diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
deleted file mode 100644--- "a/last-checkpoint/trainer_state.json"
+++ /dev/null
@@ -1,14656 +0,0 @@
-{
-  "best_metric": null,
-  "best_model_checkpoint": null,
-  "epoch": 1.9988428888866134,
-  "global_step": 24400,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.0,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 1.4445,
-      "step": 10
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 1.4332,
-      "step": 20
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 3e-06,
-      "loss": 1.4346,
-      "step": 30
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 1.4213,
-      "step": 40
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 5e-06,
-      "loss": 1.4162,
-      "step": 50
-    },
-    {
-      "epoch": 0.0,
-      "learning_rate": 6e-06,
-      "loss": 1.4093,
-      "step": 60
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 7.000000000000001e-06,
-      "loss": 1.4005,
-      "step": 70
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 1.405,
-      "step": 80
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 9e-06,
-      "loss": 1.4112,
-      "step": 90
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1e-05,
-      "loss": 1.4123,
-      "step": 100
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.1000000000000001e-05,
-      "loss": 1.4037,
-      "step": 110
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.2e-05,
-      "loss": 1.402,
-      "step": 120
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.3000000000000001e-05,
-      "loss": 1.3856,
-      "step": 130
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.4000000000000001e-05,
-      "loss": 1.392,
-      "step": 140
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.5e-05,
-      "loss": 1.3968,
-      "step": 150
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 1.393,
-      "step": 160
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.7000000000000003e-05,
-      "loss": 1.3897,
-      "step": 170
-    },
-    {
-      "epoch": 0.01,
-      "learning_rate": 1.8e-05,
-      "loss": 1.388,
-      "step": 180
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 1.9e-05,
-      "loss": 1.3885,
-      "step": 190
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 2e-05,
-      "loss": 1.3883,
-      "step": 200
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 2.1e-05,
-      "loss": 1.39,
-      "step": 210
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 2.2000000000000003e-05,
-      "loss": 1.383,
-      "step": 220
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 2.3000000000000003e-05,
-      "loss": 1.3845,
-      "step": 230
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 2.4e-05,
-      "loss": 1.3911,
-      "step": 240
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 2.5e-05,
-      "loss": 1.3943,
-      "step": 250
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 2.6000000000000002e-05,
-      "loss": 1.3948,
-      "step": 260
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 2.7000000000000002e-05,
-      "loss": 1.3793,
-      "step": 270
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 2.8000000000000003e-05,
-      "loss": 1.3741,
-      "step": 280
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 2.9e-05,
-      "loss": 1.3827,
-      "step": 290
-    },
-    {
-      "epoch": 0.02,
-      "learning_rate": 3e-05,
-      "loss": 1.3754,
-      "step": 300
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 3.1e-05,
-      "loss": 1.3793,
-      "step": 310
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 3.2000000000000005e-05,
-      "loss": 1.3758,
-      "step": 320
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 3.3e-05,
-      "loss": 1.38,
-      "step": 330
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 1.3739,
-      "step": 340
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 3.5e-05,
-      "loss": 1.3756,
-      "step": 350
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 3.6e-05,
-      "loss": 1.3786,
-      "step": 360
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 3.7e-05,
-      "loss": 1.3792,
-      "step": 370
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 3.8e-05,
-      "loss": 1.3749,
-      "step": 380
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 3.9000000000000006e-05,
-      "loss": 1.3622,
-      "step": 390
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 4e-05,
-      "loss": 1.3694,
-      "step": 400
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 4.1e-05,
-      "loss": 1.3616,
-      "step": 410
-    },
-    {
-      "epoch": 0.03,
-      "learning_rate": 4.2e-05,
-      "loss": 1.3652,
-      "step": 420
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 4.3e-05,
-      "loss": 1.3703,
-      "step": 430
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 4.4000000000000006e-05,
-      "loss": 1.3698,
-      "step": 440
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 4.5e-05,
-      "loss": 1.3639,
-      "step": 450
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 4.600000000000001e-05,
-      "loss": 1.3676,
-      "step": 460
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 4.7e-05,
-      "loss": 1.3584,
-      "step": 470
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 4.8e-05,
-      "loss": 1.365,
-      "step": 480
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 4.9e-05,
-      "loss": 1.3636,
-      "step": 490
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 5e-05,
-      "loss": 1.3658,
-      "step": 500
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 5.1000000000000006e-05,
-      "loss": 1.3788,
-      "step": 510
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 5.2000000000000004e-05,
-      "loss": 1.3577,
-      "step": 520
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 5.300000000000001e-05,
-      "loss": 1.3524,
-      "step": 530
-    },
-    {
-      "epoch": 0.04,
-      "learning_rate": 5.4000000000000005e-05,
-      "loss": 1.3547,
-      "step": 540
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 5.500000000000001e-05,
-      "loss": 1.3554,
-      "step": 550
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 5.6000000000000006e-05,
-      "loss": 1.363,
-      "step": 560
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 1.3627,
-      "step": 570
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 5.8e-05,
-      "loss": 1.3427,
-      "step": 580
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 5.9e-05,
-      "loss": 1.3639,
-      "step": 590
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 6e-05,
-      "loss": 1.3492,
-      "step": 600
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 6.1e-05,
-      "loss": 1.3495,
-      "step": 610
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 6.2e-05,
-      "loss": 1.3559,
-      "step": 620
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 6.3e-05,
-      "loss": 1.3601,
-      "step": 630
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 6.400000000000001e-05,
-      "loss": 1.3537,
-      "step": 640
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 6.500000000000001e-05,
-      "loss": 1.3578,
-      "step": 650
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 6.6e-05,
-      "loss": 1.3409,
-      "step": 660
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 6.7e-05,
-      "loss": 1.3482,
-      "step": 670
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 1.3566,
-      "step": 680
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 6.9e-05,
-      "loss": 1.3623,
-      "step": 690
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 7e-05,
-      "loss": 1.3521,
-      "step": 700
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 7.1e-05,
-      "loss": 1.35,
-      "step": 710
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 7.2e-05,
-      "loss": 1.3595,
-      "step": 720
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 7.3e-05,
-      "loss": 1.3451,
-      "step": 730
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 7.4e-05,
-      "loss": 1.3467,
-      "step": 740
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 1.3524,
-      "step": 750
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 7.6e-05,
-      "loss": 1.3504,
-      "step": 760
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 7.7e-05,
-      "loss": 1.3565,
-      "step": 770
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 7.800000000000001e-05,
-      "loss": 1.349,
-      "step": 780
-    },
-    {
-      "epoch": 0.06,
-      "learning_rate": 7.900000000000001e-05,
-      "loss": 1.3511,
-      "step": 790
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 8e-05,
-      "loss": 1.3439,
-      "step": 800
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 8.1e-05,
-      "loss": 1.3416,
-      "step": 810
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 8.2e-05,
-      "loss": 1.3474,
-      "step": 820
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 8.3e-05,
-      "loss": 1.3409,
-      "step": 830
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 8.4e-05,
-      "loss": 1.3471,
-      "step": 840
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 8.5e-05,
-      "loss": 1.3486,
-      "step": 850
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 8.6e-05,
-      "loss": 1.3487,
-      "step": 860
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 8.7e-05,
-      "loss": 1.3467,
-      "step": 870
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 8.800000000000001e-05,
-      "loss": 1.3426,
-      "step": 880
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 8.900000000000001e-05,
-      "loss": 1.3457,
-      "step": 890
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9e-05,
-      "loss": 1.3449,
-      "step": 900
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 9.1e-05,
-      "loss": 1.3366,
-      "step": 910
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.200000000000001e-05,
-      "loss": 1.3445,
-      "step": 920
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.300000000000001e-05,
-      "loss": 1.339,
-      "step": 930
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.4e-05,
-      "loss": 1.343,
-      "step": 940
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.5e-05,
-      "loss": 1.3424,
-      "step": 950
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.6e-05,
-      "loss": 1.3332,
-      "step": 960
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.7e-05,
-      "loss": 1.3484,
-      "step": 970
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.8e-05,
-      "loss": 1.3389,
-      "step": 980
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.900000000000001e-05,
-      "loss": 1.3467,
-      "step": 990
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 0.0001,
-      "loss": 1.333,
-      "step": 1000
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.995729050995132e-05,
-      "loss": 1.3338,
-      "step": 1010
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.991458101990263e-05,
-      "loss": 1.34,
-      "step": 1020
-    },
-    {
-      "epoch": 0.08,
-      "learning_rate": 9.987187152985393e-05,
-      "loss": 1.3392,
-      "step": 1030
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.982916203980525e-05,
-      "loss": 1.3398,
-      "step": 1040
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.978645254975656e-05,
-      "loss": 1.3393,
-      "step": 1050
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.974374305970788e-05,
-      "loss": 1.3393,
-      "step": 1060
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.970103356965919e-05,
-      "loss": 1.3324,
-      "step": 1070
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.96583240796105e-05,
-      "loss": 1.3361,
-      "step": 1080
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.96156145895618e-05,
-      "loss": 1.3329,
-      "step": 1090
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.957290509951312e-05,
-      "loss": 1.3183,
-      "step": 1100
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.953019560946442e-05,
-      "loss": 1.3287,
-      "step": 1110
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.948748611941574e-05,
-      "loss": 1.3339,
-      "step": 1120
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.944477662936705e-05,
-      "loss": 1.3291,
-      "step": 1130
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.940206713931836e-05,
-      "loss": 1.334,
-      "step": 1140
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 9.935935764926967e-05,
-      "loss": 1.3263,
-      "step": 1150
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.931664815922098e-05,
-      "loss": 1.3222,
-      "step": 1160
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.92739386691723e-05,
-      "loss": 1.3258,
-      "step": 1170
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.923122917912361e-05,
-      "loss": 1.3427,
-      "step": 1180
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.918851968907492e-05,
-      "loss": 1.3176,
-      "step": 1190
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.914581019902622e-05,
-      "loss": 1.3269,
-      "step": 1200
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.910310070897754e-05,
-      "loss": 1.3295,
-      "step": 1210
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.906039121892885e-05,
-      "loss": 1.3275,
-      "step": 1220
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.901768172888017e-05,
-      "loss": 1.3238,
-      "step": 1230
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.897497223883148e-05,
-      "loss": 1.3234,
-      "step": 1240
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.89322627487828e-05,
-      "loss": 1.3231,
-      "step": 1250
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.88895532587341e-05,
-      "loss": 1.3189,
-      "step": 1260
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.884684376868541e-05,
-      "loss": 1.3263,
-      "step": 1270
-    },
-    {
-      "epoch": 0.1,
-      "learning_rate": 9.880413427863672e-05,
-      "loss": 1.3207,
-      "step": 1280
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.876142478858802e-05,
-      "loss": 1.3282,
-      "step": 1290
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.871871529853934e-05,
-      "loss": 1.3217,
-      "step": 1300
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.867600580849065e-05,
-      "loss": 1.3139,
-      "step": 1310
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.863329631844195e-05,
-      "loss": 1.3108,
-      "step": 1320
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.859058682839327e-05,
-      "loss": 1.3195,
-      "step": 1330
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.854787733834458e-05,
-      "loss": 1.3139,
-      "step": 1340
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.85051678482959e-05,
-      "loss": 1.3248,
-      "step": 1350
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.846245835824721e-05,
-      "loss": 1.3188,
-      "step": 1360
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.841974886819851e-05,
-      "loss": 1.3184,
-      "step": 1370
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.837703937814983e-05,
-      "loss": 1.3144,
-      "step": 1380
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.833432988810114e-05,
-      "loss": 1.3143,
-      "step": 1390
-    },
-    {
-      "epoch": 0.11,
-      "learning_rate": 9.829162039805245e-05,
-      "loss": 1.3213,
-      "step": 1400
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.824891090800377e-05,
-      "loss": 1.3115,
-      "step": 1410
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.820620141795508e-05,
-      "loss": 1.3164,
-      "step": 1420
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.816349192790638e-05,
-      "loss": 1.315,
-      "step": 1430
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.81207824378577e-05,
-      "loss": 1.3115,
-      "step": 1440
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.807807294780901e-05,
-      "loss": 1.3159,
-      "step": 1450
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.803536345776033e-05,
-      "loss": 1.3116,
-      "step": 1460
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.799265396771163e-05,
-      "loss": 1.3138,
-      "step": 1470
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.794994447766294e-05,
-      "loss": 1.299,
-      "step": 1480
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.790723498761424e-05,
-      "loss": 1.3087,
-      "step": 1490
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.786452549756556e-05,
-      "loss": 1.3135,
-      "step": 1500
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.782181600751687e-05,
-      "loss": 1.3134,
-      "step": 1510
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 9.777910651746819e-05,
-      "loss": 1.3109,
-      "step": 1520
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.77363970274195e-05,
-      "loss": 1.3045,
-      "step": 1530
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.76936875373708e-05,
-      "loss": 1.3011,
-      "step": 1540
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.765097804732212e-05,
-      "loss": 1.2984,
-      "step": 1550
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.760826855727343e-05,
-      "loss": 1.3129,
-      "step": 1560
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.756555906722474e-05,
-      "loss": 1.3017,
-      "step": 1570
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.752284957717606e-05,
-      "loss": 1.3114,
-      "step": 1580
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.748014008712737e-05,
-      "loss": 1.3052,
-      "step": 1590
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.743743059707867e-05,
-      "loss": 1.3109,
-      "step": 1600
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.739472110702999e-05,
-      "loss": 1.3007,
-      "step": 1610
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.73520116169813e-05,
-      "loss": 1.2956,
-      "step": 1620
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.730930212693262e-05,
-      "loss": 1.311,
-      "step": 1630
-    },
-    {
-      "epoch": 0.13,
-      "learning_rate": 9.726659263688393e-05,
-      "loss": 1.3004,
-      "step": 1640
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.722388314683523e-05,
-      "loss": 1.3003,
-      "step": 1650
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.718117365678655e-05,
-      "loss": 1.2967,
-      "step": 1660
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.713846416673785e-05,
-      "loss": 1.2954,
-      "step": 1670
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.709575467668916e-05,
-      "loss": 1.3016,
-      "step": 1680
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.705304518664047e-05,
-      "loss": 1.297,
-      "step": 1690
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.701033569659179e-05,
-      "loss": 1.302,
-      "step": 1700
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.696762620654309e-05,
-      "loss": 1.3037,
-      "step": 1710
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.69249167164944e-05,
-      "loss": 1.3081,
-      "step": 1720
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.688220722644572e-05,
-      "loss": 1.3042,
-      "step": 1730
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.683949773639703e-05,
-      "loss": 1.2906,
-      "step": 1740
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.679678824634835e-05,
-      "loss": 1.2967,
-      "step": 1750
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.675407875629966e-05,
-      "loss": 1.3025,
-      "step": 1760
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 9.671136926625096e-05,
-      "loss": 1.3012,
-      "step": 1770
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.666865977620228e-05,
-      "loss": 1.301,
-      "step": 1780
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.662595028615359e-05,
-      "loss": 1.2949,
-      "step": 1790
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.658751174510978e-05,
-      "loss": 1.3007,
-      "step": 1800
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.654480225506108e-05,
-      "loss": 1.293,
-      "step": 1810
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.650209276501239e-05,
-      "loss": 1.3039,
-      "step": 1820
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.64593832749637e-05,
-      "loss": 1.2906,
-      "step": 1830
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.641667378491501e-05,
-      "loss": 1.3009,
-      "step": 1840
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.637396429486632e-05,
-      "loss": 1.301,
-      "step": 1850
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.633125480481764e-05,
-      "loss": 1.2913,
-      "step": 1860
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.628854531476894e-05,
-      "loss": 1.298,
-      "step": 1870
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.624583582472025e-05,
-      "loss": 1.3007,
-      "step": 1880
-    },
-    {
-      "epoch": 0.15,
-      "learning_rate": 9.620312633467157e-05,
-      "loss": 1.2934,
-      "step": 1890
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.616041684462288e-05,
-      "loss": 1.3003,
-      "step": 1900
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.61177073545742e-05,
-      "loss": 1.2903,
-      "step": 1910
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.607499786452551e-05,
-      "loss": 1.2884,
-      "step": 1920
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.603228837447681e-05,
-      "loss": 1.2859,
-      "step": 1930
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.598957888442812e-05,
-      "loss": 1.2918,
-      "step": 1940
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.594686939437944e-05,
-      "loss": 1.2825,
-      "step": 1950
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.590415990433075e-05,
-      "loss": 1.2877,
-      "step": 1960
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.586145041428207e-05,
-      "loss": 1.2957,
-      "step": 1970
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.581874092423338e-05,
-      "loss": 1.2911,
-      "step": 1980
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.577603143418468e-05,
-      "loss": 1.2811,
-      "step": 1990
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.5733321944136e-05,
-      "loss": 1.2849,
-      "step": 2000
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 9.56906124540873e-05,
-      "loss": 1.2797,
-      "step": 2010
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.564790296403861e-05,
-      "loss": 1.2837,
-      "step": 2020
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.560519347398993e-05,
-      "loss": 1.2848,
-      "step": 2030
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.556248398394123e-05,
-      "loss": 1.2903,
-      "step": 2040
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.551977449389254e-05,
-      "loss": 1.2938,
-      "step": 2050
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.547706500384385e-05,
-      "loss": 1.2925,
-      "step": 2060
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.543435551379517e-05,
-      "loss": 1.2857,
-      "step": 2070
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.539164602374648e-05,
-      "loss": 1.2729,
-      "step": 2080
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.53489365336978e-05,
-      "loss": 1.2799,
-      "step": 2090
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.53062270436491e-05,
-      "loss": 1.2905,
-      "step": 2100
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.526351755360041e-05,
-      "loss": 1.2886,
-      "step": 2110
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.522080806355173e-05,
-      "loss": 1.2826,
-      "step": 2120
-    },
-    {
-      "epoch": 0.17,
-      "learning_rate": 9.517809857350304e-05,
-      "loss": 1.279,
-      "step": 2130
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.513538908345436e-05,
-      "loss": 1.2801,
-      "step": 2140
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.509267959340567e-05,
-      "loss": 1.2869,
-      "step": 2150
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.504997010335697e-05,
-      "loss": 1.2845,
-      "step": 2160
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.500726061330828e-05,
-      "loss": 1.2816,
-      "step": 2170
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.49645511232596e-05,
-      "loss": 1.2763,
-      "step": 2180
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.49218416332109e-05,
-      "loss": 1.2858,
-      "step": 2190
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.487913214316221e-05,
-      "loss": 1.2813,
-      "step": 2200
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.483642265311351e-05,
-      "loss": 1.2847,
-      "step": 2210
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.479371316306483e-05,
-      "loss": 1.2767,
-      "step": 2220
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.475100367301614e-05,
-      "loss": 1.2933,
-      "step": 2230
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.470829418296746e-05,
-      "loss": 1.2721,
-      "step": 2240
-    },
-    {
-      "epoch": 0.18,
-      "learning_rate": 9.466558469291877e-05,
-      "loss": 1.2799,
-      "step": 2250
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.462287520287009e-05,
-      "loss": 1.2777,
-      "step": 2260
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.458016571282139e-05,
-      "loss": 1.27,
-      "step": 2270
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.45374562227727e-05,
-      "loss": 1.2846,
-      "step": 2280
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.449474673272402e-05,
-      "loss": 1.2744,
-      "step": 2290
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.445203724267533e-05,
-      "loss": 1.2722,
-      "step": 2300
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.440932775262664e-05,
-      "loss": 1.2755,
-      "step": 2310
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.436661826257794e-05,
-      "loss": 1.2645,
-      "step": 2320
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.432390877252926e-05,
-      "loss": 1.2826,
-      "step": 2330
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.428119928248057e-05,
-      "loss": 1.2746,
-      "step": 2340
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.423848979243189e-05,
-      "loss": 1.287,
-      "step": 2350
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.41957803023832e-05,
-      "loss": 1.2726,
-      "step": 2360
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.41530708123345e-05,
-      "loss": 1.2669,
-      "step": 2370
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.411036132228582e-05,
-      "loss": 1.2807,
-      "step": 2380
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.406765183223713e-05,
-      "loss": 1.2734,
-      "step": 2390
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.402494234218843e-05,
-      "loss": 1.2712,
-      "step": 2400
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.398223285213975e-05,
-      "loss": 1.2682,
-      "step": 2410
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.393952336209106e-05,
-      "loss": 1.2677,
-      "step": 2420
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.389681387204238e-05,
-      "loss": 1.2715,
-      "step": 2430
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.385410438199368e-05,
-      "loss": 1.2721,
-      "step": 2440
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.381139489194499e-05,
-      "loss": 1.2688,
-      "step": 2450
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.37686854018963e-05,
-      "loss": 1.2696,
-      "step": 2460
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.372597591184762e-05,
-      "loss": 1.2765,
-      "step": 2470
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.368326642179893e-05,
-      "loss": 1.2653,
-      "step": 2480
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.364055693175023e-05,
-      "loss": 1.2735,
-      "step": 2490
-    },
-    {
-      "epoch": 0.2,
-      "learning_rate": 9.359784744170155e-05,
-      "loss": 1.261,
-      "step": 2500
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.355513795165286e-05,
-      "loss": 1.2666,
-      "step": 2510
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.351242846160418e-05,
-      "loss": 1.2684,
-      "step": 2520
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.346971897155549e-05,
-      "loss": 1.2619,
-      "step": 2530
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.34270094815068e-05,
-      "loss": 1.2652,
-      "step": 2540
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.33842999914581e-05,
-      "loss": 1.263,
-      "step": 2550
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.334159050140942e-05,
-      "loss": 1.265,
-      "step": 2560
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.329888101136073e-05,
-      "loss": 1.2696,
-      "step": 2570
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.325617152131204e-05,
-      "loss": 1.2614,
-      "step": 2580
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.321346203126335e-05,
-      "loss": 1.278,
-      "step": 2590
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.317075254121466e-05,
-      "loss": 1.2659,
-      "step": 2600
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.312804305116596e-05,
-      "loss": 1.2669,
-      "step": 2610
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 9.308533356111728e-05,
-      "loss": 1.2664,
-      "step": 2620
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.30426240710686e-05,
-      "loss": 1.2643,
-      "step": 2630
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.299991458101991e-05,
-      "loss": 1.2548,
-      "step": 2640
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.295720509097122e-05,
-      "loss": 1.269,
-      "step": 2650
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.291449560092252e-05,
-      "loss": 1.2644,
-      "step": 2660
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.287178611087384e-05,
-      "loss": 1.2584,
-      "step": 2670
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.282907662082515e-05,
-      "loss": 1.2666,
-      "step": 2680
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.278636713077647e-05,
-      "loss": 1.2587,
-      "step": 2690
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.274365764072778e-05,
-      "loss": 1.2573,
-      "step": 2700
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.27009481506791e-05,
-      "loss": 1.2507,
-      "step": 2710
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.26582386606304e-05,
-      "loss": 1.2635,
-      "step": 2720
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.261552917058171e-05,
-      "loss": 1.2689,
-      "step": 2730
-    },
-    {
-      "epoch": 0.22,
-      "learning_rate": 9.257281968053302e-05,
-      "loss": 1.2584,
-      "step": 2740
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.253011019048434e-05,
-      "loss": 1.2627,
-      "step": 2750
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.248740070043564e-05,
-      "loss": 1.264,
-      "step": 2760
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.244469121038695e-05,
-      "loss": 1.2583,
-      "step": 2770
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.240198172033825e-05,
-      "loss": 1.2664,
-      "step": 2780
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.235927223028957e-05,
-      "loss": 1.2655,
-      "step": 2790
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.231656274024088e-05,
-      "loss": 1.255,
-      "step": 2800
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.22738532501922e-05,
-      "loss": 1.2639,
-      "step": 2810
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.223114376014351e-05,
-      "loss": 1.2551,
-      "step": 2820
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.218843427009481e-05,
-      "loss": 1.2704,
-      "step": 2830
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.214572478004613e-05,
-      "loss": 1.2544,
-      "step": 2840
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.210301528999744e-05,
-      "loss": 1.2583,
-      "step": 2850
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 9.206030579994875e-05,
-      "loss": 1.2521,
-      "step": 2860
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.201759630990007e-05,
-      "loss": 1.2542,
-      "step": 2870
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.197488681985138e-05,
-      "loss": 1.2478,
-      "step": 2880
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.193217732980268e-05,
-      "loss": 1.2605,
-      "step": 2890
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.1889467839754e-05,
-      "loss": 1.2652,
-      "step": 2900
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.184675834970531e-05,
-      "loss": 1.2597,
-      "step": 2910
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.180404885965663e-05,
-      "loss": 1.2572,
-      "step": 2920
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.176133936960794e-05,
-      "loss": 1.2492,
-      "step": 2930
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.171862987955924e-05,
-      "loss": 1.2667,
-      "step": 2940
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.167592038951056e-05,
-      "loss": 1.2564,
-      "step": 2950
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.163321089946186e-05,
-      "loss": 1.2546,
-      "step": 2960
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.159050140941317e-05,
-      "loss": 1.258,
-      "step": 2970
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.154779191936449e-05,
-      "loss": 1.2562,
-      "step": 2980
-    },
-    {
-      "epoch": 0.24,
-      "learning_rate": 9.15050824293158e-05,
-      "loss": 1.2615,
-      "step": 2990
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.14623729392671e-05,
-      "loss": 1.2537,
-      "step": 3000
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.141966344921841e-05,
-      "loss": 1.2573,
-      "step": 3010
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.137695395916973e-05,
-      "loss": 1.2586,
-      "step": 3020
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.133424446912104e-05,
-      "loss": 1.2503,
-      "step": 3030
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.129153497907236e-05,
-      "loss": 1.2524,
-      "step": 3040
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.124882548902367e-05,
-      "loss": 1.2524,
-      "step": 3050
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.120611599897497e-05,
-      "loss": 1.2582,
-      "step": 3060
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.116340650892629e-05,
-      "loss": 1.2553,
-      "step": 3070
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.11206970188776e-05,
-      "loss": 1.2655,
-      "step": 3080
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.107798752882892e-05,
-      "loss": 1.2537,
-      "step": 3090
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.103527803878023e-05,
-      "loss": 1.2584,
-      "step": 3100
-    },
-    {
-      "epoch": 0.25,
-      "learning_rate": 9.099256854873153e-05,
-      "loss": 1.2518,
-      "step": 3110
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.094985905868285e-05,
-      "loss": 1.2484,
-      "step": 3120
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.090714956863416e-05,
-      "loss": 1.2479,
-      "step": 3130
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.086444007858546e-05,
-      "loss": 1.2596,
