diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,72112 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.628517823639775,
+  "global_step": 12000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999906191369606,
+      "loss": 4.4817,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019998123827392122,
+      "loss": 4.0653,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999718574108818,
+      "loss": 4.1024,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019996247654784243,
+      "loss": 3.8332,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000199953095684803,
+      "loss": 3.8848,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999437148217636,
+      "loss": 3.5842,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999343339587242,
+      "loss": 3.7804,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019992495309568482,
+      "loss": 3.9365,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001999155722326454,
+      "loss": 4.056,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019990619136960603,
+      "loss": 3.7473,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019989681050656662,
+      "loss": 4.1747,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019988742964352724,
+      "loss": 3.7207,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019987804878048783,
+      "loss": 3.9834,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001998686679174484,
+      "loss": 3.5527,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000199859287054409,
+      "loss": 3.5294,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001998499061913696,
+      "loss": 3.9066,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019984052532833022,
+      "loss": 3.5959,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001998311444652908,
+      "loss": 3.4125,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019982176360225143,
+      "loss": 3.4993,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019981238273921202,
+      "loss": 3.5671,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019980300187617264,
+      "loss": 3.8916,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019979362101313323,
+      "loss": 3.5378,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019978424015009382,
+      "loss": 3.6733,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001997748592870544,
+      "loss": 3.4417,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019976547842401503,
+      "loss": 3.3094,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019975609756097562,
+      "loss": 3.341,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001997467166979362,
+      "loss": 3.4219,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019973733583489683,
+      "loss": 3.5906,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019972795497185742,
+      "loss": 3.7154,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019971857410881804,
+      "loss": 3.3878,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019970919324577863,
+      "loss": 3.3201,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019969981238273922,
+      "loss": 3.6221,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001996904315196998,
+      "loss": 3.5757,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019968105065666043,
+      "loss": 3.4569,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019967166979362102,
+      "loss": 3.4184,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019966228893058164,
+      "loss": 3.6408,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019965290806754223,
+      "loss": 3.4673,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019964352720450284,
+      "loss": 3.5535,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019963414634146344,
+      "loss": 3.5208,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019962476547842403,
+      "loss": 3.5555,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019961538461538462,
+      "loss": 3.8802,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001996060037523452,
+      "loss": 3.7051,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019959662288930583,
+      "loss": 3.6621,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019958724202626642,
+      "loss": 3.5117,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019957786116322704,
+      "loss": 3.5873,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019956848030018763,
+      "loss": 3.3219,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019955909943714824,
+      "loss": 3.5145,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019954971857410884,
+      "loss": 3.5083,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019954033771106943,
+      "loss": 3.4985,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019953095684803002,
+      "loss": 3.5682,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019952157598499064,
+      "loss": 3.6025,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019951219512195123,
+      "loss": 3.6677,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00019950281425891182,
+      "loss": 3.3639,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019949343339587244,
+      "loss": 3.4249,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019948405253283303,
+      "loss": 3.4072,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019947467166979364,
+      "loss": 3.3984,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019946529080675424,
+      "loss": 3.4417,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019945590994371483,
+      "loss": 3.4279,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019944652908067542,
+      "loss": 3.5666,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019943714821763604,
+      "loss": 3.2996,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019942776735459663,
+      "loss": 3.2709,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019941838649155724,
+      "loss": 3.6181,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019940900562851784,
+      "loss": 3.3186,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019939962476547845,
+      "loss": 3.4337,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019939024390243904,
+      "loss": 3.5828,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019938086303939964,
+      "loss": 3.5102,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019937148217636023,
+      "loss": 3.2281,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019936210131332082,
+      "loss": 3.6621,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019935272045028144,
+      "loss": 3.4608,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019934333958724203,
+      "loss": 3.4646,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019933395872420264,
+      "loss": 3.2228,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019932457786116324,
+      "loss": 3.5203,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019931519699812385,
+      "loss": 3.5169,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019930581613508444,
+      "loss": 3.3661,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019929643527204504,
+      "loss": 3.4719,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019928705440900563,
+      "loss": 3.6694,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019927767354596624,
+      "loss": 3.2621,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019926829268292683,
+      "loss": 3.4727,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019925891181988743,
+      "loss": 3.1679,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019924953095684804,
+      "loss": 3.5604,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019924015009380863,
