": 2.0296,
+      "step": 5310
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.7871557707922356e-05,
+      "loss": 2.0708,
+      "step": 5315
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.769419221256546e-05,
+      "loss": 2.0158,
+      "step": 5320
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.751730211670075e-05,
+      "loss": 1.9694,
+      "step": 5325
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.7340888583353263e-05,
+      "loss": 2.1837,
+      "step": 5330
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.716495277241463e-05,
+      "loss": 1.9613,
+      "step": 5335
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.6989495840635615e-05,
+      "loss": 2.2264,
+      "step": 5340
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6814518941618326e-05,
+      "loss": 1.9929,
+      "step": 5345
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6640023225808852e-05,
+      "loss": 2.0802,
+      "step": 5350
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6466009840489436e-05,
+      "loss": 2.1228,
+      "step": 5355
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.629247992977122e-05,
+      "loss": 1.8279,
+      "step": 5360
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.6119434634586427e-05,
+      "loss": 1.9754,
+      "step": 5365
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5946875092681134e-05,
+      "loss": 2.1167,
+      "step": 5370
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5774802438607627e-05,
+      "loss": 1.9293,
+      "step": 5375
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5603217803716938e-05,
+      "loss": 2.0927,
+      "step": 5380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.5432122316151463e-05,
+      "loss": 1.7898,
+      "step": 5385
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.5261517100837563e-05,
+      "loss": 1.8328,
+      "step": 5390
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.509140327947814e-05,
+      "loss": 1.9682,
+      "step": 5395
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.4921781970545178e-05,
+      "loss": 2.0778,
+      "step": 5400
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.4752654289272568e-05,
+      "loss": 1.8122,
+      "step": 5405
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.4584021347648645e-05,
+      "loss": 1.8171,
+      "step": 5410
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.441588425440886e-05,
+      "loss": 1.8796,
+      "step": 5415
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.424824411502856e-05,
+      "loss": 1.9946,
+      "step": 5420
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.408110203171572e-05,
+      "loss": 2.0975,
+      "step": 5425
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.3914459103403696e-05,
+      "loss": 1.9878,
+      "step": 5430
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.374831642574392e-05,
+      "loss": 1.9282,
+      "step": 5435
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3582675091098717e-05,
+      "loss": 2.0356,
+      "step": 5440
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3417536188534327e-05,
+      "loss": 2.0298,
+      "step": 5445
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3252900803813415e-05,
+      "loss": 1.8658,
+      "step": 5450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.3088770019388116e-05,
+      "loss": 1.9056,
+      "step": 5455
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.292514491439297e-05,
+      "loss": 2.1178,
+      "step": 5460
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2762026564637717e-05,
+      "loss": 2.115,
+      "step": 5465
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.259941604260024e-05,
+      "loss": 2.0306,
+      "step": 5470
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.2437314417419518e-05,
+      "loss": 2.1874,
+      "step": 5475
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.2275722754888662e-05,
+      "loss": 1.9314,
+      "step": 5480
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.211464211744787e-05,
+      "loss": 2.1194,
+      "step": 5485
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.195407356417737e-05,
+      "loss": 1.9204,
+      "step": 5490
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1794018150790507e-05,
+      "loss": 2.0758,
+      "step": 5495
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.1634476929626868e-05,
+      "loss": 2.0252,
+      "step": 5500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1475450949645325e-05,
+      "loss": 2.0451,
+      "step": 5505
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.1316941256417024e-05,
+      "loss": 2.0781,
+      "step": 5510
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.115894889211869e-05,
+      "loss": 1.8396,
+      "step": 5515
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.100147489552562e-05,
+      "loss": 1.9794,
+      "step": 5520
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.084452030200502e-05,
+      "loss": 1.8314,
+      "step": 5525
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.068808614350899e-05,
+      "loss": 2.1102,
+      "step": 5530
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.0532173448567936e-05,
+      "loss": 2.0262,
+      "step": 5535
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.037678324228366e-05,
+      "loss": 1.8992,
+      "step": 5540
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.022191654632274e-05,
+      "loss": 2.2173,
+      "step": 5545
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.0067574378909726e-05,
+      "loss": 2.1857,
+      "step": 5550
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9913757754820483e-05,
+      "loss": 1.9238,
+      "step": 5555
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.976046768537544e-05,
+      "loss": 2.0675,
+      "step": 5560
