diff --git "a/multimodal-vicuna-7b-v1.5-video-naivemc/trainer_state.json" "b/multimodal-vicuna-7b-v1.5-video-naivemc/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/multimodal-vicuna-7b-v1.5-video-naivemc/trainer_state.json"
@@ -0,0 +1,9397 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.99968,
+  "global_step": 1562,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.255319148936171e-06,
+      "loss": 1.7417,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.510638297872341e-06,
+      "loss": 1.7839,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2765957446808511e-05,
+      "loss": 1.7869,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7021276595744682e-05,
+      "loss": 1.9512,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1276595744680852e-05,
+      "loss": 1.8318,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5531914893617022e-05,
+      "loss": 1.7205,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9787234042553192e-05,
+      "loss": 1.7183,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4042553191489365e-05,
+      "loss": 1.5933,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.829787234042553e-05,
+      "loss": 1.5022,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2553191489361704e-05,
+      "loss": 1.51,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.680851063829788e-05,
+      "loss": 1.5203,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1063829787234044e-05,
+      "loss": 1.4753,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.531914893617022e-05,
+      "loss": 1.5745,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9574468085106384e-05,
+      "loss": 1.5435,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.382978723404256e-05,
+      "loss": 1.4009,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.808510638297873e-05,
+      "loss": 1.4255,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.23404255319149e-05,
+      "loss": 1.3938,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.659574468085106e-05,
+      "loss": 1.3752,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.085106382978723e-05,
+      "loss": 1.4204,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.510638297872341e-05,
+      "loss": 1.4148,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.936170212765958e-05,
+      "loss": 1.4078,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.361702127659576e-05,
+      "loss": 1.3379,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.787234042553192e-05,
+      "loss": 1.3192,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010212765957446809,
+      "loss": 1.3999,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010638297872340425,
+      "loss": 1.3838,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011063829787234043,
+      "loss": 1.3392,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011489361702127661,
+      "loss": 1.2795,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011914893617021277,
+      "loss": 1.3931,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012340425531914893,
+      "loss": 1.3298,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012765957446808513,
+      "loss": 1.3601,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013191489361702127,
+      "loss": 1.2849,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013617021276595746,
+      "loss": 1.3647,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014042553191489363,
+      "loss": 1.3516,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001446808510638298,
+      "loss": 1.3466,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014893617021276596,
+      "loss": 1.2629,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015319148936170213,
+      "loss": 1.3135,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015744680851063832,
+      "loss": 1.3647,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016170212765957446,
+      "loss": 1.2788,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00016595744680851065,
+      "loss": 1.3026,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017021276595744682,
+      "loss": 1.3643,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017446808510638298,
+      "loss": 1.3149,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017872340425531915,
+      "loss": 1.3228,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018297872340425532,
+      "loss": 1.266,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001872340425531915,
+      "loss": 1.2443,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019148936170212768,
+      "loss": 1.2927,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019574468085106384,
+      "loss": 1.3774,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 1.309,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999978499709475,
+      "loss": 1.2808,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999913998930348,
+      "loss": 1.2878,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999806497939982,
+      "loss": 1.2429,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999655997200635,
+      "loss": 1.2528,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999462497359466,
+      "loss": 1.2949,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999225999248538,
+      "loss": 1.3453,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019998946503884809,
+      "loss": 1.3411,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999862401247012,
+      "loss": 1.2302,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998258526391207,
+      "loss": 1.3079,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999785004721968,
+      "loss": 1.3201,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999739857671202,
+      "loss": 1.2678,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996904116809585,
+      "loss": 1.3325,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996366669638573,
+      "loss": 1.3556,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999578623751004,
+      "loss": 1.2568,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995162822919883,
+      "loss": 1.3364,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994496428548812,
+      "loss": 1.2809,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999378705726237,
+      "loss": 1.2058,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993034712110887,
+      "loss": 1.1538,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019992239396329498,
+      "loss": 1.3418,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991401113338104,
+      "loss": 1.2815,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990519866741371,
+      "loss": 1.313,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001998959566032871,
+      "loss": 1.3596,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019988628498074262,
+      "loss": 1.2872,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019987618384136879,
+      "loss": 1.2368,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986565322860115,
+      "loss": 1.2751,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019985469318772192,
+      "loss": 1.2997,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019984330376585988,
+      "loss": 1.3271,
+      "step": 74
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983148501199026,
+      "loss": 1.2793,
+      "step": 75
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019981923697693437,
+      "loss": 1.2528,
+      "step": 76
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980655971335945,
+      "loss": 1.3113,
+      "step": 77
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997934532757785,
+      "loss": 1.3623,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977991772054992,
+      "loss": 1.3097,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019976595310587742,
+      "loss": 1.3203,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975155949180967,
+      "loss": 1.2419,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973673694024,
+      "loss": 1.297,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019972148551490628,
+      "loss": 1.3119,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997058052813905,
+      "loss": 1.2587,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001996896963071186,
+      "loss": 1.2898,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996731586613601,
+      "loss": 1.2974,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996561924152278,
+      "loss": 1.3253,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963879764167762,
+      "loss": 1.2562,
+      "step": 88
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019962097441550802,
+      "loss": 1.3362,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019960272281335995,
+      "loss": 1.2628,
+      "step": 90
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958404291371635,
+      "loss": 1.2172,
+      "step": 91
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995649347969019,
+      "loss": 1.295,
+      "step": 92
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019954539854508254,
+      "loss": 1.198,
+      "step": 93
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019952543424226534,
+      "loss": 1.2078,
+      "step": 94
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019950504197429795,
+      "loss": 1.2612,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019948422182886833,
+      "loss": 1.2709,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946297389550433,
+      "loss": 1.2277,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994412982655732,
+      "loss": 1.296,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019941919503228153,
+      "loss": 1.23,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019939666429067443,
+      "loss": 1.3438,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019937370613763543,
+      "loss": 1.3506,
+      "step": 101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993503206718859,
+      "loss": 1.333,
+      "step": 102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019932650799398474,
+      "loss": 1.3064,
+      "step": 103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019930226820632778,
+      "loss": 1.2507,
+      "step": 104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019927760141314756,
+      "loss": 1.2402,
+      "step": 105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925250772051276,
+      "loss": 1.3029,
+      "step": 106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922698723632767,
+      "loss": 1.3087,
+      "step": 107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019920104007033185,
+      "loss": 1.2673,
+      "step": 108
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019917466633409965,
+      "loss": 1.2455,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991478661410396,
+      "loss": 1.2919,
+      "step": 110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001991206396063942,
+      "loss": 1.2332,
+      "step": 111
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909298684723904,
+      "loss": 1.2898,
+      "step": 112
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019906490798248265,
+      "loss": 1.2184,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019903640313286572,
+      "loss": 1.3071,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019900747242096082,
+      "loss": 1.2631,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019897811597117168,
+      "loss": 1.2346,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019894833390973266,
+      "loss": 1.2219,
+      "step": 117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019891812636470846,
+      "loss": 1.2968,
+      "step": 118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019888749346599321,
+      "loss": 1.2018,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019885643534531022,
+      "loss": 1.1738,
+      "step": 120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019882495213621116,
+      "loss": 1.262,
+      "step": 121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987930439740757,
+      "loss": 1.2281,
+      "step": 122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987607109961107,
+      "loss": 1.2307,
+      "step": 123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019872795334134998,
+      "loss": 1.2157,
+      "step": 124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019869477115065325,
+      "loss": 1.162,
+      "step": 125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986611645667059,
+      "loss": 1.26,
+      "step": 126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986271337340182,
+      "loss": 1.3362,
+      "step": 127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985926787989247,
+      "loss": 1.2592,
+      "step": 128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985577999095836,
+      "loss": 1.2189,
+      "step": 129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019852249721597618,
+      "loss": 1.1444,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019848677086990605,
+      "loss": 1.2634,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984506210249986,
+      "loss": 1.3064,
+      "step": 132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019841404783670026,
+      "loss": 1.2454,
+      "step": 133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019837705146227784,
+      "loss": 1.2474,
+      "step": 134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019833963206081795,
+      "loss": 1.1831,
+      "step": 135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019830178979322614,
+      "loss": 1.2302,
+      "step": 136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826352482222638,
+      "loss": 1.2761,
+      "step": 137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019822483731236025,
+      "loss": 1.2333,
+      "step": 138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019818572742998634,
+      "loss": 1.2295,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019814619534327935,
+      "loss": 1.1975,
+      "step": 140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981062412222296,
+      "loss": 1.2122,
+      "step": 141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980658652386421,
+      "loss": 1.235,
+      "step": 142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019802506756613596,
+      "loss": 1.2147,
+      "step": 143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001979838483801435,
+      "loss": 1.2522,
+      "step": 144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019794220785790967,
+      "loss": 1.2301,
+      "step": 145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019790014617849106,
+      "loss": 1.2216,
+      "step": 146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785766352275542,
+      "loss": 1.2888,
+      "step": 147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019781476007338058,
+      "loss": 1.2786,
+      "step": 148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019777143601485387,
+      "loss": 1.2614,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019772769153347127,
+      "loss": 1.2789,
+      "step": 150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019768352681733662,
+      "loss": 1.2466,
+      "step": 151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019763894205636072,
+      "loss": 1.2343,
+      "step": 152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019759393744226065,
+      "loss": 1.2729,
+      "step": 153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019754851316855886,
+      "loss": 1.2662,
+      "step": 154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019750266943058237,
+      "loss": 1.2278,
+      "step": 155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019745640642546196,
+      "loss": 1.2419,
+      "step": 156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019740972435213115,
+      "loss": 1.2341,
+      "step": 157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019736262341132563,
+      "loss": 1.3319,
+      "step": 158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019731510380558218,
+      "loss": 1.248,
+      "step": 159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019726716573923784,
+      "loss": 1.1666,
+      "step": 160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019721880941842913,
+      "loss": 1.3455,
+      "step": 161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019717003505109095,
+      "loss": 1.3564,
+      "step": 162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019712084284695603,
+      "loss": 1.1661,
+      "step": 163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001970712330175536,
+      "loss": 1.2573,
+      "step": 164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019702120577620887,
+      "loss": 1.2416,
+      "step": 165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019697076133804185,
+      "loss": 1.1995,
+      "step": 166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019691989991996663,
+      "loss": 1.2102,
+      "step": 167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019686862174069015,
+      "loss": 1.2968,
+      "step": 168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019681692702071163,
+      "loss": 1.3381,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019676481598232139,
+      "loss": 1.248,
+      "step": 170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019671228884959987,
+      "loss": 1.28,
+      "step": 171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019665934584841682,
+      "loss": 1.1851,
+      "step": 172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001966059872064302,
+      "loss": 1.1873,
+      "step": 173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019655221315308528,
+      "loss": 1.2014,
+      "step": 174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019649802391961363,
+      "loss": 1.2319,
+      "step": 175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019644341973903208,
+      "loss": 1.2565,
+      "step": 176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019638840084614182,
+      "loss": 1.212,
+      "step": 177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019633296747752723,
+      "loss": 1.1847,
+      "step": 178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019627711987155503,
+      "loss": 1.2024,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019622085826837324,
+      "loss": 1.3152,
+      "step": 180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019616418290990993,
+      "loss": 1.2759,
+      "step": 181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019610709403987246,
+      "loss": 1.1858,
+      "step": 182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019604959190374632,
+      "loss": 1.1573,
+      "step": 183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195991676748794,
+      "loss": 1.3047,
+      "step": 184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019593334882405408,
+      "loss": 1.214,
+      "step": 185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019587460838033996,
+      "loss": 1.2463,
+      "step": 186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195815455670239,
+      "loss": 1.1788,
+      "step": 187
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019575589094811128,
+      "loss": 1.1427,
+      "step": 188
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001956959144700886,
+      "loss": 1.2283,
+      "step": 189
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019563552649407324,
+      "loss": 1.2657,
+      "step": 190
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019557472727973707,
+      "loss": 1.2681,
+      "step": 191
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955135170885202,
+      "loss": 1.2141,
+      "step": 192
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019545189618363001,
+      "loss": 1.2653,
+      "step": 193
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019538986483004,
+      "loss": 1.2444,
+      "step": 194
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019532742329448854,
+      "loss": 1.2416,
+      "step": 195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019526457184547793,
+      "loss": 1.3125,
+      "step": 196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019520131075327298,
+      "loss": 1.2513,
+      "step": 197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019513764028990016,
+      "loss": 1.235,
+      "step": 198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019507356072914607,
+      "loss": 1.238,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001950090723465566,
+      "loss": 1.214,
+      "step": 200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019494417541943547,
+      "loss": 1.2566,
+      "step": 201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019487887022684336,
+      "loss": 1.2634,
+      "step": 202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001948131570495963,
+      "loss": 1.2087,
+      "step": 203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019474703617026478,
+      "loss": 1.3126,
+      "step": 204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019468050787317242,
+      "loss": 1.2074,
+      "step": 205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019461357244439479,
+      "loss": 1.2638,
+      "step": 206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019454623017175812,
+      "loss": 1.2639,
+      "step": 207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001944784813448381,
+      "loss": 1.2058,
+      "step": 208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019441032625495857,
+      "loss": 1.2854,
+      "step": 209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019434176519519047,
+      "loss": 1.2042,
+      "step": 210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019427279846035025,
+      "loss": 1.2153,