-      "step": 3140
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.082173058853677e-05,
-      "loss": 1.2611,
-      "step": 3150
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.077902109848809e-05,
-      "loss": 1.2551,
-      "step": 3160
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.073631160843939e-05,
-      "loss": 1.2474,
-      "step": 3170
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.06936021183907e-05,
-      "loss": 1.2485,
-      "step": 3180
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.065089262834202e-05,
-      "loss": 1.2395,
-      "step": 3190
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.060818313829333e-05,
-      "loss": 1.2519,
-      "step": 3200
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.056547364824465e-05,
-      "loss": 1.2581,
-      "step": 3210
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.052276415819596e-05,
-      "loss": 1.2537,
-      "step": 3220
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 9.048005466814726e-05,
-      "loss": 1.253,
-      "step": 3230
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.043734517809858e-05,
-      "loss": 1.2543,
-      "step": 3240
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.039463568804989e-05,
-      "loss": 1.2481,
-      "step": 3250
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.03519261980012e-05,
-      "loss": 1.2545,
-      "step": 3260
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.030921670795252e-05,
-      "loss": 1.2505,
-      "step": 3270
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.026650721790382e-05,
-      "loss": 1.2513,
-      "step": 3280
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.022379772785513e-05,
-      "loss": 1.2367,
-      "step": 3290
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.018108823780645e-05,
-      "loss": 1.2508,
-      "step": 3300
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.013837874775776e-05,
-      "loss": 1.2573,
-      "step": 3310
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.009566925770906e-05,
-      "loss": 1.2462,
-      "step": 3320
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.005295976766038e-05,
-      "loss": 1.2463,
-      "step": 3330
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 9.001025027761169e-05,
-      "loss": 1.2407,
-      "step": 3340
-    },
-    {
-      "epoch": 0.27,
-      "learning_rate": 8.996754078756299e-05,
-      "loss": 1.2492,
-      "step": 3350
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.992483129751431e-05,
-      "loss": 1.2548,
-      "step": 3360
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.988212180746562e-05,
-      "loss": 1.238,
-      "step": 3370
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.983941231741694e-05,
-      "loss": 1.2538,
-      "step": 3380
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.979670282736825e-05,
-      "loss": 1.2501,
-      "step": 3390
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.975399333731955e-05,
-      "loss": 1.2465,
-      "step": 3400
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.971128384727086e-05,
-      "loss": 1.2487,
-      "step": 3410
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.966857435722218e-05,
-      "loss": 1.2407,
-      "step": 3420
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.96258648671735e-05,
-      "loss": 1.238,
-      "step": 3430
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.958315537712481e-05,
-      "loss": 1.2363,
-      "step": 3440
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.954044588707611e-05,
-      "loss": 1.249,
-      "step": 3450
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.949773639702742e-05,
-      "loss": 1.2392,
-      "step": 3460
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 8.945502690697874e-05,
-      "loss": 1.2435,
-      "step": 3470
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.941231741693005e-05,
-      "loss": 1.2383,
-      "step": 3480
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.936960792688137e-05,
-      "loss": 1.2288,
-      "step": 3490
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.932689843683268e-05,
-      "loss": 1.241,
-      "step": 3500
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.928418894678398e-05,
-      "loss": 1.2397,
-      "step": 3510
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.92414794567353e-05,
-      "loss": 1.2441,
-      "step": 3520
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.91987699666866e-05,
-      "loss": 1.2502,
-      "step": 3530
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.915606047663791e-05,
-      "loss": 1.2454,
-      "step": 3540
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.911335098658922e-05,
-      "loss": 1.2559,
-      "step": 3550
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.907064149654053e-05,
-      "loss": 1.2366,
-      "step": 3560
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.902793200649184e-05,
-      "loss": 1.2456,
-      "step": 3570
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.898522251644315e-05,
-      "loss": 1.2403,
-      "step": 3580
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.894251302639447e-05,
-      "loss": 1.2401,
-      "step": 3590
-    },
-    {
-      "epoch": 0.29,
-      "learning_rate": 8.889980353634578e-05,
-      "loss": 1.2376,
-      "step": 3600
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.88570940462971e-05,
-      "loss": 1.2359,
-      "step": 3610
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.88143845562484e-05,
-      "loss": 1.2458,
-      "step": 3620
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.877167506619971e-05,
-      "loss": 1.2394,
-      "step": 3630
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.872896557615103e-05,
-      "loss": 1.2493,
-      "step": 3640
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.868625608610234e-05,
-      "loss": 1.2275,
-      "step": 3650
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.864354659605365e-05,
-      "loss": 1.2352,
-      "step": 3660
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.860083710600497e-05,
-      "loss": 1.2311,
-      "step": 3670
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.855812761595627e-05,
-      "loss": 1.2365,
-      "step": 3680
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.851541812590758e-05,
-      "loss": 1.2467,
-      "step": 3690
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.84727086358589e-05,
-      "loss": 1.2436,
-      "step": 3700
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.84299991458102e-05,
-      "loss": 1.2319,
-      "step": 3710
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 8.838728965576151e-05,
-      "loss": 1.2344,
-      "step": 3720
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.834458016571281e-05,
-      "loss": 1.243,
-      "step": 3730
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.830187067566413e-05,
-      "loss": 1.2497,
-      "step": 3740
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.825916118561544e-05,
-      "loss": 1.228,
-      "step": 3750
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.821645169556676e-05,
-      "loss": 1.2383,
-      "step": 3760
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.817374220551807e-05,
-      "loss": 1.2445,
-      "step": 3770
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.813103271546939e-05,
-      "loss": 1.2399,
-      "step": 3780
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.808832322542069e-05,
-      "loss": 1.2209,
-      "step": 3790
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.8045613735372e-05,
-      "loss": 1.2378,
-      "step": 3800
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.800290424532331e-05,
-      "loss": 1.244,
-      "step": 3810
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.796019475527463e-05,
-      "loss": 1.2338,
-      "step": 3820
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.791748526522594e-05,
-      "loss": 1.2349,
-      "step": 3830
-    },
-    {
-      "epoch": 0.31,
-      "learning_rate": 8.787477577517726e-05,
-      "loss": 1.24,
-      "step": 3840
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.783206628512856e-05,
-      "loss": 1.2374,
-      "step": 3850
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.778935679507987e-05,
-      "loss": 1.2246,
-      "step": 3860
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.774664730503119e-05,
-      "loss": 1.2336,
-      "step": 3870
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.77039378149825e-05,
-      "loss": 1.2411,
-      "step": 3880
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.76612283249338e-05,
-      "loss": 1.2344,
-      "step": 3890
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.761851883488512e-05,
-      "loss": 1.2338,
-      "step": 3900
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.757580934483642e-05,
-      "loss": 1.2428,
-      "step": 3910
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.753309985478773e-05,
-      "loss": 1.2339,
-      "step": 3920
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.749039036473905e-05,
-      "loss": 1.2342,
-      "step": 3930
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.744768087469036e-05,
-      "loss": 1.2282,
-      "step": 3940
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.740497138464167e-05,
-      "loss": 1.2413,
-      "step": 3950
-    },
-    {
-      "epoch": 0.32,
-      "learning_rate": 8.736226189459298e-05,
-      "loss": 1.235,
-      "step": 3960
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.731955240454429e-05,
-      "loss": 1.2306,
-      "step": 3970
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.72768429144956e-05,
-      "loss": 1.2436,
-      "step": 3980
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.723413342444692e-05,
-      "loss": 1.2285,
-      "step": 3990
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.719142393439823e-05,
-      "loss": 1.2224,
-      "step": 4000
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.714871444434955e-05,
-      "loss": 1.2229,
-      "step": 4010
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.710600495430085e-05,
-      "loss": 1.2256,
-      "step": 4020
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.706329546425216e-05,
-      "loss": 1.2383,
-      "step": 4030
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.702058597420348e-05,
-      "loss": 1.2291,
-      "step": 4040
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.697787648415479e-05,
-      "loss": 1.2327,
-      "step": 4050
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.69351669941061e-05,
-      "loss": 1.2321,
-      "step": 4060
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.68924575040574e-05,
-      "loss": 1.2307,
-      "step": 4070
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 8.684974801400872e-05,
-      "loss": 1.2276,
-      "step": 4080
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.680703852396002e-05,
-      "loss": 1.2267,
-      "step": 4090
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.676432903391133e-05,
-      "loss": 1.2331,
-      "step": 4100
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.672161954386265e-05,
-      "loss": 1.2298,
-      "step": 4110
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.667891005381396e-05,
-      "loss": 1.2274,
-      "step": 4120
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.663620056376526e-05,
-      "loss": 1.2183,
-      "step": 4130
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.659349107371658e-05,
-      "loss": 1.2264,
-      "step": 4140
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.655078158366789e-05,
-      "loss": 1.2245,
-      "step": 4150
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.650807209361921e-05,
-      "loss": 1.2341,
-      "step": 4160
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.646536260357052e-05,
-      "loss": 1.2279,
-      "step": 4170
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.642265311352182e-05,
-      "loss": 1.237,
-      "step": 4180
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.637994362347314e-05,
-      "loss": 1.2259,
-      "step": 4190
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.633723413342445e-05,
-      "loss": 1.227,
-      "step": 4200
-    },
-    {
-      "epoch": 0.34,
-      "learning_rate": 8.629452464337576e-05,
-      "loss": 1.2173,
-      "step": 4210
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.625181515332708e-05,
-      "loss": 1.2208,
-      "step": 4220
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.62091056632784e-05,
-      "loss": 1.2251,
-      "step": 4230
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.61663961732297e-05,
-      "loss": 1.2288,
-      "step": 4240
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.612368668318101e-05,
-      "loss": 1.2303,
-      "step": 4250
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.608097719313232e-05,
-      "loss": 1.239,
-      "step": 4260
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.603826770308362e-05,
-      "loss": 1.2266,
-      "step": 4270
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.599555821303494e-05,
-      "loss": 1.226,
-      "step": 4280
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.595284872298625e-05,
-      "loss": 1.2273,
-      "step": 4290
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.591013923293755e-05,
-      "loss": 1.2342,
-      "step": 4300
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.586742974288887e-05,
-      "loss": 1.2282,
-      "step": 4310
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.582472025284018e-05,
-      "loss": 1.2191,
-      "step": 4320
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 8.57820107627915e-05,
-      "loss": 1.2267,
-      "step": 4330
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.573930127274281e-05,
-      "loss": 1.2289,
-      "step": 4340
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.569659178269411e-05,
-      "loss": 1.2235,
-      "step": 4350
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.565388229264543e-05,
-      "loss": 1.2312,
-      "step": 4360
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.561117280259674e-05,
-      "loss": 1.2297,
-      "step": 4370
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.556846331254805e-05,
-      "loss": 1.2191,
-      "step": 4380
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.552575382249937e-05,
-      "loss": 1.2209,
-      "step": 4390
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.548304433245068e-05,
-      "loss": 1.2286,
-      "step": 4400
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.544033484240198e-05,
-      "loss": 1.2293,
-      "step": 4410
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.53976253523533e-05,
-      "loss": 1.2262,
-      "step": 4420
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.535491586230461e-05,
-      "loss": 1.2211,
-      "step": 4430
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.531220637225593e-05,
-      "loss": 1.2215,
-      "step": 4440
-    },
-    {
-      "epoch": 0.36,
-      "learning_rate": 8.526949688220724e-05,
-      "loss": 1.2173,
-      "step": 4450
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.522678739215854e-05,
-      "loss": 1.2212,
-      "step": 4460
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.518407790210986e-05,
-      "loss": 1.2199,
-      "step": 4470
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.514136841206116e-05,
-      "loss": 1.2212,
-      "step": 4480
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.509865892201247e-05,
-      "loss": 1.2233,
-      "step": 4490
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.505594943196378e-05,
-      "loss": 1.2265,
-      "step": 4500
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.50132399419151e-05,
-      "loss": 1.2246,
-      "step": 4510
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.49705304518664e-05,
-      "loss": 1.2228,
-      "step": 4520
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.492782096181771e-05,
-      "loss": 1.2248,
-      "step": 4530
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.488511147176903e-05,
-      "loss": 1.2272,
-      "step": 4540
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.484240198172034e-05,
-      "loss": 1.2233,
-      "step": 4550
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.479969249167166e-05,
-      "loss": 1.2291,
-      "step": 4560
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 8.475698300162297e-05,
-      "loss": 1.2143,
-      "step": 4570
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.471427351157427e-05,
-      "loss": 1.2248,
-      "step": 4580
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.467156402152559e-05,
-      "loss": 1.2254,
-      "step": 4590
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.46288545314769e-05,
-      "loss": 1.2211,
-      "step": 4600
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.458614504142821e-05,
-      "loss": 1.2174,
-      "step": 4610
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.454343555137953e-05,
-      "loss": 1.2187,
-      "step": 4620
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.450072606133084e-05,
-      "loss": 1.2146,
-      "step": 4630
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.445801657128214e-05,
-      "loss": 1.2177,
-      "step": 4640
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.441530708123346e-05,
-      "loss": 1.2241,
-      "step": 4650
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.437259759118476e-05,
-      "loss": 1.2176,
-      "step": 4660
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.432988810113607e-05,
-      "loss": 1.2165,
-      "step": 4670
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.428717861108739e-05,
-      "loss": 1.2148,
-      "step": 4680
-    },
-    {
-      "epoch": 0.38,
-      "learning_rate": 8.424446912103869e-05,
-      "loss": 1.2162,
-      "step": 4690
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.420175963099e-05,
-      "loss": 1.2119,
-      "step": 4700
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.415905014094132e-05,
-      "loss": 1.2246,
-      "step": 4710
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.411634065089263e-05,
-      "loss": 1.2224,
-      "step": 4720
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.407363116084395e-05,
-      "loss": 1.2139,
-      "step": 4730
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.403092167079526e-05,
-      "loss": 1.2163,
-      "step": 4740
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.398821218074656e-05,
-      "loss": 1.2145,
-      "step": 4750
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.394550269069788e-05,
-      "loss": 1.219,
-      "step": 4760
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.390279320064919e-05,
-      "loss": 1.2164,
-      "step": 4770
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.38600837106005e-05,
-      "loss": 1.2289,
-      "step": 4780
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.381737422055182e-05,
-      "loss": 1.2198,
-      "step": 4790
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.377466473050312e-05,
-      "loss": 1.2191,
-      "step": 4800
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.373195524045443e-05,
-      "loss": 1.2193,
-      "step": 4810
-    },
-    {
-      "epoch": 0.39,
-      "learning_rate": 8.368924575040575e-05,
-      "loss": 1.223,
-      "step": 4820
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.364653626035706e-05,
-      "loss": 1.2248,
-      "step": 4830
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.360382677030836e-05,
-      "loss": 1.218,
-      "step": 4840
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.356111728025968e-05,
-      "loss": 1.2185,
-      "step": 4850
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.351840779021098e-05,
-      "loss": 1.2132,
-      "step": 4860
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.347569830016229e-05,
-      "loss": 1.2168,
-      "step": 4870
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.34329888101136e-05,
-      "loss": 1.2148,
-      "step": 4880
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.339027932006492e-05,
-      "loss": 1.2219,
-      "step": 4890
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.334756983001623e-05,
-      "loss": 1.2075,
-      "step": 4900
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.330486033996755e-05,
-      "loss": 1.2018,
-      "step": 4910
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.326215084991885e-05,
-      "loss": 1.2174,
-      "step": 4920
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.321944135987016e-05,
-      "loss": 1.2092,
-      "step": 4930
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 8.317673186982148e-05,
-      "loss": 1.2055,
-      "step": 4940
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.313402237977279e-05,
-      "loss": 1.2177,
-      "step": 4950
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.309131288972411e-05,
-      "loss": 1.2174,
-      "step": 4960
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.304860339967541e-05,
-      "loss": 1.2003,
-      "step": 4970
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.300589390962672e-05,
-      "loss": 1.2103,
-      "step": 4980
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.296318441957804e-05,
-      "loss": 1.2204,
-      "step": 4990
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.292047492952935e-05,
-      "loss": 1.2144,
-      "step": 5000
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.287776543948067e-05,
-      "loss": 1.221,
-      "step": 5010
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.283505594943197e-05,
-      "loss": 1.2127,
-      "step": 5020
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.279234645938328e-05,
-      "loss": 1.2113,
-      "step": 5030
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.274963696933458e-05,
-      "loss": 1.2185,
-      "step": 5040
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.27069274792859e-05,
-      "loss": 1.2109,
-      "step": 5050
-    },
-    {
-      "epoch": 0.41,
-      "learning_rate": 8.266848893824208e-05,
-      "loss": 1.2133,
-      "step": 5060
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.26257794481934e-05,
-      "loss": 1.216,
-      "step": 5070
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.25830699581447e-05,
-      "loss": 1.208,
-      "step": 5080
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.254036046809601e-05,
-      "loss": 1.2089,
-      "step": 5090
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.249765097804733e-05,
-      "loss": 1.2079,
-      "step": 5100
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.245494148799864e-05,
-      "loss": 1.2045,
-      "step": 5110
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.241223199794995e-05,
-      "loss": 1.2084,
-      "step": 5120
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.236952250790127e-05,
-      "loss": 1.2289,
-      "step": 5130
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.232681301785257e-05,
-      "loss": 1.2087,
-      "step": 5140
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.228410352780388e-05,
-      "loss": 1.2107,
-      "step": 5150
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.224566498676006e-05,
-      "loss": 1.213,
-      "step": 5160
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.220295549671137e-05,
-      "loss": 1.2091,
-      "step": 5170
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 8.216024600666269e-05,
-      "loss": 1.2112,
-      "step": 5180
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.211753651661399e-05,
-      "loss": 1.2141,
-      "step": 5190
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.20748270265653e-05,
-      "loss": 1.2061,
-      "step": 5200
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.203211753651661e-05,
-      "loss": 1.2066,
-      "step": 5210
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.198940804646793e-05,
-      "loss": 1.2031,
-      "step": 5220
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.194669855641924e-05,
-      "loss": 1.2009,
-      "step": 5230
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.190398906637056e-05,
-      "loss": 1.2178,
-      "step": 5240
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.186127957632186e-05,
-      "loss": 1.2093,
-      "step": 5250
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.181857008627317e-05,
-      "loss": 1.2083,
-      "step": 5260
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.177586059622449e-05,
-      "loss": 1.2142,
-      "step": 5270
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.17331511061758e-05,
-      "loss": 1.2128,
-      "step": 5280
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.169044161612712e-05,
-      "loss": 1.21,
-      "step": 5290
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.164773212607842e-05,
-      "loss": 1.2087,
-      "step": 5300
-    },
-    {
-      "epoch": 0.43,
-      "learning_rate": 8.160502263602973e-05,
-      "loss": 1.2167,
-      "step": 5310
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.156231314598105e-05,
-      "loss": 1.2131,
-      "step": 5320
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.151960365593236e-05,
-      "loss": 1.2165,
-      "step": 5330
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.147689416588366e-05,
-      "loss": 1.209,
-      "step": 5340
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.143418467583497e-05,
-      "loss": 1.2144,
-      "step": 5350
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.139147518578628e-05,
-      "loss": 1.1937,
-      "step": 5360
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.134876569573759e-05,
-      "loss": 1.2172,
-      "step": 5370
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.13060562056889e-05,
-      "loss": 1.2102,
-      "step": 5380
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.126334671564022e-05,
-      "loss": 1.2022,
-      "step": 5390
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.122063722559153e-05,
-      "loss": 1.2095,
-      "step": 5400
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.117792773554285e-05,
-      "loss": 1.2023,
-      "step": 5410
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.113521824549415e-05,
-      "loss": 1.2059,
-      "step": 5420
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 8.109250875544546e-05,
-      "loss": 1.2009,
-      "step": 5430
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.104979926539678e-05,
-      "loss": 1.208,
-      "step": 5440
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.100708977534809e-05,
-      "loss": 1.2034,
-      "step": 5450
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.09643802852994e-05,
-      "loss": 1.2007,
-      "step": 5460
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.09216707952507e-05,
-      "loss": 1.2089,
-      "step": 5470
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.087896130520202e-05,
-      "loss": 1.1964,
-      "step": 5480
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.083625181515333e-05,
-      "loss": 1.2031,
-      "step": 5490
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.079354232510465e-05,
-      "loss": 1.2173,
-      "step": 5500
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.075083283505596e-05,
-      "loss": 1.196,
-      "step": 5510
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.070812334500726e-05,
-      "loss": 1.2038,
-      "step": 5520
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.066541385495858e-05,
-      "loss": 1.2055,
-      "step": 5530
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.062270436490988e-05,
-      "loss": 1.2061,
-      "step": 5540
-    },
-    {
-      "epoch": 0.45,
-      "learning_rate": 8.057999487486119e-05,
-      "loss": 1.2074,
-      "step": 5550
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.053728538481251e-05,
-      "loss": 1.2097,
-      "step": 5560
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.049457589476382e-05,
-      "loss": 1.2064,
-      "step": 5570
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.045186640471512e-05,
-      "loss": 1.2137,
-      "step": 5580
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.040915691466644e-05,
-      "loss": 1.2102,
-      "step": 5590
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.036644742461775e-05,
-      "loss": 1.2072,
-      "step": 5600
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.032373793456907e-05,
-      "loss": 1.2008,
-      "step": 5610
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.028102844452038e-05,
-      "loss": 1.2103,
-      "step": 5620
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.02383189544717e-05,
-      "loss": 1.202,
-      "step": 5630
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.0195609464423e-05,
-      "loss": 1.203,
-      "step": 5640
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.015289997437431e-05,
-      "loss": 1.2071,
-      "step": 5650
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.011019048432562e-05,
-      "loss": 1.2073,
-      "step": 5660
-    },
-    {
-      "epoch": 0.46,
-      "learning_rate": 8.006748099427694e-05,
-      "loss": 1.2152,
-      "step": 5670
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 8.002477150422825e-05,
-      "loss": 1.2071,
-      "step": 5680
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.998206201417957e-05,
-      "loss": 1.2018,
-      "step": 5690
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.993935252413087e-05,
-      "loss": 1.211,
-      "step": 5700
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.989664303408218e-05,
-      "loss": 1.2033,
-      "step": 5710
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.985393354403348e-05,
-      "loss": 1.2085,
-      "step": 5720
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.98112240539848e-05,
-      "loss": 1.201,
-      "step": 5730
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.976851456393611e-05,
-      "loss": 1.2092,
-      "step": 5740
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.972580507388741e-05,
-      "loss": 1.198,
-      "step": 5750
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.968309558383873e-05,
-      "loss": 1.2077,
-      "step": 5760
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.964038609379004e-05,
-      "loss": 1.2217,
-      "step": 5770
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.959767660374135e-05,
-      "loss": 1.1959,
-      "step": 5780
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 7.955496711369267e-05,
-      "loss": 1.1974,
-      "step": 5790
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.951225762364398e-05,
-      "loss": 1.202,
-      "step": 5800
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.946954813359528e-05,
-      "loss": 1.2059,
-      "step": 5810
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.94268386435466e-05,
-      "loss": 1.197,
-      "step": 5820
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.938412915349791e-05,
-      "loss": 1.1993,
-      "step": 5830
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.934141966344923e-05,
-      "loss": 1.1999,
-      "step": 5840
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.929871017340054e-05,
-      "loss": 1.1941,
-      "step": 5850
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.925600068335185e-05,
-      "loss": 1.2017,
-      "step": 5860
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.921329119330316e-05,
-      "loss": 1.2044,
-      "step": 5870
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.917058170325447e-05,
-      "loss": 1.2011,
-      "step": 5880
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.912787221320578e-05,
-      "loss": 1.2129,
-      "step": 5890
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.908516272315708e-05,
-      "loss": 1.2043,
-      "step": 5900
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.90424532331084e-05,
-      "loss": 1.2002,
-      "step": 5910
-    },
-    {
-      "epoch": 0.48,
-      "learning_rate": 7.89997437430597e-05,
-      "loss": 1.2053,
-      "step": 5920
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.895703425301101e-05,
-      "loss": 1.197,
-      "step": 5930
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.891432476296233e-05,
-      "loss": 1.1957,
-      "step": 5940
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.887161527291364e-05,
-      "loss": 1.2001,
-      "step": 5950
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.882890578286496e-05,
-      "loss": 1.1963,
-      "step": 5960
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.878619629281627e-05,
-      "loss": 1.1977,
-      "step": 5970
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.874348680276757e-05,
-      "loss": 1.193,
-      "step": 5980
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.870077731271889e-05,
-      "loss": 1.199,
-      "step": 5990
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.86580678226702e-05,
-      "loss": 1.2077,
-      "step": 6000
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.861535833262152e-05,
-      "loss": 1.192,
-      "step": 6010
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.857264884257283e-05,
-      "loss": 1.2055,
-      "step": 6020
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.852993935252414e-05,
-      "loss": 1.2001,
-      "step": 6030
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 7.848722986247544e-05,
-      "loss": 1.2101,
-      "step": 6040
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.844452037242676e-05,
-      "loss": 1.202,
-      "step": 6050
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.840181088237807e-05,
-      "loss": 1.2055,
-      "step": 6060
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.835910139232939e-05,
-      "loss": 1.2029,
-      "step": 6070
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.831639190228069e-05,
-      "loss": 1.2039,
-      "step": 6080
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.8273682412232e-05,
-      "loss": 1.2103,
-      "step": 6090
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.82309729221833e-05,
-      "loss": 1.1948,
-      "step": 6100
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.818826343213462e-05,
-      "loss": 1.2038,
-      "step": 6110
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.814555394208593e-05,
-      "loss": 1.1938,
-      "step": 6120
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.810284445203725e-05,
-      "loss": 1.1869,
-      "step": 6130
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.806013496198856e-05,
-      "loss": 1.199,
-      "step": 6140
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.801742547193986e-05,
-      "loss": 1.2025,
-      "step": 6150
-    },
-    {
-      "epoch": 0.5,
-      "learning_rate": 7.797471598189118e-05,
-      "loss": 1.1925,
-      "step": 6160
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.793200649184249e-05,
-      "loss": 1.198,
-      "step": 6170
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.78892970017938e-05,
-      "loss": 1.1954,
-      "step": 6180
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.784658751174512e-05,
-      "loss": 1.1948,
-      "step": 6190
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.780387802169642e-05,
-      "loss": 1.1922,
-      "step": 6200
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.776116853164773e-05,