+      "loss": 3.0779,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019923076923076925,
+      "loss": 3.3398,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019922138836772984,
+      "loss": 3.2669,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019921200750469043,
+      "loss": 3.4007,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019920262664165103,
+      "loss": 3.3207,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019919324577861164,
+      "loss": 3.3342,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019918386491557223,
+      "loss": 3.537,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019917448405253285,
+      "loss": 3.4133,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019916510318949344,
+      "loss": 3.3378,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019915572232645406,
+      "loss": 3.4997,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019914634146341465,
+      "loss": 3.3006,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019913696060037524,
+      "loss": 3.3124,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019912757973733583,
+      "loss": 3.456,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019911819887429643,
+      "loss": 3.3201,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019910881801125704,
+      "loss": 3.457,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019909943714821763,
+      "loss": 3.3134,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019909005628517825,
+      "loss": 3.3756,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019908067542213884,
+      "loss": 3.2911,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019907129455909946,
+      "loss": 3.3038,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019906191369606005,
+      "loss": 3.2558,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019905253283302064,
+      "loss": 3.4112,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019904315196998123,
+      "loss": 3.2516,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019903377110694185,
+      "loss": 3.3807,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019902439024390244,
+      "loss": 3.3388,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019901500938086306,
+      "loss": 3.5326,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019900562851782365,
+      "loss": 3.1055,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019899624765478424,
+      "loss": 3.334,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019898686679174486,
+      "loss": 3.1486,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019897748592870545,
+      "loss": 3.2901,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019896810506566604,
+      "loss": 3.5328,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019895872420262663,
+      "loss": 3.3431,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019894934333958725,
+      "loss": 3.2561,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019893996247654784,
+      "loss": 3.32,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019893058161350846,
+      "loss": 3.3777,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019892120075046905,
+      "loss": 3.258,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019891181988742967,
+      "loss": 3.4829,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019890243902439026,
+      "loss": 3.4526,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019889305816135085,
+      "loss": 3.6007,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019888367729831144,
+      "loss": 3.1199,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019887429643527203,
+      "loss": 3.1662,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019886491557223265,
+      "loss": 3.078,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019885553470919324,
+      "loss": 3.3478,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019884615384615386,
+      "loss": 3.7095,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019883677298311445,
+      "loss": 3.3086,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019882739212007507,
+      "loss": 3.2416,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019881801125703566,
+      "loss": 3.379,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019880863039399625,
+      "loss": 3.5442,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019879924953095684,
+      "loss": 3.476,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019878986866791746,
+      "loss": 3.3531,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019878048780487805,
+      "loss": 3.2609,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019877110694183867,
+      "loss": 3.6832,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019876172607879926,
+      "loss": 3.4571,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019875234521575985,
+      "loss": 3.217,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019874296435272047,
+      "loss": 3.409,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019873358348968106,
+      "loss": 3.1102,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019872420262664165,
+      "loss": 3.1755,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019871482176360224,
+      "loss": 3.1601,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019870544090056286,
+      "loss": 3.2392,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019869606003752345,
+      "loss": 2.9883,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019868667917448407,
+      "loss": 3.0923,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019867729831144466,
+      "loss": 3.1029,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019866791744840528,
+      "loss": 3.1164,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019865853658536587,
+      "loss": 3.346,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001986491557223265,
+      "loss": 3.1804,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019863977485928705,
+      "loss": 3.4769,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019863039399624764,
+      "loss": 3.1296,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019862101313320826,
+      "loss": 2.9615,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019861163227016885,
+      "loss": 3.4868,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019860225140712947,
+      "loss": 3.329,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019859287054409006,
+      "loss": 3.4755,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019858348968105068,
+      "loss": 3.153,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019857410881801127,
+      "loss": 3.2699,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001985647279549719,
+      "loss": 3.2815,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019855534709193245,
+      "loss": 3.0975,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019854596622889307,
+      "loss": 3.3787,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019853658536585366,
+      "loss": 3.2386,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019852720450281428,
+      "loss": 3.3616,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019851782363977487,
+      "loss": 3.1501,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019850844277673546,
+      "loss": 3.6069,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019849906191369608,
+      "loss": 3.2656,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019848968105065667,
+      "loss": 3.3203,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984803001876173,
+      "loss": 3.1733,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019847091932457785,
+      "loss": 3.204,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019846153846153847,