+      "step": 211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001942034263469989,
+      "loss": 1.2743,
+      "step": 212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001941336491534406,
+      "loss": 1.2437,
+      "step": 213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019406346717972125,
+      "loss": 1.1423,
+      "step": 214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019399288072762747,
+      "loss": 1.332,
+      "step": 215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019392189010068508,
+      "loss": 1.2505,
+      "step": 216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019385049560415794,
+      "loss": 1.3014,
+      "step": 217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019377869754504644,
+      "loss": 1.2224,
+      "step": 218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019370649623208651,
+      "loss": 1.2295,
+      "step": 219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019363389197574798,
+      "loss": 1.2991,
+      "step": 220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001935608850882333,
+      "loss": 1.2339,
+      "step": 221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019348747588347637,
+      "loss": 1.1903,
+      "step": 222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019341366467714104,
+      "loss": 1.2069,
+      "step": 223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019333945178661978,
+      "loss": 1.2621,
+      "step": 224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019326483753103236,
+      "loss": 1.2079,
+      "step": 225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019318982223122437,
+      "loss": 1.276,
+      "step": 226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019311440620976597,
+      "loss": 1.2046,
+      "step": 227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019303858979095043,
+      "loss": 1.3154,
+      "step": 228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019296237330079278,
+      "loss": 1.3109,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019288575706702835,
+      "loss": 1.2039,
+      "step": 230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019280874141911137,
+      "loss": 1.2771,
+      "step": 231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019273132668821364,
+      "loss": 1.2169,
+      "step": 232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019265351320722296,
+      "loss": 1.2961,
+      "step": 233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019257530131074185,
+      "loss": 1.228,
+      "step": 234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192496691335086,
+      "loss": 1.2474,
+      "step": 235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001924176836182829,
+      "loss": 1.2238,
+      "step": 236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019233827850007027,
+      "loss": 1.2428,
+      "step": 237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019225847632189474,
+      "loss": 1.2717,
+      "step": 238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019217827742691037,
+      "loss": 1.2773,
+      "step": 239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192097682159977,
+      "loss": 1.2352,
+      "step": 240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019201669086765902,
+      "loss": 1.2473,
+      "step": 241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019193530389822363,
+      "loss": 1.1792,
+      "step": 242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001918535216016396,
+      "loss": 1.2341,
+      "step": 243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001917713443295755,
+      "loss": 1.1545,
+      "step": 244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001916887724353984,
+      "loss": 1.2286,
+      "step": 245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019160580627417223,
+      "loss": 1.2582,
+      "step": 246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915224462026563,
+      "loss": 1.1648,
+      "step": 247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019143869257930377,
+      "loss": 1.3004,
+      "step": 248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001913545457642601,
+      "loss": 1.2527,
+      "step": 249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019127000611936148,
+      "loss": 1.175,
+      "step": 250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019118507400813325,
+      "loss": 1.2183,
+      "step": 251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910997497957885,
+      "loss": 1.2338,
+      "step": 252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019101403384922625,
+      "loss": 1.2175,
+      "step": 253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019092792653703007,
+      "loss": 1.1876,
+      "step": 254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001908414282294664,
+      "loss": 1.2152,
+      "step": 255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000190754539298483,
+      "loss": 1.1963,
+      "step": 256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019066726011770726,
+      "loss": 1.3168,
+      "step": 257
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001905795910624448,
+      "loss": 1.2391,
+      "step": 258
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019049153250967767,
+      "loss": 1.2212,
+      "step": 259
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019040308483806267,
+      "loss": 1.1311,
+      "step": 260
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019031424842793,
+      "loss": 1.1852,
+      "step": 261
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019022502366128135,
+      "loss": 1.2108,
+      "step": 262
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019013541092178843,
+      "loss": 1.1906,
+      "step": 263
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019004541059479123,
+      "loss": 1.2416,
+      "step": 264
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018995502306729634,
+      "loss": 1.2911,
+      "step": 265
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898642487279754,
+      "loss": 1.1554,
+      "step": 266
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897730879671634,
+      "loss": 1.2697,
+      "step": 267
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018968154117685683,
+      "loss": 1.2675,
+      "step": 268
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018958960875071226,
+      "loss": 1.2365,
+      "step": 269
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001894972910840444,
+      "loss": 1.1556,
+      "step": 270
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018940458857382467,
+      "loss": 1.1615,
+      "step": 271
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018931150161867916,
+      "loss": 1.2577,
+      "step": 272
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018921803061888728,
+      "loss": 1.2688,
+      "step": 273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018912417597637967,
+      "loss": 1.2346,
+      "step": 274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001890299380947368,
+      "loss": 1.2354,
+      "step": 275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018893531737918702,
+      "loss": 1.1978,
+      "step": 276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888403142366049,
+      "loss": 1.1801,
+      "step": 277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001887449290755095,
+      "loss": 1.2578,
+      "step": 278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018864916230606254,
+      "loss": 1.1904,
+      "step": 279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018855301434006667,
+      "loss": 1.2335,
+      "step": 280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018845648559096377,
+      "loss": 1.1714,
+      "step": 281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018835957647383303,
+      "loss": 1.2479,
+      "step": 282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001882622874053893,
+      "loss": 1.2498,
+      "step": 283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018816461880398125,
+      "loss": 1.2019,
+      "step": 284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018806657108958952,
+      "loss": 1.2397,
+      "step": 285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018796814468382498,
+      "loss": 1.2969,
+      "step": 286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018786934000992688,
+      "loss": 1.2114,
+      "step": 287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001877701574927611,
+      "loss": 1.2269,
+      "step": 288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018767059755881818,
+      "loss": 1.2316,
+      "step": 289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018757066063621166,
+      "loss": 1.1359,
+      "step": 290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018747034715467612,
+      "loss": 1.1882,
+      "step": 291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018736965754556528,
+      "loss": 1.2671,
+      "step": 292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001872685922418504,
+      "loss": 1.1952,
+      "step": 293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001871671516781181,
+      "loss": 1.2418,
+      "step": 294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018706533629056873,
+      "loss": 1.2175,
+      "step": 295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018696314651701437,
+      "loss": 1.2578,
+      "step": 296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018686058279687698,
+      "loss": 1.2942,
+      "step": 297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018675764557118653,
+      "loss": 1.1584,
+      "step": 298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018665433528257904,
+      "loss": 1.2019,
+      "step": 299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018655065237529477,
+      "loss": 1.1372,
+      "step": 300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018644659729517623,
+      "loss": 1.2638,
+      "step": 301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018634217048966637,
+      "loss": 1.2821,
+      "step": 302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018623737240780646,
+      "loss": 1.2134,
+      "step": 303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018613220350023434,
+      "loss": 1.2821,
+      "step": 304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018602666421918245,
+      "loss": 1.1929,
+      "step": 305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018592075501847584,
+      "loss": 1.2036,
+      "step": 306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858144763535302,
+      "loss": 1.3065,
+      "step": 307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018570782868134998,
+      "loss": 1.1853,
+      "step": 308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018560081246052633,
+      "loss": 1.194,
+      "step": 309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018549342815123528,
+      "loss": 1.2194,
+      "step": 310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001853856762152356,
+      "loss": 1.2305,
+      "step": 311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018527755711586678,
+      "loss": 1.2203,
+      "step": 312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018516907131804734,
+      "loss": 1.1858,
+      "step": 313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018506021928827243,
+      "loss": 1.1719,
+      "step": 314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001849510014946122,
+      "loss": 1.2834,
+      "step": 315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001848414184067094,
+      "loss": 1.224,
+      "step": 316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018473147049577774,
+      "loss": 1.3112,
+      "step": 317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018462115823459962,
+      "loss": 1.2651,
+      "step": 318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018451048209752418,
+      "loss": 1.1859,
+      "step": 319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018439944256046518,
+      "loss": 1.2177,
+      "step": 320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001842880401008992,
+      "loss": 1.2284,
+      "step": 321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018417627519786315,
+      "loss": 1.2012,
+      "step": 322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018406414833195268,
+      "loss": 1.2,
+      "step": 323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018395165998531986,
+      "loss": 1.2393,
+      "step": 324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018383881064167103,
+      "loss": 1.196,
+      "step": 325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000183725600786265,
+      "loss": 1.1853,
+      "step": 326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018361203090591071,
+      "loss": 1.1794,
+      "step": 327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018349810148896522,
+      "loss": 1.1421,
+      "step": 328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018338381302533162,
+      "loss": 1.1948,
+      "step": 329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018326916600645702,
+      "loss": 1.2379,
+      "step": 330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018315416092533023,
+      "loss": 1.3147,
+      "step": 331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018303879827647975,
+      "loss": 1.1538,
+      "step": 332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018292307855597172,
+      "loss": 1.2203,
+      "step": 333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018280700226140764,
+      "loss": 1.1741,
+      "step": 334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001826905698919223,
+      "loss": 1.302,
+      "step": 335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001825737819481817,
+      "loss": 1.2623,
+      "step": 336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018245663893238075,
+      "loss": 1.3401,
+      "step": 337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018233914134824122,
+      "loss": 1.1852,
+      "step": 338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018222128970100956,
+      "loss": 1.2277,
+      "step": 339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018210308449745472,
+      "loss": 1.2296,
+      "step": 340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018198452624586592,
+      "loss": 1.1051,
+      "step": 341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018186561545605054,
+      "loss": 1.2346,
+      "step": 342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001817463526393319,
+      "loss": 1.2339,
+      "step": 343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000181626738308547,
+      "loss": 1.1985,
+      "step": 344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018150677297804445,
+      "loss": 1.1785,
+      "step": 345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018138645716368212,
+      "loss": 1.1619,
+      "step": 346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018126579138282503,
+      "loss": 1.1221,
+      "step": 347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018114477615434303,
+      "loss": 1.2717,
+      "step": 348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018102341199860864,
+      "loss": 1.3185,
+      "step": 349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 1.1874,
+      "step": 350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001807796389943725,
+      "loss": 1.1949,
+      "step": 351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018065723119410884,
+      "loss": 1.276,
+      "step": 352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018053447656306445,
+      "loss": 1.1644,
+      "step": 353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018041137562909137,
+      "loss": 1.2693,
+      "step": 354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018028792892153076,
+      "loss": 1.1438,
+      "step": 355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018016413697121065,
+      "loss": 1.1836,
+      "step": 356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001800400003104436,
+      "loss": 1.2102,
+      "step": 357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017991551947302452,
+      "loss": 1.1578,
+      "step": 358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001797906949942282,
+      "loss": 1.2866,
+      "step": 359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017966552741080708,
+      "loss": 1.2307,
+      "step": 360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017954001726098917,
+      "loss": 1.1588,
+      "step": 361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017941416508447536,
+      "loss": 1.1719,
+      "step": 362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001792879714224373,
+      "loss": 1.2384,
+      "step": 363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001791614368175151,
+      "loss": 1.3005,
+      "step": 364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017903456181381485,
+      "loss": 1.2584,
+      "step": 365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017890734695690652,
+      "loss": 1.2102,
+      "step": 366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017877979279382135,
+      "loss": 1.2107,
+      "step": 367
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017865189987304963,
+      "loss": 1.3347,
+      "step": 368
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017852366874453838,
+      "loss": 1.1805,
+      "step": 369
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001783950999596889,
+      "loss": 1.1626,
+      "step": 370
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017826619407135445,
+      "loss": 1.2205,
+      "step": 371
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781369516338378,
+      "loss": 1.1646,
+      "step": 372
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000178007373202889,
+      "loss": 1.2349,
+      "step": 373
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017787745933570278,
+      "loss": 1.252,
+      "step": 374
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017774721059091633,
+      "loss": 1.197,
+      "step": 375
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017761662752860678,
+      "loss": 1.2517,
+      "step": 376
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177485710710289,
+      "loss": 1.2853,
+      "step": 377
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001773544606989128,
+      "loss": 1.2175,
+      "step": 378
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001772228780588609,
+      "loss": 1.2812,
+      "step": 379
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017709096335594634,
+      "loss": 1.2227,
+      "step": 380
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017695871715740994,
+      "loss": 1.2107,
+      "step": 381
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017682614003191807,
+      "loss": 1.2279,
+      "step": 382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017669323254956005,
+      "loss": 1.2261,
+      "step": 383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001765599952818458,
+      "loss": 1.2397,
+      "step": 384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017642642880170333,
+      "loss": 1.156,
+      "step": 385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017629253368347622,
+      "loss": 1.2394,
+      "step": 386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001761583105029213,
+      "loss": 1.2703,
+      "step": 387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000176023759837206,
+      "loss": 1.193,
+      "step": 388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017588888226490604,
+      "loss": 1.1993,
+      "step": 389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001757536783660028,
+      "loss": 1.1201,
+      "step": 390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017561814872188092,
+      "loss": 1.2258,
+      "step": 391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017548229391532572,
+      "loss": 1.1379,
+      "step": 392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017534611453052075,
+      "loss": 1.212,
+      "step": 393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017520961115304534,
+      "loss": 1.1478,
+      "step": 394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001750727843698719,
+      "loss": 1.1945,
+      "step": 395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017493563476936351,
+      "loss": 1.2498,
+      "step": 396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017479816294127152,
+      "loss": 1.2142,
+      "step": 397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017466036947673267,
+      "loss": 1.2241,
+      "step": 398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017452225496826697,
+      "loss": 1.2344,
+      "step": 399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017438382000977475,
+      "loss": 1.2201,
+      "step": 400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017424506519653438,
+      "loss": 1.1334,
+      "step": 401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001741059911251997,
+      "loss": 1.2565,
+      "step": 402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017396659839379722,
+      "loss": 1.215,
+      "step": 403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017382688760172375,
+      "loss": 1.1726,
+      "step": 404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017368685934974394,
+      "loss": 1.2939,
+      "step": 405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017354651423998733,
+      "loss": 1.1736,
+      "step": 406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017340585287594604,
+      "loss": 1.2247,
+      "step": 407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001732648758624721,