-      "loss": 1.1916,
-      "step": 6210
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.771845904159905e-05,
-      "loss": 1.2,
-      "step": 6220
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.767574955155036e-05,
-      "loss": 1.1999,
-      "step": 6230
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.763304006150168e-05,
-      "loss": 1.1978,
-      "step": 6240
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.759033057145299e-05,
-      "loss": 1.1997,
-      "step": 6250
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.754762108140429e-05,
-      "loss": 1.196,
-      "step": 6260
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.75049115913556e-05,
-      "loss": 1.2066,
-      "step": 6270
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 7.746220210130692e-05,
-      "loss": 1.1902,
-      "step": 6280
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.741949261125822e-05,
-      "loss": 1.1993,
-      "step": 6290
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.737678312120953e-05,
-      "loss": 1.2002,
-      "step": 6300
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.733407363116085e-05,
-      "loss": 1.1987,
-      "step": 6310
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.729136414111215e-05,
-      "loss": 1.1984,
-      "step": 6320
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.724865465106346e-05,
-      "loss": 1.1796,
-      "step": 6330
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.720594516101478e-05,
-      "loss": 1.2016,
-      "step": 6340
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.716323567096609e-05,
-      "loss": 1.1868,
-      "step": 6350
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.712052618091741e-05,
-      "loss": 1.2034,
-      "step": 6360
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.707781669086871e-05,
-      "loss": 1.189,
-      "step": 6370
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.703510720082002e-05,
-      "loss": 1.2033,
-      "step": 6380
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.699239771077134e-05,
-      "loss": 1.2059,
-      "step": 6390
-    },
-    {
-      "epoch": 0.52,
-      "learning_rate": 7.694968822072265e-05,
-      "loss": 1.2017,
-      "step": 6400
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.690697873067397e-05,
-      "loss": 1.1965,
-      "step": 6410
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.686426924062528e-05,
-      "loss": 1.1888,
-      "step": 6420
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.682155975057658e-05,
-      "loss": 1.1999,
-      "step": 6430
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.67788502605279e-05,
-      "loss": 1.204,
-      "step": 6440
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.673614077047921e-05,
-      "loss": 1.1921,
-      "step": 6450
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.669343128043052e-05,
-      "loss": 1.1859,
-      "step": 6460
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.665072179038182e-05,
-      "loss": 1.1915,
-      "step": 6470
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.660801230033314e-05,
-      "loss": 1.1956,
-      "step": 6480
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.656530281028444e-05,
-      "loss": 1.187,
-      "step": 6490
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.652259332023575e-05,
-      "loss": 1.1954,
-      "step": 6500
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.647988383018707e-05,
-      "loss": 1.1944,
-      "step": 6510
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.643717434013838e-05,
-      "loss": 1.1921,
-      "step": 6520
-    },
-    {
-      "epoch": 0.53,
-      "learning_rate": 7.63944648500897e-05,
-      "loss": 1.1879,
-      "step": 6530
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.6351755360041e-05,
-      "loss": 1.1877,
-      "step": 6540
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.630904586999231e-05,
-      "loss": 1.1841,
-      "step": 6550
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.626633637994363e-05,
-      "loss": 1.1973,
-      "step": 6560
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.622362688989494e-05,
-      "loss": 1.1958,
-      "step": 6570
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.618091739984625e-05,
-      "loss": 1.1964,
-      "step": 6580
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.613820790979757e-05,
-      "loss": 1.1851,
-      "step": 6590
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.609549841974887e-05,
-      "loss": 1.1839,
-      "step": 6600
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.605278892970018e-05,
-      "loss": 1.2004,
-      "step": 6610
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.60100794396515e-05,
-      "loss": 1.1923,
-      "step": 6620
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.596736994960281e-05,
-      "loss": 1.1958,
-      "step": 6630
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.592466045955413e-05,
-      "loss": 1.2023,
-      "step": 6640
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 7.588195096950543e-05,
-      "loss": 1.1875,
-      "step": 6650
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.583924147945674e-05,
-      "loss": 1.1929,
-      "step": 6660
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.579653198940804e-05,
-      "loss": 1.1926,
-      "step": 6670
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.575382249935936e-05,
-      "loss": 1.1955,
-      "step": 6680
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.571111300931067e-05,
-      "loss": 1.1755,
-      "step": 6690
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.566840351926198e-05,
-      "loss": 1.1858,
-      "step": 6700
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.562569402921329e-05,
-      "loss": 1.1991,
-      "step": 6710
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.55829845391646e-05,
-      "loss": 1.195,
-      "step": 6720
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.554027504911591e-05,
-      "loss": 1.1958,
-      "step": 6730
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.549756555906723e-05,
-      "loss": 1.1851,
-      "step": 6740
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.545485606901854e-05,
-      "loss": 1.1831,
-      "step": 6750
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.541214657896986e-05,
-      "loss": 1.1844,
-      "step": 6760
-    },
-    {
-      "epoch": 0.55,
-      "learning_rate": 7.536943708892116e-05,
-      "loss": 1.1892,
-      "step": 6770
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.532672759887247e-05,
-      "loss": 1.184,
-      "step": 6780
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.528401810882379e-05,
-      "loss": 1.1804,
-      "step": 6790
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.52413086187751e-05,
-      "loss": 1.188,
-      "step": 6800
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.519859912872642e-05,
-      "loss": 1.1927,
-      "step": 6810
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.515588963867772e-05,
-      "loss": 1.1916,
-      "step": 6820
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.511318014862903e-05,
-      "loss": 1.1849,
-      "step": 6830
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.507047065858034e-05,
-      "loss": 1.1966,
-      "step": 6840
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.502776116853165e-05,
-      "loss": 1.202,
-      "step": 6850
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.498505167848296e-05,
-      "loss": 1.1859,
-      "step": 6860
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.494234218843427e-05,
-      "loss": 1.1868,
-      "step": 6870
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.489963269838557e-05,
-      "loss": 1.1801,
-      "step": 6880
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 7.485692320833689e-05,
-      "loss": 1.1808,
-      "step": 6890
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.48142137182882e-05,
-      "loss": 1.1828,
-      "step": 6900
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.477150422823952e-05,
-      "loss": 1.1811,
-      "step": 6910
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.472879473819083e-05,
-      "loss": 1.1922,
-      "step": 6920
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.468608524814215e-05,
-      "loss": 1.189,
-      "step": 6930
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.464337575809345e-05,
-      "loss": 1.1906,
-      "step": 6940
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.460066626804476e-05,
-      "loss": 1.1954,
-      "step": 6950
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.455795677799608e-05,
-      "loss": 1.1918,
-      "step": 6960
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.451524728794739e-05,
-      "loss": 1.1947,
-      "step": 6970
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.44725377978987e-05,
-      "loss": 1.1924,
-      "step": 6980
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.442982830785e-05,
-      "loss": 1.1905,
-      "step": 6990
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.438711881780132e-05,
-      "loss": 1.1861,
-      "step": 7000
-    },
-    {
-      "epoch": 0.57,
-      "learning_rate": 7.434440932775263e-05,
-      "loss": 1.1913,
-      "step": 7010
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.430169983770395e-05,
-      "loss": 1.1949,
-      "step": 7020
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.425899034765525e-05,
-      "loss": 1.19,
-      "step": 7030
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.421628085760656e-05,
-      "loss": 1.1858,
-      "step": 7040
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.417357136755788e-05,
-      "loss": 1.1817,
-      "step": 7050
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.413086187750918e-05,
-      "loss": 1.1895,
-      "step": 7060
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.408815238746049e-05,
-      "loss": 1.1869,
-      "step": 7070
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.40454428974118e-05,
-      "loss": 1.1828,
-      "step": 7080
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.400273340736312e-05,
-      "loss": 1.1829,
-      "step": 7090
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.396002391731443e-05,
-      "loss": 1.1945,
-      "step": 7100
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.391731442726574e-05,
-      "loss": 1.175,
-      "step": 7110
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.387460493721705e-05,
-      "loss": 1.1806,
-      "step": 7120
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.383189544716836e-05,
-      "loss": 1.1862,
-      "step": 7130
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 7.378918595711968e-05,
-      "loss": 1.1808,
-      "step": 7140
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.374647646707099e-05,
-      "loss": 1.184,
-      "step": 7150
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.37037669770223e-05,
-      "loss": 1.187,
-      "step": 7160
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.366105748697361e-05,
-      "loss": 1.1816,
-      "step": 7170
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.361834799692492e-05,
-      "loss": 1.1897,
-      "step": 7180
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.357563850687624e-05,
-      "loss": 1.1866,
-      "step": 7190
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.353292901682755e-05,
-      "loss": 1.1762,
-      "step": 7200
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.349021952677887e-05,
-      "loss": 1.19,
-      "step": 7210
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.344751003673017e-05,
-      "loss": 1.1787,
-      "step": 7220
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.340480054668148e-05,
-      "loss": 1.1863,
-      "step": 7230
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.336209105663278e-05,
-      "loss": 1.1777,
-      "step": 7240
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.33193815665841e-05,
-      "loss": 1.1838,
-      "step": 7250
-    },
-    {
-      "epoch": 0.59,
-      "learning_rate": 7.327667207653541e-05,
-      "loss": 1.1835,
-      "step": 7260
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.323396258648672e-05,
-      "loss": 1.1849,
-      "step": 7270
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.319125309643802e-05,
-      "loss": 1.1826,
-      "step": 7280
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.314854360638934e-05,
-      "loss": 1.1791,
-      "step": 7290
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.310583411634065e-05,
-      "loss": 1.1831,
-      "step": 7300
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.306312462629197e-05,
-      "loss": 1.1797,
-      "step": 7310
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.302041513624328e-05,
-      "loss": 1.1875,
-      "step": 7320
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.297770564619458e-05,
-      "loss": 1.1842,
-      "step": 7330
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.29349961561459e-05,
-      "loss": 1.1756,
-      "step": 7340
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.289228666609721e-05,
-      "loss": 1.1908,
-      "step": 7350
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.284957717604853e-05,
-      "loss": 1.193,
-      "step": 7360
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.280686768599984e-05,
-      "loss": 1.1811,
-      "step": 7370
-    },
-    {
-      "epoch": 0.6,
-      "learning_rate": 7.276415819595115e-05,
-      "loss": 1.186,
-      "step": 7380
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.272144870590245e-05,
-      "loss": 1.1715,
-      "step": 7390
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.267873921585377e-05,
-      "loss": 1.1878,
-      "step": 7400
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.263602972580508e-05,
-      "loss": 1.1885,
-      "step": 7410
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.259332023575638e-05,
-      "loss": 1.188,
-      "step": 7420
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.25506107457077e-05,
-      "loss": 1.1706,
-      "step": 7430
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.2507901255659e-05,
-      "loss": 1.1858,
-      "step": 7440
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.246519176561031e-05,
-      "loss": 1.1776,
-      "step": 7450
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.242248227556163e-05,
-      "loss": 1.1843,
-      "step": 7460
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.237977278551294e-05,
-      "loss": 1.1893,
-      "step": 7470
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.233706329546426e-05,
-      "loss": 1.1833,
-      "step": 7480
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.229435380541557e-05,
-      "loss": 1.1808,
-      "step": 7490
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 7.225164431536687e-05,
-      "loss": 1.1814,
-      "step": 7500
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.220893482531819e-05,
-      "loss": 1.1797,
-      "step": 7510
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.21662253352695e-05,
-      "loss": 1.1764,
-      "step": 7520
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.212351584522081e-05,
-      "loss": 1.1747,
-      "step": 7530
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.208080635517213e-05,
-      "loss": 1.1863,
-      "step": 7540
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.203809686512344e-05,
-      "loss": 1.1851,
-      "step": 7550
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.199538737507474e-05,
-      "loss": 1.1828,
-      "step": 7560
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.195267788502606e-05,
-      "loss": 1.1829,
-      "step": 7570
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.190996839497737e-05,
-      "loss": 1.168,
-      "step": 7580
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.186725890492869e-05,
-      "loss": 1.1865,
-      "step": 7590
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.182454941487999e-05,
-      "loss": 1.1849,
-      "step": 7600
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.17818399248313e-05,
-      "loss": 1.1903,
-      "step": 7610
-    },
-    {
-      "epoch": 0.62,
-      "learning_rate": 7.17391304347826e-05,
-      "loss": 1.1834,
-      "step": 7620
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.169642094473392e-05,
-      "loss": 1.1969,
-      "step": 7630
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.165371145468523e-05,
-      "loss": 1.1864,
-      "step": 7640
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.161100196463655e-05,
-      "loss": 1.1769,
-      "step": 7650
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.156829247458786e-05,
-      "loss": 1.181,
-      "step": 7660
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.152558298453916e-05,
-      "loss": 1.1719,
-      "step": 7670
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.148287349449047e-05,
-      "loss": 1.1743,
-      "step": 7680
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.144016400444179e-05,
-      "loss": 1.1877,
-      "step": 7690
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.13974545143931e-05,
-      "loss": 1.1742,
-      "step": 7700
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.135474502434442e-05,
-      "loss": 1.1822,
-      "step": 7710
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.131203553429573e-05,
-      "loss": 1.1834,
-      "step": 7720
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.126932604424703e-05,
-      "loss": 1.1818,
-      "step": 7730
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.122661655419835e-05,
-      "loss": 1.1812,
-      "step": 7740
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 7.118390706414966e-05,
-      "loss": 1.1704,
-      "step": 7750
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.114119757410098e-05,
-      "loss": 1.1797,
-      "step": 7760
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.109848808405229e-05,
-      "loss": 1.1857,
-      "step": 7770
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.105577859400359e-05,
-      "loss": 1.1847,
-      "step": 7780
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.10130691039549e-05,
-      "loss": 1.1759,
-      "step": 7790
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.09703596139062e-05,
-      "loss": 1.1687,
-      "step": 7800
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.092765012385752e-05,
-      "loss": 1.1855,
-      "step": 7810
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.088494063380883e-05,
-      "loss": 1.1792,
-      "step": 7820
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.084223114376015e-05,
-      "loss": 1.1677,
-      "step": 7830
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.079952165371145e-05,
-      "loss": 1.1726,
-      "step": 7840
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.075681216366276e-05,
-      "loss": 1.1758,
-      "step": 7850
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.071410267361408e-05,
-      "loss": 1.1792,
-      "step": 7860
-    },
-    {
-      "epoch": 0.64,
-      "learning_rate": 7.067139318356539e-05,
-      "loss": 1.1873,
-      "step": 7870
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.06286836935167e-05,
-      "loss": 1.1796,
-      "step": 7880
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.058597420346802e-05,
-      "loss": 1.1751,
-      "step": 7890
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.054326471341932e-05,
-      "loss": 1.1742,
-      "step": 7900
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.050055522337064e-05,
-      "loss": 1.1834,
-      "step": 7910
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.045784573332195e-05,
-      "loss": 1.1735,
-      "step": 7920
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.041513624327326e-05,
-      "loss": 1.1731,
-      "step": 7930
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.037242675322458e-05,
-      "loss": 1.1763,
-      "step": 7940
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.032971726317588e-05,
-      "loss": 1.1804,
-      "step": 7950
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.02870077731272e-05,
-      "loss": 1.1755,
-      "step": 7960
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.024429828307851e-05,
-      "loss": 1.1868,
-      "step": 7970
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.020158879302981e-05,
-      "loss": 1.1744,
-      "step": 7980
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 7.015887930298112e-05,
-      "loss": 1.1846,
-      "step": 7990
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.011616981293244e-05,
-      "loss": 1.1745,
-      "step": 8000
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.007346032288374e-05,
-      "loss": 1.1771,
-      "step": 8010
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 7.003075083283505e-05,
-      "loss": 1.1796,
-      "step": 8020
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 6.998804134278637e-05,
-      "loss": 1.1721,
-      "step": 8030
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 6.994533185273768e-05,
-      "loss": 1.1774,
-      "step": 8040
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 6.9902622362689e-05,
-      "loss": 1.1738,
-      "step": 8050
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 6.98599128726403e-05,
-      "loss": 1.1793,
-      "step": 8060
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 6.981720338259161e-05,
-      "loss": 1.175,
-      "step": 8070
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 6.977449389254292e-05,
-      "loss": 1.1893,
-      "step": 8080
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 6.973178440249424e-05,
-      "loss": 1.181,
-      "step": 8090
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 6.968907491244555e-05,
-      "loss": 1.1774,
-      "step": 8100
-    },
-    {
-      "epoch": 0.66,
-      "learning_rate": 6.964636542239687e-05,
-      "loss": 1.1709,
-      "step": 8110
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.960365593234817e-05,
-      "loss": 1.1756,
-      "step": 8120
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.956094644229948e-05,
-      "loss": 1.1785,
-      "step": 8130
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.95182369522508e-05,
-      "loss": 1.1676,
-      "step": 8140
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.947552746220211e-05,
-      "loss": 1.1758,
-      "step": 8150
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.943281797215343e-05,
-      "loss": 1.1812,
-      "step": 8160
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.939010848210473e-05,
-      "loss": 1.1715,
-      "step": 8170
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.934739899205604e-05,
-      "loss": 1.1765,
-      "step": 8180
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.930468950200734e-05,
-      "loss": 1.1778,
-      "step": 8190
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.926198001195866e-05,
-      "loss": 1.1793,
-      "step": 8200
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.921927052190997e-05,
-      "loss": 1.1706,
-      "step": 8210
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.917656103186128e-05,
-      "loss": 1.1724,
-      "step": 8220
-    },
-    {
-      "epoch": 0.67,
-      "learning_rate": 6.913385154181258e-05,
-      "loss": 1.1742,
-      "step": 8230
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.90911420517639e-05,
-      "loss": 1.1767,
-      "step": 8240
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.904843256171521e-05,
-      "loss": 1.1688,
-      "step": 8250
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.900572307166653e-05,
-      "loss": 1.1764,
-      "step": 8260
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.896301358161784e-05,
-      "loss": 1.1668,
-      "step": 8270
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.892030409156916e-05,
-      "loss": 1.1767,
-      "step": 8280
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.887759460152046e-05,
-      "loss": 1.1737,
-      "step": 8290
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.883488511147177e-05,
-      "loss": 1.1755,
-      "step": 8300
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.879217562142309e-05,
-      "loss": 1.1741,
-      "step": 8310
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.87494661313744e-05,
-      "loss": 1.1722,
-      "step": 8320
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.870675664132571e-05,
-      "loss": 1.172,
-      "step": 8330
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.866404715127703e-05,
-      "loss": 1.1781,
-      "step": 8340
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.862133766122833e-05,
-      "loss": 1.1812,
-      "step": 8350
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 6.857862817117964e-05,
-      "loss": 1.1745,
-      "step": 8360
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.853591868113094e-05,
-      "loss": 1.1632,
-      "step": 8370
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.849320919108226e-05,
-      "loss": 1.1758,
-      "step": 8380
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.845049970103357e-05,
-      "loss": 1.1752,
-      "step": 8390
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.840779021098487e-05,
-      "loss": 1.1688,
-      "step": 8400
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.836508072093619e-05,
-      "loss": 1.1622,
-      "step": 8410
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.83223712308875e-05,
-      "loss": 1.164,
-      "step": 8420
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.827966174083882e-05,
-      "loss": 1.1801,
-      "step": 8430
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.823695225079013e-05,
-      "loss": 1.1814,
-      "step": 8440
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.819424276074145e-05,
-      "loss": 1.1755,
-      "step": 8450
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.815153327069275e-05,
-      "loss": 1.1645,
-      "step": 8460
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.810882378064406e-05,
-      "loss": 1.1714,
-      "step": 8470
-    },
-    {
-      "epoch": 0.69,
-      "learning_rate": 6.806611429059537e-05,
-      "loss": 1.1768,
-      "step": 8480
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.802340480054669e-05,
-      "loss": 1.1751,
-      "step": 8490
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.7980695310498e-05,
-      "loss": 1.1713,
-      "step": 8500
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.793798582044932e-05,
-      "loss": 1.1725,
-      "step": 8510
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.789527633040062e-05,
-      "loss": 1.1695,
-      "step": 8520
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.785256684035193e-05,
-      "loss": 1.1708,
-      "step": 8530
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.780985735030325e-05,
-      "loss": 1.1774,
-      "step": 8540
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.776714786025455e-05,
-      "loss": 1.17,
-      "step": 8550
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.772443837020586e-05,
-      "loss": 1.1822,
-      "step": 8560
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.768172888015716e-05,
-      "loss": 1.1681,
-      "step": 8570
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.763901939010848e-05,
-      "loss": 1.1691,
-      "step": 8580
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.759630990005979e-05,
-      "loss": 1.1729,
-      "step": 8590
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 6.75536004100111e-05,
-      "loss": 1.177,
-      "step": 8600
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.751089091996242e-05,
-      "loss": 1.1613,
-      "step": 8610
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.746818142991373e-05,
-      "loss": 1.173,
-      "step": 8620
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.742547193986504e-05,
-      "loss": 1.1769,
-      "step": 8630
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.738276244981635e-05,
-      "loss": 1.1616,
-      "step": 8640
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.734005295976766e-05,
-      "loss": 1.1718,
-      "step": 8650
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.729734346971898e-05,
-      "loss": 1.1806,
-      "step": 8660
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.725463397967029e-05,
-      "loss": 1.1751,
-      "step": 8670
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.721192448962159e-05,
-      "loss": 1.1695,
-      "step": 8680
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.716921499957291e-05,
-      "loss": 1.1713,
-      "step": 8690
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.712650550952422e-05,
-      "loss": 1.1745,
-      "step": 8700
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.708379601947554e-05,
-      "loss": 1.1753,
-      "step": 8710
-    },
-    {
-      "epoch": 0.71,
-      "learning_rate": 6.704108652942685e-05,
-      "loss": 1.1618,
-      "step": 8720
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.699837703937815e-05,
-      "loss": 1.1677,
-      "step": 8730
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.695566754932947e-05,
-      "loss": 1.1655,
-      "step": 8740
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.691295805928077e-05,
-      "loss": 1.1753,
-      "step": 8750
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.687024856923208e-05,
-      "loss": 1.1748,
-      "step": 8760
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.68275390791834e-05,
-      "loss": 1.1719,
-      "step": 8770
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.678482958913471e-05,
-      "loss": 1.1734,
-      "step": 8780
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.674212009908602e-05,
-      "loss": 1.161,
-      "step": 8790
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.669941060903732e-05,
-      "loss": 1.1633,
-      "step": 8800
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.665670111898864e-05,
-      "loss": 1.1744,
-      "step": 8810
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.661399162893995e-05,
-      "loss": 1.1628,
-      "step": 8820
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.657128213889127e-05,
-      "loss": 1.177,
-      "step": 8830
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.652857264884258e-05,
-      "loss": 1.1606,
-      "step": 8840
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 6.648586315879388e-05,
-      "loss": 1.1664,
-      "step": 8850
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.64431536687452e-05,
-      "loss": 1.1588,
-      "step": 8860
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.640044417869651e-05,
-      "loss": 1.155,
-      "step": 8870
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.635773468864782e-05,
-      "loss": 1.1737,
-      "step": 8880
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.631502519859914e-05,
-      "loss": 1.178,
-      "step": 8890
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.627231570855045e-05,
-      "loss": 1.1668,
-      "step": 8900
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.622960621850175e-05,
-      "loss": 1.164,
-      "step": 8910
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.618689672845307e-05,
-      "loss": 1.1669,
-      "step": 8920
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.614418723840437e-05,
-      "loss": 1.1732,
-      "step": 8930
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.610147774835568e-05,
-      "loss": 1.1694,
-      "step": 8940
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.6058768258307e-05,
-      "loss": 1.1607,
-      "step": 8950
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.601605876825831e-05,
-      "loss": 1.1629,
-      "step": 8960
-    },
-    {
-      "epoch": 0.73,
-      "learning_rate": 6.597334927820961e-05,
-      "loss": 1.1653,
-      "step": 8970
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.593063978816093e-05,
-      "loss": 1.1643,
-      "step": 8980
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.588793029811224e-05,
-      "loss": 1.168,
-      "step": 8990
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.584522080806356e-05,
-      "loss": 1.1705,
-      "step": 9000
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.580251131801487e-05,
-      "loss": 1.1682,
-      "step": 9010
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.575980182796617e-05,
-      "loss": 1.1736,
-      "step": 9020
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.571709233791749e-05,
-      "loss": 1.1762,
-      "step": 9030
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.56743828478688e-05,
-      "loss": 1.174,
-      "step": 9040
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.563167335782011e-05,
-      "loss": 1.1619,
-      "step": 9050
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.558896386777143e-05,
-      "loss": 1.1662,
-      "step": 9060
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.554625437772274e-05,
-      "loss": 1.1651,
-      "step": 9070
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.550354488767404e-05,
-      "loss": 1.1621,
-      "step": 9080
-    },
-    {
-      "epoch": 0.74,
-      "learning_rate": 6.546083539762536e-05,
-      "loss": 1.163,
-      "step": 9090
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.541812590757667e-05,
-      "loss": 1.181,
-      "step": 9100
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.537541641752799e-05,
-      "loss": 1.1684,
-      "step": 9110
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.533270692747929e-05,
-      "loss": 1.1715,
-      "step": 9120
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.52899974374306e-05,
-      "loss": 1.1626,
-      "step": 9130
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.52472879473819e-05,
-      "loss": 1.164,
-      "step": 9140
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.520457845733322e-05,
-      "loss": 1.1632,
-      "step": 9150
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.516186896728453e-05,
-      "loss": 1.1666,
-      "step": 9160
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.511915947723584e-05,
-      "loss": 1.1658,
-      "step": 9170
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.507644998718716e-05,
-      "loss": 1.1707,
-      "step": 9180
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.503374049713846e-05,
-      "loss": 1.1703,
-      "step": 9190
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.499103100708977e-05,
-      "loss": 1.151,
-      "step": 9200
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 6.494832151704109e-05,
-      "loss": 1.1683,
-      "step": 9210
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.49056120269924e-05,
-      "loss": 1.1607,
-      "step": 9220
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.486290253694372e-05,
-      "loss": 1.1676,
-      "step": 9230
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.482019304689503e-05,
-      "loss": 1.1624,
-      "step": 9240
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.477748355684633e-05,
-      "loss": 1.1648,
-      "step": 9250
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.473477406679765e-05,
-      "loss": 1.1679,
-      "step": 9260
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.469206457674896e-05,
-      "loss": 1.1697,
-      "step": 9270
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.464935508670027e-05,