+      "loss": 3.2255,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019845215759849906,
+      "loss": 3.2185,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019844277673545968,
+      "loss": 3.3166,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019843339587242027,
+      "loss": 3.3379,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984240150093809,
+      "loss": 3.2231,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019841463414634148,
+      "loss": 3.5794,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984052532833021,
+      "loss": 3.1749,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001983958724202627,
+      "loss": 2.8704,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019838649155722325,
+      "loss": 3.4093,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019837711069418387,
+      "loss": 2.8694,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019836772983114446,
+      "loss": 3.1669,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019835834896810508,
+      "loss": 3.1269,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019834896810506567,
+      "loss": 3.5762,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001983395872420263,
+      "loss": 3.0951,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019833020637898688,
+      "loss": 3.4021,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001983208255159475,
+      "loss": 3.3154,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001983114446529081,
+      "loss": 3.1113,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019830206378986868,
+      "loss": 3.4401,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019829268292682927,
+      "loss": 3.3473,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982833020637899,
+      "loss": 3.2356,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019827392120075048,
+      "loss": 3.142,
+      "step": 184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826454033771107,
+      "loss": 3.4575,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982551594746717,
+      "loss": 3.2937,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019824577861163228,
+      "loss": 2.9108,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982363977485929,
+      "loss": 2.9836,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982270168855535,
+      "loss": 3.3146,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019821763602251408,
+      "loss": 3.437,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019820825515947467,
+      "loss": 3.4605,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981988742964353,
+      "loss": 3.2878,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019818949343339588,
+      "loss": 3.0226,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981801125703565,
+      "loss": 3.3413,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981707317073171,
+      "loss": 3.146,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981613508442777,
+      "loss": 3.2567,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981519699812383,
+      "loss": 3.0675,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981425891181989,
+      "loss": 3.0851,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019813320825515948,
+      "loss": 2.9115,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019812382739212007,
+      "loss": 3.43,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019811444652908069,
+      "loss": 3.4139,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019810506566604128,
+      "loss": 3.1939,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001980956848030019,
+      "loss": 3.002,
+      "step": 203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019808630393996249,
+      "loss": 3.2066,
+      "step": 204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001980769230769231,
+      "loss": 3.3469,
+      "step": 205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001980675422138837,
+      "loss": 3.0805,
+      "step": 206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019805816135084429,
+      "loss": 2.8329,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019804878048780488,
+      "loss": 2.9739,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001980393996247655,
+      "loss": 3.1715,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019803001876172609,
+      "loss": 3.2862,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019802063789868668,
+      "loss": 3.1153,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001980112570356473,
+      "loss": 3.1485,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019800187617260789,
+      "loss": 3.2315,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001979924953095685,
+      "loss": 3.3701,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001979831144465291,
+      "loss": 3.0637,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019797373358348969,
+      "loss": 3.2054,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019796435272045028,
+      "loss": 2.8963,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001979549718574109,
+      "loss": 3.048,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019794559099437149,
+      "loss": 3.3968,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001979362101313321,
+      "loss": 3.3146,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001979268292682927,
+      "loss": 3.1418,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001979174484052533,
+      "loss": 3.3573,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001979080675422139,
+      "loss": 3.1969,
+      "step": 223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001978986866791745,
+      "loss": 3.2168,
+      "step": 224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019788930581613509,
+      "loss": 3.2909,
+      "step": 225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019787992495309568,
+      "loss": 3.1886,
+      "step": 226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001978705440900563,
+      "loss": 3.2871,
+      "step": 227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019786116322701689,
+      "loss": 3.1421,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001978517823639775,
+      "loss": 3.3196,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001978424015009381,
+      "loss": 3.2007,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001978330206378987,
+      "loss": 2.9367,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001978236397748593,
+      "loss": 3.1745,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001978142589118199,
+      "loss": 3.2418,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019780487804878049,
+      "loss": 3.0981,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001977954971857411,
+      "loss": 3.3985,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001977861163227017,
+      "loss": 2.9821,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019777673545966229,
+      "loss": 3.2717,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001977673545966229,
+      "loss": 3.6389,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001977579737335835,
+      "loss": 3.1307,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001977485928705441,
+      "loss": 3.2556,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001977392120075047,
+      "loss": 3.2301,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001977298311444653,
+      "loss": 3.1238,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019772045028142589,
+      "loss": 3.1224,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001977110694183865,
+      "loss": 2.9565,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001977016885553471,
