diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.2222222222222223,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9268292682926833e-06,
+      "loss": 2.0292,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.853658536585367e-06,
+      "loss": 2.0417,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.780487804878048e-06,
+      "loss": 1.5471,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1707317073170733e-05,
+      "loss": 1.4769,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 1.3682,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7560975609756096e-05,
+      "loss": 1.3256,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 1.317,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3414634146341466e-05,
+      "loss": 1.3441,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6341463414634148e-05,
+      "loss": 1.3028,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 1.3212,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 1.2535,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.512195121951219e-05,
+      "loss": 1.2933,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 1.2811,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.097560975609756e-05,
+      "loss": 1.304,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.390243902439024e-05,
+      "loss": 1.3205,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.682926829268293e-05,
+      "loss": 1.2825,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975609756097561e-05,
+      "loss": 1.2774,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.2682926829268296e-05,
+      "loss": 1.2641,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.560975609756098e-05,
+      "loss": 1.2429,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.853658536585366e-05,
+      "loss": 1.2503,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.146341463414634e-05,
+      "loss": 1.2882,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.439024390243903e-05,
+      "loss": 1.3068,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.731707317073171e-05,
+      "loss": 1.2592,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.024390243902439e-05,
+      "loss": 1.2538,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.317073170731707e-05,
+      "loss": 1.3129,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.609756097560976e-05,
+      "loss": 1.2578,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.902439024390244e-05,
+      "loss": 1.2534,
+      "step": 27
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.195121951219513e-05,
+      "loss": 1.3043,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.48780487804878e-05,
+      "loss": 1.2677,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.780487804878048e-05,
+      "loss": 1.2535,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.073170731707318e-05,
+      "loss": 1.2639,
+      "step": 31
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.365853658536587e-05,
+      "loss": 1.3038,
+      "step": 32
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.658536585365855e-05,
+      "loss": 1.2966,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.951219512195122e-05,
+      "loss": 1.2355,
+      "step": 34
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001024390243902439,
+      "loss": 1.2861,
+      "step": 35
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010536585365853659,
+      "loss": 1.3257,
+      "step": 36
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00010829268292682928,
+      "loss": 1.3195,
+      "step": 37
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00011121951219512196,
+      "loss": 1.3274,
+      "step": 38
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011414634146341463,
+      "loss": 1.2388,
+      "step": 39
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011707317073170732,
+      "loss": 1.2686,
+      "step": 40
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00012,
+      "loss": 1.274,
+      "step": 41
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00011999982720089112,
+      "loss": 1.2754,
+      "step": 42
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011999930880455974,
+      "loss": 1.2609,
+      "step": 43
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011999844481399185,
+      "loss": 1.3151,
+      "step": 44
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011999723523416397,
+      "loss": 1.3002,
+      "step": 45
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011999568007204328,
+      "loss": 1.2993,
+      "step": 46
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00011999377933658745,
+      "loss": 1.301,
+      "step": 47
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011999153303874466,
+      "loss": 1.3308,
+      "step": 48
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011998894119145353,
+      "loss": 1.2533,
+      "step": 49
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011998600380964302,
+      "loss": 1.2932,
+      "step": 50
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00011998272091023235,
+      "loss": 1.2577,
+      "step": 51
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011997909251213094,
+      "loss": 1.3185,
+      "step": 52
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011997511863623823,
+      "loss": 1.2698,
+      "step": 53
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011997079930544366,
+      "loss": 1.2681,
+      "step": 54
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00011996613454462643,
+      "loss": 1.2688,
+      "step": 55
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001199611243806554,
+      "loss": 1.2947,
+      "step": 56
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000119955768842389,
+      "loss": 1.3007,
+      "step": 57
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00011995006796067497,
+      "loss": 1.2609,
+      "step": 58
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00011994402176835021,
+      "loss": 1.3019,
+      "step": 59
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001199376303002406,
+      "loss": 1.3291,
+      "step": 60
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011993089359316082,
+      "loss": 1.271,
+      "step": 61
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011992381168591412,
+      "loss": 1.3238,
+      "step": 62
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011991638461929203,
+      "loss": 1.325,
+      "step": 63
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011990861243607424,
+      "loss": 1.249,
+      "step": 64
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00011990049518102833,
+      "loss": 1.2804,
+      "step": 65
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011989203290090944,
+      "loss": 1.2872,
+      "step": 66
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011988322564446003,
+      "loss": 1.3123,
+      "step": 67
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011987407346240964,
+      "loss": 1.3108,
+      "step": 68
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00011986457640747457,
+      "loss": 1.3069,
+      "step": 69
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00011985473453435758,
+      "loss": 1.2944,
+      "step": 70
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00011984454789974758,
+      "loss": 1.2525,
+      "step": 71
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00011983401656231926,
+      "loss": 1.2582,
+      "step": 72
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001198231405827328,
+      "loss": 1.3009,
+      "step": 73
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00011981192002363357,
+      "loss": 1.3136,
+      "step": 74
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011980035494965159,
+      "loss": 1.2795,
+      "step": 75
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001197884454274014,
+      "loss": 1.3302,
+      "step": 76
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00011977619152548147,
+      "loss": 1.3062,
+      "step": 77
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001197635933144739,
+      "loss": 1.2965,
+      "step": 78
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011975065086694404,
+      "loss": 1.27,
+      "step": 79
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011973736425743998,
+      "loss": 1.2587,
+      "step": 80
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001197237335624922,
+      "loss": 1.3065,
+      "step": 81
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011970975886061309,
+      "loss": 1.2612,
+      "step": 82
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00011969544023229654,
+      "loss": 1.317,
+      "step": 83
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011968077776001742,
+      "loss": 1.3093,
+      "step": 84
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011966577152823111,
+      "loss": 1.2632,
+      "step": 85
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011965042162337308,
+      "loss": 1.2846,
+      "step": 86
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00011963472813385833,
+      "loss": 1.3133,
+      "step": 87
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00011961869115008088,
+      "loss": 1.3272,
+      "step": 88
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001196023107644133,
+      "loss": 1.2728,
+      "step": 89
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001195855870712061,
+      "loss": 1.2478,
+      "step": 90
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00011956852016678727,
+      "loss": 1.2854,
+      "step": 91
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00011955111014946166,
+      "loss": 1.2712,
+      "step": 92
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00011953335711951047,
+      "loss": 1.2605,
+      "step": 93
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00011951526117919063,
+      "loss": 1.3021,
+      "step": 94
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00011949682243273419,
+      "loss": 1.2482,
+      "step": 95
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001194780409863478,
+      "loss": 1.2801,
+      "step": 96
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011945891694821206,
+      "loss": 1.2746,
+      "step": 97
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001194394504284808,
+      "loss": 1.28,
+      "step": 98
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011941964153928065,
+      "loss": 1.2296,
+      "step": 99
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011939949039471018,
+      "loss": 1.2531,
+      "step": 100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00011937899711083942,
+      "loss": 1.2641,
+      "step": 101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00011935816180570905,
+      "loss": 1.3043,
+      "step": 102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00011933698459932983,
+      "loss": 1.2511,
+      "step": 103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00011931546561368184,
+      "loss": 1.2838,
+      "step": 104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00011929360497271377,
+      "loss": 1.2616,
+      "step": 105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001192714028023423,
+      "loss": 1.3006,
+      "step": 106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011924885923045124,
+      "loss": 1.2762,
+      "step": 107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011922597438689093,
+      "loss": 1.2797,
+      "step": 108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011920274840347734,
+      "loss": 1.2526,
+      "step": 109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00011917918141399149,
+      "loss": 1.2742,
+      "step": 110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001191552735541785,
+      "loss": 1.2698,
+      "step": 111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011913102496174698,
+      "loss": 1.2516,
+      "step": 112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011910643577636807,
+      "loss": 1.2814,
+      "step": 113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00011908150613967473,
+      "loss": 1.3506,
+      "step": 114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011905623619526097,
+      "loss": 1.3044,
+      "step": 115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001190306260886809,
+      "loss": 1.2427,
+      "step": 116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011900467596744797,
+      "loss": 1.3192,
+      "step": 117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011897838598103412,
+      "loss": 1.2457,
+      "step": 118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00011895175628086887,
+      "loss": 1.2686,
+      "step": 119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001189247870203385,
+      "loss": 1.2782,
+      "step": 120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011889747835478518,
+      "loss": 1.3098,
+      "step": 121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011886983044150598,
+      "loss": 1.266,
+      "step": 122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00011884184343975209,
+      "loss": 1.2918,
+      "step": 123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011881351751072778,
+      "loss": 1.2611,
+      "step": 124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011878485281758958,
+      "loss": 1.2192,
+      "step": 125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011875584952544527,
+      "loss": 1.3018,
+      "step": 126
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011872650780135294,
+      "loss": 1.304,
+      "step": 127
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00011869682781432005,
+      "loss": 1.3303,
+      "step": 128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011866680973530246,
+      "loss": 1.2869,
+      "step": 129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011863645373720338,
+      "loss": 1.2533,
+      "step": 130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011860575999487249,
+      "loss": 1.2678,
+      "step": 131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00011857472868510483,