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9607705178433124e-05,
+      "loss": 1.8647,
+      "step": 5565
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9455471238383394e-05,
+      "loss": 2.0518,
+      "step": 5570
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9303766866140794e-05,
+      "loss": 2.0077,
+      "step": 5575
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9152593059138036e-05,
+      "loss": 2.1028,
+      "step": 5580
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9001950811319624e-05,
+      "loss": 1.9315,
+      "step": 5585
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.885184111313494e-05,
+      "loss": 2.1597,
+      "step": 5590
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.870226495153199e-05,
+      "loss": 1.9542,
+      "step": 5595
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8553223309950907e-05,
+      "loss": 1.7678,
+      "step": 5600
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8404717168317444e-05,
+      "loss": 2.0222,
+      "step": 5605
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8256747503036465e-05,
+      "loss": 2.2029,
+      "step": 5610
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.8109315286985575e-05,
+      "loss": 1.7966,
+      "step": 5615
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7962421489508797e-05,
+      "loss": 2.2122,
+      "step": 5620
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7816067076410138e-05,
+      "loss": 1.87,
+      "step": 5625
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7670253009947146e-05,
+      "loss": 2.1665,
+      "step": 5630
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7524980248824806e-05,
+      "loss": 2.2845,
+      "step": 5635
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.738024974818896e-05,
+      "loss": 2.0288,
+      "step": 5640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7236062459620306e-05,
+      "loss": 1.9568,
+      "step": 5645
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7092419331127894e-05,
+      "loss": 2.128,
+      "step": 5650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6949321307143096e-05,
+      "loss": 2.0602,
+      "step": 5655
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6806769328513226e-05,
+      "loss": 1.9605,
+      "step": 5660
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.666476433249552e-05,
+      "loss": 1.9966,
+      "step": 5665
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6523307252750787e-05,
+      "loss": 2.0614,
+      "step": 5670
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6382399019337493e-05,
+      "loss": 2.1478,
+      "step": 5675
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6242040558705386e-05,
+      "loss": 2.0078,
+      "step": 5680
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6102232793689652e-05,
+      "loss": 2.0039,
+      "step": 5685
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5962976643504734e-05,
+      "loss": 2.0423,
+      "step": 5690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5824273023738223e-05,
+      "loss": 2.0835,
+      "step": 5695
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.5686122846344932e-05,
+      "loss": 2.1321,
+      "step": 5700
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.55485270196409e-05,
+      "loss": 1.832,
+      "step": 5705
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.541148644829743e-05,
+      "loss": 2.2623,
+      "step": 5710
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5275002033335016e-05,
+      "loss": 2.0061,
+      "step": 5715
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5139074672117514e-05,
+      "loss": 2.0495,
+      "step": 5720
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.500370525834639e-05,
+      "loss": 1.8332,
+      "step": 5725
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4868894682054535e-05,
+      "loss": 2.0413,
+      "step": 5730
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.473464382960057e-05,
+      "loss": 1.9287,
+      "step": 5735
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4600953583663114e-05,
+      "loss": 1.9631,
+      "step": 5740
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4467824823234843e-05,
+      "loss": 1.8987,
+      "step": 5745
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.4335258423616737e-05,
+      "loss": 1.9999,
+      "step": 5750
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.4203255256412318e-05,
+      "loss": 2.071,
+      "step": 5755
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.407181618952199e-05,
+      "loss": 2.0848,
+      "step": 5760
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.394094208713732e-05,
+      "loss": 2.0245,
+      "step": 5765
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.3810633809735196e-05,
+      "loss": 1.9967,
+      "step": 5770
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3680892214072405e-05,
+      "loss": 2.1751,
+      "step": 5775
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3551718153179871e-05,
+      "loss": 1.9032,
+      "step": 5780
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3423112476357036e-05,
+      "loss": 2.0877,
+      "step": 5785
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3295076029166265e-05,
+      "loss": 1.9883,
+      "step": 5790
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.3167609653427426e-05,
+      "loss": 2.0494,
+      "step": 5795
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.304071418721221e-05,
+      "loss": 2.0913,
+      "step": 5800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2914390464838655e-05,
+      "loss": 1.9322,
+      "step": 5805
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.2788639316865635e-05,
+      "loss": 1.6949,
+      "step": 5810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.266346157008753e-05,