+      "loss": 1.2283,
+      "step": 408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017312358380577493,
+      "loss": 1.1475,
+      "step": 409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001729819773134185,
+      "loss": 1.1941,
+      "step": 410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017284005699431896,
+      "loss": 1.2371,
+      "step": 411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017269782345874203,
+      "loss": 1.2217,
+      "step": 412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001725552773183001,
+      "loss": 1.292,
+      "step": 413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017241241918594983,
+      "loss": 1.1865,
+      "step": 414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017226924967598956,
+      "loss": 1.168,
+      "step": 415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017212576940405647,
+      "loss": 1.1677,
+      "step": 416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017198197898712404,
+      "loss": 1.3153,
+      "step": 417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017183787904349944,
+      "loss": 1.1591,
+      "step": 418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001716934701928208,
+      "loss": 1.1223,
+      "step": 419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017154875305605457,
+      "loss": 1.1827,
+      "step": 420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017140372825549284,
+      "loss": 1.2881,
+      "step": 421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017125839641475072,
+      "loss": 1.2413,
+      "step": 422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017111275815876348,
+      "loss": 1.2347,
+      "step": 423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017096681411378416,
+      "loss": 1.2043,
+      "step": 424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017082056490738063,
+      "loss": 1.1754,
+      "step": 425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017067401116843296,
+      "loss": 1.1959,
+      "step": 426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017052715352713075,
+      "loss": 1.2229,
+      "step": 427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017037999261497035,
+      "loss": 1.1536,
+      "step": 428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017023252906475227,
+      "loss": 1.2147,
+      "step": 429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017008476351057836,
+      "loss": 1.2196,
+      "step": 430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016993669658784904,
+      "loss": 1.188,
+      "step": 431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016978832893326074,
+      "loss": 1.2643,
+      "step": 432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016963966118480292,
+      "loss": 1.2505,
+      "step": 433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016949069398175564,
+      "loss": 1.1752,
+      "step": 434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001693414279646864,
+      "loss": 1.2156,
+      "step": 435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016919186377544788,
+      "loss": 1.2576,
+      "step": 436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001690420020571747,
+      "loss": 1.1948,
+      "step": 437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016889184345428095,
+      "loss": 1.2086,
+      "step": 438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001687413886124574,
+      "loss": 1.1836,
+      "step": 439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016859063817866859,
+      "loss": 1.1949,
+      "step": 440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016843959280115015,
+      "loss": 1.1963,
+      "step": 441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016828825312940592,
+      "loss": 1.2238,
+      "step": 442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016813661981420538,
+      "loss": 1.2253,
+      "step": 443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016798469350758056,
+      "loss": 1.2057,
+      "step": 444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016783247486282335,
+      "loss": 1.1617,
+      "step": 445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016767996453448283,
+      "loss": 1.1896,
+      "step": 446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016752716317836229,
+      "loss": 1.1548,
+      "step": 447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016737407145151636,
+      "loss": 1.1881,
+      "step": 448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016722069001224842,
+      "loss": 1.1876,
+      "step": 449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016706701952010758,
+      "loss": 1.2465,
+      "step": 450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016691306063588583,
+      "loss": 1.2247,
+      "step": 451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016675881402161536,
+      "loss": 1.1149,
+      "step": 452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001666042803405656,
+      "loss": 1.1575,
+      "step": 453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016644946025724027,
+      "loss": 1.1538,
+      "step": 454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001662943544373748,
+      "loss": 1.1682,
+      "step": 455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001661389635479332,
+      "loss": 1.1312,
+      "step": 456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016598328825710533,
+      "loss": 1.1553,
+      "step": 457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000165827329234304,
+      "loss": 1.1284,
+      "step": 458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016567108715016205,
+      "loss": 1.2048,
+      "step": 459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016551456267652952,
+      "loss": 1.1881,
+      "step": 460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016535775648647075,
+      "loss": 1.1561,
+      "step": 461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016520066925426144,
+      "loss": 1.1252,
+      "step": 462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001650433016553859,
+      "loss": 1.2258,
+      "step": 463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001648856543665338,
+      "loss": 1.2365,
+      "step": 464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016472772806559778,
+      "loss": 1.1802,
+      "step": 465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016456952343167007,
+      "loss": 1.1545,
+      "step": 466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001644110411450398,
+      "loss": 1.1172,
+      "step": 467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016425228188718995,
+      "loss": 1.2764,
+      "step": 468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001640932463407946,
+      "loss": 1.1831,
+      "step": 469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016393393518971587,
+      "loss": 1.1617,
+      "step": 470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001637743491190009,
+      "loss": 1.1534,
+      "step": 471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016361448881487914,
+      "loss": 1.2035,
+      "step": 472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016345435496475908,
+      "loss": 1.186,
+      "step": 473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016329394825722568,
+      "loss": 1.283,
+      "step": 474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001631332693820371,
+      "loss": 1.1425,
+      "step": 475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001629723190301218,
+      "loss": 1.2441,
+      "step": 476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001628110978935756,
+      "loss": 1.1863,
+      "step": 477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016264960666565882,
+      "loss": 1.1378,
+      "step": 478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016248784604079315,
+      "loss": 1.1266,
+      "step": 479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016232581671455861,
+      "loss": 1.1106,
+      "step": 480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016216351938369073,
+      "loss": 1.2808,
+      "step": 481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016200095474607753,
+      "loss": 1.1763,
+      "step": 482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001618381235007563,
+      "loss": 1.1654,
+      "step": 483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016167502634791095,
+      "loss": 1.1053,
+      "step": 484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016151166398886861,
+      "loss": 1.1648,
+      "step": 485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000161348037126097,
+      "loss": 1.1927,
+      "step": 486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001611841464632011,
+      "loss": 1.1971,
+      "step": 487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016101999270492034,
+      "loss": 1.1814,
+      "step": 488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016085557655712532,
+      "loss": 1.1801,
+      "step": 489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001606908987268151,
+      "loss": 1.1608,
+      "step": 490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016052595992211387,
+      "loss": 1.1726,
+      "step": 491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016036076085226814,
+      "loss": 1.2643,
+      "step": 492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016019530222764346,
+      "loss": 1.2025,
+      "step": 493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016002958475972156,
+      "loss": 1.209,
+      "step": 494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015986360916109713,
+      "loss": 1.1636,
+      "step": 495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015969737614547494,
+      "loss": 1.1713,
+      "step": 496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595308864276666,
+      "loss": 1.1997,
+      "step": 497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015936414072358757,
+      "loss": 1.163,
+      "step": 498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015919713975025407,
+      "loss": 1.1831,
+      "step": 499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015902988422577998,
+      "loss": 1.1898,
+      "step": 500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015886237486937378,
+      "loss": 1.2146,
+      "step": 501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001586946124013354,
+      "loss": 1.1581,
+      "step": 502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001585265975430533,
+      "loss": 1.2067,
+      "step": 503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015835833101700102,
+      "loss": 1.2665,
+      "step": 504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015818981354673447,
+      "loss": 1.1948,
+      "step": 505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015802104585688851,
+      "loss": 1.2046,
+      "step": 506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015785202867317407,
+      "loss": 1.1941,
+      "step": 507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015768276272237484,
+      "loss": 1.1608,
+      "step": 508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001575132487323442,
+      "loss": 1.2275,
+      "step": 509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001573434874320022,
+      "loss": 1.0887,
+      "step": 510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015717347955133233,
+      "loss": 1.2151,
+      "step": 511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015700322582137827,
+      "loss": 1.1537,
+      "step": 512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015683272697424104,
+      "loss": 1.194,
+      "step": 513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015666198374307552,
+      "loss": 1.2704,
+      "step": 514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015649099686208755,
+      "loss": 1.1606,
+      "step": 515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015631976706653063,
+      "loss": 1.2455,
+      "step": 516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561482950927029,
+      "loss": 1.1924,
+      "step": 517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015597658167794374,
+      "loss": 1.1241,
+      "step": 518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015580462756063082,
+      "loss": 1.2062,
+      "step": 519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015563243348017691,
+      "loss": 1.2021,
+      "step": 520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015546000017702648,
+      "loss": 1.1777,
+      "step": 521
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015528732839265272,
+      "loss": 1.1935,
+      "step": 522
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015511441886955444,
+      "loss": 1.1775,
+      "step": 523
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001549412723512526,
+      "loss": 1.1912,
+      "step": 524
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015476788958228725,
+      "loss": 1.1858,
+      "step": 525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015459427130821442,
+      "loss": 1.1986,
+      "step": 526
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015442041827560274,
+      "loss": 1.218,
+      "step": 527
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015424633123203036,
+      "loss": 1.1241,
+      "step": 528
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015407201092608173,
+      "loss": 1.2119,
+      "step": 529
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001538974581073442,
+      "loss": 1.2456,
+      "step": 530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015372267352640513,
+      "loss": 1.1842,
+      "step": 531
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015354765793484834,
+      "loss": 1.1135,
+      "step": 532
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015337241208525097,
+      "loss": 1.1587,
+      "step": 533
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015319693673118047,
+      "loss": 1.2242,
+      "step": 534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000153021232627191,
+      "loss": 1.1139,
+      "step": 535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015284530052882045,
+      "loss": 1.1373,
+      "step": 536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000152669141192587,
+      "loss": 1.2007,
+      "step": 537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001524927553759861,
+      "loss": 1.1783,
+      "step": 538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015231614383748698,
+      "loss": 1.1786,
+      "step": 539
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001521393073365295,
+      "loss": 1.2511,
+      "step": 540
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015196224663352093,
+      "loss": 1.1642,
+      "step": 541
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015178496248983254,
+      "loss": 1.1476,
+      "step": 542
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015160745566779652,
+      "loss": 1.2125,
+      "step": 543
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001514297269307024,
+      "loss": 1.16,
+      "step": 544
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015125177704279418,
+      "loss": 1.187,
+      "step": 545
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015107360676926666,
+      "loss": 1.1085,
+      "step": 546
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015089521687626243,
+      "loss": 1.217,
+      "step": 547
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015071660813086837,
+      "loss": 1.1141,
+      "step": 548
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001505377813011124,
+      "loss": 1.1608,
+      "step": 549
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015035873715596036,
+      "loss": 1.1924,
+      "step": 550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001501794764653124,
+      "loss": 1.2295,
+      "step": 551
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.2363,
+      "step": 552
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014982030853178234,
+      "loss": 1.1823,
+      "step": 553
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00014964040283334316,
+      "loss": 1.2605,
+      "step": 554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014946028367828745,
+      "loss": 1.1935,
+      "step": 555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000149279951841138,
+      "loss": 1.2208,
+      "step": 556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014909940809733222,
+      "loss": 1.116,
+      "step": 557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001489186532232187,
+      "loss": 1.2063,
+      "step": 558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001487376879960539,
+      "loss": 1.177,
+      "step": 559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014855651319399882,
+      "loss": 1.1899,
+      "step": 560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001483751295961156,
+      "loss": 1.1343,
+      "step": 561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014819353798236427,
+      "loss": 1.2297,
+      "step": 562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001480117391335993,
+      "loss": 1.1555,
+      "step": 563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014782973383156635,
+      "loss": 1.1316,
+      "step": 564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014764752285889875,
+      "loss": 1.2272,
+      "step": 565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014746510699911432,
+      "loss": 1.2361,
+      "step": 566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014728248703661182,
+      "loss": 1.2084,
+      "step": 567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001470996637566677,
+      "loss": 1.1774,
+      "step": 568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001469166379454327,
+      "loss": 1.2804,
+      "step": 569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001467334103899284,
+      "loss": 1.1555,
+      "step": 570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000146549981878044,
+      "loss": 1.1625,
+      "step": 571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014636635319853275,
+      "loss": 1.1508,
+      "step": 572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014618252514100858,
+      "loss": 1.1898,
+      "step": 573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001459984984959429,
+      "loss": 1.2384,
+      "step": 574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001458142740546609,
+      "loss": 1.2347,
+      "step": 575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014562985260933845,
+      "loss": 1.2174,
+      "step": 576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014544523495299842,
+      "loss": 1.1958,
+      "step": 577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001452604218795075,
+      "loss": 1.2289,
+      "step": 578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014507541418357265,
+      "loss": 1.095,
+      "step": 579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014489021266073768,
+      "loss": 1.1451,
+      "step": 580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001447048181073799,
+      "loss": 1.1649,
+      "step": 581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001445192313207067,
+      "loss": 1.1453,
+      "step": 582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014433345309875203,
+      "loss": 1.1285,
+      "step": 583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014414748424037302,
+      "loss": 1.1593,
+      "step": 584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001439613255452466,
+      "loss": 1.1799,
+      "step": 585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001437749778138659,
+      "loss": 1.1652,
+      "step": 586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014358844184753712,
+      "loss": 1.1847,
+      "step": 587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014340171844837567,
+      "loss": 1.1749,
+      "step": 588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014321480841930298,
+      "loss": 1.2343,
+      "step": 589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014302771256404311,
+      "loss": 1.219,
+      "step": 590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014284043168711906,
+      "loss": 1.212,
+      "step": 591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014265296659384956,
+      "loss": 1.1803,
+      "step": 592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014246531809034528,
+      "loss": 1.1622,
+      "step": 593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014227748698350583,
+      "loss": 1.2043,
+      "step": 594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014208947408101576,
+      "loss": 1.2637,
+      "step": 595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014190128019134153,
+      "loss": 1.2462,
+      "step": 596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001417129061237278,
+      "loss": 1.1798,
+      "step": 597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000141524352688194,
+      "loss": 1.243,
+      "step": 598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001413356206955309,
+      "loss": 1.1082,
+      "step": 599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014114671095729695,
+      "loss": 1.2122,
+      "step": 600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014095762428581506,
+      "loss": 1.1753,
+      "step": 601
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014076836149416887,
+      "loss": 1.1422,
+      "step": 602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014057892339619944,
+      "loss": 1.1818,
+      "step": 603