+      "loss": 1.2895,
+      "step": 132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001185433599866398,
+      "loss": 1.3199,
+      "step": 133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001185116540801602,
+      "loss": 1.264,
+      "step": 134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00011847961114829109,
+      "loss": 1.2979,
+      "step": 135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001184472313755988,
+      "loss": 1.2764,
+      "step": 136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001184145149485899,
+      "loss": 1.286,
+      "step": 137
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011838146205571,
+      "loss": 1.2782,
+      "step": 138
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011834807288734277,
+      "loss": 1.2893,
+      "step": 139
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00011831434763580886,
+      "loss": 1.2874,
+      "step": 140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001182802864953647,
+      "loss": 1.3005,
+      "step": 141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011824588966220147,
+      "loss": 1.2885,
+      "step": 142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011821115733444388,
+      "loss": 1.291,
+      "step": 143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011817608971214912,
+      "loss": 1.2475,
+      "step": 144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011814068699730562,
+      "loss": 1.2787,
+      "step": 145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00011810494939383203,
+      "loss": 1.2816,
+      "step": 146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011806887710757583,
+      "loss": 1.3126,
+      "step": 147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011803247034631235,
+      "loss": 1.3111,
+      "step": 148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00011799572931974343,
+      "loss": 1.2751,
+      "step": 149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001179586542394963,
+      "loss": 1.232,
+      "step": 150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011792124531912233,
+      "loss": 1.2673,
+      "step": 151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011788350277409578,
+      "loss": 1.2299,
+      "step": 152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011784542682181257,
+      "loss": 1.2662,
+      "step": 153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001178070176815891,
+      "loss": 1.2464,
+      "step": 154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00011776827557466086,
+      "loss": 1.2414,
+      "step": 155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011772920072418121,
+      "loss": 1.254,
+      "step": 156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011768979335522015,
+      "loss": 1.2713,
+      "step": 157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011765005369476294,
+      "loss": 1.2481,
+      "step": 158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00011760998197170885,
+      "loss": 1.2539,
+      "step": 159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011756957841686985,
+      "loss": 1.2687,
+      "step": 160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011752884326296917,
+      "loss": 1.2749,
+      "step": 161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011748777674464008,
+      "loss": 1.2518,
+      "step": 162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011744637909842455,
+      "loss": 1.3132,
+      "step": 163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00011740465056277176,
+      "loss": 1.3026,
+      "step": 164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011736259137803685,
+      "loss": 1.2225,
+      "step": 165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011732020178647945,
+      "loss": 1.2805,
+      "step": 166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001172774820322624,
+      "loss": 1.2332,
+      "step": 167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00011723443236145015,
+      "loss": 1.2702,
+      "step": 168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011719105302200757,
+      "loss": 1.2665,
+      "step": 169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011714734426379837,
+      "loss": 1.2261,
+      "step": 170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011710330633858367,
+      "loss": 1.2395,
+      "step": 171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00011705893950002063,
+      "loss": 1.2588,
+      "step": 172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001170142440036609,
+      "loss": 1.2609,
+      "step": 173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011696922010694925,
+      "loss": 1.2454,
+      "step": 174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011692386806922196,
+      "loss": 1.2901,
+      "step": 175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00011687818815170541,
+      "loss": 1.2764,
+      "step": 176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001168321806175146,
+      "loss": 1.2632,
+      "step": 177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011678584573165155,
+      "loss": 1.2594,
+      "step": 178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001167391837610038,
+      "loss": 1.2609,
+      "step": 179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011669219497434297,
+      "loss": 1.2928,
+      "step": 180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011664487964232302,
+      "loss": 1.2612,
+      "step": 181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00011659723803747888,
+      "loss": 1.3024,
+      "step": 182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011654927043422479,
+      "loss": 1.2663,
+      "step": 183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001165009771088527,
+      "loss": 1.2345,
+      "step": 184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011645235833953074,
+      "loss": 1.3054,
+      "step": 185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00011640341440630155,
+      "loss": 1.2414,
+      "step": 186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011635414559108078,
+      "loss": 1.304,
+      "step": 187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011630455217765531,
+      "loss": 1.2611,
+      "step": 188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011625463445168175,
+      "loss": 1.2403,
+      "step": 189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011620439270068469,
+      "loss": 1.2718,
+      "step": 190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00011615382721405513,
+      "loss": 1.2647,
+      "step": 191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001161029382830488,
+      "loss": 1.2481,
+      "step": 192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011605172620078439,
+      "loss": 1.25,
+      "step": 193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000116000191262242,
+      "loss": 1.2761,
+      "step": 194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00011594833376426134,
+      "loss": 1.3131,
+      "step": 195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011589615400554007,
+      "loss": 1.222,
+      "step": 196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011584365228663202,
+      "loss": 1.2354,
+      "step": 197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011579082890994557,
+      "loss": 1.2224,
+      "step": 198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011573768417974176,
+      "loss": 1.2783,
+      "step": 199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00011568421840213267,
+      "loss": 1.2999,
+      "step": 200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011563043188507961,
+      "loss": 1.2273,
+      "step": 201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001155763249383913,
+      "loss": 1.2849,
+      "step": 202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011552189787372217,
+      "loss": 1.2166,
+      "step": 203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011546715100457046,
+      "loss": 1.2472,
+      "step": 204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011541208464627652,
+      "loss": 1.2523,
+      "step": 205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011535669911602097,
+      "loss": 1.2413,
+      "step": 206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011530099473282279,
+      "loss": 1.2395,
+      "step": 207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011524497181753759,
+      "loss": 1.2717,
+      "step": 208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011518863069285567,
+      "loss": 1.2152,
+      "step": 209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011513197168330026,
+      "loss": 1.2583,
+      "step": 210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011507499511522556,
+      "loss": 1.2311,
+      "step": 211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011501770131681491,
+      "loss": 1.276,
+      "step": 212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001149600906180789,
+      "loss": 1.2171,
+      "step": 213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011490216335085345,
+      "loss": 1.2381,
+      "step": 214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011484391984879785,
+      "loss": 1.2577,
+      "step": 215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000114785360447393,
+      "loss": 1.2659,
+      "step": 216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011472648548393928,
+      "loss": 1.2436,
+      "step": 217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011466729529755472,
+      "loss": 1.2542,
+      "step": 218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011460779022917307,
+      "loss": 1.2525,
+      "step": 219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011454797062154173,
+      "loss": 1.2511,
+      "step": 220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011448783681921988,
+      "loss": 1.2654,
+      "step": 221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00011442738916857643,
+      "loss": 1.2532,
+      "step": 222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011436662801778805,
+      "loss": 1.2714,
+      "step": 223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011430555371683716,
+      "loss": 1.2708,
+      "step": 224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011424416661750994,
+      "loss": 1.1997,
+      "step": 225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011418246707339422,
+      "loss": 1.2324,
+      "step": 226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00011412045543987757,
+      "loss": 1.2788,
+      "step": 227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011405813207414514,
+      "loss": 1.2543,
+      "step": 228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001139954973351777,
+      "loss": 1.2854,
+      "step": 229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011393255158374945,
+      "loss": 1.2034,
+      "step": 230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00011386929518242606,
+      "loss": 1.2724,
+      "step": 231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011380572849556251,
+      "loss": 1.2631,
+      "step": 232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011374185188930107,
+      "loss": 1.281,
+      "step": 233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011367766573156905,
+      "loss": 1.2162,
+      "step": 234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011361317039207682,
+      "loss": 1.2456,
+      "step": 235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00011354836624231564,
+      "loss": 1.2486,
+      "step": 236
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00011348325365555547,
+      "loss": 1.2923,
+      "step": 237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00011341783300684288,
+      "loss": 1.2268,
+      "step": 238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00011335210467299887,
+      "loss": 1.2583,
+      "step": 239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00011328606903261669,
+      "loss": 1.2731,
+      "step": 240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00011321972646605965,
+      "loss": 1.2224,
+      "step": 241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00011315307735545897,
+      "loss": 1.2495,
+      "step": 242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00011308612208471157,
+      "loss": 1.2641,
+      "step": 243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0001130188610394778,
+      "loss": 1.2308,
+      "step": 244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00011295129460717928,
+      "loss": 1.2406,
+      "step": 245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00011288342317699666,
+      "loss": 1.252,
+      "step": 246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00011281524713986736,
+      "loss": 1.2578,
+      "step": 247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00011274676688848332,
+      "loss": 1.2609,
+      "step": 248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00011267798281728878,
+      "loss": 1.2398,
+      "step": 249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00011260889532247793,
+      "loss": 1.2689,
+      "step": 250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00011253950480199267,
+      "loss": 1.2193,
+      "step": 251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00011246981165552038,
+      "loss": 1.2361,
+      "step": 252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00011239981628449148,
+      "loss": 1.2999,
+      "step": 253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00011232951909207721,
+      "loss": 1.2357,
+      "step": 254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00011225892048318737,
+      "loss": 1.2478,
+      "step": 255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0001121880208644678,
+      "loss": 1.2718,
+      "step": 256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00011211682064429823,
+      "loss": 1.2786,
+      "step": 257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00011204532023278979,