+      "loss": 2.1381,
+      "step": 5815
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2538858047528646e-05,
+      "loss": 1.9244,
+      "step": 5820
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2414829568437825e-05,
+      "loss": 1.9664,
+      "step": 5825
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2291376948283139e-05,
+      "loss": 1.8674,
+      "step": 5830
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2168500998746435e-05,
+      "loss": 1.8768,
+      "step": 5835
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.2046202527718076e-05,
+      "loss": 2.1325,
+      "step": 5840
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1924482339291554e-05,
+      "loss": 1.9095,
+      "step": 5845
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1803341233758291e-05,
+      "loss": 2.0019,
+      "step": 5850
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1682780007602268e-05,
+      "loss": 1.9443,
+      "step": 5855
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.1562799453494899e-05,
+      "loss": 2.0703,
+      "step": 5860
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.144340036028978e-05,
+      "loss": 1.9226,
+      "step": 5865
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.132458351301744e-05,
+      "loss": 2.1047,
+      "step": 5870
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.1206349692880236e-05,
+      "loss": 2.039,
+      "step": 5875
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.1088699677247238e-05,
+      "loss": 1.8703,
+      "step": 5880
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.097163423964912e-05,
+      "loss": 2.0559,
+      "step": 5885
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0855154149772994e-05,
+      "loss": 1.9953,
+      "step": 5890
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0739260173457355e-05,
+      "loss": 2.0058,
+      "step": 5895
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0623953072687265e-05,
+      "loss": 2.1952,
+      "step": 5900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0509233605588997e-05,
+      "loss": 2.051,
+      "step": 5905
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.0395102526425282e-05,
+      "loss": 2.0063,
+      "step": 5910
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.0281560585590311e-05,
+      "loss": 1.8081,
+      "step": 5915
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.0168608529604783e-05,
+      "loss": 2.0337,
+      "step": 5920
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.0056247101110972e-05,
+      "loss": 2.0572,
+      "step": 5925
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.944477038867838e-06,
+      "loss": 2.115,
+      "step": 5930
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.833299077746261e-06,
+      "loss": 2.1213,
+      "step": 5935
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.72271394872416e-06,
+      "loss": 1.9652,
+      "step": 5940
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.612722378881578e-06,
+      "loss": 1.9341,
+      "step": 5945
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.503325091396098e-06,
+      "loss": 1.9136,
+      "step": 5950
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.394522805537931e-06,
+      "loss": 1.8324,
+      "step": 5955
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.286316236665271e-06,
+      "loss": 2.1958,
+      "step": 5960
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.178706096219547e-06,
+      "loss": 1.9716,
+      "step": 5965
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.0716930917208e-06,
+      "loss": 2.1187,
+      "step": 5970
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.965277926762916e-06,
+      "loss": 1.8652,
+      "step": 5975
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.859461301009186e-06,
+      "loss": 2.0256,
+      "step": 5980
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.754243910187498e-06,
+      "loss": 1.9776,
+      "step": 5985
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.649626446085945e-06,
+      "loss": 1.8688,
+      "step": 5990
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.545609596548121e-06,
+      "loss": 2.1567,
+      "step": 5995
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 8.442194045468733e-06,
+      "loss": 2.0054,
+      "step": 6000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.339380472789016e-06,
+      "loss": 2.0207,
+      "step": 6005
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.237169554492297e-06,
+      "loss": 1.8489,
+      "step": 6010
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.135561962599514e-06,
+      "loss": 1.9376,
+      "step": 6015
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 8.034558365164868e-06,
+      "loss": 1.8307,
+      "step": 6020
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.934159426271403e-06,
+      "loss": 1.9554,
+      "step": 6025
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.834365806026578e-06,
+      "loss": 2.028,
+      "step": 6030
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.735178160557943e-06,
+      "loss": 1.913,
+      "step": 6035
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.636597142009017e-06,
+      "loss": 2.0417,
+      "step": 6040
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 7.538623398534661e-06,
+      "loss": 2.0136,
+      "step": 6045
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.441257574297089e-06,
+      "loss": 1.9664,
+      "step": 6050
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.344500309461511e-06,
+      "loss": 1.843,
+      "step": 6055
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.248352240192002e-06,
+      "loss": 1.995,
+      "step": 6060
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 7.15281399864719e-06,
+      "loss": 2.1217,
+      "step": 6065