-      "loss": 1.166,
-      "step": 9280
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.460664559665159e-05,
-      "loss": 1.1687,
-      "step": 9290
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.456393610660289e-05,
-      "loss": 1.1715,
-      "step": 9300
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.45212266165542e-05,
-      "loss": 1.1687,
-      "step": 9310
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.44785171265055e-05,
-      "loss": 1.1605,
-      "step": 9320
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 6.443580763645682e-05,
-      "loss": 1.1683,
-      "step": 9330
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.439309814640813e-05,
-      "loss": 1.1457,
-      "step": 9340
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.435038865635945e-05,
-      "loss": 1.1618,
-      "step": 9350
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.430767916631075e-05,
-      "loss": 1.1609,
-      "step": 9360
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.426496967626206e-05,
-      "loss": 1.1666,
-      "step": 9370
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.422226018621338e-05,
-      "loss": 1.1662,
-      "step": 9380
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.417955069616469e-05,
-      "loss": 1.17,
-      "step": 9390
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.4136841206116e-05,
-      "loss": 1.1666,
-      "step": 9400
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.409413171606732e-05,
-      "loss": 1.1695,
-      "step": 9410
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.405142222601862e-05,
-      "loss": 1.1614,
-      "step": 9420
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.400871273596994e-05,
-      "loss": 1.164,
-      "step": 9430
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.396600324592125e-05,
-      "loss": 1.1633,
-      "step": 9440
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.392329375587256e-05,
-      "loss": 1.1572,
-      "step": 9450
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 6.388058426582388e-05,
-      "loss": 1.1613,
-      "step": 9460
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.383787477577518e-05,
-      "loss": 1.1691,
-      "step": 9470
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.379516528572649e-05,
-      "loss": 1.1658,
-      "step": 9480
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.375245579567781e-05,
-      "loss": 1.1664,
-      "step": 9490
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.370974630562911e-05,
-      "loss": 1.1584,
-      "step": 9500
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.366703681558042e-05,
-      "loss": 1.1652,
-      "step": 9510
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.362432732553174e-05,
-      "loss": 1.1541,
-      "step": 9520
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.358161783548304e-05,
-      "loss": 1.1545,
-      "step": 9530
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.353890834543435e-05,
-      "loss": 1.1614,
-      "step": 9540
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.349619885538567e-05,
-      "loss": 1.1616,
-      "step": 9550
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.345348936533698e-05,
-      "loss": 1.1755,
-      "step": 9560
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.34107798752883e-05,
-      "loss": 1.1566,
-      "step": 9570
-    },
-    {
-      "epoch": 0.78,
-      "learning_rate": 6.336807038523961e-05,
-      "loss": 1.1672,
-      "step": 9580
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.332536089519091e-05,
-      "loss": 1.1615,
-      "step": 9590
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.328265140514222e-05,
-      "loss": 1.1649,
-      "step": 9600
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.323994191509354e-05,
-      "loss": 1.1586,
-      "step": 9610
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.319723242504485e-05,
-      "loss": 1.1652,
-      "step": 9620
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.315452293499617e-05,
-      "loss": 1.1659,
-      "step": 9630
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.311181344494747e-05,
-      "loss": 1.1585,
-      "step": 9640
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.306910395489878e-05,
-      "loss": 1.159,
-      "step": 9650
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.30263944648501e-05,
-      "loss": 1.1634,
-      "step": 9660
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.298368497480141e-05,
-      "loss": 1.1545,
-      "step": 9670
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.294097548475271e-05,
-      "loss": 1.1578,
-      "step": 9680
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.289826599470403e-05,
-      "loss": 1.1619,
-      "step": 9690
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 6.285555650465533e-05,
-      "loss": 1.1582,
-      "step": 9700
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.281284701460664e-05,
-      "loss": 1.1545,
-      "step": 9710
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.277013752455795e-05,
-      "loss": 1.1599,
-      "step": 9720
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.272742803450927e-05,
-      "loss": 1.1584,
-      "step": 9730
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.268471854446058e-05,
-      "loss": 1.1582,
-      "step": 9740
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.26420090544119e-05,
-      "loss": 1.1676,
-      "step": 9750
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.25992995643632e-05,
-      "loss": 1.1628,
-      "step": 9760
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.255659007431451e-05,
-      "loss": 1.162,
-      "step": 9770
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.251388058426583e-05,
-      "loss": 1.1634,
-      "step": 9780
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.247117109421714e-05,
-      "loss": 1.1684,
-      "step": 9790
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.242846160416846e-05,
-      "loss": 1.1546,
-      "step": 9800
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.238575211411976e-05,
-      "loss": 1.1594,
-      "step": 9810
-    },
-    {
-      "epoch": 0.8,
-      "learning_rate": 6.234304262407107e-05,
-      "loss": 1.1606,
-      "step": 9820
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.230033313402239e-05,
-      "loss": 1.1554,
-      "step": 9830
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.22576236439737e-05,
-      "loss": 1.1597,
-      "step": 9840
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.221491415392501e-05,
-      "loss": 1.158,
-      "step": 9850
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.217220466387631e-05,
-      "loss": 1.1517,
-      "step": 9860
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.212949517382763e-05,
-      "loss": 1.1605,
-      "step": 9870
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.208678568377894e-05,
-      "loss": 1.1543,
-      "step": 9880
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.204407619373024e-05,
-      "loss": 1.1566,
-      "step": 9890
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.200136670368156e-05,
-      "loss": 1.1562,
-      "step": 9900
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.195865721363287e-05,
-      "loss": 1.16,
-      "step": 9910
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.191594772358417e-05,
-      "loss": 1.1643,
-      "step": 9920
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.187323823353549e-05,
-      "loss": 1.1578,
-      "step": 9930
-    },
-    {
-      "epoch": 0.81,
-      "learning_rate": 6.18305287434868e-05,
-      "loss": 1.1633,
-      "step": 9940
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.178781925343812e-05,
-      "loss": 1.1619,
-      "step": 9950
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.174510976338943e-05,
-      "loss": 1.1585,
-      "step": 9960
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.170240027334074e-05,
-      "loss": 1.1556,
-      "step": 9970
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.165969078329205e-05,
-      "loss": 1.1608,
-      "step": 9980
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.161698129324336e-05,
-      "loss": 1.1602,
-      "step": 9990
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.157427180319467e-05,
-      "loss": 1.1575,
-      "step": 10000
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.153156231314599e-05,
-      "loss": 1.1537,
-      "step": 10010
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.14888528230973e-05,
-      "loss": 1.154,
-      "step": 10020
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.144614333304862e-05,
-      "loss": 1.1614,
-      "step": 10030
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.140343384299992e-05,
-      "loss": 1.1546,
-      "step": 10040
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.136072435295123e-05,
-      "loss": 1.1595,
-      "step": 10050
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.131801486290255e-05,
-      "loss": 1.1672,
-      "step": 10060
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 6.127530537285385e-05,
-      "loss": 1.1534,
-      "step": 10070
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.123259588280516e-05,
-      "loss": 1.1541,
-      "step": 10080
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.118988639275646e-05,
-      "loss": 1.1628,
-      "step": 10090
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.114717690270778e-05,
-      "loss": 1.161,
-      "step": 10100
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.110446741265909e-05,
-      "loss": 1.1647,
-      "step": 10110
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.10617579226104e-05,
-      "loss": 1.1506,
-      "step": 10120
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.101904843256172e-05,
-      "loss": 1.1509,
-      "step": 10130
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.0976338942513034e-05,
-      "loss": 1.1549,
-      "step": 10140
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.0933629452464334e-05,
-      "loss": 1.1578,
-      "step": 10150
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.089091996241565e-05,
-      "loss": 1.1632,
-      "step": 10160
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.084821047236696e-05,
-      "loss": 1.1605,
-      "step": 10170
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.080550098231828e-05,
-      "loss": 1.155,
-      "step": 10180
-    },
-    {
-      "epoch": 0.83,
-      "learning_rate": 6.076279149226959e-05,
-      "loss": 1.1491,
-      "step": 10190
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.07200820022209e-05,
-      "loss": 1.1641,
-      "step": 10200
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.0677372512172207e-05,
-      "loss": 1.1641,
-      "step": 10210
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.0634663022123514e-05,
-      "loss": 1.1594,
-      "step": 10220
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.059195353207483e-05,
-      "loss": 1.1575,
-      "step": 10230
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.054924404202614e-05,
-      "loss": 1.1593,
-      "step": 10240
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.050653455197746e-05,
-      "loss": 1.1633,
-      "step": 10250
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.046382506192876e-05,
-      "loss": 1.1663,
-      "step": 10260
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.042111557188007e-05,
-      "loss": 1.1573,
-      "step": 10270
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.0378406081831386e-05,
-      "loss": 1.1627,
-      "step": 10280
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.03356965917827e-05,
-      "loss": 1.1574,
-      "step": 10290
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.029298710173401e-05,
-      "loss": 1.1528,
-      "step": 10300
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 6.025027761168532e-05,
-      "loss": 1.1603,
-      "step": 10310
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.020756812163662e-05,
-      "loss": 1.1499,
-      "step": 10320
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.016485863158794e-05,
-      "loss": 1.1549,
-      "step": 10330
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.012214914153925e-05,
-      "loss": 1.1537,
-      "step": 10340
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.0079439651490566e-05,
-      "loss": 1.1634,
-      "step": 10350
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 6.003673016144188e-05,
-      "loss": 1.1631,
-      "step": 10360
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.9994020671393195e-05,
-      "loss": 1.1472,
-      "step": 10370
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.9951311181344495e-05,
-      "loss": 1.1601,
-      "step": 10380
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.990860169129581e-05,
-      "loss": 1.149,
-      "step": 10390
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.986589220124712e-05,
-      "loss": 1.1524,
-      "step": 10400
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.982318271119843e-05,
-      "loss": 1.153,
-      "step": 10410
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.9780473221149746e-05,
-      "loss": 1.1593,
-      "step": 10420
-    },
-    {
-      "epoch": 0.85,
-      "learning_rate": 5.973776373110105e-05,
-      "loss": 1.155,
-      "step": 10430
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.969505424105236e-05,
-      "loss": 1.1582,
-      "step": 10440
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.9652344751003675e-05,
-      "loss": 1.1522,
-      "step": 10450
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.960963526095499e-05,
-      "loss": 1.1527,
-      "step": 10460
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.9566925770906304e-05,
-      "loss": 1.1545,
-      "step": 10470
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.952421628085761e-05,
-      "loss": 1.1531,
-      "step": 10480
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.948150679080892e-05,
-      "loss": 1.1543,
-      "step": 10490
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.9438797300760226e-05,
-      "loss": 1.1599,
-      "step": 10500
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.939608781071154e-05,
-      "loss": 1.1542,
-      "step": 10510
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.9353378320662855e-05,
-      "loss": 1.1574,
-      "step": 10520
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.931066883061417e-05,
-      "loss": 1.1553,
-      "step": 10530
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.926795934056547e-05,
-      "loss": 1.1605,
-      "step": 10540
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 5.9225249850516784e-05,
-      "loss": 1.1524,
-      "step": 10550
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.91825403604681e-05,
-      "loss": 1.1556,
-      "step": 10560
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.913983087041941e-05,
-      "loss": 1.1686,
-      "step": 10570
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.909712138037072e-05,
-      "loss": 1.1512,
-      "step": 10580
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.9054411890322035e-05,
-      "loss": 1.166,
-      "step": 10590
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.9011702400273336e-05,
-      "loss": 1.1527,
-      "step": 10600
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.896899291022465e-05,
-      "loss": 1.1559,
-      "step": 10610
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.8926283420175964e-05,
-      "loss": 1.1558,
-      "step": 10620
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.888357393012728e-05,
-      "loss": 1.1529,
-      "step": 10630
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.884086444007859e-05,
-      "loss": 1.1579,
-      "step": 10640
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.879815495002991e-05,
-      "loss": 1.1535,
-      "step": 10650
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.875544545998121e-05,
-      "loss": 1.1636,
-      "step": 10660
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.871273596993252e-05,
-      "loss": 1.1468,
-      "step": 10670
-    },
-    {
-      "epoch": 0.87,
-      "learning_rate": 5.867002647988383e-05,
-      "loss": 1.1538,
-      "step": 10680
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.8627316989835144e-05,
-      "loss": 1.1528,
-      "step": 10690
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.858460749978646e-05,
-      "loss": 1.1531,
-      "step": 10700
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.854189800973776e-05,
-      "loss": 1.1613,
-      "step": 10710
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.849918851968907e-05,
-      "loss": 1.1589,
-      "step": 10720
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.845647902964039e-05,
-      "loss": 1.1531,
-      "step": 10730
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.84137695395917e-05,
-      "loss": 1.1547,
-      "step": 10740
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.8371060049543016e-05,
-      "loss": 1.1539,
-      "step": 10750
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.8328350559494324e-05,
-      "loss": 1.1522,
-      "step": 10760
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.828564106944563e-05,
-      "loss": 1.1535,
-      "step": 10770
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.824293157939694e-05,
-      "loss": 1.1628,
-      "step": 10780
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.820022208934825e-05,
-      "loss": 1.1528,
-      "step": 10790
-    },
-    {
-      "epoch": 0.88,
-      "learning_rate": 5.815751259929957e-05,
-      "loss": 1.1496,
-      "step": 10800
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.811480310925088e-05,
-      "loss": 1.143,
-      "step": 10810
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.8072093619202196e-05,
-      "loss": 1.1527,
-      "step": 10820
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.80293841291535e-05,
-      "loss": 1.1501,
-      "step": 10830
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.798667463910481e-05,
-      "loss": 1.1347,
-      "step": 10840
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.7943965149056125e-05,
-      "loss": 1.1521,
-      "step": 10850
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.790125565900743e-05,
-      "loss": 1.1574,
-      "step": 10860
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.785854616895875e-05,
-      "loss": 1.1556,
-      "step": 10870
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.7815836678910055e-05,
-      "loss": 1.1571,
-      "step": 10880
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.777312718886136e-05,
-      "loss": 1.153,
-      "step": 10890
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.7730417698812676e-05,
-      "loss": 1.1532,
-      "step": 10900
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.768770820876399e-05,
-      "loss": 1.1531,
-      "step": 10910
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 5.7644998718715305e-05,
-      "loss": 1.1568,
-      "step": 10920
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.760228922866662e-05,
-      "loss": 1.153,
-      "step": 10930
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.755957973861792e-05,
-      "loss": 1.1576,
-      "step": 10940
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.7516870248569234e-05,
-      "loss": 1.1541,
-      "step": 10950
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.747416075852055e-05,
-      "loss": 1.1529,
-      "step": 10960
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.7431451268471856e-05,
-      "loss": 1.1503,
-      "step": 10970
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.738874177842317e-05,
-      "loss": 1.1479,
-      "step": 10980
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.7346032288374485e-05,
-      "loss": 1.1449,
-      "step": 10990
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.7303322798325786e-05,
-      "loss": 1.1403,
-      "step": 11000
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.72606133082771e-05,
-      "loss": 1.1539,
-      "step": 11010
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.7217903818228414e-05,
-      "loss": 1.1459,
-      "step": 11020
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.717519432817973e-05,
-      "loss": 1.1522,
-      "step": 11030
-    },
-    {
-      "epoch": 0.9,
-      "learning_rate": 5.713248483813104e-05,
-      "loss": 1.1492,
-      "step": 11040
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.7089775348082344e-05,
-      "loss": 1.1475,
-      "step": 11050
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.704706585803366e-05,
-      "loss": 1.1596,
-      "step": 11060
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.7004356367984965e-05,
-      "loss": 1.1543,
-      "step": 11070
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.696164687793628e-05,
-      "loss": 1.1526,
-      "step": 11080
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.6918937387887594e-05,
-      "loss": 1.152,
-      "step": 11090
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.687622789783891e-05,
-      "loss": 1.1501,
-      "step": 11100
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.683351840779021e-05,
-      "loss": 1.1377,
-      "step": 11110
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.679080891774152e-05,
-      "loss": 1.1551,
-      "step": 11120
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.674809942769284e-05,
-      "loss": 1.1492,
-      "step": 11130
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.670538993764415e-05,
-      "loss": 1.1569,
-      "step": 11140
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.666268044759546e-05,
-      "loss": 1.1514,
-      "step": 11150
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 5.661997095754677e-05,
-      "loss": 1.1505,
-      "step": 11160
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.6577261467498074e-05,
-      "loss": 1.1576,
-      "step": 11170
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.653455197744939e-05,
-      "loss": 1.1459,
-      "step": 11180
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.64918424874007e-05,
-      "loss": 1.1484,
-      "step": 11190
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.644913299735202e-05,
-      "loss": 1.1561,
-      "step": 11200
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.640642350730333e-05,
-      "loss": 1.1493,
-      "step": 11210
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.636371401725463e-05,
-      "loss": 1.1438,
-      "step": 11220
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.632100452720595e-05,
-      "loss": 1.1515,
-      "step": 11230
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.627829503715726e-05,
-      "loss": 1.1479,
-      "step": 11240
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.623558554710857e-05,
-      "loss": 1.1595,
-      "step": 11250
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.619287605705988e-05,
-      "loss": 1.1503,
-      "step": 11260
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.61501665670112e-05,
-      "loss": 1.1511,
-      "step": 11270
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.61074570769625e-05,
-      "loss": 1.1507,
-      "step": 11280
-    },
-    {
-      "epoch": 0.92,
-      "learning_rate": 5.606474758691381e-05,
-      "loss": 1.1529,
-      "step": 11290
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.6022038096865127e-05,
-      "loss": 1.1469,
-      "step": 11300
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.597932860681644e-05,
-      "loss": 1.1513,
-      "step": 11310
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.5936619116767755e-05,
-      "loss": 1.1453,
-      "step": 11320
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.5893909626719056e-05,
-      "loss": 1.1603,
-      "step": 11330
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.585120013667037e-05,
-      "loss": 1.1519,
-      "step": 11340
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.580849064662168e-05,
-      "loss": 1.1538,
-      "step": 11350
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.576578115657299e-05,
-      "loss": 1.1507,
-      "step": 11360
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.5723071666524306e-05,
-      "loss": 1.1477,
-      "step": 11370
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.568036217647562e-05,
-      "loss": 1.152,
-      "step": 11380
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.563765268642692e-05,
-      "loss": 1.1574,
-      "step": 11390
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.5594943196378236e-05,
-      "loss": 1.1595,
-      "step": 11400
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 5.555223370632955e-05,
-      "loss": 1.1427,
-      "step": 11410
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.5509524216280864e-05,
-      "loss": 1.1506,
-      "step": 11420
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.546681472623217e-05,
-      "loss": 1.1395,
-      "step": 11430
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.5424105236183486e-05,
-      "loss": 1.1535,
-      "step": 11440
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.538139574613479e-05,
-      "loss": 1.1442,
-      "step": 11450
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.53386862560861e-05,
-      "loss": 1.1484,
-      "step": 11460
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.5295976766037415e-05,
-      "loss": 1.1507,
-      "step": 11470
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.525326727598873e-05,
-      "loss": 1.151,
-      "step": 11480
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.5210557785940044e-05,
-      "loss": 1.1509,
-      "step": 11490
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.5167848295891345e-05,
-      "loss": 1.144,
-      "step": 11500
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.512513880584266e-05,
-      "loss": 1.1432,
-      "step": 11510
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.508242931579397e-05,
-      "loss": 1.1328,
-      "step": 11520
-    },
-    {
-      "epoch": 0.94,
-      "learning_rate": 5.503971982574528e-05,
-      "loss": 1.1509,
-      "step": 11530
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.4997010335696595e-05,
-      "loss": 1.1465,
-      "step": 11540
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.495430084564791e-05,
-      "loss": 1.1495,
-      "step": 11550
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.491159135559921e-05,
-      "loss": 1.1473,
-      "step": 11560
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.4868881865550525e-05,
-      "loss": 1.1452,
-      "step": 11570
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.482617237550184e-05,
-      "loss": 1.1383,
-      "step": 11580
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.478346288545315e-05,
-      "loss": 1.1514,
-      "step": 11590
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.474075339540447e-05,
-      "loss": 1.1496,
-      "step": 11600
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.4698043905355775e-05,
-      "loss": 1.1401,
-      "step": 11610
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.465533441530708e-05,
-      "loss": 1.1485,
-      "step": 11620
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.461262492525839e-05,
-      "loss": 1.1611,
-      "step": 11630
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.4569915435209704e-05,
-      "loss": 1.1459,
-      "step": 11640
-    },
-    {
-      "epoch": 0.95,
-      "learning_rate": 5.452720594516102e-05,
-      "loss": 1.1445,
-      "step": 11650
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.448449645511233e-05,
-      "loss": 1.1439,
-      "step": 11660
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.4441786965063634e-05,
-      "loss": 1.1481,
-      "step": 11670
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.439907747501495e-05,
-      "loss": 1.1502,
-      "step": 11680
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.435636798496626e-05,
-      "loss": 1.1508,
-      "step": 11690
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.4313658494917577e-05,
-      "loss": 1.1477,
-      "step": 11700
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.4270949004868884e-05,
-      "loss": 1.1437,
-      "step": 11710
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.42282395148202e-05,
-      "loss": 1.1447,
-      "step": 11720
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.4185530024771506e-05,
-      "loss": 1.1532,
-      "step": 11730
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.4142820534722813e-05,
-      "loss": 1.1466,
-      "step": 11740
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.410011104467413e-05,
-      "loss": 1.1437,
-      "step": 11750
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.405740155462544e-05,
-      "loss": 1.144,
-      "step": 11760
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 5.4014692064576756e-05,
-      "loss": 1.138,
-      "step": 11770
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.397198257452806e-05,
-      "loss": 1.137,
-      "step": 11780
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.392927308447937e-05,
-      "loss": 1.1454,
-      "step": 11790
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.3886563594430686e-05,
-      "loss": 1.146,
-      "step": 11800
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.3843854104382e-05,
-      "loss": 1.1469,
-      "step": 11810
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.380114461433331e-05,
-      "loss": 1.1452,
-      "step": 11820
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.375843512428462e-05,
-      "loss": 1.1456,
-      "step": 11830
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.371572563423592e-05,
-      "loss": 1.1454,
-      "step": 11840
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.367301614418724e-05,
-      "loss": 1.1503,
-      "step": 11850
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.363030665413855e-05,
-      "loss": 1.1389,
-      "step": 11860
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.3587597164089865e-05,
-      "loss": 1.1487,
-      "step": 11870
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.354488767404118e-05,
-      "loss": 1.1453,
-      "step": 11880
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.3502178183992494e-05,
-      "loss": 1.1458,
-      "step": 11890
-    },
-    {
-      "epoch": 0.97,
-      "learning_rate": 5.3459468693943795e-05,
-      "loss": 1.1415,
-      "step": 11900
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.341675920389511e-05,
-      "loss": 1.1392,
-      "step": 11910
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.337404971384642e-05,
-      "loss": 1.1422,
-      "step": 11920
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.333134022379773e-05,
-      "loss": 1.135,
-      "step": 11930
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.3288630733749045e-05,
-      "loss": 1.1432,
-      "step": 11940
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.3245921243700346e-05,
-      "loss": 1.1484,
-      "step": 11950
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.320321175365166e-05,
-      "loss": 1.1548,
-      "step": 11960
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.3160502263602975e-05,
-      "loss": 1.144,
-      "step": 11970
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.311779277355429e-05,
-      "loss": 1.1347,
-      "step": 11980
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.30750832835056e-05,
-      "loss": 1.147,
-      "step": 11990
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.303237379345691e-05,
-      "loss": 1.1352,
-      "step": 12000
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.298966430340822e-05,
-      "loss": 1.1597,
-      "step": 12010
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 5.2946954813359526e-05,
-      "loss": 1.1489,
-      "step": 12020
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.290424532331084e-05,
-      "loss": 1.1426,
-      "step": 12030
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.2861535833262154e-05,
-      "loss": 1.1396,
-      "step": 12040
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.281882634321347e-05,
-      "loss": 1.1462,
-      "step": 12050
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.277611685316478e-05,
-      "loss": 1.1371,
-      "step": 12060
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.2733407363116084e-05,
-      "loss": 1.1509,
-      "step": 12070
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.26906978730674e-05,
-      "loss": 1.1439,
-      "step": 12080
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.264798838301871e-05,
-      "loss": 1.146,
-      "step": 12090
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.260527889297002e-05,
-      "loss": 1.149,
-      "step": 12100
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.2562569402921334e-05,
-      "loss": 1.1462,
-      "step": 12110
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.2519859912872635e-05,
-      "loss": 1.1392,
-      "step": 12120
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.247715042282395e-05,
-      "loss": 1.1403,
-      "step": 12130
-    },
-    {
-      "epoch": 0.99,
-      "learning_rate": 5.2434440932775263e-05,
-      "loss": 1.1497,
-      "step": 12140
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.239173144272658e-05,
-      "loss": 1.155,
-      "step": 12150
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.234902195267789e-05,
-      "loss": 1.1471,
-      "step": 12160
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.2306312462629206e-05,
-      "loss": 1.1432,
-      "step": 12170
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.226360297258051e-05,
-      "loss": 1.1374,
-      "step": 12180
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.222089348253182e-05,
-      "loss": 1.1348,
-      "step": 12190
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.217818399248313e-05,
-      "loss": 1.1548,
-      "step": 12200
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.213547450243444e-05,
-      "loss": 1.1635,
-      "step": 12210
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.209276501238576e-05,
-      "loss": 1.1367,
-      "step": 12220
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.205005552233707e-05,
-      "loss": 1.1436,
-      "step": 12230
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.200734603228837e-05,
-      "loss": 1.1445,
-      "step": 12240
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.196463654223969e-05,
-      "loss": 1.1382,
-      "step": 12250
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 5.1921927052191e-05,
-      "loss": 1.1419,
-      "step": 12260
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.1879217562142315e-05,
-      "loss": 1.1395,
-      "step": 12270
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.183650807209362e-05,
-      "loss": 1.1472,
-      "step": 12280
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.179379858204493e-05,
-      "loss": 1.1438,
-      "step": 12290
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.175108909199624e-05,
-      "loss": 1.1414,
-      "step": 12300
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.170837960194755e-05,
-      "loss": 1.1509,
-      "step": 12310
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.166567011189887e-05,