+      "loss": 1.2636,
+      "step": 258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00011197352004178271,
+      "loss": 1.2662,
+      "step": 259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00011190142048484403,
+      "loss": 1.2616,
+      "step": 260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00011182902197726497,
+      "loss": 1.2511,
+      "step": 261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00011175632493605883,
+      "loss": 1.2794,
+      "step": 262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00011168332977995841,
+      "loss": 1.2709,
+      "step": 263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00011161003692941364,
+      "loss": 1.2583,
+      "step": 264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00011153644680658915,
+      "loss": 1.2573,
+      "step": 265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00011146255983536184,
+      "loss": 1.2251,
+      "step": 266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001113883764413185,
+      "loss": 1.2647,
+      "step": 267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00011131389705175328,
+      "loss": 1.2252,
+      "step": 268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00011123912209566525,
+      "loss": 1.2373,
+      "step": 269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00011116405200375591,
+      "loss": 1.2447,
+      "step": 270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00011108868720842679,
+      "loss": 1.2294,
+      "step": 271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00011101302814377686,
+      "loss": 1.2497,
+      "step": 272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011093707524560006,
+      "loss": 1.2151,
+      "step": 273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011086082895138288,
+      "loss": 1.2533,
+      "step": 274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011078428970030167,
+      "loss": 1.2424,
+      "step": 275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00011070745793322026,
+      "loss": 1.2656,
+      "step": 276
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011063033409268734,
+      "loss": 1.2047,
+      "step": 277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011055291862293394,
+      "loss": 1.2705,
+      "step": 278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011047521196987087,
+      "loss": 1.2819,
+      "step": 279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011039721458108616,
+      "loss": 1.2561,
+      "step": 280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00011031892690584239,
+      "loss": 1.2291,
+      "step": 281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011024034939507433,
+      "loss": 1.2234,
+      "step": 282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011016148250138605,
+      "loss": 1.247,
+      "step": 283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011008232667904853,
+      "loss": 1.2593,
+      "step": 284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00011000288238399695,
+      "loss": 1.255,
+      "step": 285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001099231500738281,
+      "loss": 1.2471,
+      "step": 286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010984313020779771,
+      "loss": 1.2817,
+      "step": 287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010976282324681785,
+      "loss": 1.2424,
+      "step": 288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00010968222965345421,
+      "loss": 1.2042,
+      "step": 289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001096013498919235,
+      "loss": 1.2324,
+      "step": 290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010952018442809074,
+      "loss": 1.2063,
+      "step": 291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010943873372946661,
+      "loss": 1.2345,
+      "step": 292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010935699826520467,
+      "loss": 1.2373,
+      "step": 293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00010927497850609882,
+      "loss": 1.2694,
+      "step": 294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010919267492458041,
+      "loss": 1.2393,
+      "step": 295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010911008799471562,
+      "loss": 1.2307,
+      "step": 296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010902721819220271,
+      "loss": 1.2317,
+      "step": 297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0001089440659943693,
+      "loss": 1.2129,
+      "step": 298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00010886063188016958,
+      "loss": 1.2341,
+      "step": 299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00010877691633018154,
+      "loss": 1.2456,
+      "step": 300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001086929198266043,
+      "loss": 1.2511,
+      "step": 301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001086086428532552,
+      "loss": 1.247,
+      "step": 302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0001085240858955671,
+      "loss": 1.2702,
+      "step": 303
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010843924944058557,
+      "loss": 1.2601,
+      "step": 304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0001083541339769661,
+      "loss": 1.1809,
+      "step": 305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010826873999497118,
+      "loss": 1.2388,
+      "step": 306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010818306798646766,
+      "loss": 1.2608,
+      "step": 307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00010809711844492373,
+      "loss": 1.2479,
+      "step": 308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010801089186540621,
+      "loss": 1.2148,
+      "step": 309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010792438874457763,
+      "loss": 1.225,
+      "step": 310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010783760958069341,
+      "loss": 1.2321,
+      "step": 311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00010775055487359894,
+      "loss": 1.2556,
+      "step": 312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010766322512472675,
+      "loss": 1.2122,
+      "step": 313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010757562083709362,
+      "loss": 1.2606,
+      "step": 314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010748774251529763,
+      "loss": 1.2528,
+      "step": 315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010739959066551528,
+      "loss": 1.2228,
+      "step": 316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00010731116579549864,
+      "loss": 1.1934,
+      "step": 317
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00010722246841457232,
+      "loss": 1.2169,
+      "step": 318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001071334990336306,
+      "loss": 1.2657,
+      "step": 319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001070442581651345,
+      "loss": 1.2457,
+      "step": 320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00010695474632310871,
+      "loss": 1.2288,
+      "step": 321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010686496402313882,
+      "loss": 1.2224,
+      "step": 322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010677491178236823,
+      "loss": 1.2083,
+      "step": 323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010668459011949512,
+      "loss": 1.2031,
+      "step": 324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010659399955476964,
+      "loss": 1.2406,
+      "step": 325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00010650314060999073,
+      "loss": 1.1586,
+      "step": 326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010641201380850319,
+      "loss": 1.2393,
+      "step": 327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010632061967519473,
+      "loss": 1.1987,
+      "step": 328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010622895873649281,
+      "loss": 1.1982,
+      "step": 329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00010613703152036172,
+      "loss": 1.2368,
+      "step": 330
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010604483855629952,
+      "loss": 1.2315,
+      "step": 331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010595238037533491,
+      "loss": 1.1949,
+      "step": 332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001058596575100243,
+      "loss": 1.2115,
+      "step": 333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00010576667049444861,
+      "loss": 1.2225,
+      "step": 334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001056734198642103,
+      "loss": 1.2243,
+      "step": 335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010557990615643023,
+      "loss": 1.2686,
+      "step": 336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010548612990974458,
+      "loss": 1.2224,
+      "step": 337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010539209166430176,
+      "loss": 1.2389,
+      "step": 338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00010529779196175924,
+      "loss": 1.2085,
+      "step": 339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010520323134528051,
+      "loss": 1.2501,
+      "step": 340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010510841035953194,
+      "loss": 1.2202,
+      "step": 341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010501332955067958,
+      "loss": 1.2156,
+      "step": 342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010491798946638606,
+      "loss": 1.2211,
+      "step": 343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00010482239065580742,
+      "loss": 1.2209,
+      "step": 344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00010472653366958998,
+      "loss": 1.2249,
+      "step": 345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00010463041905986715,
+      "loss": 1.1985,
+      "step": 346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001045340473802562,
+      "loss": 1.2689,
+      "step": 347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00010443741918585517,
+      "loss": 1.2237,
+      "step": 348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00010434053503323955,
+      "loss": 1.2322,
+      "step": 349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00010424339548045921,
+      "loss": 1.195,
+      "step": 350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001041460010870351,
+      "loss": 1.1922,
+      "step": 351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00010404835241395601,
+      "loss": 1.2476,
+      "step": 352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00010395045002367541,
+      "loss": 1.2289,
+      "step": 353
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010385229448010814,
+      "loss": 1.249,
+      "step": 354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010375388634862723,
+      "loss": 1.2138,
+      "step": 355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010365522619606062,
+      "loss": 1.1953,
+      "step": 356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010355631459068779,
+      "loss": 1.2433,
+      "step": 357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010345715210223671,
+      "loss": 1.1922,
+      "step": 358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010335773930188036,
+      "loss": 1.2071,
+      "step": 359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001032580767622335,
+      "loss": 1.2189,
+      "step": 360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001031581650573494,
+      "loss": 1.2081,
+      "step": 361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010305800476271651,
+      "loss": 1.2028,
+      "step": 362
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010295759645525515,
+      "loss": 1.2179,
+      "step": 363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010285694071331422,
+      "loss": 1.2041,
+      "step": 364
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010275603811666778,
+      "loss": 1.2169,
+      "step": 365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00010265488924651176,
+      "loss": 1.2272,
+      "step": 366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010255349468546072,
+      "loss": 1.2104,
+      "step": 367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010245185501754425,
+      "loss": 1.2381,
+      "step": 368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010234997082820383,
+      "loss": 1.2231,
+      "step": 369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010224784270428942,
+      "loss": 1.2252,
+      "step": 370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00010214547123405592,
+      "loss": 1.2611,
+      "step": 371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00010204285700715998,
+      "loss": 1.2557,
+      "step": 372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00010194000061465648,
+      "loss": 1.2176,
+      "step": 373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00010183690264899521,
+      "loss": 1.2211,
+      "step": 374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00010173356370401741,
+      "loss": 1.2117,
+      "step": 375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010162998437495228,
+      "loss": 1.2404,
+      "step": 376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001015261652584137,
+      "loss": 1.2181,
+      "step": 377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001014221069523967,
+      "loss": 1.2391,
+      "step": 378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010131781005627406,
+      "loss": 1.2204,
+      "step": 379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00010121327517079276,
+      "loss": 1.2545,
+      "step": 380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010110850289807066,
+      "loss": 1.2036,
+      "step": 381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010100349384159291,
+      "loss": 1.2642,
+      "step": 382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010089824860620861,
+      "loss": 1.2331,
+      "step": 383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00010079276779812714,