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 7.057886212976239e-06,
+      "loss": 1.9386,
+      "step": 6070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.963569507314627e-06,
+      "loss": 2.0779,
+      "step": 6075
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.8698645017801325e-06,
+      "loss": 1.9562,
+      "step": 6080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.776771812468618e-06,
+      "loss": 1.9248,
+      "step": 6085
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.684292051450147e-06,
+      "loss": 1.9028,
+      "step": 6090
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.592425826764781e-06,
+      "loss": 2.0701,
+      "step": 6095
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.501173742418753e-06,
+      "loss": 1.8953,
+      "step": 6100
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.410536398380385e-06,
+      "loss": 2.0693,
+      "step": 6105
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.320514390576193e-06,
+      "loss": 2.1479,
+      "step": 6110
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.231108310886924e-06,
+      "loss": 1.9884,
+      "step": 6115
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.142318747143716e-06,
+      "loss": 2.0673,
+      "step": 6120
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.054146283124218e-06,
+      "loss": 2.0731,
+      "step": 6125
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.966591498548724e-06,
+      "loss": 2.0371,
+      "step": 6130
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.8796549690763645e-06,
+      "loss": 2.0572,
+      "step": 6135
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.79333726630138e-06,
+      "loss": 1.9367,
+      "step": 6140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.7076389577493175e-06,
+      "loss": 2.1085,
+      "step": 6145
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.622560606873262e-06,
+      "loss": 1.891,
+      "step": 6150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.538102773050235e-06,
+      "loss": 1.9659,
+      "step": 6155
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.454266011577369e-06,
+      "loss": 2.1352,
+      "step": 6160
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.371050873668437e-06,
+      "loss": 2.007,
+      "step": 6165
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.2884579064500615e-06,
+      "loss": 1.8625,
+      "step": 6170
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.206487652958214e-06,
+      "loss": 2.1092,
+      "step": 6175
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.125140652134652e-06,
+      "loss": 2.0602,
+      "step": 6180
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.044417438823279e-06,
+      "loss": 1.8894,
+      "step": 6185
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.964318543766733e-06,
+      "loss": 1.9809,
+      "step": 6190
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.884844493602847e-06,
+      "loss": 1.9969,
+      "step": 6195
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.805995810861219e-06,
+      "loss": 2.0476,
+      "step": 6200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.727773013959702e-06,
+      "loss": 1.8422,
+      "step": 6205
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.650176617201074e-06,
+      "loss": 2.1646,
+      "step": 6210
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.573207130769663e-06,
+      "loss": 1.8574,
+      "step": 6215
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.496865060727917e-06,
+      "loss": 1.9283,
+      "step": 6220
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.421150909013094e-06,
+      "loss": 2.149,
+      "step": 6225
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.346065173434055e-06,
+      "loss": 1.8866,
+      "step": 6230
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.271608347667888e-06,
+      "loss": 2.2319,
+      "step": 6235
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.197780921256678e-06,
+      "loss": 1.9323,
+      "step": 6240
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.1245833796043184e-06,
+      "loss": 1.8525,
+      "step": 6245
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.052016203973319e-06,
+      "loss": 1.8321,
+      "step": 6250
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.9800798714816566e-06,
+      "loss": 2.0823,
+      "step": 6255
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.908774855099529e-06,
+      "loss": 2.2925,
+      "step": 6260
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.838101623646429e-06,
+      "loss": 2.0427,
+      "step": 6265
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.768060641787874e-06,
+      "loss": 1.9859,
+      "step": 6270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.698652370032496e-06,
+      "loss": 1.9979,
+      "step": 6275
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.6298772647289204e-06,
+      "loss": 2.0323,
+      "step": 6280
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.561735778062847e-06,
+      "loss": 1.8849,
+      "step": 6285
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.4942283580539747e-06,
+      "loss": 2.0051,
+      "step": 6290
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.427355448553149e-06,
+      "loss": 1.9764,
+      "step": 6295
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.3611174892393848e-06,
+      "loss": 2.0537,
+      "step": 6300
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.2955149156170373e-06,
+      "loss": 2.0022,
+      "step": 6305
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.230548159012836e-06,
+      "loss": 1.9045,
+      "step": 6310
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.1662176465731776e-06,
+      "loss": 2.0595,
+      "step": 6315
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.1025238012612146e-06,