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014038931080650156,
+      "loss": 1.1827,
+      "step": 604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014019952454042035,
+      "loss": 1.1517,
+      "step": 605
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014000956541404785,
+      "loss": 1.2177,
+      "step": 606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013981943424421932,
+      "loss": 1.1884,
+      "step": 607
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013962913184850979,
+      "loss": 1.1519,
+      "step": 608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013943865904523066,
+      "loss": 1.1736,
+      "step": 609
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013924801665342603,
+      "loss": 1.2433,
+      "step": 610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013905720549286932,
+      "loss": 1.1155,
+      "step": 611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013886622638405952,
+      "loss": 1.1417,
+      "step": 612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013867508014821797,
+      "loss": 1.2524,
+      "step": 613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013848376760728455,
+      "loss": 1.1986,
+      "step": 614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001382922895839143,
+      "loss": 1.2299,
+      "step": 615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013810064690147387,
+      "loss": 1.1937,
+      "step": 616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013790884038403795,
+      "loss": 1.1978,
+      "step": 617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013771687085638564,
+      "loss": 1.2552,
+      "step": 618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013752473914399713,
+      "loss": 1.1432,
+      "step": 619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013733244607304988,
+      "loss": 1.163,
+      "step": 620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013713999247041533,
+      "loss": 1.1996,
+      "step": 621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013694737916365517,
+      "loss": 1.1993,
+      "step": 622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013675460698101772,
+      "loss": 1.1851,
+      "step": 623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013656167675143462,
+      "loss": 1.2172,
+      "step": 624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001363685893045171,
+      "loss": 1.1332,
+      "step": 625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013617534547055236,
+      "loss": 1.2618,
+      "step": 626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001359819460805001,
+      "loss": 1.2021,
+      "step": 627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013578839196598899,
+      "loss": 1.2086,
+      "step": 628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001355946839593129,
+      "loss": 1.2122,
+      "step": 629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013540082289342756,
+      "loss": 1.1428,
+      "step": 630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001352068096019468,
+      "loss": 1.1406,
+      "step": 631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013501264491913906,
+      "loss": 1.1794,
+      "step": 632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013481832967992378,
+      "loss": 1.0834,
+      "step": 633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013462386471986773,
+      "loss": 1.1808,
+      "step": 634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013442925087518153,
+      "loss": 1.1782,
+      "step": 635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001342344889827161,
+      "loss": 1.1171,
+      "step": 636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013403957987995882,
+      "loss": 1.1755,
+      "step": 637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013384452440503018,
+      "loss": 1.1349,
+      "step": 638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013364932339668008,
+      "loss": 1.1179,
+      "step": 639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013345397769428413,
+      "loss": 1.2069,
+      "step": 640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001332584881378403,
+      "loss": 1.1396,
+      "step": 641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013306285556796495,
+      "loss": 1.2068,
+      "step": 642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013286708082588955,
+      "loss": 1.2002,
+      "step": 643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001326711647534568,
+      "loss": 1.1348,
+      "step": 644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013247510819311732,
+      "loss": 1.1887,
+      "step": 645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001322789119879256,
+      "loss": 1.1902,
+      "step": 646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013208257698153677,
+      "loss": 1.1693,
+      "step": 647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013188610401820277,
+      "loss": 1.1429,
+      "step": 648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001316894939427687,
+      "loss": 1.1323,
+      "step": 649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013149274760066942,
+      "loss": 1.1664,
+      "step": 650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001312958658379255,
+      "loss": 1.1743,
+      "step": 651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013109884950114007,
+      "loss": 1.2094,
+      "step": 652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013090169943749476,
+      "loss": 1.1506,
+      "step": 653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001307044164947463,
+      "loss": 1.1667,
+      "step": 654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013050700152122285,
+      "loss": 1.0879,
+      "step": 655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013030945536582025,
+      "loss": 1.176,
+      "step": 656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013011177887799845,
+      "loss": 1.223,
+      "step": 657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001299139729077778,
+      "loss": 1.1235,
+      "step": 658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001297160383057355,
+      "loss": 1.179,
+      "step": 659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012951797592300186,
+      "loss": 1.248,
+      "step": 660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012931978661125655,
+      "loss": 1.1429,
+      "step": 661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012912147122272523,
+      "loss": 1.2206,
+      "step": 662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001289230306101755,
+      "loss": 1.1935,
+      "step": 663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0001287244656269136,
+      "loss": 1.1617,
+      "step": 664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012852577712678045,
+      "loss": 1.2507,
+      "step": 665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012832696596414817,
+      "loss": 1.1566,
+      "step": 666
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012812803299391628,
+      "loss": 1.2056,
+      "step": 667
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012792897907150817,
+      "loss": 1.1694,
+      "step": 668
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001277298050528672,
+      "loss": 1.2231,
+      "step": 669
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012753051179445325,
+      "loss": 1.188,
+      "step": 670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012733110015323898,
+      "loss": 1.2972,
+      "step": 671
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001271315709867059,
+      "loss": 1.0586,
+      "step": 672
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001269319251528411,
+      "loss": 1.1582,
+      "step": 673
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001267321635101333,
+      "loss": 1.1554,
+      "step": 674
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012653228691756913,
+      "loss": 1.1897,
+      "step": 675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012633229623462951,
+      "loss": 1.2316,
+      "step": 676
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012613219232128608,
+      "loss": 1.1713,
+      "step": 677
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012593197603799725,
+      "loss": 1.1412,
+      "step": 678
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012573164824570464,
+      "loss": 1.1844,
+      "step": 679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012553120980582946,
+      "loss": 1.2214,
+      "step": 680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012533066158026862,
+      "loss": 1.1602,
+      "step": 681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012513000443139112,
+      "loss": 1.1747,
+      "step": 682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001249292392220344,
+      "loss": 1.2472,
+      "step": 683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012472836681550048,
+      "loss": 1.1359,
+      "step": 684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012452738807555244,
+      "loss": 1.1903,
+      "step": 685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001243263038664105,
+      "loss": 1.2522,
+      "step": 686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012412511505274844,
+      "loss": 1.0997,
+      "step": 687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012392382249968987,
+      "loss": 1.1289,
+      "step": 688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012372242707280447,
+      "loss": 1.1274,
+      "step": 689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012352092963810425,
+      "loss": 1.1438,
+      "step": 690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012331933106203986,
+      "loss": 1.1536,
+      "step": 691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000123117632211497,
+      "loss": 1.1329,
+      "step": 692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012291583395379228,
+      "loss": 1.2712,
+      "step": 693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012271393715667013,
+      "loss": 1.1494,
+      "step": 694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001225119426882984,
+      "loss": 1.1537,
+      "step": 695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012230985141726498,
+      "loss": 1.2388,
+      "step": 696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001221076642125742,
+      "loss": 1.2141,
+      "step": 697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001219053819436427,
+      "loss": 1.1442,
+      "step": 698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012170300548029605,
+      "loss": 1.1599,
+      "step": 699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012150053569276478,
+      "loss": 1.228,
+      "step": 700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012129797345168073,
+      "loss": 1.2166,
+      "step": 701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012109531962807332,
+      "loss": 1.1168,
+      "step": 702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012089257509336577,
+      "loss": 1.2246,
+      "step": 703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012068974071937136,
+      "loss": 1.3116,
+      "step": 704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012048681737828962,
+      "loss": 1.0999,
+      "step": 705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012028380594270283,
+      "loss": 1.1261,
+      "step": 706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012008070728557186,
+      "loss": 1.1099,
+      "step": 707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011987752228023277,
+      "loss": 1.1195,
+      "step": 708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001196742518003929,
+      "loss": 1.1169,
+      "step": 709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001194708967201271,
+      "loss": 1.1305,
+      "step": 710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011926745791387406,
+      "loss": 1.2178,
+      "step": 711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011906393625643244,
+      "loss": 1.089,
+      "step": 712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001188603326229572,
+      "loss": 1.2487,
+      "step": 713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011865664788895578,
+      "loss": 1.2024,
+      "step": 714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011845288293028445,
+      "loss": 1.2987,
+      "step": 715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011824903862314427,
+      "loss": 1.1744,
+      "step": 716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011804511584407763,
+      "loss": 1.1953,
+      "step": 717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011784111546996439,
+      "loss": 1.1957,
+      "step": 718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011763703837801794,
+      "loss": 1.2052,
+      "step": 719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011743288544578166,
+      "loss": 1.2065,
+      "step": 720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011722865755112504,
+      "loss": 1.208,
+      "step": 721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011702435557223987,
+      "loss": 1.2378,
+      "step": 722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011681998038763653,
+      "loss": 1.2872,
+      "step": 723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001166155328761402,
+      "loss": 1.2098,
+      "step": 724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011641101391688707,
+      "loss": 1.173,
+      "step": 725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001162064243893205,
+      "loss": 1.2261,
+      "step": 726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011600176517318741,
+      "loss": 1.1428,
+      "step": 727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011579703714853425,
+      "loss": 1.21,
+      "step": 728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001155922411957035,
+      "loss": 1.1725,
+      "step": 729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011538737819532958,
+      "loss": 1.1587,
+      "step": 730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011518244902833537,
+      "loss": 1.1833,
+      "step": 731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011497745457592816,
+      "loss": 1.1714,
+      "step": 732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000114772395719596,
+      "loss": 1.1599,
+      "step": 733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011456727334110389,
+      "loss": 1.1924,
+      "step": 734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011436208832248997,
+      "loss": 1.2112,
+      "step": 735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011415684154606177,
+      "loss": 1.1709,
+      "step": 736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011395153389439233,
+      "loss": 1.1292,
+      "step": 737
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011374616625031647,
+      "loss": 1.1881,
+      "step": 738
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011354073949692703,
+      "loss": 1.153,
+      "step": 739
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011333525451757094,
+      "loss": 1.2166,
+      "step": 740
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011312971219584563,
+      "loss": 1.1292,
+      "step": 741
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001129241134155949,
+      "loss": 1.1156,
+      "step": 742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011271845906090557,
+      "loss": 1.1687,
+      "step": 743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011251275001610327,
+      "loss": 1.0724,
+      "step": 744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011230698716574886,
+      "loss": 1.1212,
+      "step": 745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011210117139463452,
+      "loss": 1.1542,
+      "step": 746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011189530358778005,
+      "loss": 1.1841,
+      "step": 747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011168938463042895,
+      "loss": 1.1019,
+      "step": 748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011148341540804472,
+      "loss": 1.1106,
+      "step": 749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011127739680630705,
+      "loss": 1.1577,
+      "step": 750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011107132971110779,
+      "loss": 1.1444,
+      "step": 751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011086521500854745,
+      "loss": 1.1184,
+      "step": 752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011065905358493132,
+      "loss": 1.1562,
+      "step": 753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011045284632676536,
+      "loss": 1.1566,
+      "step": 754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011024659412075284,
+      "loss": 1.1202,
+      "step": 755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011004029785379024,
+      "loss": 1.1932,
+      "step": 756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010983395841296348,
+      "loss": 1.1973,
+      "step": 757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010962757668554413,
+      "loss": 1.1412,
+      "step": 758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010942115355898563,
+      "loss": 1.2078,
+      "step": 759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010921468992091941,
+      "loss": 1.2145,
+      "step": 760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010900818665915109,
+      "loss": 1.1747,
+      "step": 761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010880164466165674,
+      "loss": 1.1938,
+      "step": 762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010859506481657889,
+      "loss": 1.1536,
+      "step": 763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010838844801222295,
+      "loss": 1.0934,
+      "step": 764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010818179513705311,
+      "loss": 1.123,
+      "step": 765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010797510707968878,
+      "loss": 1.1267,
+      "step": 766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010776838472890065,
+      "loss": 1.208,
+      "step": 767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010756162897360675,
+      "loss": 1.0935,
+      "step": 768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010735484070286892,
+      "loss": 1.1809,
+      "step": 769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010714802080588872,
+      "loss": 1.1965,
+      "step": 770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010694117017200372,
+      "loss": 1.2223,
+      "step": 771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010673428969068364,
+      "loss": 1.1519,
+      "step": 772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010652738025152661,
+      "loss": 1.0735,
+      "step": 773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010632044274425524,
+      "loss": 1.0942,
+      "step": 774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010611347805871277,
+      "loss": 1.1316,
+      "step": 775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010590648708485946,
+      "loss": 1.2067,
+      "step": 776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010569947071276847,
+      "loss": 1.1517,
+      "step": 777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010549242983262227,
+      "loss": 1.1459,
+      "step": 778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010528536533470863,
+      "loss": 1.1318,
+      "step": 779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010507827810941695,
+      "loss": 1.1545,
+      "step": 780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010487116904723433,
+      "loss": 1.1475,
+      "step": 781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010466403903874176,
+      "loss": 1.2015,
+      "step": 782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010445688897461032,
+      "loss": 1.1841,
+      "step": 783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010424971974559732,
+      "loss": 1.1901,
+      "step": 784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001040425322425425,
+      "loss": 1.0631,
+      "step": 785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010383532735636411,
+      "loss": 1.1538,
+      "step": 786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010362810597805526,
+      "loss": 1.1512,
+      "step": 787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010342086899867991,
+      "loss": 1.1804,
+      "step": 788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010321361730936904,
+      "loss": 1.0977,
+      "step": 789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010300635180131708,
+      "loss": 1.1124,
+      "step": 790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010279907336577765,
+      "loss": 1.1691,
+      "step": 791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010259178289406011,
+      "loss": 1.1335,
+      "step": 792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010238448127752554,
+      "loss": 1.1041,
+      "step": 793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010217716940758291,
+      "loss": 1.1401,
+      "step": 794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010196984817568534,
+      "loss": 1.1738,
+      "step": 795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010176251847332614,
+      "loss": 1.1201,
+      "step": 796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001015551811920351,
+      "loss": 1.1411,
+      "step": 797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010134783722337455,
+      "loss": 1.1713,
+      "step": 798
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010114048745893567,