-      "loss": 1.1474,
-      "step": 12320
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.162296062185018e-05,
-      "loss": 1.1362,
-      "step": 12330
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.1580251131801495e-05,
-      "loss": 1.1427,
-      "step": 12340
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.1537541641752796e-05,
-      "loss": 1.1348,
-      "step": 12350
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.149483215170411e-05,
-      "loss": 1.1415,
-      "step": 12360
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.1452122661655425e-05,
-      "loss": 1.139,
-      "step": 12370
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.140941317160673e-05,
-      "loss": 1.1335,
-      "step": 12380
-    },
-    {
-      "epoch": 1.01,
-      "learning_rate": 5.1366703681558046e-05,
-      "loss": 1.1454,
-      "step": 12390
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.132399419150935e-05,
-      "loss": 1.1481,
-      "step": 12400
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.128128470146066e-05,
-      "loss": 1.148,
-      "step": 12410
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.1238575211411976e-05,
-      "loss": 1.1379,
-      "step": 12420
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.119586572136329e-05,
-      "loss": 1.1442,
-      "step": 12430
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.1153156231314604e-05,
-      "loss": 1.1248,
-      "step": 12440
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.111044674126592e-05,
-      "loss": 1.1394,
-      "step": 12450
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.106773725121722e-05,
-      "loss": 1.1411,
-      "step": 12460
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.1025027761168534e-05,
-      "loss": 1.147,
-      "step": 12470
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.098231827111984e-05,
-      "loss": 1.1229,
-      "step": 12480
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.0939608781071156e-05,
-      "loss": 1.1456,
-      "step": 12490
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.089689929102247e-05,
-      "loss": 1.1411,
-      "step": 12500
-    },
-    {
-      "epoch": 1.02,
-      "learning_rate": 5.0854189800973784e-05,
-      "loss": 1.1343,
-      "step": 12510
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.0811480310925085e-05,
-      "loss": 1.1445,
-      "step": 12520
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.07687708208764e-05,
-      "loss": 1.137,
-      "step": 12530
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.0726061330827714e-05,
-      "loss": 1.1398,
-      "step": 12540
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.068335184077903e-05,
-      "loss": 1.1408,
-      "step": 12550
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.0640642350730335e-05,
-      "loss": 1.1448,
-      "step": 12560
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.059793286068164e-05,
-      "loss": 1.1407,
-      "step": 12570
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.055522337063295e-05,
-      "loss": 1.1355,
-      "step": 12580
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.0512513880584265e-05,
-      "loss": 1.1319,
-      "step": 12590
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.046980439053558e-05,
-      "loss": 1.1468,
-      "step": 12600
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.042709490048689e-05,
-      "loss": 1.1376,
-      "step": 12610
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.038438541043821e-05,
-      "loss": 1.1406,
-      "step": 12620
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 5.034167592038951e-05,
-      "loss": 1.1374,
-      "step": 12630
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 5.029896643034082e-05,
-      "loss": 1.1357,
-      "step": 12640
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 5.025625694029214e-05,
-      "loss": 1.1344,
-      "step": 12650
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 5.0213547450243444e-05,
-      "loss": 1.1371,
-      "step": 12660
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 5.017083796019476e-05,
-      "loss": 1.1378,
-      "step": 12670
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 5.012812847014607e-05,
-      "loss": 1.1319,
-      "step": 12680
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 5.0085418980097374e-05,
-      "loss": 1.1331,
-      "step": 12690
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 5.004270949004869e-05,
-      "loss": 1.1277,
-      "step": 12700
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 5e-05,
-      "loss": 1.1423,
-      "step": 12710
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 4.995729050995132e-05,
-      "loss": 1.1326,
-      "step": 12720
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 4.9914581019902624e-05,
-      "loss": 1.1486,
-      "step": 12730
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 4.987187152985394e-05,
-      "loss": 1.1272,
-      "step": 12740
-    },
-    {
-      "epoch": 1.04,
-      "learning_rate": 4.982916203980525e-05,
-      "loss": 1.143,
-      "step": 12750
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.978645254975656e-05,
-      "loss": 1.133,
-      "step": 12760
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.974374305970787e-05,
-      "loss": 1.1358,
-      "step": 12770
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.970103356965918e-05,
-      "loss": 1.1437,
-      "step": 12780
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.965832407961049e-05,
-      "loss": 1.1401,
-      "step": 12790
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.9615614589561804e-05,
-      "loss": 1.1385,
-      "step": 12800
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.957290509951311e-05,
-      "loss": 1.1404,
-      "step": 12810
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.9530195609464426e-05,
-      "loss": 1.1344,
-      "step": 12820
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.948748611941574e-05,
-      "loss": 1.1378,
-      "step": 12830
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.944477662936705e-05,
-      "loss": 1.134,
-      "step": 12840
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.940206713931836e-05,
-      "loss": 1.14,
-      "step": 12850
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.935935764926967e-05,
-      "loss": 1.1407,
-      "step": 12860
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.931664815922098e-05,
-      "loss": 1.137,
-      "step": 12870
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.927393866917229e-05,
-      "loss": 1.1342,
-      "step": 12880
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.9231229179123606e-05,
-      "loss": 1.1454,
-      "step": 12890
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.918851968907491e-05,
-      "loss": 1.1349,
-      "step": 12900
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.914581019902623e-05,
-      "loss": 1.1451,
-      "step": 12910
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.910310070897754e-05,
-      "loss": 1.1393,
-      "step": 12920
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.906039121892885e-05,
-      "loss": 1.1384,
-      "step": 12930
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.9017681728880164e-05,
-      "loss": 1.1381,
-      "step": 12940
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.897497223883147e-05,
-      "loss": 1.1405,
-      "step": 12950
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.893226274878278e-05,
-      "loss": 1.1362,
-      "step": 12960
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.888955325873409e-05,
-      "loss": 1.1395,
-      "step": 12970
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.88468437686854e-05,
-      "loss": 1.1334,
-      "step": 12980
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.8804134278636715e-05,
-      "loss": 1.1473,
-      "step": 12990
-    },
-    {
-      "epoch": 1.06,
-      "learning_rate": 4.876142478858803e-05,
-      "loss": 1.1305,
-      "step": 13000
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.8718715298539337e-05,
-      "loss": 1.1405,
-      "step": 13010
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.867600580849065e-05,
-      "loss": 1.1328,
-      "step": 13020
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.8633296318441965e-05,
-      "loss": 1.1356,
-      "step": 13030
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.859485777739814e-05,
-      "loss": 1.1272,
-      "step": 13040
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.855214828734945e-05,
-      "loss": 1.1398,
-      "step": 13050
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.850943879730076e-05,
-      "loss": 1.1395,
-      "step": 13060
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.8466729307252075e-05,
-      "loss": 1.1334,
-      "step": 13070
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.842401981720339e-05,
-      "loss": 1.1327,
-      "step": 13080
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.8381310327154696e-05,
-      "loss": 1.1358,
-      "step": 13090
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.8338600837106004e-05,
-      "loss": 1.1318,
-      "step": 13100
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.829589134705732e-05,
-      "loss": 1.1379,
-      "step": 13110
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.8253181857008626e-05,
-      "loss": 1.1329,
-      "step": 13120
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.821047236695994e-05,
-      "loss": 1.1352,
-      "step": 13130
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.8167762876911254e-05,
-      "loss": 1.139,
-      "step": 13140
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.812505338686256e-05,
-      "loss": 1.1379,
-      "step": 13150
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.8082343896813876e-05,
-      "loss": 1.1323,
-      "step": 13160
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.803963440676519e-05,
-      "loss": 1.1445,
-      "step": 13170
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.79969249167165e-05,
-      "loss": 1.1266,
-      "step": 13180
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.7954215426667806e-05,
-      "loss": 1.1425,
-      "step": 13190
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.791150593661912e-05,
-      "loss": 1.1363,
-      "step": 13200
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.786879644657043e-05,
-      "loss": 1.1366,
-      "step": 13210
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.782608695652174e-05,
-      "loss": 1.1297,
-      "step": 13220
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.778337746647305e-05,
-      "loss": 1.1444,
-      "step": 13230
-    },
-    {
-      "epoch": 1.08,
-      "learning_rate": 4.7740667976424364e-05,
-      "loss": 1.1306,
-      "step": 13240
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.769795848637568e-05,
-      "loss": 1.1368,
-      "step": 13250
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.7655248996326985e-05,
-      "loss": 1.141,
-      "step": 13260
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.76125395062783e-05,
-      "loss": 1.1349,
-      "step": 13270
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.7569830016229614e-05,
-      "loss": 1.1304,
-      "step": 13280
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.752712052618092e-05,
-      "loss": 1.1374,
-      "step": 13290
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.748441103613223e-05,
-      "loss": 1.1321,
-      "step": 13300
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.744170154608354e-05,
-      "loss": 1.1426,
-      "step": 13310
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.739899205603485e-05,
-      "loss": 1.135,
-      "step": 13320
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.7356282565986165e-05,
-      "loss": 1.1314,
-      "step": 13330
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.731357307593747e-05,
-      "loss": 1.1334,
-      "step": 13340
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.727086358588879e-05,
-      "loss": 1.1385,
-      "step": 13350
-    },
-    {
-      "epoch": 1.09,
-      "learning_rate": 4.72281540958401e-05,
-      "loss": 1.1371,
-      "step": 13360
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.718544460579141e-05,
-      "loss": 1.1369,
-      "step": 13370
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.714273511574272e-05,
-      "loss": 1.1347,
-      "step": 13380
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.710002562569403e-05,
-      "loss": 1.1283,
-      "step": 13390
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.705731613564534e-05,
-      "loss": 1.1356,
-      "step": 13400
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.701460664559665e-05,
-      "loss": 1.137,
-      "step": 13410
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.697189715554797e-05,
-      "loss": 1.1391,
-      "step": 13420
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.6929187665499274e-05,
-      "loss": 1.1325,
-      "step": 13430
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.688647817545059e-05,
-      "loss": 1.128,
-      "step": 13440
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.68437686854019e-05,
-      "loss": 1.1409,
-      "step": 13450
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.680105919535321e-05,
-      "loss": 1.1345,
-      "step": 13460
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.6758349705304525e-05,
-      "loss": 1.1256,
-      "step": 13470
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.671564021525583e-05,
-      "loss": 1.1313,
-      "step": 13480
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.667293072520714e-05,
-      "loss": 1.1323,
-      "step": 13490
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.6630221235158454e-05,
-      "loss": 1.1285,
-      "step": 13500
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.658751174510976e-05,
-      "loss": 1.1369,
-      "step": 13510
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.6544802255061076e-05,
-      "loss": 1.1326,
-      "step": 13520
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.650209276501239e-05,
-      "loss": 1.1303,
-      "step": 13530
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.64593832749637e-05,
-      "loss": 1.127,
-      "step": 13540
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.641667378491501e-05,
-      "loss": 1.1367,
-      "step": 13550
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.6373964294866326e-05,
-      "loss": 1.1301,
-      "step": 13560
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.6331254804817634e-05,
-      "loss": 1.1327,
-      "step": 13570
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.628854531476894e-05,
-      "loss": 1.1335,
-      "step": 13580
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.6245835824720256e-05,
-      "loss": 1.1318,
-      "step": 13590
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.620312633467156e-05,
-      "loss": 1.1353,
-      "step": 13600
-    },
-    {
-      "epoch": 1.11,
-      "learning_rate": 4.616041684462288e-05,
-      "loss": 1.1272,
-      "step": 13610
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.611770735457419e-05,
-      "loss": 1.13,
-      "step": 13620
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.60749978645255e-05,
-      "loss": 1.1327,
-      "step": 13630
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.6032288374476814e-05,
-      "loss": 1.1295,
-      "step": 13640
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.598957888442812e-05,
-      "loss": 1.1369,
-      "step": 13650
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.5946869394379435e-05,
-      "loss": 1.126,
-      "step": 13660
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.590415990433074e-05,
-      "loss": 1.1248,
-      "step": 13670
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.586145041428205e-05,
-      "loss": 1.1302,
-      "step": 13680
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.5818740924233365e-05,
-      "loss": 1.1293,
-      "step": 13690
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.577603143418468e-05,
-      "loss": 1.1281,
-      "step": 13700
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.5733321944135987e-05,
-      "loss": 1.1398,
-      "step": 13710
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.56906124540873e-05,
-      "loss": 1.1365,
-      "step": 13720
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.5647902964038615e-05,
-      "loss": 1.1397,
-      "step": 13730
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.560519347398992e-05,
-      "loss": 1.1332,
-      "step": 13740
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.556248398394124e-05,
-      "loss": 1.1341,
-      "step": 13750
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.5519774493892545e-05,
-      "loss": 1.1365,
-      "step": 13760
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.547706500384385e-05,
-      "loss": 1.1414,
-      "step": 13770
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.5434355513795166e-05,
-      "loss": 1.1347,
-      "step": 13780
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.539164602374648e-05,
-      "loss": 1.1314,
-      "step": 13790
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.534893653369779e-05,
-      "loss": 1.1179,
-      "step": 13800
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.53062270436491e-05,
-      "loss": 1.1307,
-      "step": 13810
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.526351755360041e-05,
-      "loss": 1.127,
-      "step": 13820
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.5220808063551724e-05,
-      "loss": 1.1371,
-      "step": 13830
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.517809857350304e-05,
-      "loss": 1.1316,
-      "step": 13840
-    },
-    {
-      "epoch": 1.13,
-      "learning_rate": 4.5135389083454346e-05,
-      "loss": 1.1285,
-      "step": 13850
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.5092679593405654e-05,
-      "loss": 1.1366,
-      "step": 13860
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.504997010335697e-05,
-      "loss": 1.1348,
-      "step": 13870
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.5007260613308275e-05,
-      "loss": 1.1302,
-      "step": 13880
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.496455112325959e-05,
-      "loss": 1.1305,
-      "step": 13890
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.4921841633210904e-05,
-      "loss": 1.1299,
-      "step": 13900
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.487913214316221e-05,
-      "loss": 1.1303,
-      "step": 13910
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.4836422653113526e-05,
-      "loss": 1.1313,
-      "step": 13920
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.479371316306484e-05,
-      "loss": 1.1299,
-      "step": 13930
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.475100367301615e-05,
-      "loss": 1.1266,
-      "step": 13940
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.4708294182967455e-05,
-      "loss": 1.1316,
-      "step": 13950
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.466558469291876e-05,
-      "loss": 1.1317,
-      "step": 13960
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.462287520287008e-05,
-      "loss": 1.131,
-      "step": 13970
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.458016571282139e-05,
-      "loss": 1.1291,
-      "step": 13980
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.45374562227727e-05,
-      "loss": 1.1353,
-      "step": 13990
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.449474673272401e-05,
-      "loss": 1.1275,
-      "step": 14000
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.445203724267533e-05,
-      "loss": 1.1253,
-      "step": 14010
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.4409327752626635e-05,
-      "loss": 1.1276,
-      "step": 14020
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.436661826257795e-05,
-      "loss": 1.1215,
-      "step": 14030
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.432390877252926e-05,
-      "loss": 1.1322,
-      "step": 14040
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.4281199282480564e-05,
-      "loss": 1.1404,
-      "step": 14050
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.423848979243188e-05,
-      "loss": 1.131,
-      "step": 14060
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.419578030238319e-05,
-      "loss": 1.1193,
-      "step": 14070
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.41530708123345e-05,
-      "loss": 1.1284,
-      "step": 14080
-    },
-    {
-      "epoch": 1.15,
-      "learning_rate": 4.4110361322285815e-05,
-      "loss": 1.1307,
-      "step": 14090
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.406765183223713e-05,
-      "loss": 1.1283,
-      "step": 14100
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.402494234218844e-05,
-      "loss": 1.1317,
-      "step": 14110
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.398223285213975e-05,
-      "loss": 1.1374,
-      "step": 14120
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.393952336209106e-05,
-      "loss": 1.1369,
-      "step": 14130
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.389681387204237e-05,
-      "loss": 1.1251,
-      "step": 14140
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.385410438199368e-05,
-      "loss": 1.1425,
-      "step": 14150
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.381139489194499e-05,
-      "loss": 1.1243,
-      "step": 14160
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.37686854018963e-05,
-      "loss": 1.1349,
-      "step": 14170
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.3725975911847616e-05,
-      "loss": 1.1356,
-      "step": 14180
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.3683266421798924e-05,
-      "loss": 1.1276,
-      "step": 14190
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.364055693175024e-05,
-      "loss": 1.1348,
-      "step": 14200
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.359784744170155e-05,
-      "loss": 1.1342,
-      "step": 14210
-    },
-    {
-      "epoch": 1.16,
-      "learning_rate": 4.355513795165286e-05,
-      "loss": 1.1273,
-      "step": 14220
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.3512428461604174e-05,
-      "loss": 1.1295,
-      "step": 14230
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.346971897155548e-05,
-      "loss": 1.1371,
-      "step": 14240
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.342700948150679e-05,
-      "loss": 1.1416,
-      "step": 14250
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.3384299991458104e-05,
-      "loss": 1.1224,
-      "step": 14260
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.334159050140941e-05,
-      "loss": 1.132,
-      "step": 14270
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.3298881011360726e-05,
-      "loss": 1.1274,
-      "step": 14280
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.325617152131204e-05,
-      "loss": 1.1314,
-      "step": 14290
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.321346203126335e-05,
-      "loss": 1.1315,
-      "step": 14300
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.317075254121466e-05,
-      "loss": 1.1254,
-      "step": 14310
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.3128043051165976e-05,
-      "loss": 1.1254,
-      "step": 14320
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.3085333561117283e-05,
-      "loss": 1.1298,
-      "step": 14330
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.304262407106859e-05,
-      "loss": 1.1273,
-      "step": 14340
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.2999914581019905e-05,
-      "loss": 1.1379,
-      "step": 14350
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.295720509097121e-05,
-      "loss": 1.1315,
-      "step": 14360
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.291449560092253e-05,
-      "loss": 1.1332,
-      "step": 14370
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.287178611087384e-05,
-      "loss": 1.1276,
-      "step": 14380
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.282907662082515e-05,
-      "loss": 1.1272,
-      "step": 14390
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.278636713077646e-05,
-      "loss": 1.1193,
-      "step": 14400
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.274365764072778e-05,
-      "loss": 1.1316,
-      "step": 14410
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.2700948150679085e-05,
-      "loss": 1.127,
-      "step": 14420
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.265823866063039e-05,
-      "loss": 1.1305,
-      "step": 14430
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.26155291705817e-05,
-      "loss": 1.1327,
-      "step": 14440
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.2572819680533014e-05,
-      "loss": 1.1347,
-      "step": 14450
-    },
-    {
-      "epoch": 1.18,
-      "learning_rate": 4.253011019048433e-05,
-      "loss": 1.1171,
-      "step": 14460
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.2487400700435636e-05,
-      "loss": 1.1296,
-      "step": 14470
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.244469121038695e-05,
-      "loss": 1.1248,
-      "step": 14480
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.2401981720338265e-05,
-      "loss": 1.1288,
-      "step": 14490
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.235927223028957e-05,
-      "loss": 1.1378,
-      "step": 14500
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.231656274024089e-05,
-      "loss": 1.1255,
-      "step": 14510
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.2273853250192194e-05,
-      "loss": 1.1325,
-      "step": 14520
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.22311437601435e-05,
-      "loss": 1.1237,
-      "step": 14530
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.2188434270094816e-05,
-      "loss": 1.1283,
-      "step": 14540
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.214572478004613e-05,
-      "loss": 1.1234,
-      "step": 14550
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.210301528999744e-05,
-      "loss": 1.1292,
-      "step": 14560
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.206030579994875e-05,
-      "loss": 1.1255,
-      "step": 14570
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.201759630990006e-05,
-      "loss": 1.1261,
-      "step": 14580
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.1974886819851374e-05,
-      "loss": 1.1249,
-      "step": 14590
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.193217732980269e-05,
-      "loss": 1.128,
-      "step": 14600
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.1889467839753996e-05,
-      "loss": 1.1369,
-      "step": 14610
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.18467583497053e-05,
-      "loss": 1.1178,
-      "step": 14620
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.180404885965662e-05,
-      "loss": 1.1284,
-      "step": 14630
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.1761339369607925e-05,
-      "loss": 1.1245,
-      "step": 14640
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.171862987955924e-05,
-      "loss": 1.1205,
-      "step": 14650
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.1675920389510554e-05,
-      "loss": 1.1294,
-      "step": 14660
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.163321089946186e-05,
-      "loss": 1.1409,
-      "step": 14670
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.1590501409413176e-05,
-      "loss": 1.1385,
-      "step": 14680
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.154779191936449e-05,
-      "loss": 1.1245,
-      "step": 14690
-    },
-    {
-      "epoch": 1.2,
-      "learning_rate": 4.15050824293158e-05,
-      "loss": 1.1242,
-      "step": 14700
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.1462372939267105e-05,
-      "loss": 1.1292,
-      "step": 14710
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.141966344921842e-05,
-      "loss": 1.1252,
-      "step": 14720
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.137695395916973e-05,
-      "loss": 1.133,
-      "step": 14730
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.133424446912104e-05,
-      "loss": 1.1275,
-      "step": 14740
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.129153497907235e-05,
-      "loss": 1.1215,
-      "step": 14750
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.124882548902366e-05,
-      "loss": 1.1268,
-      "step": 14760
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.120611599897498e-05,
-      "loss": 1.1306,
-      "step": 14770
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.1163406508926285e-05,
-      "loss": 1.132,
-      "step": 14780
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.11206970188776e-05,
-      "loss": 1.1303,
-      "step": 14790
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.1077987528828906e-05,
-      "loss": 1.1205,
-      "step": 14800
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.1035278038780214e-05,
-      "loss": 1.135,
-      "step": 14810
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.099256854873153e-05,
-      "loss": 1.1249,
-      "step": 14820
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.094985905868284e-05,
-      "loss": 1.1217,
-      "step": 14830
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.090714956863415e-05,
-      "loss": 1.1166,
-      "step": 14840
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.0864440078585464e-05,
-      "loss": 1.1312,
-      "step": 14850
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.082173058853678e-05,
-      "loss": 1.1283,
-      "step": 14860
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.0779021098488086e-05,
-      "loss": 1.1305,
-      "step": 14870
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.07363116084394e-05,
-      "loss": 1.1281,
-      "step": 14880
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.069360211839071e-05,
-      "loss": 1.1205,
-      "step": 14890
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.0650892628342016e-05,
-      "loss": 1.1365,
-      "step": 14900
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.060818313829333e-05,
-      "loss": 1.1174,
-      "step": 14910
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.056547364824464e-05,
-      "loss": 1.1161,
-      "step": 14920
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.052276415819595e-05,
-      "loss": 1.1231,
-      "step": 14930
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.0480054668147266e-05,
-      "loss": 1.1267,
-      "step": 14940
-    },
-    {
-      "epoch": 1.22,
-      "learning_rate": 4.0437345178098574e-05,
-      "loss": 1.1237,
-      "step": 14950
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 4.039463568804989e-05,
-      "loss": 1.1275,
-      "step": 14960
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 4.03519261980012e-05,
-      "loss": 1.1288,
-      "step": 14970
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 4.030921670795251e-05,
-      "loss": 1.1163,
-      "step": 14980
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 4.026650721790382e-05,
-      "loss": 1.1297,
-      "step": 14990
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 4.022379772785513e-05,
-      "loss": 1.1255,
-      "step": 15000
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 4.018108823780644e-05,
-      "loss": 1.1328,
-      "step": 15010
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 4.013837874775775e-05,
-      "loss": 1.1258,
-      "step": 15020
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 4.009566925770907e-05,
-      "loss": 1.1274,
-      "step": 15030
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 4.0052959767660375e-05,
-      "loss": 1.1249,
-      "step": 15040
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 4.001025027761169e-05,
-      "loss": 1.1274,
-      "step": 15050
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 3.9967540787563e-05,
-      "loss": 1.128,
-      "step": 15060
-    },
-    {
-      "epoch": 1.23,
-      "learning_rate": 3.992483129751431e-05,
-      "loss": 1.1248,
-      "step": 15070
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.9882121807465626e-05,
-      "loss": 1.1258,
-      "step": 15080
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.983941231741693e-05,
-      "loss": 1.1291,
-      "step": 15090
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.979670282736824e-05,
-      "loss": 1.1221,
-      "step": 15100
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.9753993337319555e-05,
-      "loss": 1.1292,
-      "step": 15110
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.971128384727086e-05,
-      "loss": 1.1243,
-      "step": 15120
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.966857435722218e-05,
-      "loss": 1.1222,
-      "step": 15130
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.962586486717349e-05,
-      "loss": 1.1264,
-      "step": 15140
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.95831553771248e-05,
-      "loss": 1.126,
-      "step": 15150
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.954044588707611e-05,
-      "loss": 1.131,
-      "step": 15160
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.949773639702743e-05,
-      "loss": 1.1256,
-      "step": 15170
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.9455026906978735e-05,
-      "loss": 1.1328,
-      "step": 15180
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 3.941231741693004e-05,
-      "loss": 1.1394,
-      "step": 15190
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.936960792688135e-05,
-      "loss": 1.1147,
-      "step": 15200
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.9326898436832664e-05,
-      "loss": 1.1202,
-      "step": 15210
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.928418894678398e-05,
-      "loss": 1.1207,
-      "step": 15220
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.9241479456735286e-05,
-      "loss": 1.1357,
-      "step": 15230
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.91987699666866e-05,
-      "loss": 1.1182,
-      "step": 15240
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.9156060476637914e-05,
-      "loss": 1.1114,
-      "step": 15250
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.911335098658922e-05,
-      "loss": 1.1173,
-      "step": 15260
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.9070641496540536e-05,
-      "loss": 1.1235,
-      "step": 15270
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.9027932006491844e-05,
-      "loss": 1.1203,
-      "step": 15280
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.898522251644315e-05,
-      "loss": 1.1334,
-      "step": 15290
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.8942513026394466e-05,
-      "loss": 1.1172,
-      "step": 15300
-    },
-    {
-      "epoch": 1.25,
-      "learning_rate": 3.889980353634578e-05,
-      "loss": 1.1201,
-      "step": 15310
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.885709404629709e-05,
-      "loss": 1.1234,
-      "step": 15320
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.88143845562484e-05,
-      "loss": 1.1241,
-      "step": 15330
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.8771675066199716e-05,
-      "loss": 1.121,
-      "step": 15340
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.8728965576151024e-05,