+      "loss": 1.2063,
+      "step": 384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010068705202491485,
+      "loss": 1.1969,
+      "step": 385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010058110189549143,
+      "loss": 1.2161,
+      "step": 386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010047491802012648,
+      "loss": 1.2371,
+      "step": 387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.000100368501010436,
+      "loss": 1.2154,
+      "step": 388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00010026185147937877,
+      "loss": 1.2139,
+      "step": 389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010015497004125293,
+      "loss": 1.2437,
+      "step": 390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00010004785731169242,
+      "loss": 1.2077,
+      "step": 391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.994051390766333e-05,
+      "loss": 1.2164,
+      "step": 392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.983294044746051e-05,
+      "loss": 1.2177,
+      "step": 393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.97251375507039e-05,
+      "loss": 1.2081,
+      "step": 394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.961710583833494e-05,
+      "loss": 1.2373,
+      "step": 395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.950884593261315e-05,
+      "loss": 1.218,
+      "step": 396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.940035845711232e-05,
+      "loss": 1.24,
+      "step": 397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.929164403671711e-05,
+      "loss": 1.1771,
+      "step": 398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.918270329761933e-05,
+      "loss": 1.2131,
+      "step": 399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.907353686731444e-05,
+      "loss": 1.2335,
+      "step": 400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.89641453745978e-05,
+      "loss": 1.2427,
+      "step": 401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.885452944956118e-05,
+      "loss": 1.2184,
+      "step": 402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.874468972358904e-05,
+      "loss": 1.2157,
+      "step": 403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.863462682935493e-05,
+      "loss": 1.2399,
+      "step": 404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.852434140081789e-05,
+      "loss": 1.221,
+      "step": 405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.841383407321866e-05,
+      "loss": 1.2568,
+      "step": 406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.830310548307622e-05,
+      "loss": 1.2178,
+      "step": 407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.819215626818392e-05,
+      "loss": 1.2101,
+      "step": 408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.808098706760595e-05,
+      "loss": 1.212,
+      "step": 409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.796959852167363e-05,
+      "loss": 1.2028,
+      "step": 410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.785799127198162e-05,
+      "loss": 1.1922,
+      "step": 411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.77461659613844e-05,
+      "loss": 1.1668,
+      "step": 412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.763412323399245e-05,
+      "loss": 1.1926,
+      "step": 413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.752186373516853e-05,
+      "loss": 1.2085,
+      "step": 414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.740938811152401e-05,
+      "loss": 1.2137,
+      "step": 415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.729669701091517e-05,
+      "loss": 1.2701,
+      "step": 416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.718379108243939e-05,
+      "loss": 1.2591,
+      "step": 417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.707067097643147e-05,
+      "loss": 1.2277,
+      "step": 418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.695733734445982e-05,
+      "loss": 1.2128,
+      "step": 419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.684379083932286e-05,
+      "loss": 1.2091,
+      "step": 420
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.673003211504503e-05,
+      "loss": 1.2067,
+      "step": 421
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.661606182687324e-05,
+      "loss": 1.191,
+      "step": 422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.650188063127296e-05,
+      "loss": 1.1973,
+      "step": 423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.638748918592445e-05,
+      "loss": 1.2405,
+      "step": 424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.627288814971908e-05,
+      "loss": 1.2487,
+      "step": 425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.615807818275539e-05,
+      "loss": 1.1887,
+      "step": 426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.604305994633539e-05,
+      "loss": 1.2214,
+      "step": 427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.592783410296071e-05,
+      "loss": 1.2136,
+      "step": 428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.581240131632876e-05,
+      "loss": 1.2074,
+      "step": 429
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.569676225132898e-05,
+      "loss": 1.1687,
+      "step": 430
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.558091757403897e-05,
+      "loss": 1.1988,
+      "step": 431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.54648679517206e-05,
+      "loss": 1.2025,
+      "step": 432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.534861405281625e-05,
+      "loss": 1.2027,
+      "step": 433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.523215654694493e-05,
+      "loss": 1.2192,
+      "step": 434
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.511549610489844e-05,
+      "loss": 1.2245,
+      "step": 435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.499863339863741e-05,
+      "loss": 1.1959,
+      "step": 436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.48815691012876e-05,
+      "loss": 1.2241,
+      "step": 437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.476430388713586e-05,
+      "loss": 1.2185,
+      "step": 438
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.464683843162635e-05,
+      "loss": 1.202,
+      "step": 439
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.45291734113566e-05,
+      "loss": 1.2031,
+      "step": 440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.441130950407367e-05,
+      "loss": 1.2341,
+      "step": 441
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.42932473886701e-05,
+      "loss": 1.2146,
+      "step": 442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.417498774518019e-05,
+      "loss": 1.2221,
+      "step": 443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.4056531254776e-05,
+      "loss": 1.2147,
+      "step": 444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.393787859976338e-05,
+      "loss": 1.229,
+      "step": 445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.381903046357809e-05,
+      "loss": 1.2305,
+      "step": 446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.369998753078188e-05,
+      "loss": 1.2403,
+      "step": 447
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.35807504870585e-05,
+      "loss": 1.2102,
+      "step": 448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.346132001920977e-05,
+      "loss": 1.1927,
+      "step": 449
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.334169681515164e-05,
+      "loss": 1.1759,
+      "step": 450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.322188156391023e-05,
+      "loss": 0.9853,
+      "step": 451
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.31018749556178e-05,
+      "loss": 0.9395,
+      "step": 452
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.29816776815089e-05,
+      "loss": 0.9399,
+      "step": 453
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.28612904339162e-05,
+      "loss": 0.8974,
+      "step": 454
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.27407139062667e-05,
+      "loss": 0.9517,
+      "step": 455
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.261994879307761e-05,
+      "loss": 0.889,
+      "step": 456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.24989957899524e-05,
+      "loss": 0.9045,
+      "step": 457
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.237785559357675e-05,
+      "loss": 0.8965,
+      "step": 458
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.225652890171464e-05,
+      "loss": 0.9576,
+      "step": 459
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.213501641320418e-05,
+      "loss": 0.9374,
+      "step": 460
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.20133188279537e-05,
+      "loss": 0.8864,
+      "step": 461
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.189143684693768e-05,
+      "loss": 0.8912,
+      "step": 462
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.176937117219272e-05,
+      "loss": 0.8981,
+      "step": 463
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.164712250681344e-05,
+      "loss": 0.892,
+      "step": 464
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.152469155494857e-05,
+      "loss": 0.8896,
+      "step": 465
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.140207902179673e-05,
+      "loss": 0.9112,
+      "step": 466
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.127928561360246e-05,
+      "loss": 0.8909,
+      "step": 467
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.115631203765218e-05,
+      "loss": 0.9034,
+      "step": 468
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.103315900226999e-05,
+      "loss": 0.903,
+      "step": 469
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.090982721681376e-05,
+      "loss": 0.9069,
+      "step": 470
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.07863173916709e-05,
+      "loss": 0.877,
+      "step": 471
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.06626302382543e-05,
+      "loss": 0.874,
+      "step": 472
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.05387664689983e-05,
+      "loss": 0.8924,
+      "step": 473
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.041472679735459e-05,
+      "loss": 0.9288,
+      "step": 474
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.029051193778793e-05,
+      "loss": 0.8858,
+      "step": 475
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.016612260577223e-05,
+      "loss": 0.869,
+      "step": 476
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.004155951778635e-05,
+      "loss": 0.8812,
+      "step": 477
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.991682339130999e-05,
+      "loss": 0.8824,
+      "step": 478
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.979191494481956e-05,
+      "loss": 0.891,
+      "step": 479
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.966683489778394e-05,
+      "loss": 0.8898,
+      "step": 480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.954158397066053e-05,
+      "loss": 0.8971,
+      "step": 481
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.941616288489093e-05,
+      "loss": 0.8956,
+      "step": 482
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.929057236289687e-05,
+      "loss": 0.903,
+      "step": 483
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.916481312807606e-05,
+      "loss": 0.9051,
+      "step": 484
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.90388859047979e-05,
+      "loss": 0.8458,
+      "step": 485
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.891279141839948e-05,
+      "loss": 0.8767,
+      "step": 486
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.878653039518131e-05,
+      "loss": 0.878,
+      "step": 487
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.866010356240313e-05,
+      "loss": 0.9164,
+      "step": 488
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.853351164827973e-05,
+      "loss": 0.8943,
+      "step": 489
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.840675538197676e-05,
+      "loss": 0.8845,
+      "step": 490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.827983549360659e-05,
+      "loss": 0.8685,
+      "step": 491
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.815275271422398e-05,
+      "loss": 0.898,
+      "step": 492
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.802550777582197e-05,
+      "loss": 0.8824,
+      "step": 493
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.789810141132762e-05,
+      "loss": 0.9068,
+      "step": 494
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.777053435459781e-05,
+      "loss": 0.9178,
+      "step": 495
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.7642807340415e-05,
+      "loss": 0.886,
+      "step": 496
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.7514921104483e-05,
+      "loss": 0.8724,
+      "step": 497
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.738687638342273e-05,
+      "loss": 0.8978,
+      "step": 498
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.725867391476798e-05,
+      "loss": 0.9092,
+      "step": 499
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.713031443696114e-05,
+      "loss": 0.9074,
+      "step": 500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.700179868934902e-05,
+      "loss": 0.9011,
+      "step": 501
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.687312741217851e-05,
+      "loss": 0.8765,
+      "step": 502
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.67443013465923e-05,
+      "loss": 0.8833,
+      "step": 503
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.661532123462474e-05,
+      "loss": 0.904,
+      "step": 504
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.648618781919745e-05,
+      "loss": 0.9168,
+      "step": 505
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.635690184411505e-05,
+      "loss": 0.9153,
+      "step": 506
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.62274640540609e-05,
+      "loss": 0.8865,
+      "step": 507
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.609787519459285e-05,