+      "loss": 2.1485,
+      "step": 6320
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.039467041854105e-06,
+      "loss": 2.084,
+      "step": 6325
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.97704778294029e-06,
+      "loss": 2.011,
+      "step": 6330
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.9152664349167415e-06,
+      "loss": 1.853,
+      "step": 6335
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.854123403986253e-06,
+      "loss": 1.9935,
+      "step": 6340
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.793619092154787e-06,
+      "loss": 1.817,
+      "step": 6345
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.7337538972287967e-06,
+      "loss": 2.1709,
+      "step": 6350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.674528212812721e-06,
+      "loss": 2.1739,
+      "step": 6355
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.6159424283062507e-06,
+      "loss": 2.1036,
+      "step": 6360
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.557996928901829e-06,
+      "loss": 1.8333,
+      "step": 6365
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.5006920955821465e-06,
+      "loss": 1.6949,
+      "step": 6370
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.4440283051176405e-06,
+      "loss": 1.9627,
+      "step": 6375
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.388005930063941e-06,
+      "loss": 2.1238,
+      "step": 6380
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.3326253387594753e-06,
+      "loss": 1.9462,
+      "step": 6385
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.277886895323078e-06,
+      "loss": 1.922,
+      "step": 6390
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.2237909596515396e-06,
+      "loss": 2.2586,
+      "step": 6395
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1703378874172507e-06,
+      "loss": 1.9842,
+      "step": 6400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.117528030065907e-06,
+      "loss": 2.0181,
+      "step": 6405
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0653617348141084e-06,
+      "loss": 1.951,
+      "step": 6410
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.013839344647217e-06,
+      "loss": 2.0486,
+      "step": 6415
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.962961198316937e-06,
+      "loss": 1.9864,
+      "step": 6420
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.912727630339217e-06,
+      "loss": 2.0762,
+      "step": 6425
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.8631389709919843e-06,
+      "loss": 2.1018,
+      "step": 6430
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.8141955463129912e-06,
+      "loss": 1.9501,
+      "step": 6435
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7658976780976944e-06,
+      "loss": 2.0425,
+      "step": 6440
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7182456838971016e-06,
+      "loss": 2.0135,
+      "step": 6445
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6712398770156734e-06,
+      "loss": 2.0289,
+      "step": 6450
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6248805665093348e-06,
+      "loss": 2.0349,
+      "step": 6455
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5791680571833667e-06,
+      "loss": 2.1509,
+      "step": 6460
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5341026495904409e-06,
+      "loss": 1.9185,
+      "step": 6465
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4896846400286323e-06,
+      "loss": 1.8242,
+      "step": 6470
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4459143205394876e-06,
+      "loss": 2.0976,
+      "step": 6475
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4027919789060818e-06,
+      "loss": 2.0517,
+      "step": 6480
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.36031789865112e-06,
+      "loss": 2.0073,
+      "step": 6485
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.3184923590351062e-06,
+      "loss": 2.011,
+      "step": 6490
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.27731563505451e-06,
+      "loss": 1.9835,
+      "step": 6495
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.236787997439892e-06,
+      "loss": 1.9704,
+      "step": 6500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.196909712654204e-06,
+      "loss": 2.0033,
+      "step": 6505
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1576810428910012e-06,
+      "loss": 1.9454,
+      "step": 6510
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.1191022460727007e-06,
+      "loss": 2.0045,
+      "step": 6515
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0811735758489372e-06,
+      "loss": 1.7296,
+      "step": 6520
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.04389528159482e-06,
+      "loss": 2.1404,
+      "step": 6525
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.0072676084093902e-06,
+      "loss": 2.0602,
+      "step": 6530
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.712907971139218e-07,
+      "loss": 1.8307,
+      "step": 6535
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.359650842503565e-07,
+      "loss": 2.0053,
+      "step": 6540
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.012907020798156e-07,
+      "loss": 1.9717,
+      "step": 6545
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.672678785809796e-07,
+      "loss": 2.053,
+      "step": 6550
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.338968374486555e-07,
+      "loss": 1.9612,
+      "step": 6555
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.011777980922564e-07,
+      "loss": 2.073,
+      "step": 6560
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.691109756344128e-07,
+      "loss": 2.309,
+      "step": 6565
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.376965809095193e-07,
+      "loss": 2.1699,
+      "step": 6570