+      "loss": 1.1046,
+      "step": 799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010093313279033441,
+      "loss": 1.1963,
+      "step": 800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010072577410920794,
+      "loss": 1.2045,
+      "step": 801
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010051841230721065,
+      "loss": 1.1401,
+      "step": 802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010031104827601027,
+      "loss": 1.1497,
+      "step": 803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010010368290728427,
+      "loss": 1.1224,
+      "step": 804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989631709271571e-05,
+      "loss": 1.1826,
+      "step": 805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.968895172398974e-05,
+      "loss": 1.1705,
+      "step": 806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948158769278939e-05,
+      "loss": 1.1644,
+      "step": 807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.927422589079206e-05,
+      "loss": 1.1825,
+      "step": 808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906686720966561e-05,
+      "loss": 1.1396,
+      "step": 809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.885951254106437e-05,
+      "loss": 1.1647,
+      "step": 810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865216277662545e-05,
+      "loss": 1.1077,
+      "step": 811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844481880796491e-05,
+      "loss": 1.1812,
+      "step": 812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82374815266739e-05,
+      "loss": 1.0654,
+      "step": 813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803015182431468e-05,
+      "loss": 1.1079,
+      "step": 814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.782283059241711e-05,
+      "loss": 1.1233,
+      "step": 815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761551872247449e-05,
+      "loss": 1.1553,
+      "step": 816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740821710593989e-05,
+      "loss": 1.2006,
+      "step": 817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.720092663422237e-05,
+      "loss": 1.1814,
+      "step": 818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.699364819868295e-05,
+      "loss": 1.1859,
+      "step": 819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678638269063094e-05,
+      "loss": 1.1329,
+      "step": 820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657913100132011e-05,
+      "loss": 1.1667,
+      "step": 821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637189402194476e-05,
+      "loss": 1.2103,
+      "step": 822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61646726436359e-05,
+      "loss": 1.1074,
+      "step": 823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595746775745753e-05,
+      "loss": 1.1871,
+      "step": 824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.575028025440272e-05,
+      "loss": 1.1433,
+      "step": 825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.554311102538966e-05,
+      "loss": 1.137,
+      "step": 826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.533596096125825e-05,
+      "loss": 1.1602,
+      "step": 827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.512883095276569e-05,
+      "loss": 1.1733,
+      "step": 828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492172189058304e-05,
+      "loss": 1.125,
+      "step": 829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.471463466529139e-05,
+      "loss": 1.1597,
+      "step": 830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450757016737776e-05,
+      "loss": 1.1364,
+      "step": 831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430052928723153e-05,
+      "loss": 1.1344,
+      "step": 832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409351291514054e-05,
+      "loss": 1.0674,
+      "step": 833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388652194128726e-05,
+      "loss": 1.1382,
+      "step": 834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36795572557448e-05,
+      "loss": 1.1605,
+      "step": 835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347261974847341e-05,
+      "loss": 1.1862,
+      "step": 836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326571030931637e-05,
+      "loss": 1.184,
+      "step": 837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.305882982799633e-05,
+      "loss": 1.1393,
+      "step": 838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28519791941113e-05,
+      "loss": 1.1614,
+      "step": 839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264515929713109e-05,
+      "loss": 1.1273,
+      "step": 840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243837102639328e-05,
+      "loss": 1.1093,
+      "step": 841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223161527109937e-05,
+      "loss": 1.1691,
+      "step": 842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.202489292031123e-05,
+      "loss": 1.1337,
+      "step": 843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181820486294693e-05,
+      "loss": 1.1567,
+      "step": 844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161155198777707e-05,
+      "loss": 1.1486,
+      "step": 845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140493518342113e-05,
+      "loss": 1.1324,
+      "step": 846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.119835533834331e-05,
+      "loss": 1.1495,
+      "step": 847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.099181334084892e-05,
+      "loss": 1.1776,
+      "step": 848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078531007908062e-05,
+      "loss": 1.2368,
+      "step": 849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05788464410144e-05,
+      "loss": 1.0144,
+      "step": 850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.037242331445588e-05,
+      "loss": 1.1685,
+      "step": 851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016604158703654e-05,
+      "loss": 1.1536,
+      "step": 852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995970214620977e-05,
+      "loss": 1.1459,
+      "step": 853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.975340587924716e-05,
+      "loss": 1.1815,
+      "step": 854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-05,
+      "loss": 1.1047,
+      "step": 855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934094641506873e-05,
+      "loss": 1.1449,
+      "step": 856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913478499145254e-05,
+      "loss": 1.1432,
+      "step": 857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892867028889223e-05,
+      "loss": 1.1549,
+      "step": 858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872260319369301e-05,
+      "loss": 1.1686,
+      "step": 859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.851658459195528e-05,
+      "loss": 1.2246,
+      "step": 860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831061536957107e-05,
+      "loss": 1.1605,
+      "step": 861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810469641222001e-05,
+      "loss": 1.1598,
+      "step": 862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789882860536549e-05,
+      "loss": 1.2153,
+      "step": 863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769301283425117e-05,
+      "loss": 1.1487,
+      "step": 864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.748724998389674e-05,
+      "loss": 1.1018,
+      "step": 865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728154093909441e-05,
+      "loss": 1.116,
+      "step": 866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.707588658440511e-05,
+      "loss": 1.1493,
+      "step": 867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.687028780415442e-05,
+      "loss": 1.1542,
+      "step": 868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.666474548242907e-05,
+      "loss": 1.1676,
+      "step": 869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645926050307299e-05,
+      "loss": 1.1742,
+      "step": 870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.625383374968357e-05,
+      "loss": 1.1565,
+      "step": 871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604846610560771e-05,
+      "loss": 1.1178,
+      "step": 872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584315845393827e-05,
+      "loss": 1.1311,
+      "step": 873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563791167751008e-05,
+      "loss": 1.1128,
+      "step": 874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.543272665889612e-05,
+      "loss": 1.1713,
+      "step": 875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522760428040402e-05,
+      "loss": 1.0631,
+      "step": 876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502254542407186e-05,
+      "loss": 1.145,
+      "step": 877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481755097166463e-05,
+      "loss": 1.2001,
+      "step": 878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461262180467043e-05,
+      "loss": 1.1534,
+      "step": 879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.440775880429653e-05,
+      "loss": 1.1454,
+      "step": 880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.420296285146574e-05,
+      "loss": 1.1576,
+      "step": 881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399823482681262e-05,
+      "loss": 1.1447,
+      "step": 882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379357561067952e-05,
+      "loss": 1.1235,
+      "step": 883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.358898608311296e-05,
+      "loss": 1.1724,
+      "step": 884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338446712385982e-05,
+      "loss": 1.1528,
+      "step": 885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318001961236349e-05,
+      "loss": 1.1079,
+      "step": 886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297564442776014e-05,
+      "loss": 1.1935,
+      "step": 887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277134244887499e-05,
+      "loss": 1.1636,
+      "step": 888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256711455421835e-05,
+      "loss": 1.1444,
+      "step": 889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236296162198206e-05,
+      "loss": 1.1599,
+      "step": 890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215888453003562e-05,
+      "loss": 1.1545,
+      "step": 891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195488415592238e-05,
+      "loss": 1.1512,
+      "step": 892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.175096137685575e-05,
+      "loss": 1.1985,
+      "step": 893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154711706971557e-05,
+      "loss": 1.1396,
+      "step": 894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134335211104423e-05,
+      "loss": 1.1888,
+      "step": 895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11396673770428e-05,
+      "loss": 1.2313,
+      "step": 896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.093606374356759e-05,
+      "loss": 1.1218,
+      "step": 897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.073254208612597e-05,
+      "loss": 1.1675,
+      "step": 898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05291032798729e-05,
+      "loss": 1.1619,
+      "step": 899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032574819960713e-05,
+      "loss": 1.2344,
+      "step": 900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012247771976726e-05,
+      "loss": 1.1226,
+      "step": 901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.991929271442817e-05,
+      "loss": 1.0907,
+      "step": 902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.971619405729719e-05,
+      "loss": 1.1901,
+      "step": 903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951318262171039e-05,
+      "loss": 1.1099,
+      "step": 904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.931025928062869e-05,
+      "loss": 1.1261,
+      "step": 905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.910742490663425e-05,
+      "loss": 1.1609,
+      "step": 906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89046803719267e-05,
+      "loss": 1.2235,
+      "step": 907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.870202654831931e-05,
+      "loss": 1.151,
+      "step": 908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.849946430723525e-05,
+      "loss": 1.2247,
+      "step": 909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829699451970397e-05,
+      "loss": 1.1696,
+      "step": 910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809461805635734e-05,
+      "loss": 1.1638,
+      "step": 911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789233578742582e-05,
+      "loss": 1.1042,
+      "step": 912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.769014858273506e-05,
+      "loss": 1.1649,
+      "step": 913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.748805731170168e-05,
+      "loss": 1.1675,
+      "step": 914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728606284332988e-05,
+      "loss": 1.1241,
+      "step": 915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.708416604620772e-05,
+      "loss": 1.082,
+      "step": 916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688236778850306e-05,
+      "loss": 1.1375,
+      "step": 917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.668066893796014e-05,
+      "loss": 1.2039,
+      "step": 918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647907036189579e-05,
+      "loss": 1.1631,
+      "step": 919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627757292719558e-05,
+      "loss": 1.1604,
+      "step": 920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607617750031014e-05,
+      "loss": 1.0842,
+      "step": 921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587488494725157e-05,
+      "loss": 1.1527,
+      "step": 922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567369613358953e-05,
+      "loss": 1.0852,
+      "step": 923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547261192444755e-05,
+      "loss": 1.1232,
+      "step": 924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.527163318449953e-05,
+      "loss": 1.1283,
+      "step": 925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507076077796565e-05,
+      "loss": 1.2274,
+      "step": 926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48699955686089e-05,
+      "loss": 1.1752,
+      "step": 927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46693384197314e-05,
+      "loss": 1.1863,
+      "step": 928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446879019417059e-05,
+      "loss": 1.1312,
+      "step": 929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426835175429537e-05,
+      "loss": 1.1844,
+      "step": 930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.406802396200279e-05,
+      "loss": 1.173,
+      "step": 931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386780767871397e-05,
+      "loss": 1.1541,
+      "step": 932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366770376537048e-05,
+      "loss": 1.1309,
+      "step": 933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34677130824309e-05,
+      "loss": 1.1582,
+      "step": 934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326783648986671e-05,
+      "loss": 1.1874,
+      "step": 935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.306807484715888e-05,
+      "loss": 1.1975,
+      "step": 936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.286842901329412e-05,
+      "loss": 1.1273,
+      "step": 937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266889984676109e-05,
+      "loss": 1.0885,
+      "step": 938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246948820554675e-05,
+      "loss": 1.1217,
+      "step": 939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227019494713282e-05,
+      "loss": 1.1392,
+      "step": 940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207102092849187e-05,
+      "loss": 1.1562,
+      "step": 941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187196700608373e-05,
+      "loss": 1.1219,
+      "step": 942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.167303403585186e-05,
+      "loss": 1.1326,
+      "step": 943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147422287321957e-05,
+      "loss": 1.1615,
+      "step": 944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.12755343730864e-05,
+      "loss": 1.1578,
+      "step": 945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107696938982451e-05,
+      "loss": 1.2216,
+      "step": 946
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087852877727481e-05,
+      "loss": 1.1921,
+      "step": 947
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.068021338874344e-05,
+      "loss": 1.1387,
+      "step": 948
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048202407699816e-05,
+      "loss": 1.0919,
+      "step": 949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.028396169426452e-05,
+      "loss": 1.1836,
+      "step": 950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00860270922222e-05,
+      "loss": 1.2598,
+      "step": 951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988822112200156e-05,
+      "loss": 1.1233,
+      "step": 952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969054463417976e-05,
+      "loss": 1.0995,
+      "step": 953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.949299847877712e-05,
+      "loss": 1.1595,
+      "step": 954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92955835052537e-05,
+      "loss": 1.1337,
+      "step": 955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-05,
+      "loss": 1.1672,
+      "step": 956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890115049885994e-05,
+      "loss": 1.1196,
+      "step": 957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.870413416207451e-05,
+      "loss": 1.1234,
+      "step": 958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850725239933063e-05,
+      "loss": 1.1017,
+      "step": 959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.831050605723128e-05,
+      "loss": 1.176,
+      "step": 960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811389598179725e-05,
+      "loss": 1.1055,
+      "step": 961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791742301846326e-05,
+      "loss": 1.109,
+      "step": 962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772108801207442e-05,
+      "loss": 1.0842,
+      "step": 963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.75248918068827e-05,
+      "loss": 1.1461,
+      "step": 964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.732883524654318e-05,
+      "loss": 1.129,
+      "step": 965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.71329191741105e-05,
+      "loss": 1.1416,
+      "step": 966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693714443203507e-05,
+      "loss": 1.1243,
+      "step": 967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.674151186215973e-05,
+      "loss": 1.118,
+      "step": 968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654602230571589e-05,
+      "loss": 1.155,
+      "step": 969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635067660331993e-05,
+      "loss": 1.1504,
+      "step": 970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615547559496984e-05,
+      "loss": 1.0894,
+      "step": 971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59604201200412e-05,
+      "loss": 1.1758,
+      "step": 972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.57655110172839e-05,
+      "loss": 1.1959,
+      "step": 973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557074912481846e-05,
+      "loss": 1.1593,
+      "step": 974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.537613528013232e-05,
+      "loss": 1.1343,
+      "step": 975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518167032007623e-05,
+      "loss": 1.1719,
+      "step": 976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498735508086093e-05,
+      "loss": 1.1345,
+      "step": 977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479319039805323e-05,
+      "loss": 1.1636,
+      "step": 978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.459917710657246e-05,
+      "loss": 1.193,
+      "step": 979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440531604068713e-05,
+      "loss": 1.1906,
+      "step": 980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421160803401105e-05,
+      "loss": 1.1815,
+      "step": 981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40180539194999e-05,
+      "loss": 1.1255,
+      "step": 982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382465452944768e-05,
+      "loss": 1.1794,
+      "step": 983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.363141069548293e-05,
+      "loss": 1.1478,
+      "step": 984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.343832324856538e-05,
+      "loss": 1.1276,
+      "step": 985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.324539301898229e-05,
+      "loss": 1.1556,
+      "step": 986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305262083634488e-05,
+      "loss": 1.1263,
+      "step": 987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.286000752958467e-05,
+      "loss": 1.1423,
+      "step": 988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266755392695011e-05,
+      "loss": 1.1251,
+      "step": 989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.247526085600291e-05,
+      "loss": 1.2281,
+      "step": 990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228312914361435e-05,
+      "loss": 1.1592,
+      "step": 991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.209115961596208e-05,
+      "loss": 1.1982,