-      "loss": 1.1449,
-      "step": 15350
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.868625608610234e-05,
-      "loss": 1.1228,
-      "step": 15360
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.8643546596053645e-05,
-      "loss": 1.12,
-      "step": 15370
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.860083710600495e-05,
-      "loss": 1.1175,
-      "step": 15380
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.855812761595627e-05,
-      "loss": 1.1306,
-      "step": 15390
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.8515418125907575e-05,
-      "loss": 1.1184,
-      "step": 15400
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.847270863585889e-05,
-      "loss": 1.1252,
-      "step": 15410
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.84299991458102e-05,
-      "loss": 1.1158,
-      "step": 15420
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.838728965576151e-05,
-      "loss": 1.1221,
-      "step": 15430
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 3.8344580165712825e-05,
-      "loss": 1.1198,
-      "step": 15440
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.830187067566414e-05,
-      "loss": 1.1262,
-      "step": 15450
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.825916118561545e-05,
-      "loss": 1.1257,
-      "step": 15460
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.8216451695566755e-05,
-      "loss": 1.1167,
-      "step": 15470
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.817374220551807e-05,
-      "loss": 1.1186,
-      "step": 15480
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.8131032715469376e-05,
-      "loss": 1.12,
-      "step": 15490
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.808832322542069e-05,
-      "loss": 1.12,
-      "step": 15500
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.8045613735372e-05,
-      "loss": 1.1223,
-      "step": 15510
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.800290424532331e-05,
-      "loss": 1.1229,
-      "step": 15520
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.796019475527463e-05,
-      "loss": 1.1147,
-      "step": 15530
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.7917485265225934e-05,
-      "loss": 1.1209,
-      "step": 15540
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.787477577517725e-05,
-      "loss": 1.1241,
-      "step": 15550
-    },
-    {
-      "epoch": 1.27,
-      "learning_rate": 3.7832066285128556e-05,
-      "loss": 1.1221,
-      "step": 15560
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.7789356795079864e-05,
-      "loss": 1.1165,
-      "step": 15570
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.774664730503118e-05,
-      "loss": 1.117,
-      "step": 15580
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.770393781498249e-05,
-      "loss": 1.1199,
-      "step": 15590
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.76612283249338e-05,
-      "loss": 1.1301,
-      "step": 15600
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.7618518834885114e-05,
-      "loss": 1.1178,
-      "step": 15610
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.757580934483643e-05,
-      "loss": 1.1179,
-      "step": 15620
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.7533099854787736e-05,
-      "loss": 1.1164,
-      "step": 15630
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.749039036473905e-05,
-      "loss": 1.12,
-      "step": 15640
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.744768087469036e-05,
-      "loss": 1.1247,
-      "step": 15650
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.7404971384641665e-05,
-      "loss": 1.1253,
-      "step": 15660
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.736226189459298e-05,
-      "loss": 1.1168,
-      "step": 15670
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 3.731955240454429e-05,
-      "loss": 1.1177,
-      "step": 15680
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.72768429144956e-05,
-      "loss": 1.1126,
-      "step": 15690
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.7234133424446916e-05,
-      "loss": 1.1251,
-      "step": 15700
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.719142393439822e-05,
-      "loss": 1.1263,
-      "step": 15710
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.714871444434954e-05,
-      "loss": 1.127,
-      "step": 15720
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.710600495430085e-05,
-      "loss": 1.127,
-      "step": 15730
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.706329546425216e-05,
-      "loss": 1.1136,
-      "step": 15740
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.702058597420347e-05,
-      "loss": 1.1178,
-      "step": 15750
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.697787648415478e-05,
-      "loss": 1.1237,
-      "step": 15760
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.693516699410609e-05,
-      "loss": 1.1154,
-      "step": 15770
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.68924575040574e-05,
-      "loss": 1.123,
-      "step": 15780
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.684974801400872e-05,
-      "loss": 1.1167,
-      "step": 15790
-    },
-    {
-      "epoch": 1.29,
-      "learning_rate": 3.6807038523960025e-05,
-      "loss": 1.1246,
-      "step": 15800
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.676432903391134e-05,
-      "loss": 1.1247,
-      "step": 15810
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.672161954386265e-05,
-      "loss": 1.1203,
-      "step": 15820
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.667891005381396e-05,
-      "loss": 1.1248,
-      "step": 15830
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.663620056376527e-05,
-      "loss": 1.1237,
-      "step": 15840
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.6593491073716576e-05,
-      "loss": 1.1178,
-      "step": 15850
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.655078158366789e-05,
-      "loss": 1.1242,
-      "step": 15860
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.6508072093619205e-05,
-      "loss": 1.1241,
-      "step": 15870
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.646536260357051e-05,
-      "loss": 1.1255,
-      "step": 15880
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.6422653113521826e-05,
-      "loss": 1.126,
-      "step": 15890
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.637994362347314e-05,
-      "loss": 1.1195,
-      "step": 15900
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.633723413342445e-05,
-      "loss": 1.1183,
-      "step": 15910
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.629452464337576e-05,
-      "loss": 1.1202,
-      "step": 15920
-    },
-    {
-      "epoch": 1.3,
-      "learning_rate": 3.625181515332707e-05,
-      "loss": 1.1221,
-      "step": 15930
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.620910566327838e-05,
-      "loss": 1.1206,
-      "step": 15940
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.616639617322969e-05,
-      "loss": 1.127,
-      "step": 15950
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.6123686683181006e-05,
-      "loss": 1.1111,
-      "step": 15960
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.6080977193132314e-05,
-      "loss": 1.1129,
-      "step": 15970
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.603826770308363e-05,
-      "loss": 1.1185,
-      "step": 15980
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.5995558213034936e-05,
-      "loss": 1.1216,
-      "step": 15990
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.5957119671991116e-05,
-      "loss": 1.1247,
-      "step": 16000
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.591441018194243e-05,
-      "loss": 1.1201,
-      "step": 16010
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.587170069189374e-05,
-      "loss": 1.1185,
-      "step": 16020
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.582899120184505e-05,
-      "loss": 1.1174,
-      "step": 16030
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.5786281711796366e-05,
-      "loss": 1.1173,
-      "step": 16040
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 3.5743572221747674e-05,
-      "loss": 1.1223,
-      "step": 16050
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.570086273169899e-05,
-      "loss": 1.1261,
-      "step": 16060
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.5658153241650295e-05,
-      "loss": 1.1192,
-      "step": 16070
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.561544375160161e-05,
-      "loss": 1.122,
-      "step": 16080
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.557273426155292e-05,
-      "loss": 1.1257,
-      "step": 16090
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.5530024771504225e-05,
-      "loss": 1.1152,
-      "step": 16100
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.548731528145554e-05,
-      "loss": 1.1232,
-      "step": 16110
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.5444605791406853e-05,
-      "loss": 1.1247,
-      "step": 16120
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.540189630135816e-05,
-      "loss": 1.1147,
-      "step": 16130
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.5359186811309475e-05,
-      "loss": 1.1156,
-      "step": 16140
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.531647732126079e-05,
-      "loss": 1.1185,
-      "step": 16150
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.52737678312121e-05,
-      "loss": 1.1321,
-      "step": 16160
-    },
-    {
-      "epoch": 1.32,
-      "learning_rate": 3.523105834116341e-05,
-      "loss": 1.1127,
-      "step": 16170
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.518834885111472e-05,
-      "loss": 1.1121,
-      "step": 16180
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.5145639361066026e-05,
-      "loss": 1.1302,
-      "step": 16190
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.510292987101734e-05,
-      "loss": 1.1275,
-      "step": 16200
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.506022038096865e-05,
-      "loss": 1.1188,
-      "step": 16210
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.501751089091996e-05,
-      "loss": 1.1144,
-      "step": 16220
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.497480140087128e-05,
-      "loss": 1.1246,
-      "step": 16230
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.4932091910822584e-05,
-      "loss": 1.1225,
-      "step": 16240
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.48893824207739e-05,
-      "loss": 1.1213,
-      "step": 16250
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.484667293072521e-05,
-      "loss": 1.1202,
-      "step": 16260
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.480396344067652e-05,
-      "loss": 1.1258,
-      "step": 16270
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.476125395062783e-05,
-      "loss": 1.1129,
-      "step": 16280
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 3.471854446057914e-05,
-      "loss": 1.1199,
-      "step": 16290
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.467583497053045e-05,
-      "loss": 1.1146,
-      "step": 16300
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.4633125480481764e-05,
-      "loss": 1.1187,
-      "step": 16310
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.459041599043308e-05,
-      "loss": 1.1154,
-      "step": 16320
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.4547706500384386e-05,
-      "loss": 1.1107,
-      "step": 16330
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.45049970103357e-05,
-      "loss": 1.1112,
-      "step": 16340
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.4462287520287015e-05,
-      "loss": 1.1292,
-      "step": 16350
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.441957803023832e-05,
-      "loss": 1.117,
-      "step": 16360
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.437686854018963e-05,
-      "loss": 1.1187,
-      "step": 16370
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.433415905014094e-05,
-      "loss": 1.1154,
-      "step": 16380
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.429144956009225e-05,
-      "loss": 1.1268,
-      "step": 16390
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.4248740070043566e-05,
-      "loss": 1.1149,
-      "step": 16400
-    },
-    {
-      "epoch": 1.34,
-      "learning_rate": 3.420603057999487e-05,
-      "loss": 1.1121,
-      "step": 16410
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.416332108994619e-05,
-      "loss": 1.1137,
-      "step": 16420
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.41206115998975e-05,
-      "loss": 1.1109,
-      "step": 16430
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.407790210984881e-05,
-      "loss": 1.1191,
-      "step": 16440
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.4035192619800124e-05,
-      "loss": 1.1205,
-      "step": 16450
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.399248312975143e-05,
-      "loss": 1.1232,
-      "step": 16460
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.3949773639702746e-05,
-      "loss": 1.1254,
-      "step": 16470
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.390706414965405e-05,
-      "loss": 1.1147,
-      "step": 16480
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.386435465960537e-05,
-      "loss": 1.1158,
-      "step": 16490
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.3821645169556675e-05,
-      "loss": 1.1218,
-      "step": 16500
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.377893567950799e-05,
-      "loss": 1.1215,
-      "step": 16510
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.37362261894593e-05,
-      "loss": 1.1127,
-      "step": 16520
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.369351669941061e-05,
-      "loss": 1.1232,
-      "step": 16530
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 3.3650807209361925e-05,
-      "loss": 1.1181,
-      "step": 16540
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.360809771931323e-05,
-      "loss": 1.1133,
-      "step": 16550
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.356538822926455e-05,
-      "loss": 1.1207,
-      "step": 16560
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.3522678739215855e-05,
-      "loss": 1.1118,
-      "step": 16570
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.347996924916716e-05,
-      "loss": 1.1162,
-      "step": 16580
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.3437259759118476e-05,
-      "loss": 1.1118,
-      "step": 16590
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.339455026906979e-05,
-      "loss": 1.1205,
-      "step": 16600
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.33518407790211e-05,
-      "loss": 1.1288,
-      "step": 16610
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.330913128897241e-05,
-      "loss": 1.1231,
-      "step": 16620
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.326642179892373e-05,
-      "loss": 1.1185,
-      "step": 16630
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.3223712308875034e-05,
-      "loss": 1.1097,
-      "step": 16640
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.318100281882635e-05,
-      "loss": 1.1153,
-      "step": 16650
-    },
-    {
-      "epoch": 1.36,
-      "learning_rate": 3.3138293328777656e-05,
-      "loss": 1.1257,
-      "step": 16660
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.3095583838728964e-05,
-      "loss": 1.117,
-      "step": 16670
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.305287434868028e-05,
-      "loss": 1.1111,
-      "step": 16680
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.3010164858631586e-05,
-      "loss": 1.114,
-      "step": 16690
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.29674553685829e-05,
-      "loss": 1.1188,
-      "step": 16700
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.2924745878534214e-05,
-      "loss": 1.1187,
-      "step": 16710
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.288203638848552e-05,
-      "loss": 1.1096,
-      "step": 16720
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.2839326898436836e-05,
-      "loss": 1.1129,
-      "step": 16730
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.279661740838815e-05,
-      "loss": 1.1232,
-      "step": 16740
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.275390791833946e-05,
-      "loss": 1.1213,
-      "step": 16750
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.2711198428290765e-05,
-      "loss": 1.1205,
-      "step": 16760
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.266848893824208e-05,
-      "loss": 1.1112,
-      "step": 16770
-    },
-    {
-      "epoch": 1.37,
-      "learning_rate": 3.262577944819339e-05,
-      "loss": 1.12,
-      "step": 16780
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.25830699581447e-05,
-      "loss": 1.1149,
-      "step": 16790
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.2540360468096016e-05,
-      "loss": 1.1179,
-      "step": 16800
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.249765097804732e-05,
-      "loss": 1.1127,
-      "step": 16810
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.245494148799864e-05,
-      "loss": 1.1107,
-      "step": 16820
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.2412231997949945e-05,
-      "loss": 1.1094,
-      "step": 16830
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.236952250790126e-05,
-      "loss": 1.1158,
-      "step": 16840
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.232681301785257e-05,
-      "loss": 1.1184,
-      "step": 16850
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.2284103527803874e-05,
-      "loss": 1.119,
-      "step": 16860
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.224139403775519e-05,
-      "loss": 1.1251,
-      "step": 16870
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.21986845477065e-05,
-      "loss": 1.1202,
-      "step": 16880
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.215597505765781e-05,
-      "loss": 1.1122,
-      "step": 16890
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 3.2113265567609125e-05,
-      "loss": 1.112,
-      "step": 16900
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.207055607756044e-05,
-      "loss": 1.1261,
-      "step": 16910
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.202784658751175e-05,
-      "loss": 1.1131,
-      "step": 16920
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.198513709746306e-05,
-      "loss": 1.1183,
-      "step": 16930
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.194242760741437e-05,
-      "loss": 1.1116,
-      "step": 16940
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.1899718117365676e-05,
-      "loss": 1.1117,
-      "step": 16950
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.185700862731699e-05,
-      "loss": 1.1184,
-      "step": 16960
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.1814299137268305e-05,
-      "loss": 1.1185,
-      "step": 16970
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.177158964721961e-05,
-      "loss": 1.1124,
-      "step": 16980
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.1728880157170926e-05,
-      "loss": 1.124,
-      "step": 16990
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.1686170667122234e-05,
-      "loss": 1.1215,
-      "step": 17000
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.164346117707355e-05,
-      "loss": 1.1073,
-      "step": 17010
-    },
-    {
-      "epoch": 1.39,
-      "learning_rate": 3.160075168702486e-05,
-      "loss": 1.1123,
-      "step": 17020
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.155804219697617e-05,
-      "loss": 1.1243,
-      "step": 17030
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.151533270692748e-05,
-      "loss": 1.1139,
-      "step": 17040
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.147262321687879e-05,
-      "loss": 1.1185,
-      "step": 17050
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.14299137268301e-05,
-      "loss": 1.1189,
-      "step": 17060
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.1387204236781414e-05,
-      "loss": 1.1203,
-      "step": 17070
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.134449474673273e-05,
-      "loss": 1.119,
-      "step": 17080
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.1301785256684036e-05,
-      "loss": 1.1162,
-      "step": 17090
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.125907576663535e-05,
-      "loss": 1.1052,
-      "step": 17100
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.1216366276586664e-05,
-      "loss": 1.112,
-      "step": 17110
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.117365678653797e-05,
-      "loss": 1.1259,
-      "step": 17120
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.113094729648928e-05,
-      "loss": 1.1113,
-      "step": 17130
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.108823780644059e-05,
-      "loss": 1.1158,
-      "step": 17140
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 3.10455283163919e-05,
-      "loss": 1.1188,
-      "step": 17150
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.1002818826343215e-05,
-      "loss": 1.1138,
-      "step": 17160
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.096010933629452e-05,
-      "loss": 1.1179,
-      "step": 17170
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.091739984624584e-05,
-      "loss": 1.1151,
-      "step": 17180
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.087469035619715e-05,
-      "loss": 1.1123,
-      "step": 17190
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.083198086614846e-05,
-      "loss": 1.1198,
-      "step": 17200
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.078927137609977e-05,
-      "loss": 1.1124,
-      "step": 17210
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.074656188605108e-05,
-      "loss": 1.1222,
-      "step": 17220
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.070385239600239e-05,
-      "loss": 1.1173,
-      "step": 17230
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.06611429059537e-05,
-      "loss": 1.1185,
-      "step": 17240
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.061843341590502e-05,
-      "loss": 1.1102,
-      "step": 17250
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.0575723925856325e-05,
-      "loss": 1.1181,
-      "step": 17260
-    },
-    {
-      "epoch": 1.41,
-      "learning_rate": 3.053301443580764e-05,
-      "loss": 1.1147,
-      "step": 17270
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.0490304945758953e-05,
-      "loss": 1.1124,
-      "step": 17280
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.044759545571026e-05,
-      "loss": 1.1038,
-      "step": 17290
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.040488596566157e-05,
-      "loss": 1.1208,
-      "step": 17300
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.036217647561288e-05,
-      "loss": 1.1036,
-      "step": 17310
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.0319466985564193e-05,
-      "loss": 1.1163,
-      "step": 17320
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.0276757495515508e-05,
-      "loss": 1.1179,
-      "step": 17330
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.0234048005466815e-05,
-      "loss": 1.1134,
-      "step": 17340
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.0191338515418126e-05,
-      "loss": 1.1128,
-      "step": 17350
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.014862902536944e-05,
-      "loss": 1.1039,
-      "step": 17360
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.0105919535320748e-05,
-      "loss": 1.1056,
-      "step": 17370
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.0063210045272062e-05,
-      "loss": 1.1105,
-      "step": 17380
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 3.0020500555223373e-05,
-      "loss": 1.1024,
-      "step": 17390
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.997779106517468e-05,
-      "loss": 1.1137,
-      "step": 17400
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.9935081575125995e-05,
-      "loss": 1.1145,
-      "step": 17410
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.989237208507731e-05,
-      "loss": 1.1135,
-      "step": 17420
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.9849662595028617e-05,
-      "loss": 1.1105,
-      "step": 17430
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.9806953104979928e-05,
-      "loss": 1.1114,
-      "step": 17440
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.9764243614931235e-05,
-      "loss": 1.1267,
-      "step": 17450
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.972153412488255e-05,
-      "loss": 1.117,
-      "step": 17460
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.9678824634833864e-05,
-      "loss": 1.1119,
-      "step": 17470
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.963611514478517e-05,
-      "loss": 1.1091,
-      "step": 17480
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.9593405654736482e-05,
-      "loss": 1.1151,
-      "step": 17490
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.9550696164687797e-05,
-      "loss": 1.1206,
-      "step": 17500
-    },
-    {
-      "epoch": 1.43,
-      "learning_rate": 2.9507986674639104e-05,
-      "loss": 1.1099,
-      "step": 17510
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.946527718459042e-05,
-      "loss": 1.1222,
-      "step": 17520
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.942256769454173e-05,
-      "loss": 1.1135,
-      "step": 17530
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.9379858204493037e-05,
-      "loss": 1.117,
-      "step": 17540
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.933714871444435e-05,
-      "loss": 1.1089,
-      "step": 17550
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.9294439224395665e-05,
-      "loss": 1.1145,
-      "step": 17560
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.9251729734346973e-05,
-      "loss": 1.1186,
-      "step": 17570
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.9209020244298284e-05,
-      "loss": 1.1144,
-      "step": 17580
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.9166310754249598e-05,
-      "loss": 1.1171,
-      "step": 17590
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.9123601264200906e-05,
-      "loss": 1.107,
-      "step": 17600
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.908089177415222e-05,
-      "loss": 1.1105,
-      "step": 17610
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.9038182284103528e-05,
-      "loss": 1.1246,
-      "step": 17620
-    },
-    {
-      "epoch": 1.44,
-      "learning_rate": 2.899547279405484e-05,
-      "loss": 1.115,
-      "step": 17630
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.8952763304006153e-05,
-      "loss": 1.1168,
-      "step": 17640
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.891005381395746e-05,
-      "loss": 1.118,
-      "step": 17650
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.8867344323908775e-05,
-      "loss": 1.1048,
-      "step": 17660
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.8824634833860085e-05,
-      "loss": 1.1088,
-      "step": 17670
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.8781925343811393e-05,
-      "loss": 1.1135,
-      "step": 17680
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.8739215853762707e-05,
-      "loss": 1.1169,
-      "step": 17690
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.869650636371402e-05,
-      "loss": 1.1084,
-      "step": 17700
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.865379687366533e-05,
-      "loss": 1.1058,
-      "step": 17710
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.861108738361664e-05,
-      "loss": 1.1105,
-      "step": 17720
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.8568377893567954e-05,
-      "loss": 1.1192,
-      "step": 17730
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.8525668403519262e-05,
-      "loss": 1.1142,
-      "step": 17740
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.8482958913470576e-05,
-      "loss": 1.1164,
-      "step": 17750
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 2.8440249423421884e-05,
-      "loss": 1.1065,
-      "step": 17760
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.8397539933373195e-05,
-      "loss": 1.1172,
-      "step": 17770
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.835483044332451e-05,
-      "loss": 1.1192,
-      "step": 17780
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.8312120953275816e-05,
-      "loss": 1.1058,
-      "step": 17790
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.826941146322713e-05,
-      "loss": 1.1085,
-      "step": 17800
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.822670197317844e-05,
-      "loss": 1.1169,
-      "step": 17810
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.818399248312975e-05,
-      "loss": 1.1174,
-      "step": 17820
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.8141282993081063e-05,
-      "loss": 1.109,
-      "step": 17830
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.8098573503032378e-05,
-      "loss": 1.1121,
-      "step": 17840
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.8055864012983685e-05,
-      "loss": 1.0997,
-      "step": 17850
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.8013154522934996e-05,
-      "loss": 1.1059,
-      "step": 17860
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.797044503288631e-05,
-      "loss": 1.1062,
-      "step": 17870
-    },
-    {
-      "epoch": 1.46,
-      "learning_rate": 2.7927735542837618e-05,
-      "loss": 1.1165,
-      "step": 17880
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.7885026052788932e-05,
-      "loss": 1.114,
-      "step": 17890
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.7842316562740243e-05,
-      "loss": 1.11,
-      "step": 17900
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.7799607072691554e-05,
-      "loss": 1.1094,
-      "step": 17910
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.7756897582642865e-05,
-      "loss": 1.1086,
-      "step": 17920
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.7714188092594173e-05,
-      "loss": 1.1226,
-      "step": 17930
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.7671478602545487e-05,
-      "loss": 1.1116,
-      "step": 17940
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.76287691124968e-05,
-      "loss": 1.1158,
-      "step": 17950
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.758605962244811e-05,
-      "loss": 1.1175,
-      "step": 17960
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.754335013239942e-05,
-      "loss": 1.1127,
-      "step": 17970
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.7500640642350734e-05,
-      "loss": 1.1082,
-      "step": 17980
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.745793115230204e-05,
-      "loss": 1.1068,
-      "step": 17990
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 2.7415221662253356e-05,
-      "loss": 1.1138,
-      "step": 18000
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.7372512172204667e-05,
-      "loss": 1.1162,
-      "step": 18010
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.7329802682155974e-05,
-      "loss": 1.1159,
-      "step": 18020
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.728709319210729e-05,
-      "loss": 1.1101,
-      "step": 18030
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.7244383702058603e-05,
-      "loss": 1.1114,
-      "step": 18040
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.720167421200991e-05,
-      "loss": 1.1113,
-      "step": 18050
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.715896472196122e-05,
-      "loss": 1.1213,
-      "step": 18060
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.711625523191253e-05,
-      "loss": 1.1101,
-      "step": 18070
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.7073545741863843e-05,
-      "loss": 1.1076,
-      "step": 18080
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.7030836251815157e-05,
-      "loss": 1.11,
-      "step": 18090
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.6988126761766465e-05,
-      "loss": 1.1087,
-      "step": 18100
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.6945417271717776e-05,
-      "loss": 1.113,
-      "step": 18110
-    },
-    {
-      "epoch": 1.48,
-      "learning_rate": 2.690270778166909e-05,
-      "loss": 1.1124,
-      "step": 18120
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.6859998291620398e-05,
-      "loss": 1.1082,
-      "step": 18130
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.6817288801571712e-05,
-      "loss": 1.1087,
-      "step": 18140
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.6774579311523023e-05,
-      "loss": 1.1132,
-      "step": 18150
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.673186982147433e-05,
-      "loss": 1.1111,
-      "step": 18160
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.6689160331425645e-05,
-      "loss": 1.1062,
-      "step": 18170
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.664645084137696e-05,
-      "loss": 1.1047,
-      "step": 18180
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.6603741351328266e-05,
-      "loss": 1.1039,
-      "step": 18190
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.6561031861279577e-05,
-      "loss": 1.107,
-      "step": 18200
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.651832237123089e-05,
-      "loss": 1.1132,
-      "step": 18210
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.64756128811822e-05,
-      "loss": 1.115,
-      "step": 18220
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.6432903391133513e-05,
-      "loss": 1.1139,
-      "step": 18230
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 2.639019390108482e-05,
-      "loss": 1.1099,
-      "step": 18240
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.6347484411036132e-05,
-      "loss": 1.1137,
-      "step": 18250
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.6304774920987446e-05,
-      "loss": 1.1043,
-      "step": 18260
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.6262065430938754e-05,
-      "loss": 1.1034,
-      "step": 18270
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.6219355940890068e-05,
-      "loss": 1.1084,
-      "step": 18280
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.617664645084138e-05,
-      "loss": 1.1132,
-      "step": 18290
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.6133936960792686e-05,
-      "loss": 1.1056,
-      "step": 18300
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.6091227470744e-05,
-      "loss": 1.1174,
-      "step": 18310
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.6048517980695315e-05,
-      "loss": 1.1026,
-      "step": 18320
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.6005808490646623e-05,
-      "loss": 1.1058,
-      "step": 18330
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.5963099000597934e-05,
-      "loss": 1.1183,
-      "step": 18340
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.5920389510549248e-05,
-      "loss": 1.1111,
-      "step": 18350
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.5877680020500555e-05,
-      "loss": 1.1087,
-      "step": 18360
-    },
-    {
-      "epoch": 1.5,
-      "learning_rate": 2.583497053045187e-05,
-      "loss": 1.1042,
-      "step": 18370
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.5792261040403177e-05,