+      "loss": 0.9066,
+      "step": 508
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.596813601213889e-05,
+      "loss": 0.8977,
+      "step": 509
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.583824725399285e-05,
+      "loss": 0.8505,
+      "step": 510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.570820966831008e-05,
+      "loss": 0.9051,
+      "step": 511
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.557802400410326e-05,
+      "loss": 0.8694,
+      "step": 512
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.544769101123793e-05,
+      "loss": 0.8798,
+      "step": 513
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.531721144042826e-05,
+      "loss": 0.8827,
+      "step": 514
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.518658604323272e-05,
+      "loss": 0.8774,
+      "step": 515
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.505581557204968e-05,
+      "loss": 0.9107,
+      "step": 516
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.49249007801132e-05,
+      "loss": 0.8869,
+      "step": 517
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.479384242148856e-05,
+      "loss": 0.9126,
+      "step": 518
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.466264125106806e-05,
+      "loss": 0.9356,
+      "step": 519
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.45312980245665e-05,
+      "loss": 0.8958,
+      "step": 520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.4399813498517e-05,
+      "loss": 0.8788,
+      "step": 521
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.42681884302665e-05,
+      "loss": 0.8861,
+      "step": 522
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.413642357797148e-05,
+      "loss": 0.8971,
+      "step": 523
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.40045197005936e-05,
+      "loss": 0.882,
+      "step": 524
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.387247755789525e-05,
+      "loss": 0.909,
+      "step": 525
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.374029791043527e-05,
+      "loss": 0.8958,
+      "step": 526
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.36079815195645e-05,
+      "loss": 0.9146,
+      "step": 527
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.347552914742142e-05,
+      "loss": 0.8973,
+      "step": 528
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.334294155692774e-05,
+      "loss": 0.9057,
+      "step": 529
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.321021951178405e-05,
+      "loss": 0.8658,
+      "step": 530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.30773637764654e-05,
+      "loss": 0.9342,
+      "step": 531
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.29443751162169e-05,
+      "loss": 0.8892,
+      "step": 532
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.281125429704923e-05,
+      "loss": 0.9283,
+      "step": 533
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.26780020857344e-05,
+      "loss": 0.8811,
+      "step": 534
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.254461924980116e-05,
+      "loss": 0.8753,
+      "step": 535
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.24111065575307e-05,
+      "loss": 0.8788,
+      "step": 536
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.227746477795215e-05,
+      "loss": 0.9025,
+      "step": 537
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.21436946808382e-05,
+      "loss": 0.9066,
+      "step": 538
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.200979703670062e-05,
+      "loss": 0.8974,
+      "step": 539
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.18757726167859e-05,
+      "loss": 0.898,
+      "step": 540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.17416221930707e-05,
+      "loss": 0.9442,
+      "step": 541
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.160734653825743e-05,
+      "loss": 0.9257,
+      "step": 542
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.147294642576993e-05,
+      "loss": 0.9328,
+      "step": 543
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.133842262974885e-05,
+      "loss": 0.8714,
+      "step": 544
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.120377592504725e-05,
+      "loss": 0.8962,
+      "step": 545
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.106900708722612e-05,
+      "loss": 0.9002,
+      "step": 546
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.093411689255001e-05,
+      "loss": 0.912,
+      "step": 547
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.07991061179824e-05,
+      "loss": 0.8852,
+      "step": 548
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.066397554118136e-05,
+      "loss": 0.8716,
+      "step": 549
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.052872594049501e-05,
+      "loss": 0.9013,
+      "step": 550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.0393358094957e-05,
+      "loss": 0.9178,
+      "step": 551
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.025787278428213e-05,
+      "loss": 0.9117,
+      "step": 552
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.012227078886174e-05,
+      "loss": 0.9364,
+      "step": 553
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.998655288975931e-05,
+      "loss": 0.886,
+      "step": 554
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.985071986870591e-05,
+      "loss": 0.8811,
+      "step": 555
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.971477250809569e-05,
+      "loss": 0.9123,
+      "step": 556
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.957871159098143e-05,
+      "loss": 0.8964,
+      "step": 557
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.944253790106996e-05,
+      "loss": 0.8677,
+      "step": 558
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.930625222271768e-05,
+      "loss": 0.9193,
+      "step": 559
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.916985534092606e-05,
+      "loss": 0.9202,
+      "step": 560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.903334804133711e-05,
+      "loss": 0.8913,
+      "step": 561
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.889673111022878e-05,
+      "loss": 0.884,
+      "step": 562
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.876000533451057e-05,
+      "loss": 0.8619,
+      "step": 563
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.862317150171886e-05,
+      "loss": 0.9247,
+      "step": 564
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.848623040001246e-05,
+      "loss": 0.9068,
+      "step": 565
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.834918281816805e-05,
+      "loss": 0.8869,
+      "step": 566
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.821202954557568e-05,
+      "loss": 0.8924,
+      "step": 567
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.807477137223406e-05,
+      "loss": 0.906,
+      "step": 568
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.793740908874622e-05,
+      "loss": 0.9001,
+      "step": 569
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.779994348631484e-05,
+      "loss": 0.9018,
+      "step": 570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.76623753567377e-05,
+      "loss": 0.9202,
+      "step": 571
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.752470549240314e-05,
+      "loss": 0.8876,
+      "step": 572
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.738693468628548e-05,
+      "loss": 0.8735,
+      "step": 573
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.724906373194049e-05,
+      "loss": 0.8747,
+      "step": 574
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.711109342350075e-05,
+      "loss": 0.9046,
+      "step": 575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.697302455567116e-05,
+      "loss": 0.8478,
+      "step": 576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.683485792372427e-05,
+      "loss": 0.887,
+      "step": 577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.669659432349581e-05,
+      "loss": 0.8691,
+      "step": 578
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.655823455137998e-05,
+      "loss": 0.9361,
+      "step": 579
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.641977940432499e-05,
+      "loss": 0.8743,
+      "step": 580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.628122967982834e-05,
+      "loss": 0.8967,
+      "step": 581
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.614258617593234e-05,
+      "loss": 0.8915,
+      "step": 582
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.600384969121945e-05,
+      "loss": 0.8841,
+      "step": 583
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.586502102480773e-05,
+      "loss": 0.8665,
+      "step": 584
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.572610097634613e-05,
+      "loss": 0.8986,
+      "step": 585
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.558709034601004e-05,
+      "loss": 0.9087,
+      "step": 586
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.544798993449654e-05,
+      "loss": 0.9034,
+      "step": 587
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.53088005430199e-05,
+      "loss": 0.8983,
+      "step": 588
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.516952297330684e-05,
+      "loss": 0.9164,
+      "step": 589
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.503015802759202e-05,
+      "loss": 0.8989,
+      "step": 590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.489070650861344e-05,
+      "loss": 0.921,
+      "step": 591
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.475116921960766e-05,
+      "loss": 0.8921,
+      "step": 592
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.461154696430534e-05,
+      "loss": 0.8911,
+      "step": 593
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.447184054692651e-05,
+      "loss": 0.8925,
+      "step": 594
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.4332050772176e-05,
+      "loss": 0.9112,
+      "step": 595
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.419217844523875e-05,
+      "loss": 0.9141,
+      "step": 596
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.405222437177523e-05,
+      "loss": 0.8961,
+      "step": 597
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.391218935791671e-05,
+      "loss": 0.9301,
+      "step": 598
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.377207421026074e-05,
+      "loss": 0.8793,
+      "step": 599
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.363187973586639e-05,
+      "loss": 0.8918,
+      "step": 600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.349160674224967e-05,
+      "loss": 0.9046,
+      "step": 601
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.335125603737886e-05,
+      "loss": 0.871,
+      "step": 602
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.321082842966986e-05,
+      "loss": 0.8872,
+      "step": 603
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.307032472798151e-05,
+      "loss": 0.91,
+      "step": 604
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.292974574161098e-05,
+      "loss": 0.927,
+      "step": 605
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.278909228028903e-05,
+      "loss": 0.9181,
+      "step": 606
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.26483651541754e-05,
+      "loss": 0.8765,
+      "step": 607
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.25075651738542e-05,
+      "loss": 0.9266,
+      "step": 608
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.236669315032912e-05,
+      "loss": 0.9351,
+      "step": 609
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.222574989501881e-05,
+      "loss": 0.8773,
+      "step": 610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.208473621975227e-05,
+      "loss": 0.9041,
+      "step": 611
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.194365293676404e-05,
+      "loss": 0.9115,
+      "step": 612
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.180250085868969e-05,
+      "loss": 0.8905,
+      "step": 613
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.166128079856096e-05,
+      "loss": 0.883,
+      "step": 614
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.151999356980121e-05,
+      "loss": 0.9163,
+      "step": 615
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.137863998622067e-05,
+      "loss": 0.8888,
+      "step": 616
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.123722086201181e-05,
+      "loss": 0.9359,
+      "step": 617
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.109573701174457e-05,
+      "loss": 0.8758,
+      "step": 618
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.09541892503617e-05,
+      "loss": 0.8859,
+      "step": 619
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.081257839317415e-05,
+      "loss": 0.8991,
+      "step": 620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.067090525585621e-05,
+      "loss": 0.8943,
+      "step": 621
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.052917065444098e-05,
+      "loss": 0.8796,
+      "step": 622
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.03873754053155e-05,
+      "loss": 0.8951,
+      "step": 623
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.024552032521625e-05,
+      "loss": 0.9414,
+      "step": 624
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.010360623122425e-05,
+      "loss": 0.8948,
+      "step": 625
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.996163394076047e-05,
+      "loss": 0.8822,
+      "step": 626
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 6.98196042715811e-05,
+      "loss": 0.9204,
+      "step": 627
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.967751804177279e-05,
+      "loss": 0.8788,
+      "step": 628
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.953537606974799e-05,
+      "loss": 0.9132,
+      "step": 629
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.939317917424028e-05,
+      "loss": 0.9086,
+      "step": 630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.925092817429956e-05,
+      "loss": 0.8749,
+      "step": 631
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 6.910862388928732e-05,