+      "step": 992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189935309852612e-05,
+      "loss": 1.1608,
+      "step": 993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17077104160857e-05,
+      "loss": 1.1901,
+      "step": 994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151623239271548e-05,
+      "loss": 1.0984,
+      "step": 995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.132491985178206e-05,
+      "loss": 1.181,
+      "step": 996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113377361594049e-05,
+      "loss": 1.1445,
+      "step": 997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0942794507130707e-05,
+      "loss": 1.1573,
+      "step": 998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.075198334657399e-05,
+      "loss": 1.2152,
+      "step": 999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056134095476936e-05,
+      "loss": 1.0542,
+      "step": 1000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0370868151490244e-05,
+      "loss": 1.1953,
+      "step": 1001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018056575578075e-05,
+      "loss": 1.2288,
+      "step": 1002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999043458595215e-05,
+      "loss": 1.1121,
+      "step": 1003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.980047545957965e-05,
+      "loss": 1.0894,
+      "step": 1004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.961068919349847e-05,
+      "loss": 1.1046,
+      "step": 1005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9421076603800566e-05,
+      "loss": 1.1754,
+      "step": 1006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923163850583113e-05,
+      "loss": 1.1395,
+      "step": 1007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.904237571418496e-05,
+      "loss": 1.1428,
+      "step": 1008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885328904270306e-05,
+      "loss": 1.0903,
+      "step": 1009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866437930446913e-05,
+      "loss": 1.1772,
+      "step": 1010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8475647311806016e-05,
+      "loss": 1.0997,
+      "step": 1011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828709387627218e-05,
+      "loss": 1.1145,
+      "step": 1012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8098719808658464e-05,
+      "loss": 1.1002,
+      "step": 1013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7910525918984285e-05,
+      "loss": 1.1476,
+      "step": 1014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772251301649418e-05,
+      "loss": 1.116,
+      "step": 1015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7534681909654694e-05,
+      "loss": 1.2078,
+      "step": 1016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73470334061505e-05,
+      "loss": 1.2252,
+      "step": 1017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715956831288092e-05,
+      "loss": 1.1134,
+      "step": 1018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6972287435956895e-05,
+      "loss": 1.1069,
+      "step": 1019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6785191580697025e-05,
+      "loss": 1.0913,
+      "step": 1020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6598281551624364e-05,
+      "loss": 1.198,
+      "step": 1021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6411558152462894e-05,
+      "loss": 1.2224,
+      "step": 1022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622502218613408e-05,
+      "loss": 1.1956,
+      "step": 1023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6038674454753414e-05,
+      "loss": 1.1119,
+      "step": 1024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5852515759626976e-05,
+      "loss": 1.1057,
+      "step": 1025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5666546901247974e-05,
+      "loss": 1.1144,
+      "step": 1026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.54807686792933e-05,
+      "loss": 1.14,
+      "step": 1027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5295181892620105e-05,
+      "loss": 1.1362,
+      "step": 1028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510978733926234e-05,
+      "loss": 1.1582,
+      "step": 1029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4924585816427375e-05,
+      "loss": 1.1298,
+      "step": 1030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473957812049252e-05,
+      "loss": 1.1852,
+      "step": 1031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4554765047001613e-05,
+      "loss": 1.1534,
+      "step": 1032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4370147390661595e-05,
+      "loss": 1.1598,
+      "step": 1033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.418572594533908e-05,
+      "loss": 1.1655,
+      "step": 1034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.400150150405714e-05,
+      "loss": 1.1503,
+      "step": 1035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3817474858991446e-05,
+      "loss": 1.0911,
+      "step": 1036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363364680146725e-05,
+      "loss": 1.1611,
+      "step": 1037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345001812195603e-05,
+      "loss": 1.1061,
+      "step": 1038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3266589610071624e-05,
+      "loss": 1.1285,
+      "step": 1039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.308336205456732e-05,
+      "loss": 1.1658,
+      "step": 1040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29003362433323e-05,
+      "loss": 1.1265,
+      "step": 1041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.271751296338823e-05,
+      "loss": 1.1495,
+      "step": 1042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2534893000885676e-05,
+      "loss": 1.1719,
+      "step": 1043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235247714110124e-05,
+      "loss": 1.1809,
+      "step": 1044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217026616843366e-05,
+      "loss": 1.1127,
+      "step": 1045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.198826086640072e-05,
+      "loss": 1.1893,
+      "step": 1046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180646201763577e-05,
+      "loss": 1.1195,
+      "step": 1047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162487040388444e-05,
+      "loss": 1.1379,
+      "step": 1048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.144348680600122e-05,
+      "loss": 1.1675,
+      "step": 1049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1262312003946115e-05,
+      "loss": 1.1418,
+      "step": 1050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1081346776781316e-05,
+      "loss": 1.1956,
+      "step": 1051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090059190266779e-05,
+      "loss": 1.2166,
+      "step": 1052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072004815886202e-05,
+      "loss": 1.1348,
+      "step": 1053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.053971632171259e-05,
+      "loss": 1.0433,
+      "step": 1054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0359597166656826e-05,
+      "loss": 1.1761,
+      "step": 1055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0179691468217694e-05,
+      "loss": 1.1838,
+      "step": 1056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 1.1764,
+      "step": 1057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982052353468757e-05,
+      "loss": 1.1196,
+      "step": 1058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.964126284403969e-05,
+      "loss": 1.1193,
+      "step": 1059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946221869888764e-05,
+      "loss": 1.1338,
+      "step": 1060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9283391869131656e-05,
+      "loss": 1.2031,
+      "step": 1061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9104783123737566e-05,
+      "loss": 1.1786,
+      "step": 1062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892639323073336e-05,
+      "loss": 1.1268,
+      "step": 1063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874822295720581e-05,
+      "loss": 1.1255,
+      "step": 1064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85702730692976e-05,
+      "loss": 1.1171,
+      "step": 1065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.839254433220355e-05,
+      "loss": 1.1187,
+      "step": 1066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821503751016746e-05,
+      "loss": 1.142,
+      "step": 1067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8037753366479085e-05,
+      "loss": 1.0964,
+      "step": 1068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7860692663470505e-05,
+      "loss": 1.1437,
+      "step": 1069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.768385616251304e-05,
+      "loss": 1.1338,
+      "step": 1070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7507244624013914e-05,
+      "loss": 1.1956,
+      "step": 1071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733085880741301e-05,
+      "loss": 1.1274,
+      "step": 1072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7154699471179575e-05,
+      "loss": 1.1342,
+      "step": 1073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697876737280901e-05,
+      "loss": 1.1718,
+      "step": 1074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680306326881954e-05,
+      "loss": 1.1409,
+      "step": 1075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6627587914749035e-05,
+      "loss": 1.2383,
+      "step": 1076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645234206515171e-05,
+      "loss": 1.1102,
+      "step": 1077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6277326473594894e-05,
+      "loss": 1.1378,
+      "step": 1078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610254189265577e-05,
+      "loss": 1.1069,
+      "step": 1079
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592798907391831e-05,
+      "loss": 1.1332,
+      "step": 1080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575366876796966e-05,
+      "loss": 1.1381,
+      "step": 1081
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5579581724397255e-05,
+      "loss": 1.1187,
+      "step": 1082
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5405728691785585e-05,
+      "loss": 1.2009,
+      "step": 1083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.523211041771278e-05,
+      "loss": 1.1088,
+      "step": 1084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505872764874741e-05,
+      "loss": 1.1226,
+      "step": 1085
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488558113044555e-05,
+      "loss": 1.1039,
+      "step": 1086
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471267160734731e-05,
+      "loss": 1.1254,
+      "step": 1087
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4539999822973553e-05,
+      "loss": 1.1194,
+      "step": 1088
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436756651982311e-05,
+      "loss": 1.178,
+      "step": 1089
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419537243936917e-05,
+      "loss": 1.1473,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.402341832205631e-05,
+      "loss": 1.1101,
+      "step": 1091
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385170490729712e-05,
+      "loss": 1.1844,
+      "step": 1092
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368023293346937e-05,
+      "loss": 1.1498,
+      "step": 1093
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3509003137912476e-05,
+      "loss": 1.2166,
+      "step": 1094
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.333801625692451e-05,
+      "loss": 1.1019,
+      "step": 1095
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.316727302575899e-05,
+      "loss": 1.1077,
+      "step": 1096
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2996774178621736e-05,
+      "loss": 1.1669,
+      "step": 1097
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2826520448667696e-05,
+      "loss": 1.0706,
+      "step": 1098
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265651256799781e-05,
+      "loss": 1.1403,
+      "step": 1099
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2486751267655825e-05,
+      "loss": 1.095,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231723727762521e-05,
+      "loss": 1.1428,
+      "step": 1101
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2147971326825966e-05,
+      "loss": 1.058,
+      "step": 1102
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1978954143111515e-05,
+      "loss": 1.1794,
+      "step": 1103
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1810186453265534e-05,
+      "loss": 1.1375,
+      "step": 1104
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1641668982999005e-05,
+      "loss": 1.0833,
+      "step": 1105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1473402456946744e-05,
+      "loss": 1.1292,
+      "step": 1106
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130538759866457e-05,
+      "loss": 1.0895,
+      "step": 1107
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1137625130626256e-05,
+      "loss": 1.1876,
+      "step": 1108
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0970115774220054e-05,
+      "loss": 1.1388,
+      "step": 1109
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0802860249745936e-05,
+      "loss": 1.163,
+      "step": 1110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063585927641243e-05,
+      "loss": 1.1589,
+      "step": 1111
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046911357233343e-05,
+      "loss": 1.1545,
+      "step": 1112
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0302623854525054e-05,
+      "loss": 1.1185,
+      "step": 1113
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013639083890287e-05,
+      "loss": 1.1385,
+      "step": 1114
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997041524027846e-05,
+      "loss": 1.1622,
+      "step": 1115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9804697772356544e-05,
+      "loss": 1.1124,
+      "step": 1116
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.963923914773187e-05,
+      "loss": 1.0652,
+      "step": 1117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9474040077886134e-05,
+      "loss": 1.1329,
+      "step": 1118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9309101273184926e-05,
+      "loss": 1.1583,
+      "step": 1119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91444234428747e-05,
+      "loss": 1.1295,
+      "step": 1120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8980007295079704e-05,
+      "loss": 1.1003,
+      "step": 1121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8815853536798904e-05,
+      "loss": 1.1418,
+      "step": 1122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865196287390301e-05,
+      "loss": 1.1189,
+      "step": 1123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848833601113141e-05,
+      "loss": 1.1334,
+      "step": 1124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8324973652089056e-05,
+      "loss": 1.1342,
+      "step": 1125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816187649924372e-05,
+      "loss": 1.1097,
+      "step": 1126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79990452539225e-05,
+      "loss": 1.0721,
+      "step": 1127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7836480616309244e-05,
+      "loss": 1.1252,
+      "step": 1128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.767418328544142e-05,
+      "loss": 1.1486,
+      "step": 1129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75121539592069e-05,
+      "loss": 1.1272,
+      "step": 1130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.735039333434117e-05,
+      "loss": 1.1097,
+      "step": 1131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7188902106424416e-05,
+      "loss": 1.1287,
+      "step": 1132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7027680969878274e-05,
+      "loss": 1.0658,
+      "step": 1133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.686673061796293e-05,
+      "loss": 1.142,
+      "step": 1134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6706051742774315e-05,
+      "loss": 1.1876,
+      "step": 1135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6545645035240915e-05,
+      "loss": 1.0573,
+      "step": 1136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.638551118512089e-05,
+      "loss": 1.1575,
+      "step": 1137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.622565088099911e-05,
+      "loss": 1.101,
+      "step": 1138
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.606606481028415e-05,
+      "loss": 1.0853,
+      "step": 1139
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5906753659205395e-05,
+      "loss": 1.1421,
+      "step": 1140
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5747718112810056e-05,
+      "loss": 1.1744,
+      "step": 1141
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558895885496023e-05,
+      "loss": 1.0837,
+      "step": 1142
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543047656832994e-05,
+      "loss": 1.0829,
+      "step": 1143
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5272271934402226e-05,
+      "loss": 1.1592,
+      "step": 1144
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.511434563346621e-05,
+      "loss": 1.2019,
+      "step": 1145
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4956698344614126e-05,
+      "loss": 1.1754,
+      "step": 1146
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479933074573858e-05,
+      "loss": 1.1323,
+      "step": 1147
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.464224351352929e-05,
+      "loss": 1.165,
+      "step": 1148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.448543732347048e-05,
+      "loss": 1.1201,
+      "step": 1149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.432891284983798e-05,
+      "loss": 1.2426,
+      "step": 1150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.417267076569602e-05,
+      "loss": 1.1017,
+      "step": 1151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.401671174289469e-05,
+      "loss": 1.1931,
+      "step": 1152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.386103645206679e-05,
+      "loss": 1.1614,
+      "step": 1153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.370564556262523e-05,
+      "loss": 1.0848,
+      "step": 1154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3550539742759766e-05,
+      "loss": 1.0789,
+      "step": 1155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.339571965943442e-05,
+      "loss": 1.1305,
+      "step": 1156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324118597838464e-05,
+      "loss": 1.1694,
+      "step": 1157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-05,
+      "loss": 1.1715,
+      "step": 1158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.293298047989244e-05,
+      "loss": 1.1205,
+      "step": 1159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2779309987751586e-05,
+      "loss": 1.176,
+      "step": 1160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262592854848369e-05,
+      "loss": 1.0929,
+      "step": 1161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2472836821637744e-05,
+      "loss": 1.168,
+      "step": 1162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2320035465517176e-05,
+      "loss": 1.1996,
+      "step": 1163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2167525137176655e-05,
+      "loss": 1.1481,
+      "step": 1164
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2015306492419475e-05,
+      "loss": 1.0442,
+      "step": 1165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.186338018579463e-05,
+      "loss": 1.1133,
+      "step": 1166
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1711746870594086e-05,
+      "loss": 1.1696,
+      "step": 1167
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1560407198849884e-05,
+      "loss": 1.0999,
+      "step": 1168
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.140936182133144e-05,
+      "loss": 1.1434,
+      "step": 1169
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1258611387542614e-05,
+      "loss": 1.146,
+      "step": 1170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.110815654571907e-05,
+      "loss": 1.141,
+      "step": 1171
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0957997942825336e-05,
+      "loss": 1.1866,
+      "step": 1172
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0808136224552154e-05,
+      "loss": 1.1321,
+      "step": 1173
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0658572035313584e-05,
+      "loss": 1.0857,
+      "step": 1174
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.050930601824441e-05,
+      "loss": 1.1287,
+      "step": 1175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0360338815197097e-05,
+      "loss": 1.1174,
+      "step": 1176
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.021167106673928e-05,
+      "loss": 1.1344,
+      "step": 1177
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0063303412150957e-05,
+      "loss": 1.0624,
+      "step": 1178
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9915236489421684e-05,
+      "loss": 1.1863,
+      "step": 1179
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9767470935247733e-05,
+      "loss": 1.1361,
+      "step": 1180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9620007385029658e-05,
+      "loss": 1.1829,
+      "step": 1181
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9472846472869298e-05,
+      "loss": 1.0559,
+      "step": 1182
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.932598883156704e-05,
+      "loss": 1.1842,
+      "step": 1183
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9179435092619366e-05,