-      "loss": 1.1062,
-      "step": 18380
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.5749551550354488e-05,
-      "loss": 1.106,
-      "step": 18390
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.5706842060305802e-05,
-      "loss": 1.1193,
-      "step": 18400
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.566413257025711e-05,
-      "loss": 1.1126,
-      "step": 18410
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.5621423080208424e-05,
-      "loss": 1.1056,
-      "step": 18420
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.5578713590159735e-05,
-      "loss": 1.1103,
-      "step": 18430
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.5536004100111043e-05,
-      "loss": 1.1099,
-      "step": 18440
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.5493294610062357e-05,
-      "loss": 1.1073,
-      "step": 18450
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.545058512001367e-05,
-      "loss": 1.1032,
-      "step": 18460
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.540787562996498e-05,
-      "loss": 1.1172,
-      "step": 18470
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.536516613991629e-05,
-      "loss": 1.1118,
-      "step": 18480
-    },
-    {
-      "epoch": 1.51,
-      "learning_rate": 2.5322456649867604e-05,
-      "loss": 1.1071,
-      "step": 18490
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.527974715981891e-05,
-      "loss": 1.1084,
-      "step": 18500
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.5237037669770226e-05,
-      "loss": 1.1125,
-      "step": 18510
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.5194328179721537e-05,
-      "loss": 1.1159,
-      "step": 18520
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.5151618689672844e-05,
-      "loss": 1.104,
-      "step": 18530
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.510890919962416e-05,
-      "loss": 1.11,
-      "step": 18540
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.5066199709575466e-05,
-      "loss": 1.1037,
-      "step": 18550
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.502349021952678e-05,
-      "loss": 1.1041,
-      "step": 18560
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.498078072947809e-05,
-      "loss": 1.1094,
-      "step": 18570
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.4938071239429402e-05,
-      "loss": 1.102,
-      "step": 18580
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.4895361749380713e-05,
-      "loss": 1.1188,
-      "step": 18590
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.4852652259332024e-05,
-      "loss": 1.1087,
-      "step": 18600
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 2.480994276928334e-05,
-      "loss": 1.1141,
-      "step": 18610
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.4767233279234646e-05,
-      "loss": 1.1095,
-      "step": 18620
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.4724523789185957e-05,
-      "loss": 1.1053,
-      "step": 18630
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.468181429913727e-05,
-      "loss": 1.1151,
-      "step": 18640
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.4639104809088582e-05,
-      "loss": 1.1081,
-      "step": 18650
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.4596395319039893e-05,
-      "loss": 1.1107,
-      "step": 18660
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.45536858289912e-05,
-      "loss": 1.1031,
-      "step": 18670
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.4510976338942515e-05,
-      "loss": 1.1089,
-      "step": 18680
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.4468266848893826e-05,
-      "loss": 1.1023,
-      "step": 18690
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.4425557358845137e-05,
-      "loss": 1.1087,
-      "step": 18700
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.4382847868796447e-05,
-      "loss": 1.1147,
-      "step": 18710
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.434013837874776e-05,
-      "loss": 1.1169,
-      "step": 18720
-    },
-    {
-      "epoch": 1.53,
-      "learning_rate": 2.429742888869907e-05,
-      "loss": 1.1149,
-      "step": 18730
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.425471939865038e-05,
-      "loss": 1.117,
-      "step": 18740
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.4212009908601694e-05,
-      "loss": 1.112,
-      "step": 18750
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.4169300418553002e-05,
-      "loss": 1.1151,
-      "step": 18760
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.4126590928504313e-05,
-      "loss": 1.1103,
-      "step": 18770
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.4083881438455627e-05,
-      "loss": 1.1064,
-      "step": 18780
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.4041171948406938e-05,
-      "loss": 1.1148,
-      "step": 18790
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.399846245835825e-05,
-      "loss": 1.1005,
-      "step": 18800
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.395575296830956e-05,
-      "loss": 1.1166,
-      "step": 18810
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.391304347826087e-05,
-      "loss": 1.1067,
-      "step": 18820
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.3870333988212182e-05,
-      "loss": 1.1053,
-      "step": 18830
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.3827624498163493e-05,
-      "loss": 1.1001,
-      "step": 18840
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 2.3784915008114807e-05,
-      "loss": 1.1138,
-      "step": 18850
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.3742205518066115e-05,
-      "loss": 1.1036,
-      "step": 18860
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.3699496028017425e-05,
-      "loss": 1.1159,
-      "step": 18870
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.3656786537968736e-05,
-      "loss": 1.094,
-      "step": 18880
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.361407704792005e-05,
-      "loss": 1.1088,
-      "step": 18890
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.357136755787136e-05,
-      "loss": 1.1032,
-      "step": 18900
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.352865806782267e-05,
-      "loss": 1.1068,
-      "step": 18910
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.3485948577773983e-05,
-      "loss": 1.1078,
-      "step": 18920
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.3443239087725294e-05,
-      "loss": 1.1105,
-      "step": 18930
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.3400529597676605e-05,
-      "loss": 1.1181,
-      "step": 18940
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.3357820107627916e-05,
-      "loss": 1.1112,
-      "step": 18950
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.3315110617579227e-05,
-      "loss": 1.1067,
-      "step": 18960
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.3272401127530538e-05,
-      "loss": 1.1107,
-      "step": 18970
-    },
-    {
-      "epoch": 1.55,
-      "learning_rate": 2.322969163748185e-05,
-      "loss": 1.1173,
-      "step": 18980
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.3186982147433163e-05,
-      "loss": 1.1128,
-      "step": 18990
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.314427265738447e-05,
-      "loss": 1.0996,
-      "step": 19000
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.310156316733578e-05,
-      "loss": 1.1112,
-      "step": 19010
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.3058853677287096e-05,
-      "loss": 1.1143,
-      "step": 19020
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.3016144187238407e-05,
-      "loss": 1.1077,
-      "step": 19030
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.2973434697189718e-05,
-      "loss": 1.1066,
-      "step": 19040
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.2930725207141025e-05,
-      "loss": 1.0992,
-      "step": 19050
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.288801571709234e-05,
-      "loss": 1.0956,
-      "step": 19060
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.284530622704365e-05,
-      "loss": 1.1105,
-      "step": 19070
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.280259673699496e-05,
-      "loss": 1.1171,
-      "step": 19080
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.2759887246946272e-05,
-      "loss": 1.1019,
-      "step": 19090
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 2.2717177756897583e-05,
-      "loss": 1.106,
-      "step": 19100
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.2674468266848894e-05,
-      "loss": 1.1072,
-      "step": 19110
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.2631758776800205e-05,
-      "loss": 1.1196,
-      "step": 19120
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.258904928675152e-05,
-      "loss": 1.107,
-      "step": 19130
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.2546339796702827e-05,
-      "loss": 1.1087,
-      "step": 19140
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.2503630306654138e-05,
-      "loss": 1.1063,
-      "step": 19150
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.2460920816605452e-05,
-      "loss": 1.1132,
-      "step": 19160
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.2418211326556763e-05,
-      "loss": 1.1114,
-      "step": 19170
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.2375501836508074e-05,
-      "loss": 1.109,
-      "step": 19180
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.233279234645938e-05,
-      "loss": 1.1014,
-      "step": 19190
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.2290082856410696e-05,
-      "loss": 1.1044,
-      "step": 19200
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.2247373366362007e-05,
-      "loss": 1.1107,
-      "step": 19210
-    },
-    {
-      "epoch": 1.57,
-      "learning_rate": 2.2204663876313318e-05,
-      "loss": 1.1054,
-      "step": 19220
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.216195438626463e-05,
-      "loss": 1.0985,
-      "step": 19230
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.211924489621594e-05,
-      "loss": 1.1065,
-      "step": 19240
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.207653540616725e-05,
-      "loss": 1.1068,
-      "step": 19250
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.2033825916118565e-05,
-      "loss": 1.1057,
-      "step": 19260
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.1991116426069875e-05,
-      "loss": 1.1053,
-      "step": 19270
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.1948406936021186e-05,
-      "loss": 1.0977,
-      "step": 19280
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.1905697445972494e-05,
-      "loss": 1.1127,
-      "step": 19290
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.1862987955923808e-05,
-      "loss": 1.113,
-      "step": 19300
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.182027846587512e-05,
-      "loss": 1.0974,
-      "step": 19310
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.177756897582643e-05,
-      "loss": 1.1114,
-      "step": 19320
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.173485948577774e-05,
-      "loss": 1.1067,
-      "step": 19330
-    },
-    {
-      "epoch": 1.58,
-      "learning_rate": 2.1692149995729052e-05,
-      "loss": 1.1057,
-      "step": 19340
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.1649440505680363e-05,
-      "loss": 1.106,
-      "step": 19350
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.1606731015631674e-05,
-      "loss": 1.1003,
-      "step": 19360
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.1564021525582988e-05,
-      "loss": 1.11,
-      "step": 19370
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.1521312035534295e-05,
-      "loss": 1.101,
-      "step": 19380
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.1478602545485606e-05,
-      "loss": 1.1058,
-      "step": 19390
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.143589305543692e-05,
-      "loss": 1.1131,
-      "step": 19400
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.139318356538823e-05,
-      "loss": 1.1186,
-      "step": 19410
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.1350474075339543e-05,
-      "loss": 1.1055,
-      "step": 19420
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.130776458529085e-05,
-      "loss": 1.1108,
-      "step": 19430
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.1265055095242164e-05,
-      "loss": 1.106,
-      "step": 19440
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.1222345605193475e-05,
-      "loss": 1.1041,
-      "step": 19450
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.1179636115144786e-05,
-      "loss": 1.1034,
-      "step": 19460
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 2.1136926625096097e-05,
-      "loss": 1.1095,
-      "step": 19470
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.1094217135047408e-05,
-      "loss": 1.0965,
-      "step": 19480
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.105150764499872e-05,
-      "loss": 1.1079,
-      "step": 19490
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.100879815495003e-05,
-      "loss": 1.1108,
-      "step": 19500
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.0966088664901344e-05,
-      "loss": 1.1035,
-      "step": 19510
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.092337917485265e-05,
-      "loss": 1.1039,
-      "step": 19520
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.0880669684803963e-05,
-      "loss": 1.099,
-      "step": 19530
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.0837960194755277e-05,
-      "loss": 1.1133,
-      "step": 19540
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.0795250704706588e-05,
-      "loss": 1.1105,
-      "step": 19550
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.07525412146579e-05,
-      "loss": 1.0971,
-      "step": 19560
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.070983172460921e-05,
-      "loss": 1.1137,
-      "step": 19570
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.066712223456052e-05,
-      "loss": 1.1067,
-      "step": 19580
-    },
-    {
-      "epoch": 1.6,
-      "learning_rate": 2.062441274451183e-05,
-      "loss": 1.102,
-      "step": 19590
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.0581703254463142e-05,
-      "loss": 1.1063,
-      "step": 19600
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.0538993764414453e-05,
-      "loss": 1.1024,
-      "step": 19610
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.0496284274365764e-05,
-      "loss": 1.1123,
-      "step": 19620
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.0453574784317075e-05,
-      "loss": 1.1054,
-      "step": 19630
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.041086529426839e-05,
-      "loss": 1.1081,
-      "step": 19640
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.03681558042197e-05,
-      "loss": 1.1058,
-      "step": 19650
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.0325446314171008e-05,
-      "loss": 1.1047,
-      "step": 19660
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.028273682412232e-05,
-      "loss": 1.0948,
-      "step": 19670
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.0240027334073633e-05,
-      "loss": 1.1048,
-      "step": 19680
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.0197317844024944e-05,
-      "loss": 1.1064,
-      "step": 19690
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.0154608353976255e-05,
-      "loss": 1.1056,
-      "step": 19700
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 2.0111898863927566e-05,
-      "loss": 1.1035,
-      "step": 19710
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 2.0069189373878877e-05,
-      "loss": 1.1056,
-      "step": 19720
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 2.0026479883830188e-05,
-      "loss": 1.1043,
-      "step": 19730
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 1.99837703937815e-05,
-      "loss": 1.1097,
-      "step": 19740
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 1.9941060903732813e-05,
-      "loss": 1.1131,
-      "step": 19750
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 1.989835141368412e-05,
-      "loss": 1.1001,
-      "step": 19760
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 1.985564192363543e-05,
-      "loss": 1.1011,
-      "step": 19770
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 1.9812932433586746e-05,
-      "loss": 1.1145,
-      "step": 19780
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 1.9770222943538056e-05,
-      "loss": 1.1131,
-      "step": 19790
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 1.9727513453489367e-05,
-      "loss": 1.1009,
-      "step": 19800
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 1.9684803963440675e-05,
-      "loss": 1.1,
-      "step": 19810
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 1.964209447339199e-05,
-      "loss": 1.1063,
-      "step": 19820
-    },
-    {
-      "epoch": 1.62,
-      "learning_rate": 1.95993849833433e-05,
-      "loss": 1.1103,
-      "step": 19830
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.955667549329461e-05,
-      "loss": 1.1042,
-      "step": 19840
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.9513966003245922e-05,
-      "loss": 1.1021,
-      "step": 19850
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.9471256513197233e-05,
-      "loss": 1.1051,
-      "step": 19860
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.9428547023148544e-05,
-      "loss": 1.1074,
-      "step": 19870
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.9385837533099858e-05,
-      "loss": 1.107,
-      "step": 19880
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.934312804305117e-05,
-      "loss": 1.1041,
-      "step": 19890
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.9300418553002476e-05,
-      "loss": 1.1011,
-      "step": 19900
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.9257709062953787e-05,
-      "loss": 1.1057,
-      "step": 19910
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.92149995729051e-05,
-      "loss": 1.1056,
-      "step": 19920
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.9172290082856413e-05,
-      "loss": 1.1037,
-      "step": 19930
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.9129580592807724e-05,
-      "loss": 1.1121,
-      "step": 19940
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 1.9086871102759034e-05,
-      "loss": 1.1099,
-      "step": 19950
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.9044161612710345e-05,
-      "loss": 1.1067,
-      "step": 19960
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.9001452122661656e-05,
-      "loss": 1.1123,
-      "step": 19970
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.8958742632612967e-05,
-      "loss": 1.109,
-      "step": 19980
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.8916033142564278e-05,
-      "loss": 1.1019,
-      "step": 19990
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.887332365251559e-05,
-      "loss": 1.1069,
-      "step": 20000
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.88306141624669e-05,
-      "loss": 1.1064,
-      "step": 20010
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.8787904672418214e-05,
-      "loss": 1.1084,
-      "step": 20020
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.8745195182369525e-05,
-      "loss": 1.1083,
-      "step": 20030
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.8702485692320833e-05,
-      "loss": 1.1034,
-      "step": 20040
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.8659776202272144e-05,
-      "loss": 1.0958,
-      "step": 20050
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.8617066712223458e-05,
-      "loss": 1.0979,
-      "step": 20060
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.857435722217477e-05,
-      "loss": 1.0991,
-      "step": 20070
-    },
-    {
-      "epoch": 1.64,
-      "learning_rate": 1.853164773212608e-05,
-      "loss": 1.0949,
-      "step": 20080
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.848893824207739e-05,
-      "loss": 1.1008,
-      "step": 20090
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.84462287520287e-05,
-      "loss": 1.1072,
-      "step": 20100
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.8403519261980012e-05,
-      "loss": 1.1048,
-      "step": 20110
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.8360809771931323e-05,
-      "loss": 1.1097,
-      "step": 20120
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.8318100281882634e-05,
-      "loss": 1.1086,
-      "step": 20130
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.8275390791833945e-05,
-      "loss": 1.1041,
-      "step": 20140
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.8232681301785256e-05,
-      "loss": 1.1037,
-      "step": 20150
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.818997181173657e-05,
-      "loss": 1.1079,
-      "step": 20160
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.814726232168788e-05,
-      "loss": 1.1017,
-      "step": 20170
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.810455283163919e-05,
-      "loss": 1.114,
-      "step": 20180
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.8061843341590503e-05,
-      "loss": 1.0977,
-      "step": 20190
-    },
-    {
-      "epoch": 1.65,
-      "learning_rate": 1.8019133851541814e-05,
-      "loss": 1.1096,
-      "step": 20200
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.7976424361493125e-05,
-      "loss": 1.1036,
-      "step": 20210
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.7933714871444436e-05,
-      "loss": 1.0987,
-      "step": 20220
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.7891005381395747e-05,
-      "loss": 1.1054,
-      "step": 20230
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.7848295891347058e-05,
-      "loss": 1.1007,
-      "step": 20240
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.780558640129837e-05,
-      "loss": 1.1022,
-      "step": 20250
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.7762876911249683e-05,
-      "loss": 1.1065,
-      "step": 20260
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.7720167421200994e-05,
-      "loss": 1.1072,
-      "step": 20270
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.76774579311523e-05,
-      "loss": 1.0954,
-      "step": 20280
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.7634748441103612e-05,
-      "loss": 1.1048,
-      "step": 20290
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.7592038951054927e-05,
-      "loss": 1.1056,
-      "step": 20300
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.7549329461006237e-05,
-      "loss": 1.1015,
-      "step": 20310
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 1.750661997095755e-05,
-      "loss": 1.1068,
-      "step": 20320
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.746391048090886e-05,
-      "loss": 1.0936,
-      "step": 20330
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.742120099086017e-05,
-      "loss": 1.1084,
-      "step": 20340
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.737849150081148e-05,
-      "loss": 1.0986,
-      "step": 20350
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.7335782010762792e-05,
-      "loss": 1.104,
-      "step": 20360
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.7293072520714103e-05,
-      "loss": 1.1,
-      "step": 20370
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.7250363030665414e-05,
-      "loss": 1.1048,
-      "step": 20380
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.7207653540616725e-05,
-      "loss": 1.0993,
-      "step": 20390
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.716494405056804e-05,
-      "loss": 1.0935,
-      "step": 20400
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.712223456051935e-05,
-      "loss": 1.108,
-      "step": 20410
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.7079525070470657e-05,
-      "loss": 1.0976,
-      "step": 20420
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.703681558042197e-05,
-      "loss": 1.1066,
-      "step": 20430
-    },
-    {
-      "epoch": 1.67,
-      "learning_rate": 1.6994106090373283e-05,
-      "loss": 1.1034,
-      "step": 20440
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.6951396600324594e-05,
-      "loss": 1.1087,
-      "step": 20450
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.6908687110275905e-05,
-      "loss": 1.1021,
-      "step": 20460
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.6865977620227215e-05,
-      "loss": 1.1024,
-      "step": 20470
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.6823268130178526e-05,
-      "loss": 1.1063,
-      "step": 20480
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.6780558640129837e-05,
-      "loss": 1.1011,
-      "step": 20490
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.673784915008115e-05,
-      "loss": 1.0945,
-      "step": 20500
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.669513966003246e-05,
-      "loss": 1.0972,
-      "step": 20510
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.665243016998377e-05,
-      "loss": 1.1006,
-      "step": 20520
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.660972067993508e-05,
-      "loss": 1.1045,
-      "step": 20530
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.6567011189886395e-05,
-      "loss": 1.104,
-      "step": 20540
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.6524301699837706e-05,
-      "loss": 1.1026,
-      "step": 20550
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 1.6481592209789014e-05,
-      "loss": 1.1063,
-      "step": 20560
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.6438882719740328e-05,
-      "loss": 1.1033,
-      "step": 20570
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.639617322969164e-05,
-      "loss": 1.093,
-      "step": 20580
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.635346373964295e-05,
-      "loss": 1.1008,
-      "step": 20590
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.631075424959426e-05,
-      "loss": 1.1047,
-      "step": 20600
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.626804475954557e-05,
-      "loss": 1.0996,
-      "step": 20610
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.6225335269496882e-05,
-      "loss": 1.108,
-      "step": 20620
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.6182625779448193e-05,
-      "loss": 1.1041,
-      "step": 20630
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.6139916289399508e-05,
-      "loss": 1.096,
-      "step": 20640
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.6097206799350815e-05,
-      "loss": 1.0904,
-      "step": 20650
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.6054497309302126e-05,
-      "loss": 1.1021,
-      "step": 20660
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.6011787819253437e-05,
-      "loss": 1.1042,
-      "step": 20670
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.596907832920475e-05,
-      "loss": 1.107,
-      "step": 20680
-    },
-    {
-      "epoch": 1.69,
-      "learning_rate": 1.5926368839156062e-05,
-      "loss": 1.0975,
-      "step": 20690
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.5883659349107373e-05,
-      "loss": 1.1032,
-      "step": 20700
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.5840949859058684e-05,
-      "loss": 1.1146,
-      "step": 20710
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.5798240369009995e-05,
-      "loss": 1.0973,
-      "step": 20720
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.5755530878961306e-05,
-      "loss": 1.105,
-      "step": 20730
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.5712821388912617e-05,
-      "loss": 1.1082,
-      "step": 20740
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.5670111898863928e-05,
-      "loss": 1.1024,
-      "step": 20750
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.562740240881524e-05,
-      "loss": 1.103,
-      "step": 20760
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.558469291876655e-05,
-      "loss": 1.097,
-      "step": 20770
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.5541983428717864e-05,
-      "loss": 1.0996,
-      "step": 20780
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.5499273938669175e-05,
-      "loss": 1.1077,
-      "step": 20790
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.5456564448620482e-05,
-      "loss": 1.1033,
-      "step": 20800
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 1.5413854958571797e-05,
-      "loss": 1.1021,
-      "step": 20810
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.5371145468523108e-05,
-      "loss": 1.0975,
-      "step": 20820
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.532843597847442e-05,
-      "loss": 1.0981,
-      "step": 20830
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.528572648842573e-05,
-      "loss": 1.1019,
-      "step": 20840
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.524301699837704e-05,
-      "loss": 1.1089,
-      "step": 20850
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.5200307508328351e-05,
-      "loss": 1.1111,
-      "step": 20860
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.5157598018279662e-05,
-      "loss": 1.1091,
-      "step": 20870
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.5114888528230975e-05,
-      "loss": 1.1001,
-      "step": 20880
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.5072179038182286e-05,
-      "loss": 1.0987,
-      "step": 20890
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.5029469548133595e-05,
-      "loss": 1.107,
-      "step": 20900
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.4986760058084906e-05,
-      "loss": 1.0973,
-      "step": 20910
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.494405056803622e-05,
-      "loss": 1.0978,
-      "step": 20920
-    },
-    {
-      "epoch": 1.71,
-      "learning_rate": 1.490134107798753e-05,
-      "loss": 1.0993,
-      "step": 20930
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.485863158793884e-05,
-      "loss": 1.093,
-      "step": 20940
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.4815922097890153e-05,
-      "loss": 1.1091,
-      "step": 20950
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.4773212607841464e-05,
-      "loss": 1.1066,
-      "step": 20960
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.4730503117792775e-05,
-      "loss": 1.1056,
-      "step": 20970
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.4687793627744084e-05,
-      "loss": 1.101,
-      "step": 20980
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.4645084137695398e-05,
-      "loss": 1.1034,
-      "step": 20990
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.4602374647646707e-05,
-      "loss": 1.0953,
-      "step": 21000
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.4559665157598018e-05,
-      "loss": 1.0946,
-      "step": 21010
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.451695566754933e-05,
-      "loss": 1.0985,
-      "step": 21020
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.4474246177500642e-05,
-      "loss": 1.088,
-      "step": 21030
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.4431536687451953e-05,
-      "loss": 1.0984,
-      "step": 21040
-    },
-    {
-      "epoch": 1.72,
-      "learning_rate": 1.4388827197403262e-05,
-      "loss": 1.0939,
-      "step": 21050
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.4346117707354576e-05,
-      "loss": 1.1039,
-      "step": 21060
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.4303408217305885e-05,
-      "loss": 1.0896,
-      "step": 21070
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.4260698727257196e-05,
-      "loss": 1.1071,
-      "step": 21080
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.4217989237208509e-05,
-      "loss": 1.1027,
-      "step": 21090
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.417527974715982e-05,
-      "loss": 1.0989,
-      "step": 21100
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.413257025711113e-05,
-      "loss": 1.0983,
-      "step": 21110
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.4089860767062443e-05,
-      "loss": 1.0997,
-      "step": 21120
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.4047151277013754e-05,
-      "loss": 1.0958,
-      "step": 21130
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.4004441786965063e-05,
-      "loss": 1.1011,
-      "step": 21140
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.3961732296916374e-05,
-      "loss": 1.0983,
-      "step": 21150
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.3919022806867687e-05,
-      "loss": 1.1066,
-      "step": 21160
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 1.3876313316818998e-05,
-      "loss": 1.0998,
-      "step": 21170
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.3833603826770309e-05,
-      "loss": 1.105,
-      "step": 21180
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.3790894336721621e-05,
-      "loss": 1.1007,
-      "step": 21190
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.3748184846672932e-05,
-      "loss": 1.1024,
-      "step": 21200
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.3705475356624242e-05,
-      "loss": 1.0984,
-      "step": 21210
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.3662765866575552e-05,
-      "loss": 1.0997,
-      "step": 21220
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.3620056376526865e-05,
-      "loss": 1.113,
-      "step": 21230
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.3577346886478176e-05,
-      "loss": 1.0983,
-      "step": 21240
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.3534637396429487e-05,
-      "loss": 1.0967,
-      "step": 21250
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.34919279063808e-05,
-      "loss": 1.1054,
-      "step": 21260
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.344921841633211e-05,
-      "loss": 1.105,
-      "step": 21270
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.340650892628342e-05,
-      "loss": 1.0989,
-      "step": 21280
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.336379943623473e-05,
-      "loss": 1.1016,
-      "step": 21290
-    },
-    {
-      "epoch": 1.74,
-      "learning_rate": 1.3321089946186043e-05,
-      "loss": 1.0951,
-      "step": 21300
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.3278380456137354e-05,
-      "loss": 1.1042,
-      "step": 21310
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.3235670966088665e-05,
-      "loss": 1.0997,
-      "step": 21320
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.3192961476039978e-05,
-      "loss": 1.1112,
-      "step": 21330
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.3150251985991288e-05,
-      "loss": 1.0963,
-      "step": 21340
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.3107542495942598e-05,
-      "loss": 1.1119,
-      "step": 21350
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.3064833005893909e-05,
-      "loss": 1.0989,
-      "step": 21360
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.3022123515845221e-05,
-      "loss": 1.1022,
-      "step": 21370
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.2979414025796532e-05,
-      "loss": 1.1046,
-      "step": 21380
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.2936704535747843e-05,
-      "loss": 1.0963,
-      "step": 21390
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.2893995045699156e-05,