+      "loss": 0.8853,
+      "step": 632
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.896626713887203e-05,
+      "loss": 0.9135,
+      "step": 633
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.882385874302436e-05,
+      "loss": 0.8551,
+      "step": 634
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.868139952201243e-05,
+      "loss": 0.8678,
+      "step": 635
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 6.853889029639712e-05,
+      "loss": 0.9216,
+      "step": 636
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.839633188702733e-05,
+      "loss": 0.8722,
+      "step": 637
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.825372511503526e-05,
+      "loss": 0.9105,
+      "step": 638
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.811107080183171e-05,
+      "loss": 0.8553,
+      "step": 639
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.796836976910128e-05,
+      "loss": 0.8705,
+      "step": 640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 6.782562283879765e-05,
+      "loss": 0.9197,
+      "step": 641
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.768283083313891e-05,
+      "loss": 0.8856,
+      "step": 642
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.753999457460279e-05,
+      "loss": 0.8936,
+      "step": 643
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.739711488592188e-05,
+      "loss": 0.8796,
+      "step": 644
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 6.725419259007895e-05,
+      "loss": 0.8901,
+      "step": 645
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.711122851030217e-05,
+      "loss": 0.9187,
+      "step": 646
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.696822347006038e-05,
+      "loss": 0.8882,
+      "step": 647
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.682517829305842e-05,
+      "loss": 0.9129,
+      "step": 648
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.668209380323221e-05,
+      "loss": 0.9057,
+      "step": 649
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 6.653897082474416e-05,
+      "loss": 0.8826,
+      "step": 650
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.639581018197841e-05,
+      "loss": 0.8933,
+      "step": 651
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.625261269953598e-05,
+      "loss": 0.8698,
+      "step": 652
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.610937920223014e-05,
+      "loss": 0.9076,
+      "step": 653
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 6.596611051508155e-05,
+      "loss": 0.8564,
+      "step": 654
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.58228074633136e-05,
+      "loss": 0.8766,
+      "step": 655
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.567947087234762e-05,
+      "loss": 0.9037,
+      "step": 656
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.553610156779812e-05,
+      "loss": 0.9012,
+      "step": 657
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.539270037546804e-05,
+      "loss": 0.9266,
+      "step": 658
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 6.524926812134396e-05,
+      "loss": 0.8679,
+      "step": 659
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.510580563159145e-05,
+      "loss": 0.9083,
+      "step": 660
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.496231373255014e-05,
+      "loss": 0.8996,
+      "step": 661
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.481879325072914e-05,
+      "loss": 0.9023,
+      "step": 662
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 6.467524501280213e-05,
+      "loss": 0.9016,
+      "step": 663
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.453166984560274e-05,
+      "loss": 0.8643,
+      "step": 664
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.438806857611963e-05,
+      "loss": 0.8907,
+      "step": 665
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.424444203149187e-05,
+      "loss": 0.8973,
+      "step": 666
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.410079103900409e-05,
+      "loss": 0.9139,
+      "step": 667
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 6.395711642608172e-05,
+      "loss": 0.8845,
+      "step": 668
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.381341902028629e-05,
+      "loss": 0.8828,
+      "step": 669
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.366969964931058e-05,
+      "loss": 0.882,
+      "step": 670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.352595914097388e-05,
+      "loss": 0.8749,
+      "step": 671
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 6.338219832321725e-05,
+      "loss": 0.9002,
+      "step": 672
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.323841802409875e-05,
+      "loss": 0.8694,
+      "step": 673
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.309461907178863e-05,
+      "loss": 0.9151,
+      "step": 674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.295080229456456e-05,
+      "loss": 0.9025,
+      "step": 675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.280696852080694e-05,
+      "loss": 0.8891,
+      "step": 676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 6.2663118578994e-05,
+      "loss": 0.9178,
+      "step": 677
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.251925329769718e-05,
+      "loss": 0.9171,
+      "step": 678
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.237537350557617e-05,
+      "loss": 0.9079,
+      "step": 679
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.223148003137435e-05,
+      "loss": 0.8745,
+      "step": 680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 6.208757370391379e-05,
+      "loss": 0.9134,
+      "step": 681
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.194365535209074e-05,
+      "loss": 0.8587,
+      "step": 682
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.179972580487057e-05,
+      "loss": 0.9039,
+      "step": 683
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.165578589128323e-05,
+      "loss": 0.9315,
+      "step": 684
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.151183644041834e-05,
+      "loss": 0.9108,
+      "step": 685
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 6.136787828142047e-05,
+      "loss": 0.8866,
+      "step": 686
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.122391224348433e-05,
+      "loss": 0.8872,
+      "step": 687
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.107993915585001e-05,
+      "loss": 0.9115,
+      "step": 688
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.0935959847798226e-05,
+      "loss": 0.8776,
+      "step": 689
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.079197514864554e-05,
+      "loss": 0.8889,
+      "step": 690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.064798588773952e-05,
+      "loss": 0.922,
+      "step": 691
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.0503992894454006e-05,
+      "loss": 0.907,
+      "step": 692
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.035999699818442e-05,
+      "loss": 0.9033,
+      "step": 693
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.02159990283428e-05,
+      "loss": 0.8618,
+      "step": 694
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.0071999814353204e-05,
+      "loss": 0.8837,
+      "step": 695
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.99280001856468e-05,
+      "loss": 0.8993,
+      "step": 696
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.97840009716572e-05,
+      "loss": 0.8508,
+      "step": 697
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.964000300181559e-05,
+      "loss": 0.9043,
+      "step": 698
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.9496007105546004e-05,
+      "loss": 0.9256,
+      "step": 699
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.935201411226049e-05,
+      "loss": 0.8824,
+      "step": 700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.920802485135447e-05,
+      "loss": 0.9082,
+      "step": 701
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.9064040152201777e-05,
+      "loss": 0.9122,
+      "step": 702
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.892006084415001e-05,
+      "loss": 0.8684,
+      "step": 703
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.87760877565157e-05,
+      "loss": 0.8681,
+      "step": 704
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.863212171857953e-05,
+      "loss": 0.8688,
+      "step": 705
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8488163559581656e-05,
+      "loss": 0.867,
+      "step": 706
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8344214108716775e-05,
+      "loss": 0.8632,
+      "step": 707
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.820027419512944e-05,
+      "loss": 0.8988,
+      "step": 708
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.805634464790927e-05,
+      "loss": 0.8976,
+      "step": 709
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.791242629608622e-05,
+      "loss": 0.8957,
+      "step": 710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.7768519968625685e-05,
+      "loss": 0.8449,
+      "step": 711
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.7624626494423846e-05,
+      "loss": 0.9322,
+      "step": 712
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.748074670230282e-05,
+      "loss": 0.8835,
+      "step": 713
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.733688142100598e-05,
+      "loss": 0.9181,
+      "step": 714
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.7193031479193065e-05,
+      "loss": 0.8979,
+      "step": 715
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.704919770543544e-05,
+      "loss": 0.8659,
+      "step": 716
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.690538092821139e-05,
+      "loss": 0.8898,
+      "step": 717
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.6761581975901255e-05,
+      "loss": 0.8836,
+      "step": 718
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.661780167678277e-05,
+      "loss": 0.9026,
+      "step": 719
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.6474040859026145e-05,
+      "loss": 0.9006,
+      "step": 720
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.633030035068945e-05,
+      "loss": 0.8909,
+      "step": 721
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.6186580979713706e-05,
+      "loss": 0.9137,
+      "step": 722
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.604288357391828e-05,
+      "loss": 0.912,
+      "step": 723
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.5899208960995915e-05,
+      "loss": 0.9013,
+      "step": 724
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.575555796850813e-05,
+      "loss": 0.8905,
+      "step": 725
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.561193142388037e-05,
+      "loss": 0.9419,
+      "step": 726
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.546833015439727e-05,
+      "loss": 0.9136,
+      "step": 727
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.5324754987197876e-05,
+      "loss": 0.8949,
+      "step": 728
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.518120674927088e-05,
+      "loss": 0.8903,
+      "step": 729
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.5037686267449886e-05,
+      "loss": 0.8621,
+      "step": 730
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.489419436840856e-05,
+      "loss": 0.873,
+      "step": 731
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.475073187865603e-05,
+      "loss": 0.8902,
+      "step": 732
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.460729962453197e-05,
+      "loss": 0.8985,
+      "step": 733
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.446389843220189e-05,
+      "loss": 0.8906,
+      "step": 734
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4320529127652394e-05,
+      "loss": 0.8643,
+      "step": 735
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.417719253668641e-05,
+      "loss": 0.8695,
+      "step": 736
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.4033889484918476e-05,
+      "loss": 0.8997,
+      "step": 737
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3890620797769894e-05,
+      "loss": 0.8748,
+      "step": 738
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.374738730046401e-05,
+      "loss": 0.8905,
+      "step": 739
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.360418981802159e-05,
+      "loss": 0.9044,
+      "step": 740
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.3461029175255834e-05,
+      "loss": 0.9008,
+      "step": 741
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.33179061967678e-05,
+      "loss": 0.8648,
+      "step": 742
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.317482170694159e-05,
+      "loss": 0.8987,
+      "step": 743
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.303177652993962e-05,
+      "loss": 0.9102,
+      "step": 744
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.288877148969784e-05,
+      "loss": 0.8639,
+      "step": 745
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.274580740992107e-05,
+      "loss": 0.8713,
+      "step": 746
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.260288511407814e-05,
+      "loss": 0.8878,
+      "step": 747
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.246000542539721e-05,
+      "loss": 0.888,
+      "step": 748
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.2317169166861096e-05,
+      "loss": 0.8598,
+      "step": 749
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.217437716120237e-05,
+      "loss": 0.9094,
+      "step": 750
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.203163023089874e-05,
+      "loss": 0.8648,
+      "step": 751
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.18889291981683e-05,
+      "loss": 0.8844,
+      "step": 752
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.174627488496475e-05,
+      "loss": 0.8593,
+      "step": 753
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.160366811297269e-05,
+      "loss": 0.8916,
+      "step": 754
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.14611097036029e-05,