+      "loss": 3.3286,
+      "step": 245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976923076923077,
+      "loss": 2.9657,
+      "step": 246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976829268292683,
+      "loss": 3.2817,
+      "step": 247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019767354596622892,
+      "loss": 3.3331,
+      "step": 248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976641651031895,
+      "loss": 3.2049,
+      "step": 249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976547842401501,
+      "loss": 2.9588,
+      "step": 250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976454033771107,
+      "loss": 2.9376,
+      "step": 251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019763602251407129,
+      "loss": 3.2963,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976266416510319,
+      "loss": 3.1433,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976172607879925,
+      "loss": 3.251,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001976078799249531,
+      "loss": 3.2784,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975984990619137,
+      "loss": 3.1472,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019758911819887432,
+      "loss": 2.9169,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975797373358349,
+      "loss": 3.3255,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975703564727955,
+      "loss": 3.2086,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975609756097561,
+      "loss": 2.895,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975515947467167,
+      "loss": 3.3644,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975422138836773,
+      "loss": 2.614,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975328330206379,
+      "loss": 3.2392,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975234521575985,
+      "loss": 3.1228,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001975140712945591,
+      "loss": 3.5526,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019750469043151972,
+      "loss": 3.2653,
+      "step": 266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001974953095684803,
+      "loss": 3.0265,
+      "step": 267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001974859287054409,
+      "loss": 3.0848,
+      "step": 268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001974765478424015,
+      "loss": 2.9478,
+      "step": 269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001974671669793621,
+      "loss": 2.9934,
+      "step": 270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001974577861163227,
+      "loss": 3.105,
+      "step": 271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019744840525328332,
+      "loss": 3.3499,
+      "step": 272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001974390243902439,
+      "loss": 3.1997,
+      "step": 273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019742964352720453,
+      "loss": 3.229,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019742026266416512,
+      "loss": 3.0452,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001974108818011257,
+      "loss": 2.8139,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001974015009380863,
+      "loss": 2.9864,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001973921200750469,
+      "loss": 3.2327,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001973827392120075,
+      "loss": 3.2668,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001973733583489681,
+      "loss": 3.2585,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019736397748592872,
+      "loss": 3.3574,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001973545966228893,
+      "loss": 3.0275,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019734521575984993,
+      "loss": 3.3189,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019733583489681052,
+      "loss": 2.9936,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001973264540337711,
+      "loss": 3.213,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001973170731707317,
+      "loss": 3.1598,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019730769230769232,
+      "loss": 3.2352,
+      "step": 287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001972983114446529,
+      "loss": 2.9842,
+      "step": 288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001972889305816135,
+      "loss": 3.1276,
+      "step": 289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019727954971857412,
+      "loss": 3.3057,
+      "step": 290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001972701688555347,
+      "loss": 3.5251,
+      "step": 291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019726078799249533,
+      "loss": 2.8811,
+      "step": 292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019725140712945592,
+      "loss": 2.9672,
+      "step": 293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001972420262664165,
+      "loss": 3.4543,
+      "step": 294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001972326454033771,
+      "loss": 3.3079,
+      "step": 295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019722326454033772,
+      "loss": 3.1928,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001972138836772983,
+      "loss": 2.7911,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019720450281425893,
+      "loss": 3.2284,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019719512195121952,
+      "loss": 3.2733,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019718574108818014,
+      "loss": 3.3209,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019717636022514073,
+      "loss": 3.1679,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019716697936210132,
+      "loss": 3.2568,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001971575984990619,
+      "loss": 3.1317,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001971482176360225,
+      "loss": 3.1342,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019713883677298312,
+      "loss": 2.93,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001971294559099437,
+      "loss": 3.2792,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019712007504690433,
+      "loss": 3.3382,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019711069418386492,
+      "loss": 3.5098,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019710131332082554,
+      "loss": 3.3247,
+      "step": 309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019709193245778613,
+      "loss": 3.0962,
+      "step": 310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019708255159474675,
+      "loss": 3.1922,
+      "step": 311
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001970731707317073,
+      "loss": 3.2864,
+      "step": 312
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019706378986866793,
+      "loss": 3.6584,
+      "step": 313
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019705440900562852,
+      "loss": 2.946,
+      "step": 314
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019704502814258914,
+      "loss": 3.278,
+      "step": 315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019703564727954973,
+      "loss": 3.4231,
+      "step": 316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019702626641651032,
+      "loss": 2.9578,
+      "step": 317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019701688555347094,
+      "loss": 3.2847,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019700750469043153,
+      "loss": 3.4821,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019699812382739215,
+      "loss": 3.3185,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001969887429643527,