-      "loss": 1.0991,
-      "step": 21400
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.2851285555650467e-05,
-      "loss": 1.0999,
-      "step": 21410
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 1.2808576065601777e-05,
-      "loss": 1.0963,
-      "step": 21420
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.276586657555309e-05,
-      "loss": 1.0985,
-      "step": 21430
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.2723157085504401e-05,
-      "loss": 1.1012,
-      "step": 21440
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.268044759545571e-05,
-      "loss": 1.0908,
-      "step": 21450
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.2637738105407021e-05,
-      "loss": 1.0985,
-      "step": 21460
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.2595028615358334e-05,
-      "loss": 1.0978,
-      "step": 21470
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.2552319125309645e-05,
-      "loss": 1.0924,
-      "step": 21480
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.2509609635260956e-05,
-      "loss": 1.0952,
-      "step": 21490
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.2466900145212266e-05,
-      "loss": 1.104,
-      "step": 21500
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.2424190655163579e-05,
-      "loss": 1.0898,
-      "step": 21510
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.2381481165114888e-05,
-      "loss": 1.0985,
-      "step": 21520
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.2338771675066201e-05,
-      "loss": 1.0975,
-      "step": 21530
-    },
-    {
-      "epoch": 1.76,
-      "learning_rate": 1.229606218501751e-05,
-      "loss": 1.1053,
-      "step": 21540
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.2253352694968823e-05,
-      "loss": 1.0974,
-      "step": 21550
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.2210643204920134e-05,
-      "loss": 1.095,
-      "step": 21560
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.2167933714871445e-05,
-      "loss": 1.0974,
-      "step": 21570
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.2125224224822757e-05,
-      "loss": 1.097,
-      "step": 21580
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.2082514734774066e-05,
-      "loss": 1.1019,
-      "step": 21590
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.2039805244725379e-05,
-      "loss": 1.0976,
-      "step": 21600
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.199709575467669e-05,
-      "loss": 1.1006,
-      "step": 21610
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.1954386264628e-05,
-      "loss": 1.106,
-      "step": 21620
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.1911676774579313e-05,
-      "loss": 1.0975,
-      "step": 21630
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.1868967284530623e-05,
-      "loss": 1.1006,
-      "step": 21640
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.1826257794481935e-05,
-      "loss": 1.1044,
-      "step": 21650
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 1.1783548304433244e-05,
-      "loss": 1.0937,
-      "step": 21660
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.1740838814384557e-05,
-      "loss": 1.1016,
-      "step": 21670
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.1698129324335868e-05,
-      "loss": 1.1072,
-      "step": 21680
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.1655419834287179e-05,
-      "loss": 1.0945,
-      "step": 21690
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.1612710344238491e-05,
-      "loss": 1.0969,
-      "step": 21700
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.15700008541898e-05,
-      "loss": 1.087,
-      "step": 21710
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.1527291364141113e-05,
-      "loss": 1.0926,
-      "step": 21720
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.1484581874092423e-05,
-      "loss": 1.089,
-      "step": 21730
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.1441872384043735e-05,
-      "loss": 1.1029,
-      "step": 21740
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.1399162893995046e-05,
-      "loss": 1.1021,
-      "step": 21750
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.1356453403946357e-05,
-      "loss": 1.1008,
-      "step": 21760
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.131374391389767e-05,
-      "loss": 1.107,
-      "step": 21770
-    },
-    {
-      "epoch": 1.78,
-      "learning_rate": 1.1271034423848979e-05,
-      "loss": 1.0993,
-      "step": 21780
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.1228324933800291e-05,
-      "loss": 1.0907,
-      "step": 21790
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.1185615443751602e-05,
-      "loss": 1.1048,
-      "step": 21800
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.1142905953702913e-05,
-      "loss": 1.0968,
-      "step": 21810
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.1100196463654224e-05,
-      "loss": 1.1015,
-      "step": 21820
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.1057486973605535e-05,
-      "loss": 1.0939,
-      "step": 21830
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.1014777483556848e-05,
-      "loss": 1.0934,
-      "step": 21840
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.0972067993508157e-05,
-      "loss": 1.0963,
-      "step": 21850
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.092935850345947e-05,
-      "loss": 1.0942,
-      "step": 21860
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.088664901341078e-05,
-      "loss": 1.0919,
-      "step": 21870
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.0843939523362091e-05,
-      "loss": 1.0946,
-      "step": 21880
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.0801230033313404e-05,
-      "loss": 1.094,
-      "step": 21890
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.0758520543264713e-05,
-      "loss": 1.1054,
-      "step": 21900
-    },
-    {
-      "epoch": 1.79,
-      "learning_rate": 1.0715811053216026e-05,
-      "loss": 1.101,
-      "step": 21910
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.0673101563167335e-05,
-      "loss": 1.0871,
-      "step": 21920
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.0630392073118648e-05,
-      "loss": 1.1,
-      "step": 21930
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.0587682583069958e-05,
-      "loss": 1.0952,
-      "step": 21940
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.054497309302127e-05,
-      "loss": 1.0958,
-      "step": 21950
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.0502263602972582e-05,
-      "loss": 1.0983,
-      "step": 21960
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.0459554112923891e-05,
-      "loss": 1.1055,
-      "step": 21970
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.0416844622875204e-05,
-      "loss": 1.0974,
-      "step": 21980
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.0374135132826515e-05,
-      "loss": 1.0938,
-      "step": 21990
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.0331425642777826e-05,
-      "loss": 1.1006,
-      "step": 22000
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.0288716152729137e-05,
-      "loss": 1.0926,
-      "step": 22010
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.0246006662680447e-05,
-      "loss": 1.1046,
-      "step": 22020
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 1.020329717263176e-05,
-      "loss": 1.105,
-      "step": 22030
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 1.016058768258307e-05,
-      "loss": 1.0968,
-      "step": 22040
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 1.0117878192534382e-05,
-      "loss": 1.0926,
-      "step": 22050
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 1.0075168702485693e-05,
-      "loss": 1.0997,
-      "step": 22060
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 1.0032459212437004e-05,
-      "loss": 1.092,
-      "step": 22070
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 9.989749722388315e-06,
-      "loss": 1.0821,
-      "step": 22080
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 9.947040232339626e-06,
-      "loss": 1.0919,
-      "step": 22090
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 9.904330742290938e-06,
-      "loss": 1.0988,
-      "step": 22100
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 9.861621252242249e-06,
-      "loss": 1.0942,
-      "step": 22110
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 9.81891176219356e-06,
-      "loss": 1.1028,
-      "step": 22120
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 9.776202272144871e-06,
-      "loss": 1.0965,
-      "step": 22130
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 9.733492782096182e-06,
-      "loss": 1.0923,
-      "step": 22140
-    },
-    {
-      "epoch": 1.81,
-      "learning_rate": 9.690783292047494e-06,
-      "loss": 1.1053,
-      "step": 22150
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.648073801998804e-06,
-      "loss": 1.0927,
-      "step": 22160
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.605364311950116e-06,
-      "loss": 1.0953,
-      "step": 22170
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.562654821901427e-06,
-      "loss": 1.0936,
-      "step": 22180
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.519945331852738e-06,
-      "loss": 1.1028,
-      "step": 22190
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.477235841804049e-06,
-      "loss": 1.0942,
-      "step": 22200
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.43452635175536e-06,
-      "loss": 1.0954,
-      "step": 22210
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.391816861706672e-06,
-      "loss": 1.0987,
-      "step": 22220
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.349107371657982e-06,
-      "loss": 1.0944,
-      "step": 22230
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.306397881609294e-06,
-      "loss": 1.0964,
-      "step": 22240
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.263688391560605e-06,
-      "loss": 1.0824,
-      "step": 22250
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.220978901511916e-06,
-      "loss": 1.0956,
-      "step": 22260
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 9.178269411463227e-06,
-      "loss": 1.0859,
-      "step": 22270
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 9.135559921414538e-06,
-      "loss": 1.1046,
-      "step": 22280
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 9.09285043136585e-06,
-      "loss": 1.0886,
-      "step": 22290
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 9.050140941317161e-06,
-      "loss": 1.0904,
-      "step": 22300
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 9.007431451268472e-06,
-      "loss": 1.1018,
-      "step": 22310
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 8.964721961219783e-06,
-      "loss": 1.0969,
-      "step": 22320
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 8.922012471171094e-06,
-      "loss": 1.0947,
-      "step": 22330
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 8.879302981122407e-06,
-      "loss": 1.1037,
-      "step": 22340
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 8.836593491073716e-06,
-      "loss": 1.1002,
-      "step": 22350
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 8.793884001025029e-06,
-      "loss": 1.1097,
-      "step": 22360
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 8.75117451097634e-06,
-      "loss": 1.0976,
-      "step": 22370
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 8.70846502092765e-06,
-      "loss": 1.1031,
-      "step": 22380
-    },
-    {
-      "epoch": 1.83,
-      "learning_rate": 8.665755530878961e-06,
-      "loss": 1.1085,
-      "step": 22390
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.623046040830272e-06,
-      "loss": 1.0935,
-      "step": 22400
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.580336550781585e-06,
-      "loss": 1.1055,
-      "step": 22410
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.537627060732896e-06,
-      "loss": 1.0993,
-      "step": 22420
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.494917570684207e-06,
-      "loss": 1.095,
-      "step": 22430
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.452208080635518e-06,
-      "loss": 1.0911,
-      "step": 22440
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.409498590586829e-06,
-      "loss": 1.0951,
-      "step": 22450
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.36678910053814e-06,
-      "loss": 1.0983,
-      "step": 22460
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.32407961048945e-06,
-      "loss": 1.1006,
-      "step": 22470
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.281370120440763e-06,
-      "loss": 1.102,
-      "step": 22480
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.238660630392074e-06,
-      "loss": 1.0961,
-      "step": 22490
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.195951140343385e-06,
-      "loss": 1.0976,
-      "step": 22500
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.153241650294696e-06,
-      "loss": 1.1009,
-      "step": 22510
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 8.110532160246007e-06,
-      "loss": 1.0906,
-      "step": 22520
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 8.067822670197318e-06,
-      "loss": 1.0981,
-      "step": 22530
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 8.025113180148628e-06,
-      "loss": 1.0984,
-      "step": 22540
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 7.982403690099941e-06,
-      "loss": 1.0896,
-      "step": 22550
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 7.939694200051252e-06,
-      "loss": 1.0949,
-      "step": 22560
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 7.896984710002563e-06,
-      "loss": 1.0933,
-      "step": 22570
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 7.854275219953874e-06,
-      "loss": 1.1001,
-      "step": 22580
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 7.811565729905185e-06,
-      "loss": 1.0924,
-      "step": 22590
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 7.768856239856497e-06,
-      "loss": 1.0933,
-      "step": 22600
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 7.726146749807808e-06,
-      "loss": 1.095,
-      "step": 22610
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 7.683437259759119e-06,
-      "loss": 1.0883,
-      "step": 22620
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 7.64072776971043e-06,
-      "loss": 1.0959,
-      "step": 22630
-    },
-    {
-      "epoch": 1.85,
-      "learning_rate": 7.598018279661741e-06,
-      "loss": 1.1026,
-      "step": 22640
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.555308789613053e-06,
-      "loss": 1.0957,
-      "step": 22650
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.512599299564363e-06,
-      "loss": 1.094,
-      "step": 22660
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.4698898095156745e-06,
-      "loss": 1.0884,
-      "step": 22670
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.427180319466986e-06,
-      "loss": 1.0911,
-      "step": 22680
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.384470829418297e-06,
-      "loss": 1.0945,
-      "step": 22690
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.341761339369609e-06,
-      "loss": 1.0952,
-      "step": 22700
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.299051849320919e-06,
-      "loss": 1.0944,
-      "step": 22710
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.256342359272231e-06,
-      "loss": 1.1118,
-      "step": 22720
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.2136328692235425e-06,
-      "loss": 1.0888,
-      "step": 22730
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.170923379174853e-06,
-      "loss": 1.0838,
-      "step": 22740
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.128213889126164e-06,
-      "loss": 1.0948,
-      "step": 22750
-    },
-    {
-      "epoch": 1.86,
-      "learning_rate": 7.085504399077475e-06,
-      "loss": 1.1008,
-      "step": 22760
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 7.042794909028787e-06,
-      "loss": 1.0991,
-      "step": 22770
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 7.000085418980097e-06,
-      "loss": 1.0944,
-      "step": 22780
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.957375928931409e-06,
-      "loss": 1.1038,
-      "step": 22790
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.914666438882721e-06,
-      "loss": 1.0842,
-      "step": 22800
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.871956948834031e-06,
-      "loss": 1.0977,
-      "step": 22810
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.8292474587853425e-06,
-      "loss": 1.0954,
-      "step": 22820
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.786537968736653e-06,
-      "loss": 1.0991,
-      "step": 22830
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.743828478687965e-06,
-      "loss": 1.1029,
-      "step": 22840
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.701118988639275e-06,
-      "loss": 1.1032,
-      "step": 22850
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.658409498590587e-06,
-      "loss": 1.1004,
-      "step": 22860
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.615700008541899e-06,
-      "loss": 1.096,
-      "step": 22870
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 6.572990518493209e-06,
-      "loss": 1.0969,
-      "step": 22880
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.5302810284445205e-06,
-      "loss": 1.095,
-      "step": 22890
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.4875715383958314e-06,
-      "loss": 1.0948,
-      "step": 22900
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.444862048347143e-06,
-      "loss": 1.1051,
-      "step": 22910
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.402152558298455e-06,
-      "loss": 1.1039,
-      "step": 22920
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.359443068249765e-06,
-      "loss": 1.0959,
-      "step": 22930
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.316733578201077e-06,
-      "loss": 1.0947,
-      "step": 22940
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.274024088152388e-06,
-      "loss": 1.096,
-      "step": 22950
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.2313145981036995e-06,
-      "loss": 1.0941,
-      "step": 22960
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.18860510805501e-06,
-      "loss": 1.0825,
-      "step": 22970
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.145895618006321e-06,
-      "loss": 1.0957,
-      "step": 22980
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.103186127957632e-06,
-      "loss": 1.0968,
-      "step": 22990
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.060476637908943e-06,
-      "loss": 1.0977,
-      "step": 23000
-    },
-    {
-      "epoch": 1.88,
-      "learning_rate": 6.017767147860255e-06,
-      "loss": 1.0926,
-      "step": 23010
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.975057657811567e-06,
-      "loss": 1.1021,
-      "step": 23020
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.9323481677628775e-06,
-      "loss": 1.0946,
-      "step": 23030
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.8896386777141885e-06,
-      "loss": 1.0981,
-      "step": 23040
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.846929187665499e-06,
-      "loss": 1.1002,
-      "step": 23050
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.80421969761681e-06,
-      "loss": 1.0914,
-      "step": 23060
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.761510207568122e-06,
-      "loss": 1.0876,
-      "step": 23070
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.718800717519433e-06,
-      "loss": 1.0873,
-      "step": 23080
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.676091227470745e-06,
-      "loss": 1.0941,
-      "step": 23090
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.633381737422056e-06,
-      "loss": 1.0972,
-      "step": 23100
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.5906722473733665e-06,
-      "loss": 1.1081,
-      "step": 23110
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.5479627573246774e-06,
-      "loss": 1.0979,
-      "step": 23120
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 5.505253267275989e-06,
-      "loss": 1.0856,
-      "step": 23130
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.4625437772273e-06,
-      "loss": 1.1022,
-      "step": 23140
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.419834287178612e-06,
-      "loss": 1.0968,
-      "step": 23150
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.377124797129923e-06,
-      "loss": 1.0927,
-      "step": 23160
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.334415307081234e-06,
-      "loss": 1.1012,
-      "step": 23170
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.291705817032545e-06,
-      "loss": 1.0993,
-      "step": 23180
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.2489963269838555e-06,
-      "loss": 1.087,
-      "step": 23190
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.206286836935167e-06,
-      "loss": 1.091,
-      "step": 23200
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.163577346886478e-06,
-      "loss": 1.1037,
-      "step": 23210
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.12086785683779e-06,
-      "loss": 1.0975,
-      "step": 23220
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.078158366789101e-06,
-      "loss": 1.0912,
-      "step": 23230
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 5.035448876740412e-06,
-      "loss": 1.0884,
-      "step": 23240
-    },
-    {
-      "epoch": 1.9,
-      "learning_rate": 4.992739386691723e-06,
-      "loss": 1.1024,
-      "step": 23250
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.9500298966430344e-06,
-      "loss": 1.0989,
-      "step": 23260
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.907320406594345e-06,
-      "loss": 1.0979,
-      "step": 23270
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.864610916545657e-06,
-      "loss": 1.0996,
-      "step": 23280
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.821901426496968e-06,
-      "loss": 1.0942,
-      "step": 23290
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.779191936448279e-06,
-      "loss": 1.0908,
-      "step": 23300
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.73648244639959e-06,
-      "loss": 1.1009,
-      "step": 23310
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.693772956350902e-06,
-      "loss": 1.0937,
-      "step": 23320
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.6510634663022125e-06,
-      "loss": 1.0939,
-      "step": 23330
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.6083539762535234e-06,
-      "loss": 1.0999,
-      "step": 23340
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.565644486204835e-06,
-      "loss": 1.0967,
-      "step": 23350
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.522934996156146e-06,
-      "loss": 1.0971,
-      "step": 23360
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.480225506107457e-06,
-      "loss": 1.0912,
-      "step": 23370
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.437516016058769e-06,
-      "loss": 1.098,
-      "step": 23380
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.39480652601008e-06,
-      "loss": 1.0896,
-      "step": 23390
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.352097035961391e-06,
-      "loss": 1.0929,
-      "step": 23400
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.309387545912702e-06,
-      "loss": 1.1071,
-      "step": 23410
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.266678055864013e-06,
-      "loss": 1.1033,
-      "step": 23420
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.223968565815324e-06,
-      "loss": 1.0978,
-      "step": 23430
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.181259075766636e-06,
-      "loss": 1.0997,
-      "step": 23440
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.138549585717947e-06,
-      "loss": 1.099,
-      "step": 23450
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.095840095669258e-06,
-      "loss": 1.1018,
-      "step": 23460
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.053130605620569e-06,
-      "loss": 1.0951,
-      "step": 23470
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 4.0104211155718804e-06,
-      "loss": 1.0942,
-      "step": 23480
-    },
-    {
-      "epoch": 1.92,
-      "learning_rate": 3.967711625523191e-06,
-      "loss": 1.0955,
-      "step": 23490
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.925002135474502e-06,
-      "loss": 1.0987,
-      "step": 23500
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.882292645425814e-06,
-      "loss": 1.0936,
-      "step": 23510
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.839583155377125e-06,
-      "loss": 1.0925,
-      "step": 23520
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.7968736653284363e-06,
-      "loss": 1.0992,
-      "step": 23530
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.754164175279747e-06,
-      "loss": 1.0953,
-      "step": 23540
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.7114546852310585e-06,
-      "loss": 1.0945,
-      "step": 23550
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.6687451951823694e-06,
-      "loss": 1.0896,
-      "step": 23560
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.626035705133681e-06,
-      "loss": 1.1,
-      "step": 23570
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.583326215084992e-06,
-      "loss": 1.0902,
-      "step": 23580
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.5406167250363034e-06,
-      "loss": 1.093,
-      "step": 23590
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.4979072349876144e-06,
-      "loss": 1.0958,
-      "step": 23600
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.4551977449389253e-06,
-      "loss": 1.0884,
-      "step": 23610
-    },
-    {
-      "epoch": 1.93,
-      "learning_rate": 3.4124882548902366e-06,
-      "loss": 1.089,
-      "step": 23620
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 3.3697787648415484e-06,
-      "loss": 1.1006,
-      "step": 23630
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 3.3270692747928593e-06,
-      "loss": 1.1045,
-      "step": 23640
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 3.28435978474417e-06,
-      "loss": 1.0885,
-      "step": 23650
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 3.2416502946954815e-06,
-      "loss": 1.0965,
-      "step": 23660
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 3.1989408046467924e-06,
-      "loss": 1.0937,
-      "step": 23670
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 3.1562313145981038e-06,
-      "loss": 1.0996,
-      "step": 23680
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 3.113521824549415e-06,
-      "loss": 1.0987,
-      "step": 23690
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 3.070812334500726e-06,
-      "loss": 1.1024,
-      "step": 23700
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 3.0281028444520374e-06,
-      "loss": 1.0882,
-      "step": 23710
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 2.9853933544033487e-06,
-      "loss": 1.0858,
-      "step": 23720
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 2.9426838643546596e-06,
-      "loss": 1.0936,
-      "step": 23730
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 2.899974374305971e-06,
-      "loss": 1.0942,
-      "step": 23740
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.8572648842572823e-06,
-      "loss": 1.0987,
-      "step": 23750
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.814555394208593e-06,
-      "loss": 1.1006,
-      "step": 23760
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.7718459041599045e-06,
-      "loss": 1.0933,
-      "step": 23770
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.729136414111216e-06,
-      "loss": 1.0987,
-      "step": 23780
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.6864269240625268e-06,
-      "loss": 1.0904,
-      "step": 23790
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.643717434013838e-06,
-      "loss": 1.0878,
-      "step": 23800
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.601007943965149e-06,
-      "loss": 1.0929,
-      "step": 23810
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.5582984539164604e-06,
-      "loss": 1.1019,
-      "step": 23820
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.5155889638677717e-06,
-      "loss": 1.1028,
-      "step": 23830
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.4728794738190826e-06,
-      "loss": 1.0909,
-      "step": 23840
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.430169983770394e-06,
-      "loss": 1.0993,
-      "step": 23850
-    },
-    {
-      "epoch": 1.95,
-      "learning_rate": 2.3874604937217053e-06,
-      "loss": 1.0914,
-      "step": 23860
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 2.344751003673016e-06,
-      "loss": 1.1006,
-      "step": 23870
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 2.3020415136243275e-06,
-      "loss": 1.0929,
-      "step": 23880
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 2.259332023575639e-06,
-      "loss": 1.0984,
-      "step": 23890
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 2.2166225335269498e-06,
-      "loss": 1.091,
-      "step": 23900
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 2.173913043478261e-06,
-      "loss": 1.0904,
-      "step": 23910
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 2.131203553429572e-06,
-      "loss": 1.089,
-      "step": 23920
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 2.0884940633808834e-06,
-      "loss": 1.0969,
-      "step": 23930
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 2.0457845733321943e-06,
-      "loss": 1.0949,
-      "step": 23940
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 2.0030750832835056e-06,
-      "loss": 1.0944,
-      "step": 23950
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 1.960365593234817e-06,
-      "loss": 1.0958,
-      "step": 23960
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 1.917656103186128e-06,
-      "loss": 1.0986,
-      "step": 23970
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 1.8749466131374392e-06,
-      "loss": 1.1024,
-      "step": 23980
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.8322371230887505e-06,
-      "loss": 1.095,
-      "step": 23990
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.7895276330400616e-06,
-      "loss": 1.08,
-      "step": 24000
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.7468181429913726e-06,
-      "loss": 1.0849,
-      "step": 24010
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.704108652942684e-06,
-      "loss": 1.0995,
-      "step": 24020
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.661399162893995e-06,
-      "loss": 1.0837,
-      "step": 24030
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.6186896728453061e-06,
-      "loss": 1.0905,
-      "step": 24040
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.5759801827966177e-06,
-      "loss": 1.0942,
-      "step": 24050
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.5332706927479286e-06,
-      "loss": 1.0924,
-      "step": 24060
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.49056120269924e-06,
-      "loss": 1.0925,
-      "step": 24070
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.447851712650551e-06,
-      "loss": 1.0886,
-      "step": 24080
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.4051422226018622e-06,
-      "loss": 1.1016,
-      "step": 24090
-    },
-    {
-      "epoch": 1.97,
-      "learning_rate": 1.3624327325531735e-06,
-      "loss": 1.0909,
-      "step": 24100
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 1.3197232425044844e-06,
-      "loss": 1.1014,
-      "step": 24110
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 1.2770137524557958e-06,
-      "loss": 1.0913,
-      "step": 24120
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 1.234304262407107e-06,
-      "loss": 1.0932,
-      "step": 24130
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 1.191594772358418e-06,
-      "loss": 1.1051,
-      "step": 24140
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 1.1488852823097293e-06,
-      "loss": 1.0882,
-      "step": 24150
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 1.1061757922610405e-06,
-      "loss": 1.0917,
-      "step": 24160
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 1.0634663022123516e-06,
-      "loss": 1.0956,
-      "step": 24170
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 1.020756812163663e-06,
-      "loss": 1.0957,
-      "step": 24180
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 9.780473221149738e-07,
-      "loss": 1.0886,
-      "step": 24190
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 9.353378320662852e-07,
-      "loss": 1.0992,
-      "step": 24200
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 8.968992910224651e-07,
-      "loss": 1.0964,
-      "step": 24210
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 8.541898009737764e-07,
-      "loss": 1.0921,
-      "step": 24220
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 8.114803109250876e-07,
-      "loss": 1.0858,
-      "step": 24230
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 7.687708208763988e-07,
-      "loss": 1.0917,
-      "step": 24240
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 7.260613308277099e-07,
-      "loss": 1.0885,
-      "step": 24250
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 6.833518407790211e-07,
-      "loss": 1.0898,
-      "step": 24260
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 6.406423507303323e-07,
-      "loss": 1.0852,
-      "step": 24270
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 5.979328606816435e-07,
-      "loss": 1.0954,
-      "step": 24280
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 5.552233706329546e-07,
-      "loss": 1.1051,
-      "step": 24290
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 5.125138805842659e-07,
-      "loss": 1.0875,
-      "step": 24300
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 4.6980439053557705e-07,
-      "loss": 1.0967,
-      "step": 24310
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 4.270949004868882e-07,
-      "loss": 1.099,
-      "step": 24320
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 3.843854104381994e-07,
-      "loss": 1.0952,
-      "step": 24330
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 3.4167592038951053e-07,
-      "loss": 1.0911,
-      "step": 24340
-    },
-    {
-      "epoch": 1.99,
-      "learning_rate": 2.9896643034082176e-07,
-      "loss": 1.0909,
-      "step": 24350
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 2.5625694029213294e-07,
-      "loss": 1.0973,
-      "step": 24360
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 2.135474502434441e-07,
-      "loss": 1.0926,
-      "step": 24370
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 1.7083796019475527e-07,
-      "loss": 1.0924,
-      "step": 24380
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 1.2812847014606647e-07,
-      "loss": 1.0848,
-      "step": 24390
-    },
-    {
-      "epoch": 2.0,
-      "learning_rate": 8.541898009737763e-08,
-      "loss": 1.1018,
-      "step": 24400
-    }
-  ],
-  "max_steps": 24414,
-  "num_train_epochs": 2,
-  "total_flos": 3.296592748745349e+18,
-  "trial_name": null,
-  "trial_params": null
-}