+      "loss": 1.1078,
+      "step": 1184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.903318588621583e-05,
+      "loss": 1.1636,
+      "step": 1185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8887241841236524e-05,
+      "loss": 1.1315,
+      "step": 1186
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874160358524931e-05,
+      "loss": 1.0558,
+      "step": 1187
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8596271744507175e-05,
+      "loss": 1.0801,
+      "step": 1188
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.845124694394545e-05,
+      "loss": 1.1187,
+      "step": 1189
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8306529807179227e-05,
+      "loss": 1.0895,
+      "step": 1190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8162120956500593e-05,
+      "loss": 1.145,
+      "step": 1191
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8018021012875994e-05,
+      "loss": 1.0654,
+      "step": 1192
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7874230595943573e-05,
+      "loss": 1.1296,
+      "step": 1193
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7730750324010467e-05,
+      "loss": 1.0781,
+      "step": 1194
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.758758081405015e-05,
+      "loss": 1.1252,
+      "step": 1195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.744472268169993e-05,
+      "loss": 1.1168,
+      "step": 1196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7302176541257986e-05,
+      "loss": 1.1046,
+      "step": 1197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7159943005680997e-05,
+      "loss": 1.1561,
+      "step": 1198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7018022686581534e-05,
+      "loss": 1.145,
+      "step": 1199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.687641619422512e-05,
+      "loss": 1.0917,
+      "step": 1200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.67351241375279e-05,
+      "loss": 1.1322,
+      "step": 1201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.659414712405398e-05,
+      "loss": 1.1647,
+      "step": 1202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6453485760012718e-05,
+      "loss": 1.2217,
+      "step": 1203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6313140650256064e-05,
+      "loss": 1.1871,
+      "step": 1204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6173112398276235e-05,
+      "loss": 1.0884,
+      "step": 1205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.603340160620281e-05,
+      "loss": 1.1461,
+      "step": 1206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5894008874800325e-05,
+      "loss": 1.1003,
+      "step": 1207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5754934803465614e-05,
+      "loss": 1.1453,
+      "step": 1208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5616179990225263e-05,
+      "loss": 1.043,
+      "step": 1209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5477745031733057e-05,
+      "loss": 1.1628,
+      "step": 1210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.533963052326733e-05,
+      "loss": 1.0878,
+      "step": 1211
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5201837058728505e-05,
+      "loss": 1.1375,
+      "step": 1212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5064365230636498e-05,
+      "loss": 1.1521,
+      "step": 1213
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492721563012813e-05,
+      "loss": 1.1508,
+      "step": 1214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.479038884695468e-05,
+      "loss": 1.2063,
+      "step": 1215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4653885469479253e-05,
+      "loss": 1.1038,
+      "step": 1216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451770608467432e-05,
+      "loss": 1.0598,
+      "step": 1217
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4381851278119118e-05,
+      "loss": 1.1674,
+      "step": 1218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4246321633997225e-05,
+      "loss": 1.2076,
+      "step": 1219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4111117735093993e-05,
+      "loss": 1.1309,
+      "step": 1220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.397624016279403e-05,
+      "loss": 1.2217,
+      "step": 1221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3841689497078746e-05,
+      "loss": 1.09,
+      "step": 1222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3707466316523785e-05,
+      "loss": 1.1078,
+      "step": 1223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3573571198296706e-05,
+      "loss": 1.1259,
+      "step": 1224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3440004718154218e-05,
+      "loss": 1.1655,
+      "step": 1225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.330676745043995e-05,
+      "loss": 1.0859,
+      "step": 1226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3173859968081944e-05,
+      "loss": 1.0924,
+      "step": 1227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3041282842590097e-05,
+      "loss": 1.1421,
+      "step": 1228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290903664405367e-05,
+      "loss": 1.0713,
+      "step": 1229
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.277712194113909e-05,
+      "loss": 1.0837,
+      "step": 1230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2645539301087238e-05,
+      "loss": 1.0574,
+      "step": 1231
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.251428928971102e-05,
+      "loss": 1.1298,
+      "step": 1232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2383372471393215e-05,
+      "loss": 1.1036,
+      "step": 1233
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2252789409083708e-05,
+      "loss": 1.094,
+      "step": 1234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2122540664297254e-05,
+      "loss": 1.1522,
+      "step": 1235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1992626797111026e-05,
+      "loss": 1.1178,
+      "step": 1236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1863048366162208e-05,
+      "loss": 1.1472,
+      "step": 1237
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1733805928645567e-05,
+      "loss": 1.1826,
+      "step": 1238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1604900040311105e-05,
+      "loss": 1.1304,
+      "step": 1239
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1476331255461635e-05,
+      "loss": 1.087,
+      "step": 1240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1348100126950398e-05,
+      "loss": 1.1079,
+      "step": 1241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1220207206178688e-05,
+      "loss": 1.1205,
+      "step": 1242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1092653043093513e-05,
+      "loss": 1.1505,
+      "step": 1243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0965438186185148e-05,
+      "loss": 1.0918,
+      "step": 1244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.083856318248495e-05,
+      "loss": 1.0422,
+      "step": 1245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0712028577562738e-05,
+      "loss": 1.066,
+      "step": 1246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058583491552465e-05,
+      "loss": 1.1057,
+      "step": 1247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045998273901082e-05,
+      "loss": 1.0065,
+      "step": 1248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0334472589192933e-05,
+      "loss": 1.0826,
+      "step": 1249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.020930500577184e-05,
+      "loss": 1.1976,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.008448052697549e-05,
+      "loss": 1.1365,
+      "step": 1251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995999968955641e-05,
+      "loss": 1.1066,
+      "step": 1252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983586302878935e-05,
+      "loss": 1.2499,
+      "step": 1253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9712071078469242e-05,
+      "loss": 1.1559,
+      "step": 1254
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.958862437090865e-05,
+      "loss": 1.1011,
+      "step": 1255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.946552343693556e-05,
+      "loss": 1.1321,
+      "step": 1256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9342768805891178e-05,
+      "loss": 1.1361,
+      "step": 1257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.922036100562753e-05,
+      "loss": 1.1542,
+      "step": 1258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-05,
+      "loss": 1.1322,
+      "step": 1259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.897658800139139e-05,
+      "loss": 1.1599,
+      "step": 1260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8855223845656988e-05,
+      "loss": 1.054,
+      "step": 1261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8734208617174988e-05,
+      "loss": 1.0933,
+      "step": 1262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8613542836317888e-05,
+      "loss": 1.0936,
+      "step": 1263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.849322702195556e-05,
+      "loss": 1.2249,
+      "step": 1264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.837326169145299e-05,
+      "loss": 1.1735,
+      "step": 1265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8253647360668126e-05,
+      "loss": 1.1558,
+      "step": 1266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8134384543949478e-05,
+      "loss": 1.1237,
+      "step": 1267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.801547375413407e-05,
+      "loss": 1.1055,
+      "step": 1268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7896915502545273e-05,
+      "loss": 1.0945,
+      "step": 1269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7778710298990464e-05,
+      "loss": 1.1356,
+      "step": 1270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7660858651758792e-05,
+      "loss": 1.0895,
+      "step": 1271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.754336106761927e-05,
+      "loss": 1.1038,
+      "step": 1272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7426218051818344e-05,
+      "loss": 1.0459,
+      "step": 1273
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7309430108077707e-05,
+      "loss": 1.1342,
+      "step": 1274
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7192997738592375e-05,
+      "loss": 1.1471,
+      "step": 1275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707692144402828e-05,
+      "loss": 1.1213,
+      "step": 1276
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.696120172352025e-05,
+      "loss": 1.1418,
+      "step": 1277
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6845839074669777e-05,
+      "loss": 1.1826,
+      "step": 1278
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6730833993542982e-05,
+      "loss": 1.1608,
+      "step": 1279
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6616186974668378e-05,
+      "loss": 1.1427,
+      "step": 1280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6501898511034808e-05,
+      "loss": 1.0375,
+      "step": 1281
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6387969094089316e-05,
+      "loss": 1.1113,
+      "step": 1282
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6274399213735014e-05,
+      "loss": 1.1357,
+      "step": 1283
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6161189358328978e-05,
+      "loss": 1.0583,
+      "step": 1284
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6048340014680165e-05,
+      "loss": 1.0255,
+      "step": 1285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.593585166804732e-05,
+      "loss": 1.1014,
+      "step": 1286
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5823724802136865e-05,
+      "loss": 1.1903,
+      "step": 1287
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5711959899100838e-05,
+      "loss": 1.1096,
+      "step": 1288
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5600557439534812e-05,
+      "loss": 1.1181,
+      "step": 1289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5489517902475815e-05,
+      "loss": 1.0726,
+      "step": 1290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5378841765400386e-05,
+      "loss": 1.1566,
+      "step": 1291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526852950422226e-05,
+      "loss": 1.1924,
+      "step": 1292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5158581593290588e-05,
+      "loss": 1.1052,
+      "step": 1293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5048998505387835e-05,
+      "loss": 1.1394,
+      "step": 1294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.493978071172757e-05,
+      "loss": 1.0989,
+      "step": 1295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4830928681952672e-05,
+      "loss": 1.083,
+      "step": 1296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4722442884133214e-05,
+      "loss": 1.0513,
+      "step": 1297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.461432378476445e-05,
+      "loss": 1.1754,
+      "step": 1298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4506571848764716e-05,
+      "loss": 1.1161,
+      "step": 1299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4399187539473668e-05,
+      "loss": 1.1337,
+      "step": 1300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.429217131865005e-05,
+      "loss": 1.1,
+      "step": 1301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4185523646469822e-05,
+      "loss": 1.1135,
+      "step": 1302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4079244981524187e-05,
+      "loss": 1.1974,
+      "step": 1303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3973335780817564e-05,
+      "loss": 1.0519,
+      "step": 1304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3867796499765684e-05,
+      "loss": 1.1055,
+      "step": 1305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3762627592193566e-05,
+      "loss": 1.104,
+      "step": 1306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3657829510333654e-05,
+      "loss": 1.1268,
+      "step": 1307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.355340270482377e-05,
+      "loss": 0.9304,
+      "step": 1308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3449347624705244e-05,
+      "loss": 1.1049,
+      "step": 1309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.334566471742098e-05,
+      "loss": 1.1178,
+      "step": 1310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3242354428813497e-05,
+      "loss": 1.1276,
+      "step": 1311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3139417203123027e-05,
+      "loss": 1.1857,
+      "step": 1312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3036853482985644e-05,
+      "loss": 1.1394,
+      "step": 1313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2934663709431261e-05,
+      "loss": 1.168,
+      "step": 1314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2832848321881918e-05,
+      "loss": 1.1279,
+      "step": 1315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2731407758149639e-05,
+      "loss": 1.1697,
+      "step": 1316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.263034245443473e-05,
+      "loss": 1.1641,
+      "step": 1317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2529652845323914e-05,
+      "loss": 1.2076,
+      "step": 1318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2429339363788362e-05,
+      "loss": 1.1156,
+      "step": 1319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2329402441181815e-05,
+      "loss": 1.1279,
+      "step": 1320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2229842507238909e-05,
+      "loss": 1.1234,
+      "step": 1321
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2130659990073146e-05,
+      "loss": 1.1093,
+      "step": 1322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2031855316175033e-05,
+      "loss": 1.1171,
+      "step": 1323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1933428910410493e-05,
+      "loss": 1.0764,
+      "step": 1324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1835381196018769e-05,
+      "loss": 1.1218,
+      "step": 1325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1737712594610706e-05,
+      "loss": 1.123,
+      "step": 1326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1640423526166988e-05,
+      "loss": 1.1746,
+      "step": 1327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.154351440903626e-05,
+      "loss": 1.0978,
+      "step": 1328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1446985659933352e-05,
+      "loss": 1.17,
+      "step": 1329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1350837693937488e-05,
+      "loss": 1.1079,
+      "step": 1330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1255070924490518e-05,
+      "loss": 1.041,
+      "step": 1331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1159685763395111e-05,
+      "loss": 1.1293,
+      "step": 1332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1064682620813005e-05,
+      "loss": 1.1125,
+      "step": 1333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0970061905263229e-05,
+      "loss": 1.1263,
+      "step": 1334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.087582402362034e-05,
+      "loss": 1.1774,
+      "step": 1335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0781969381112755e-05,
+      "loss": 1.0214,
+      "step": 1336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0688498381320855e-05,
+      "loss": 1.2096,
+      "step": 1337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.059541142617535e-05,
+      "loss": 1.1316,
+      "step": 1338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0502708915955595e-05,
+      "loss": 1.0732,
+      "step": 1339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0410391249287788e-05,
+      "loss": 1.0472,
+      "step": 1340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0318458823143174e-05,
+      "loss": 1.1808,
+      "step": 1341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0226912032836611e-05,
+      "loss": 1.1432,
+      "step": 1342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0135751272024607e-05,
+      "loss": 1.0643,
+      "step": 1343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0044976932703686e-05,
+      "loss": 1.0752,
+      "step": 1344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.954589405208791e-06,
+      "loss": 1.1056,
+      "step": 1345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.864589078211572e-06,
+      "loss": 1.1063,
+      "step": 1346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.774976338718677e-06,
+      "loss": 0.9705,
+      "step": 1347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.685751572070024e-06,
+      "loss": 1.1154,
+      "step": 1348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.596915161937347e-06,
+      "loss": 1.1411,
+      "step": 1349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.50846749032237e-06,
+      "loss": 1.1222,
+      "step": 1350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.420408937555203e-06,
+      "loss": 1.0753,
+      "step": 1351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.332739882292752e-06,
+      "loss": 1.1313,
+      "step": 1352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.24546070151704e-06,
+      "loss": 1.1444,
+      "step": 1353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.158571770533619e-06,
+      "loss": 1.063,
+      "step": 1354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.072073462969944e-06,
+      "loss": 1.1243,
+      "step": 1355
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.98596615077376e-06,
+      "loss": 1.0975,
+      "step": 1356
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.900250204211514e-06,
+      "loss": 1.1384,
+      "step": 1357
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.814925991866751e-06,
+      "loss": 1.1802,
+      "step": 1358
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.729993880638555e-06,
+      "loss": 1.103,
+      "step": 1359
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645454235739903e-06,
+      "loss": 1.1226,
+      "step": 1360
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.561307420696252e-06,
+      "loss": 1.0923,
+      "step": 1361
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.47755379734373e-06,
+      "loss": 1.0684,
+      "step": 1362
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.394193725827792e-06,
+      "loss": 1.123,
+      "step": 1363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.311227564601642e-06,
+      "loss": 1.0881,
+      "step": 1364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.228655670424535e-06,
+      "loss": 1.0903,
+      "step": 1365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.146478398360414e-06,
+      "loss": 1.1017,
+      "step": 1366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.064696101776358e-06,
+      "loss": 1.1592,
+      "step": 1367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.98330913234101e-06,
+      "loss": 1.0905,
+      "step": 1368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.902317840023e-06,
+      "loss": 1.0947,
+      "step": 1369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.821722573089641e-06,
+      "loss": 1.0614,
+      "step": 1370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.741523678105257e-06,
+      "loss": 1.1547,
+      "step": 1371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.661721499929753e-06,
+      "loss": 1.071,
+      "step": 1372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.582316381717125e-06,
+      "loss": 1.0781,
+      "step": 1373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.503308664913999e-06,
+      "loss": 1.082,
+      "step": 1374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.424698689258158e-06,