+      "loss": 0.8647,
+      "step": 755
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.131860047798759e-05,
+      "loss": 0.8975,
+      "step": 756
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.117614125697564e-05,
+      "loss": 0.898,
+      "step": 757
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.103373286112797e-05,
+      "loss": 0.8703,
+      "step": 758
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.089137611071269e-05,
+      "loss": 0.865,
+      "step": 759
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.0749071825700455e-05,
+      "loss": 0.888,
+      "step": 760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.060682082575972e-05,
+      "loss": 0.8707,
+      "step": 761
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.046462393025202e-05,
+      "loss": 0.8743,
+      "step": 762
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.032248195822724e-05,
+      "loss": 0.8957,
+      "step": 763
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.0180395728418923e-05,
+      "loss": 0.8876,
+      "step": 764
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.003836605923952e-05,
+      "loss": 0.8571,
+      "step": 765
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.989639376877574e-05,
+      "loss": 0.8655,
+      "step": 766
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.9754479674783744e-05,
+      "loss": 0.8823,
+      "step": 767
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.96126245946845e-05,
+      "loss": 0.8753,
+      "step": 768
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.9470829345559044e-05,
+      "loss": 0.8835,
+      "step": 769
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.93290947441438e-05,
+      "loss": 0.8921,
+      "step": 770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.918742160682586e-05,
+      "loss": 0.876,
+      "step": 771
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.9045810749638305e-05,
+      "loss": 0.912,
+      "step": 772
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.890426298825546e-05,
+      "loss": 0.9259,
+      "step": 773
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.876277913798819e-05,
+      "loss": 0.8822,
+      "step": 774
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.862136001377932e-05,
+      "loss": 0.8957,
+      "step": 775
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.84800064301988e-05,
+      "loss": 0.8935,
+      "step": 776
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.833871920143905e-05,
+      "loss": 0.8739,
+      "step": 777
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.819749914131032e-05,
+      "loss": 0.88,
+      "step": 778
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.805634706323596e-05,
+      "loss": 0.8859,
+      "step": 779
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.7915263780247744e-05,
+      "loss": 0.852,
+      "step": 780
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.77742501049812e-05,
+      "loss": 0.8774,
+      "step": 781
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7633306849670906e-05,
+      "loss": 0.9166,
+      "step": 782
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.74924348261458e-05,
+      "loss": 0.8773,
+      "step": 783
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.73516348458246e-05,
+      "loss": 0.9,
+      "step": 784
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.7210907719710985e-05,
+      "loss": 0.8678,
+      "step": 785
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.707025425838904e-05,
+      "loss": 0.8743,
+      "step": 786
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.692967527201849e-05,
+      "loss": 0.8748,
+      "step": 787
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.678917157033015e-05,
+      "loss": 0.8972,
+      "step": 788
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.664874396262114e-05,
+      "loss": 0.9149,
+      "step": 789
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.650839325775035e-05,
+      "loss": 0.8609,
+      "step": 790
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.636812026413362e-05,
+      "loss": 0.8849,
+      "step": 791
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.622792578973926e-05,
+      "loss": 0.8396,
+      "step": 792
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.608781064208329e-05,
+      "loss": 0.8749,
+      "step": 793
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.594777562822478e-05,
+      "loss": 0.8876,
+      "step": 794
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.580782155476124e-05,
+      "loss": 0.8608,
+      "step": 795
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.566794922782401e-05,
+      "loss": 0.8541,
+      "step": 796
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.55281594530735e-05,
+      "loss": 0.8943,
+      "step": 797
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.538845303569468e-05,
+      "loss": 0.8855,
+      "step": 798
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.524883078039236e-05,
+      "loss": 0.9044,
+      "step": 799
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.5109293491386574e-05,
+      "loss": 0.8696,
+      "step": 800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4969841972407975e-05,
+      "loss": 0.8856,
+      "step": 801
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4830477026693174e-05,
+      "loss": 0.8855,
+      "step": 802
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4691199456980124e-05,
+      "loss": 0.8757,
+      "step": 803
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.455201006550346e-05,
+      "loss": 0.8659,
+      "step": 804
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.441290965398997e-05,
+      "loss": 0.8783,
+      "step": 805
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.427389902365389e-05,
+      "loss": 0.8709,
+      "step": 806
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.413497897519229e-05,
+      "loss": 0.8902,
+      "step": 807
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.399615030878056e-05,
+      "loss": 0.8592,
+      "step": 808
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.385741382406768e-05,
+      "loss": 0.8872,
+      "step": 809
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3718770320171675e-05,
+      "loss": 0.8412,
+      "step": 810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.3580220595675034e-05,
+      "loss": 0.8991,
+      "step": 811
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.344176544862003e-05,
+      "loss": 0.8699,
+      "step": 812
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.330340567650421e-05,
+      "loss": 0.899,
+      "step": 813
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.316514207627574e-05,
+      "loss": 0.8972,
+      "step": 814
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.302697544432887e-05,
+      "loss": 0.9027,
+      "step": 815
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.288890657649926e-05,
+      "loss": 0.8791,
+      "step": 816
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.275093626805952e-05,
+      "loss": 0.8567,
+      "step": 817
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.261306531371451e-05,
+      "loss": 0.8879,
+      "step": 818
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.247529450759686e-05,
+      "loss": 0.8643,
+      "step": 819
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.2337624643262304e-05,
+      "loss": 0.8892,
+      "step": 820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.220005651368516e-05,
+      "loss": 0.8808,
+      "step": 821
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.2062590911253785e-05,
+      "loss": 0.8825,
+      "step": 822
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.192522862776596e-05,
+      "loss": 0.9095,
+      "step": 823
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.178797045442435e-05,
+      "loss": 0.8632,
+      "step": 824
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.165081718183195e-05,
+      "loss": 0.8327,
+      "step": 825
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.151376959998756e-05,
+      "loss": 0.8498,
+      "step": 826
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.1376828498281155e-05,
+      "loss": 0.8993,
+      "step": 827
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.123999466548944e-05,
+      "loss": 0.8822,
+      "step": 828
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.110326888977122e-05,
+      "loss": 0.8647,
+      "step": 829
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.09666519586629e-05,
+      "loss": 0.8726,
+      "step": 830
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.083014465907394e-05,
+      "loss": 0.8798,
+      "step": 831
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0693747777282334e-05,
+      "loss": 0.909,
+      "step": 832
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.0557462098930055e-05,
+      "loss": 0.8665,
+      "step": 833
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.042128840901858e-05,
+      "loss": 0.8626,
+      "step": 834
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.028522749190431e-05,
+      "loss": 0.8765,
+      "step": 835
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.014928013129409e-05,
+      "loss": 0.8295,
+      "step": 836
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.001344711024069e-05,
+      "loss": 0.8936,
+      "step": 837
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.987772921113826e-05,
+      "loss": 0.893,
+      "step": 838
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.974212721571788e-05,
+      "loss": 0.8726,
+      "step": 839
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.960664190504301e-05,
+      "loss": 0.8881,
+      "step": 840
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.9471274059505005e-05,
+      "loss": 0.8708,
+      "step": 841
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.9336024458818635e-05,
+      "loss": 0.8602,
+      "step": 842
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.920089388201759e-05,
+      "loss": 0.8906,
+      "step": 843
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.906588310745e-05,
+      "loss": 0.8987,
+      "step": 844
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.893099291277388e-05,
+      "loss": 0.898,
+      "step": 845
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.879622407495277e-05,
+      "loss": 0.8516,
+      "step": 846
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.866157737025116e-05,
+      "loss": 0.8571,
+      "step": 847
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.852705357423007e-05,
+      "loss": 0.8516,
+      "step": 848
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.8392653461742577e-05,
+      "loss": 0.8529,
+      "step": 849
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.825837780692934e-05,
+      "loss": 0.8345,
+      "step": 850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.812422738321411e-05,
+      "loss": 0.8737,
+      "step": 851
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7990202963299366e-05,
+      "loss": 0.8744,
+      "step": 852
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.785630531916181e-05,
+      "loss": 0.8445,
+      "step": 853
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.772253522204784e-05,
+      "loss": 0.8953,
+      "step": 854
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7588893442469306e-05,
+      "loss": 0.8855,
+      "step": 855
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7455380750198846e-05,
+      "loss": 0.865,
+      "step": 856
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.7321997914265616e-05,
+      "loss": 0.8641,
+      "step": 857
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.718874570295078e-05,
+      "loss": 0.8741,
+      "step": 858
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.705562488378312e-05,
+      "loss": 0.855,
+      "step": 859
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.69226362235346e-05,
+      "loss": 0.8534,
+      "step": 860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.678978048821595e-05,
+      "loss": 0.8412,
+      "step": 861
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.665705844307227e-05,
+      "loss": 0.8799,
+      "step": 862
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.65244708525786e-05,
+      "loss": 0.88,
+      "step": 863
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.6392018480435505e-05,
+      "loss": 0.8754,
+      "step": 864
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.6259702089564735e-05,
+      "loss": 0.9,
+      "step": 865
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.612752244210476e-05,
+      "loss": 0.8999,
+      "step": 866
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.599548029940642e-05,
+      "loss": 0.8757,
+      "step": 867
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5863576422028536e-05,
+      "loss": 0.9028,
+      "step": 868
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.573181156973351e-05,
+      "loss": 0.8594,
+      "step": 869
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.560018650148302e-05,
+      "loss": 0.8576,
+      "step": 870
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5468701975433504e-05,
+      "loss": 0.8628,
+      "step": 871
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5337358748931946e-05,
+      "loss": 0.8762,
+      "step": 872
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.520615757851144e-05,
+      "loss": 0.8671,
+      "step": 873
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.507509921988682e-05,
+      "loss": 0.8806,
+      "step": 874
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4944184427950325e-05,
+      "loss": 0.8895,
+      "step": 875
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4813413956767295e-05,
+      "loss": 0.8812,
+      "step": 876
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.468278855957174e-05,
+      "loss": 0.8756,
+      "step": 877
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.455230898876205e-05,
+      "loss": 0.8863,
+      "step": 878