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.06934820462346e-07,
+      "loss": 1.9122,
+      "step": 6575
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.768258965467289e-07,
+      "loss": 1.9562,
+      "step": 6580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.473700071241484e-07,
+      "loss": 2.01,
+      "step": 6585
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.185673458625418e-07,
+      "loss": 1.9432,
+      "step": 6590
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.904181021349375e-07,
+      "loss": 1.9138,
+      "step": 6595
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.629224610182671e-07,
+      "loss": 1.8805,
+      "step": 6600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.360806032920995e-07,
+      "loss": 2.0896,
+      "step": 6605
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.09892705437498e-07,
+      "loss": 1.9198,
+      "step": 6610
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.843589396358427e-07,
+      "loss": 1.9662,
+      "step": 6615
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.5947947376767663e-07,
+      "loss": 2.1967,
+      "step": 6620
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.3525447141165023e-07,
+      "loss": 2.0899,
+      "step": 6625
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.116840918434006e-07,
+      "loss": 2.0715,
+      "step": 6630
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.887684900345301e-07,
+      "loss": 1.9081,
+      "step": 6635
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.665078166515623e-07,
+      "loss": 2.181,
+      "step": 6640
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.449022180549766e-07,
+      "loss": 1.938,
+      "step": 6645
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2395183629824186e-07,
+      "loss": 2.0899,
+      "step": 6650
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0365680912688434e-07,
+      "loss": 2.0134,
+      "step": 6655
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.840172699775656e-07,
+      "loss": 2.0707,
+      "step": 6660
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.650333479771949e-07,
+      "loss": 2.0242,
+      "step": 6665
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.467051679421406e-07,
+      "loss": 1.7765,
+      "step": 6670
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.290328503773309e-07,
+      "loss": 1.9012,
+      "step": 6675
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1201651147554347e-07,
+      "loss": 2.0647,
+      "step": 6680
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.956562631165504e-07,
+      "loss": 2.0192,
+      "step": 6685
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7995221286645215e-07,
+      "loss": 2.0953,
+      "step": 6690
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6490446397696702e-07,
+      "loss": 2.0095,
+      "step": 6695
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5051311538469837e-07,
+      "loss": 2.1963,
+      "step": 6700
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.367782617105351e-07,
+      "loss": 1.91,
+      "step": 6705
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2369999325901881e-07,
+      "loss": 2.153,
+      "step": 6710
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1127839601774437e-07,
+      "loss": 2.1033,
+      "step": 6715
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.951355165678244e-08,
+      "loss": 2.1281,
+      "step": 6720
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.840553752815783e-08,
+      "loss": 2.0639,
+      "step": 6725
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.79544266653609e-08,
+      "loss": 2.006,
+      "step": 6730
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.816028778281469e-08,
+      "loss": 1.8766,
+      "step": 6735
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.902318527547523e-08,
+      "loss": 2.1754,
+      "step": 6740
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.0543179218365265e-08,
+      "loss": 1.8537,
+      "step": 6745
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.272032536621895e-08,
+      "loss": 1.8291,
+      "step": 6750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5554675153082195e-08,
+      "loss": 1.9296,
+      "step": 6755
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9046275692012904e-08,
+      "loss": 2.2341,
+      "step": 6760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3195169774714586e-08,
+      "loss": 1.9139,
+      "step": 6765
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8001395871303228e-08,
+      "loss": 2.1894,
+      "step": 6770
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3464988130051925e-08,
+      "loss": 2.1036,
+      "step": 6775
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.585976377124439e-09,
+      "loss": 2.1144,
+      "step": 6780
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.364386116419762e-09,
+      "loss": 2.0481,
+      "step": 6785
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.800238529416688e-09,
+      "loss": 1.8588,
+      "step": 6790
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8935504749628684e-09,
+      "loss": 2.1501,
+      "step": 6795
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.443344892637093e-10,
+      "loss": 1.9769,
+      "step": 6800
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.259878569363608e-11,
+      "loss": 1.9641,
+      "step": 6805
+    },
+    {
+      "epoch": 3.0,
+      "step": 6807,
+      "total_flos": 1.620091928969216e+16,
+      "train_loss": 2.365634794292671,
+      "train_runtime": 60212.4858,
+      "train_samples_per_second": 0.452,
+      "train_steps_per_second": 0.113
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 6807,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 1.620091928969216e+16,
+  "trial_name": null,
+  "trial_params": null
+}