+      "loss": 1.1213,
+      "step": 1375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.346486792777052e-06,
+      "loss": 1.0657,
+      "step": 1376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2686733117863784e-06,
+      "loss": 1.0945,
+      "step": 1377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.191258580888638e-06,
+      "loss": 1.1382,
+      "step": 1378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.114242932971671e-06,
+      "loss": 1.1581,
+      "step": 1379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.037626699207234e-06,
+      "loss": 1.1543,
+      "step": 1380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.961410209049568e-06,
+      "loss": 1.0864,
+      "step": 1381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8855937902340576e-06,
+      "loss": 1.0708,
+      "step": 1382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.8101777687756605e-06,
+      "loss": 1.1335,
+      "step": 1383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.735162468967649e-06,
+      "loss": 1.1222,
+      "step": 1384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.660548213380224e-06,
+      "loss": 1.0801,
+      "step": 1385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.586335322858972e-06,
+      "loss": 1.1293,
+      "step": 1386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.512524116523633e-06,
+      "loss": 1.0828,
+      "step": 1387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.439114911766708e-06,
+      "loss": 1.1077,
+      "step": 1388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.36610802425206e-06,
+      "loss": 1.1335,
+      "step": 1389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.293503767913478e-06,
+      "loss": 1.0741,
+      "step": 1390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.221302454953548e-06,
+      "loss": 1.1428,
+      "step": 1391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.149504395842087e-06,
+      "loss": 1.1523,
+      "step": 1392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.078109899314921e-06,
+      "loss": 1.1409,
+      "step": 1393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.007119272372541e-06,
+      "loss": 1.16,
+      "step": 1394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.936532820278751e-06,
+      "loss": 1.1863,
+      "step": 1395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8663508465594204e-06,
+      "loss": 1.1307,
+      "step": 1396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7965736530010916e-06,
+      "loss": 1.1085,
+      "step": 1397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.727201539649762e-06,
+      "loss": 1.1204,
+      "step": 1398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.658234804809548e-06,
+      "loss": 1.1105,
+      "step": 1399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.589673745041424e-06,
+      "loss": 1.1074,
+      "step": 1400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.521518655161917e-06,
+      "loss": 1.1377,
+      "step": 1401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.453769828241872e-06,
+      "loss": 1.1239,
+      "step": 1402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.386427555605211e-06,
+      "loss": 1.1853,
+      "step": 1403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.319492126827585e-06,
+      "loss": 1.0693,
+      "step": 1404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.252963829735247e-06,
+      "loss": 1.1064,
+      "step": 1405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.186842950403737e-06,
+      "loss": 1.137,
+      "step": 1406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.121129773156663e-06,
+      "loss": 1.1108,
+      "step": 1407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.055824580564517e-06,
+      "loss": 1.106,
+      "step": 1408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.990927653443423e-06,
+      "loss": 1.127,
+      "step": 1409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.926439270853955e-06,
+      "loss": 1.0995,
+      "step": 1410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.862359710099873e-06,
+      "loss": 1.1565,
+      "step": 1411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.798689246727006e-06,
+      "loss": 1.0658,
+      "step": 1412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.735428154522092e-06,
+      "loss": 1.1161,
+      "step": 1413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.672576705511478e-06,
+      "loss": 1.1477,
+      "step": 1414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.610135169960028e-06,
+      "loss": 1.0669,
+      "step": 1415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5481038163700085e-06,
+      "loss": 1.0856,
+      "step": 1416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.486482911479839e-06,
+      "loss": 1.1489,
+      "step": 1417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.425272720262941e-06,
+      "loss": 1.1494,
+      "step": 1418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3644735059267585e-06,
+      "loss": 1.1315,
+      "step": 1419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.304085529911406e-06,
+      "loss": 1.1737,
+      "step": 1420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.244109051888712e-06,
+      "loss": 1.0747,
+      "step": 1421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.184544329761009e-06,
+      "loss": 1.1371,
+      "step": 1422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.125391619660057e-06,
+      "loss": 1.1001,
+      "step": 1423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.066651175945946e-06,
+      "loss": 1.1176,
+      "step": 1424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0083232512060035e-06,
+      "loss": 1.1436,
+      "step": 1425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.950408096253699e-06,
+      "loss": 1.0927,
+      "step": 1426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.892905960127546e-06,
+      "loss": 1.1204,
+      "step": 1427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8358170900900995e-06,
+      "loss": 1.1904,
+      "step": 1428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.77914173162679e-06,
+      "loss": 1.0658,
+      "step": 1429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.72288012844495e-06,
+      "loss": 1.1394,
+      "step": 1430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6670325224727907e-06,
+      "loss": 1.095,
+      "step": 1431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.611599153858214e-06,
+      "loss": 1.1508,
+      "step": 1432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.556580260967923e-06,
+      "loss": 1.1062,
+      "step": 1433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5019760803863823e-06,
+      "loss": 1.0555,
+      "step": 1434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4477868469147466e-06,
+      "loss": 1.0875,
+      "step": 1435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.394012793569823e-06,
+      "loss": 1.0709,
+      "step": 1436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3406541515832003e-06,
+      "loss": 1.1849,
+      "step": 1437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2877111504001522e-06,
+      "loss": 1.1282,
+      "step": 1438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2351840176786164e-06,
+      "loss": 1.0425,
+      "step": 1439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1830729792883596e-06,
+      "loss": 1.101,
+      "step": 1440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1313782593098474e-06,
+      "loss": 1.1094,
+      "step": 1441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0801000800333877e-06,
+      "loss": 1.1074,
+      "step": 1442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0292386619581337e-06,
+      "loss": 1.0964,
+      "step": 1443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.978794223791137e-06,
+      "loss": 1.1046,
+      "step": 1444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9287669824464072e-06,
+      "loss": 1.1477,
+      "step": 1445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8791571530439985e-06,
+      "loss": 1.1431,
+      "step": 1446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8299649489090475e-06,
+      "loss": 1.111,
+      "step": 1447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.781190581570903e-06,
+      "loss": 1.1109,
+      "step": 1448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.732834260762174e-06,
+      "loss": 1.183,
+      "step": 1449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.684896194417841e-06,
+      "loss": 1.1298,
+      "step": 1450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6373765886743785e-06,
+      "loss": 1.1205,
+      "step": 1451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.590275647868867e-06,
+      "loss": 1.1281,
+      "step": 1452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5435935745380835e-06,
+      "loss": 1.1438,
+      "step": 1453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.497330569417633e-06,
+      "loss": 1.0519,
+      "step": 1454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.451486831441152e-06,
+      "loss": 1.0969,
+      "step": 1455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.406062557739375e-06,
+      "loss": 1.1803,
+      "step": 1456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3610579436393e-06,
+      "loss": 1.1187,
+      "step": 1457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3164731826633925e-06,
+      "loss": 1.1517,
+      "step": 1458
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2723084665287385e-06,
+      "loss": 1.1587,
+      "step": 1459
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.228563985146137e-06,
+      "loss": 1.1232,
+      "step": 1460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1852399266194314e-06,
+      "loss": 1.1198,
+      "step": 1461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1423364772445887e-06,
+      "loss": 1.1532,
+      "step": 1462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0998538215089348e-06,
+      "loss": 1.1666,
+      "step": 1463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0577921420903535e-06,
+      "loss": 1.1318,
+      "step": 1464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0161516198564988e-06,
+      "loss": 1.1309,
+      "step": 1465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9749324338640517e-06,
+      "loss": 1.1505,
+      "step": 1466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9341347613579087e-06,
+      "loss": 1.1744,
+      "step": 1467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8937587777704157e-06,
+      "loss": 1.1532,
+      "step": 1468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8538046567206591e-06,
+      "loss": 1.1746,
+      "step": 1469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8142725700136865e-06,
+      "loss": 1.0931,
+      "step": 1470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7751626876397532e-06,
+      "loss": 1.1063,
+      "step": 1471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7364751777736332e-06,
+      "loss": 1.0747,
+      "step": 1472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6982102067738758e-06,
+      "loss": 1.0687,
+      "step": 1473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6603679391820616e-06,
+      "loss": 1.1362,
+      "step": 1474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6229485377221576e-06,
+      "loss": 1.115,
+      "step": 1475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5859521632997533e-06,
+      "loss": 1.1656,
+      "step": 1476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5493789750014031e-06,
+      "loss": 1.0537,
+      "step": 1477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5132291300939628e-06,
+      "loss": 1.0706,
+      "step": 1478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4775027840238431e-06,
+      "loss": 1.076,
+      "step": 1479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4422000904164345e-06,
+      "loss": 1.0984,
+      "step": 1480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.40732120107534e-06,
+      "loss": 1.1323,
+      "step": 1481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3728662659818204e-06,
+      "loss": 1.1809,
+      "step": 1482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3388354332941166e-06,
+      "loss": 1.1897,
+      "step": 1483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3052288493467735e-06,
+      "loss": 1.1423,
+      "step": 1484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.272046658650039e-06,
+      "loss": 1.1757,
+      "step": 1485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.239289003889288e-06,
+      "loss": 1.104,
+      "step": 1486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2069560259243328e-06,
+      "loss": 1.087,
+      "step": 1487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1750478637888473e-06,
+      "loss": 1.1382,
+      "step": 1488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.143564654689777e-06,
+      "loss": 1.0892,
+      "step": 1489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1125065340067743e-06,
+      "loss": 1.1399,
+      "step": 1490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0818736352915527e-06,
+      "loss": 1.1483,
+      "step": 1491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0516660902673448e-06,
+      "loss": 1.0756,
+      "step": 1492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0218840288283571e-06,
+      "loss": 1.1632,
+      "step": 1493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.92527579039193e-07,
+      "loss": 1.0754,
+      "step": 1494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.635968671342754e-07,
+      "loss": 1.1562,
+      "step": 1495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.350920175173694e-07,
+      "loss": 1.0787,
+      "step": 1496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.070131527609604e-07,
+      "loss": 1.1334,
+      "step": 1497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.793603936058104e-07,
+      "loss": 1.1202,
+      "step": 1498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.521338589603911e-07,
+      "loss": 1.1241,
+      "step": 1499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.253336659003741e-07,
+      "loss": 1.1545,
+      "step": 1500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.989599296681749e-07,
+      "loss": 1.1483,
+      "step": 1501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.730127636723539e-07,
+      "loss": 1.0649,
+      "step": 1502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.474922794872497e-07,
+      "loss": 1.093,
+      "step": 1503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.223985868524241e-07,
+      "loss": 1.1648,
+      "step": 1504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.977317936722294e-07,
+      "loss": 1.095,
+      "step": 1505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.734920060152861e-07,
+      "loss": 1.2029,
+      "step": 1506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.496793281141056e-07,
+      "loss": 1.1561,
+      "step": 1507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.262938623645908e-07,
+      "loss": 1.1479,
+      "step": 1508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.033357093255809e-07,
+      "loss": 1.0747,
+      "step": 1509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.808049677184846e-07,
+      "loss": 1.1157,
+      "step": 1510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.587017344267919e-07,
+      "loss": 1.167,
+      "step": 1511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.370261044956971e-07,
+      "loss": 1.1707,
+      "step": 1512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.157781711316645e-07,
+      "loss": 1.123,
+      "step": 1513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.949580257020525e-07,
+      "loss": 1.0945,
+      "step": 1514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.745657577346796e-07,
+      "loss": 1.0894,
+      "step": 1515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.546014549174915e-07,
+      "loss": 1.1667,
+      "step": 1516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3506520309813947e-07,
+      "loss": 1.1136,
+      "step": 1517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1595708628366926e-07,
+      "loss": 1.1222,
+      "step": 1518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.97277186640066e-07,
+      "loss": 1.1266,
+      "step": 1519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7902558449198765e-07,
+      "loss": 1.1196,
+      "step": 1520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.612023583223878e-07,
+      "loss": 1.0963,
+      "step": 1521
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4380758477219333e-07,
+      "loss": 1.1187,
+      "step": 1522
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2684133863991604e-07,
+      "loss": 1.1245,
+      "step": 1523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1030369288139736e-07,
+      "loss": 1.0632,
+      "step": 1524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.941947186094973e-07,
+      "loss": 1.1425,
+      "step": 1525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7851448509372826e-07,
+      "loss": 1.0514,
+      "step": 1526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6326305976001055e-07,
+      "loss": 1.1483,
+      "step": 1527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.484405081903507e-07,
+      "loss": 1.1021,
+      "step": 1528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3404689412258596e-07,
+      "loss": 1.1647,
+      "step": 1529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2008227945009563e-07,
+      "loss": 1.0427,
+      "step": 1530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0654672422152356e-07,
+      "loss": 1.1174,
+      "step": 1531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9344028664056713e-07,
+      "loss": 1.0723,
+      "step": 1532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.807630230656443e-07,
+      "loss": 1.1381,
+      "step": 1533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6851498800973808e-07,
+      "loss": 1.1365,
+      "step": 1534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5669623414011901e-07,
+      "loss": 1.0962,
+      "step": 1535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4530681227808984e-07,
+      "loss": 1.0846,
+      "step": 1536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3434677139885222e-07,
+      "loss": 1.1132,
+      "step": 1537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2381615863120699e-07,
+      "loss": 1.1,
+      "step": 1538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.137150192573988e-07,
+      "loss": 1.1193,
+      "step": 1539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0404339671291618e-07,
+      "loss": 1.1172,
+      "step": 1540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.480133258630286e-08,
+      "loss": 1.0637,
+      "step": 1541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.598886661895788e-08,
+      "loss": 1.0847,
+      "step": 1542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.760603670501354e-08,
+      "loss": 1.1638,
+      "step": 1543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.965287889112438e-08,
+      "loss": 1.0765,
+      "step": 1544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.212942737632288e-08,
+      "loss": 1.0667,
+      "step": 1545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.503571451188627e-08,
+      "loss": 1.1353,
+      "step": 1546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.837177080119215e-08,
+      "loss": 1.1317,
+      "step": 1547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2137624899596387e-08,
+      "loss": 1.1282,
+      "step": 1548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.633330361427767e-08,
+      "loss": 1.0785,
+      "step": 1549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0958831904170926e-08,
+      "loss": 1.0884,
+      "step": 1550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.601423287978966e-08,
+      "loss": 1.1924,
+      "step": 1551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1499527803214846e-08,
+      "loss": 1.0063,
+      "step": 1552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7414736087950633e-08,
+      "loss": 1.1067,
+      "step": 1553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3759875298813286e-08,
+      "loss": 1.1364,
+      "step": 1554
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.05349611519312e-08,
+      "loss": 1.1656,
+      "step": 1555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.740007514622782e-09,
+      "loss": 1.1687,
+      "step": 1556
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.375026405352035e-09,
+      "loss": 1.1046,
+      "step": 1557
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4400279936619427e-09,
+      "loss": 1.1127,
+      "step": 1558
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.935020600174475e-09,
+      "loss": 1.0908,
+      "step": 1559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.600106965128696e-10,
+      "loss": 1.0944,
+      "step": 1560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1500290525722223e-10,
+      "loss": 1.1345,
+      "step": 1561
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.1343,
+      "step": 1562
+    },
+    {
+      "epoch": 1.0,
+      "step": 1562,
+      "total_flos": 1820945008820224.0,
+      "train_loss": 1.1780791337603034,
+      "train_runtime": 25885.7958,
+      "train_samples_per_second": 7.727,
+      "train_steps_per_second": 0.06
+    }
+  ],
+  "max_steps": 1562,
+  "num_train_epochs": 1,
+  "total_flos": 1820945008820224.0,
+  "trial_name": null,
+  "trial_params": null
+}