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4421975995896726e-05,
+      "loss": 0.8704,
+      "step": 879
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.429179033168992e-05,
+      "loss": 0.8714,
+      "step": 880
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.416175274600717e-05,
+      "loss": 0.8869,
+      "step": 881
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.403186398786112e-05,
+      "loss": 0.8608,
+      "step": 882
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.3902124805407154e-05,
+      "loss": 0.8984,
+      "step": 883
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.377253594593912e-05,
+      "loss": 0.8477,
+      "step": 884
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.364309815588499e-05,
+      "loss": 0.8735,
+      "step": 885
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.351381218080258e-05,
+      "loss": 0.8592,
+      "step": 886
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.3384678765375257e-05,
+      "loss": 0.837,
+      "step": 887
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.325569865340771e-05,
+      "loss": 0.8653,
+      "step": 888
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.312687258782151e-05,
+      "loss": 0.8385,
+      "step": 889
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.2998201310650995e-05,
+      "loss": 0.8794,
+      "step": 890
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.286968556303887e-05,
+      "loss": 0.8761,
+      "step": 891
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.274132608523204e-05,
+      "loss": 0.882,
+      "step": 892
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.261312361657727e-05,
+      "loss": 0.8849,
+      "step": 893
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.2485078895517e-05,
+      "loss": 0.8571,
+      "step": 894
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.235719265958498e-05,
+      "loss": 0.8554,
+      "step": 895
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.222946564540217e-05,
+      "loss": 0.8459,
+      "step": 896
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.210189858867238e-05,
+      "loss": 0.8627,
+      "step": 897
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.197449222417804e-05,
+      "loss": 0.8543,
+      "step": 898
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.184724728577603e-05,
+      "loss": 0.8538,
+      "step": 899
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.1720164506393426e-05,
+      "loss": 0.8591,
+      "step": 900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.1593244618023246e-05,
+      "loss": 0.4844,
+      "step": 901
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.14664883517203e-05,
+      "loss": 0.4434,
+      "step": 902
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.13398964375969e-05,
+      "loss": 0.4157,
+      "step": 903
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.121346960481869e-05,
+      "loss": 0.392,
+      "step": 904
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.108720858160052e-05,
+      "loss": 0.4112,
+      "step": 905
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.0961114095202115e-05,
+      "loss": 0.4132,
+      "step": 906
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.083518687192397e-05,
+      "loss": 0.4043,
+      "step": 907
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.070942763710314e-05,
+      "loss": 0.3904,
+      "step": 908
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.0583837115109085e-05,
+      "loss": 0.3975,
+      "step": 909
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.045841602933947e-05,
+      "loss": 0.3928,
+      "step": 910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.0333165102216057e-05,
+      "loss": 0.3887,
+      "step": 911
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.020808505518045e-05,
+      "loss": 0.3765,
+      "step": 912
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.0083176608689983e-05,
+      "loss": 0.3679,
+      "step": 913
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9958440482213635e-05,
+      "loss": 0.4016,
+      "step": 914
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.9833877394227778e-05,
+      "loss": 0.3845,
+      "step": 915
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9709488062212084e-05,
+      "loss": 0.3582,
+      "step": 916
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9585273202645425e-05,
+      "loss": 0.3709,
+      "step": 917
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9461233531001697e-05,
+      "loss": 0.3715,
+      "step": 918
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9337369761745715e-05,
+      "loss": 0.3608,
+      "step": 919
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.9213682608329127e-05,
+      "loss": 0.3733,
+      "step": 920
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.9090172783186234e-05,
+      "loss": 0.3788,
+      "step": 921
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.8966840997729996e-05,
+      "loss": 0.3656,
+      "step": 922
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.884368796234783e-05,
+      "loss": 0.3637,
+      "step": 923
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.8720714386397548e-05,
+      "loss": 0.3615,
+      "step": 924
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.8597920978203288e-05,
+      "loss": 0.3734,
+      "step": 925
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.8475308445051456e-05,
+      "loss": 0.3603,
+      "step": 926
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.835287749318658e-05,
+      "loss": 0.3749,
+      "step": 927
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.8230628827807295e-05,
+      "loss": 0.3683,
+      "step": 928
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.8108563153062326e-05,
+      "loss": 0.3673,
+      "step": 929
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.7986681172046306e-05,
+      "loss": 0.3817,
+      "step": 930
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.7864983586795808e-05,
+      "loss": 0.3785,
+      "step": 931
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.774347109828535e-05,
+      "loss": 0.3589,
+      "step": 932
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.7622144406423235e-05,
+      "loss": 0.3586,
+      "step": 933
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.7501004210047614e-05,
+      "loss": 0.3766,
+      "step": 934
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.73800512069224e-05,
+      "loss": 0.359,
+      "step": 935
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.7259286093733313e-05,
+      "loss": 0.3778,
+      "step": 936
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.7138709566083814e-05,
+      "loss": 0.3664,
+      "step": 937
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.701832231849112e-05,
+      "loss": 0.3484,
+      "step": 938
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6898125044382186e-05,
+      "loss": 0.3625,
+      "step": 939
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6778118436089773e-05,
+      "loss": 0.3585,
+      "step": 940
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6658303184848363e-05,
+      "loss": 0.3774,
+      "step": 941
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.6538679980790244e-05,
+      "loss": 0.3574,
+      "step": 942
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6419249512941523e-05,
+      "loss": 0.342,
+      "step": 943
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.630001246921814e-05,
+      "loss": 0.3684,
+      "step": 944
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6180969536421928e-05,
+      "loss": 0.373,
+      "step": 945
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.6062121400236623e-05,
+      "loss": 0.3772,
+      "step": 946
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.5943468745224004e-05,
+      "loss": 0.3624,
+      "step": 947
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.5825012254819793e-05,
+      "loss": 0.3545,
+      "step": 948
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.5706752611329903e-05,
+      "loss": 0.3572,
+      "step": 949
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.5588690495926343e-05,
+      "loss": 0.3541,
+      "step": 950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.5470826588643393e-05,
+      "loss": 0.3597,
+      "step": 951
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.5353161568373653e-05,
+      "loss": 0.3567,
+      "step": 952
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.523569611286415e-05,
+      "loss": 0.3684,
+      "step": 953
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.5118430898712418e-05,
+      "loss": 0.3586,
+      "step": 954
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.5001366601362606e-05,
+      "loss": 0.3616,
+      "step": 955
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.488450389510157e-05,
+      "loss": 0.3661,
+      "step": 956
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.4767843453055065e-05,
+      "loss": 0.3443,
+      "step": 957
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.4651385947183756e-05,
+      "loss": 0.3586,
+      "step": 958
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.4535132048279413e-05,
+      "loss": 0.3522,
+      "step": 959
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.4419082425961047e-05,
+      "loss": 0.3683,
+      "step": 960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.4303237748671032e-05,
+      "loss": 0.3437,
+      "step": 961
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.418759868367126e-05,
+      "loss": 0.3581,
+      "step": 962
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.4072165897039327e-05,
+      "loss": 0.376,
+      "step": 963
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.3956940053664616e-05,
+      "loss": 0.3558,
+      "step": 964
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.384192181724462e-05,
+      "loss": 0.3449,
+      "step": 965
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3727111850280917e-05,
+      "loss": 0.3712,
+      "step": 966
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.361251081407555e-05,
+      "loss": 0.3631,
+      "step": 967
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3498119368727052e-05,
+      "loss": 0.3532,
+      "step": 968
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.3383938173126764e-05,
+      "loss": 0.3545,
+      "step": 969
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.3269967884954974e-05,
+      "loss": 0.3638,
+      "step": 970
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.315620916067716e-05,
+      "loss": 0.3532,
+      "step": 971
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.304266265554019e-05,
+      "loss": 0.3485,
+      "step": 972
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.2929329023568543e-05,
+      "loss": 0.3604,
+      "step": 973
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.281620891756061e-05,
+      "loss": 0.3547,
+      "step": 974
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.2703302989084833e-05,
+      "loss": 0.3621,
+      "step": 975
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.2590611888475993e-05,
+      "loss": 0.3608,
+      "step": 976
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.2478136264831488e-05,
+      "loss": 0.3623,
+      "step": 977
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.236587676600757e-05,
+      "loss": 0.3622,
+      "step": 978
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.225383403861562e-05,
+      "loss": 0.3664,
+      "step": 979
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.214200872801841e-05,
+      "loss": 0.3603,
+      "step": 980
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.2030401478326398e-05,
+      "loss": 0.3437,
+      "step": 981
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.191901293239406e-05,
+      "loss": 0.3715,
+      "step": 982
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.180784373181609e-05,
+      "loss": 0.3613,
+      "step": 983
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1696894516923776e-05,
+      "loss": 0.3566,
+      "step": 984
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1586165926781324e-05,
+      "loss": 0.3476,
+      "step": 985
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.147565859918211e-05,
+      "loss": 0.3506,
+      "step": 986
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1365373170645063e-05,
+      "loss": 0.3543,
+      "step": 987
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.1255310276410968e-05,
+      "loss": 0.3517,
+      "step": 988
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.114547055043883e-05,
+      "loss": 0.3567,
+      "step": 989
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.103585462540221e-05,
+      "loss": 0.3647,
+      "step": 990
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.0926463132685555e-05,
+      "loss": 0.364,
+      "step": 991
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.0817296702380662e-05,
+      "loss": 0.3489,
+      "step": 992
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.07083559632829e-05,
+      "loss": 0.3516,
+      "step": 993
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.059964154288769e-05,
+      "loss": 0.3576,
+      "step": 994
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.049115406738687e-05,
+      "loss": 0.3573,
+      "step": 995
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.0382894161665065e-05,
+      "loss": 0.3637,
+      "step": 996
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.0274862449296132e-05,
+      "loss": 0.348,
+      "step": 997
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.016705955253951e-05,
+      "loss": 0.361,
+      "step": 998
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.005948609233668e-05,
+      "loss": 0.3559,
+      "step": 999
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.99521426883076e-05,
+      "loss": 0.3608,
+      "step": 1000
+    }
+  ],
+  "max_steps": 1350,
+  "num_train_epochs": 3,
+  "total_flos": 9.936494717011231e+17,
+  "trial_name": null,
+  "trial_params": null
+}