+      "loss": 3.2078,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019697936210131333,
+      "loss": 3.2401,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019696998123827392,
+      "loss": 3.1292,
+      "step": 323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019696060037523454,
+      "loss": 3.3149,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019695121951219513,
+      "loss": 3.1736,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019694183864915575,
+      "loss": 3.0693,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019693245778611634,
+      "loss": 3.2557,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019692307692307696,
+      "loss": 2.9683,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019691369606003755,
+      "loss": 3.0178,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001969043151969981,
+      "loss": 3.183,
+      "step": 330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019689493433395873,
+      "loss": 3.1946,
+      "step": 331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019688555347091932,
+      "loss": 3.2796,
+      "step": 332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019687617260787994,
+      "loss": 3.1918,
+      "step": 333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019686679174484053,
+      "loss": 3.0988,
+      "step": 334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019685741088180115,
+      "loss": 2.9437,
+      "step": 335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019684803001876174,
+      "loss": 3.4726,
+      "step": 336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019683864915572236,
+      "loss": 3.2602,
+      "step": 337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019682926829268295,
+      "loss": 3.2375,
+      "step": 338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019681988742964354,
+      "loss": 3.1893,
+      "step": 339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019681050656660413,
+      "loss": 2.9893,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019680112570356475,
+      "loss": 3.0255,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019679174484052534,
+      "loss": 3.3126,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019678236397748593,
+      "loss": 3.1722,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019677298311444655,
+      "loss": 3.3442,
+      "step": 344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019676360225140714,
+      "loss": 3.4173,
+      "step": 345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019675422138836776,
+      "loss": 3.1724,
+      "step": 346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019674484052532835,
+      "loss": 3.1334,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019673545966228894,
+      "loss": 3.1981,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019672607879924953,
+      "loss": 3.3249,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019671669793621015,
+      "loss": 3.182,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019670731707317074,
+      "loss": 2.9724,
+      "step": 351
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019669793621013136,
+      "loss": 3.108,
+      "step": 352
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019668855534709195,
+      "loss": 3.3069,
+      "step": 353
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019667917448405256,
+      "loss": 2.8407,
+      "step": 354
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019666979362101316,
+      "loss": 3.2126,
+      "step": 355
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019666041275797375,
+      "loss": 2.917,
+      "step": 356
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019665103189493434,
+      "loss": 3.2031,
+      "step": 357
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019664165103189493,
+      "loss": 2.9737,
+      "step": 358
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019663227016885555,
+      "loss": 3.2185,
+      "step": 359
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019662288930581614,
+      "loss": 3.1448,
+      "step": 360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019661350844277676,
+      "loss": 3.2242,
+      "step": 361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019660412757973735,
+      "loss": 3.025,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019659474671669796,
+      "loss": 2.9939,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019658536585365856,
+      "loss": 3.2972,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019657598499061915,
+      "loss": 3.3134,
+      "step": 365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019656660412757974,
+      "loss": 3.1917,
+      "step": 366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019655722326454035,
+      "loss": 3.2138,
+      "step": 367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019654784240150095,
+      "loss": 3.1665,
+      "step": 368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019653846153846154,
+      "loss": 3.3689,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019652908067542215,
+      "loss": 3.3078,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019651969981238275,
+      "loss": 3.1689,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019651031894934336,
+      "loss": 3.0549,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019650093808630395,
+      "loss": 2.8325,
+      "step": 373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019649155722326455,
+      "loss": 3.0392,
+      "step": 374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019648217636022514,
+      "loss": 2.895,
+      "step": 375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019647279549718575,
+      "loss": 3.2956,
+      "step": 376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019646341463414635,
+      "loss": 3.1983,
+      "step": 377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019645403377110696,
+      "loss": 2.7943,
+      "step": 378
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019644465290806755,
+      "loss": 3.2294,
+      "step": 379
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019643527204502817,
+      "loss": 3.2686,
+      "step": 380
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019642589118198876,
+      "loss": 3.3026,
+      "step": 381
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019641651031894935,
+      "loss": 3.0165,
+      "step": 382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019640712945590995,
+      "loss": 3.4815,
+      "step": 383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019639774859287054,
+      "loss": 3.3266,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019638836772983115,
+      "loss": 2.9623,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019637898686679175,
+      "loss": 3.4484,
+      "step": 386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019636960600375236,
+      "loss": 3.0611,
+      "step": 387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019636022514071295,
+      "loss": 3.3492,
+      "step": 388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019635084427767357,
+      "loss": 3.1666,
+      "step": 389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019634146341463416,
+      "loss": 3.1794,
+      "step": 390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019633208255159475,
+      "loss": 3.1826,
+      "step": 391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019632270168855535,
+      "loss": 3.1122,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019631332082551596,
+      "loss": 3.2412,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00019630393996247655,
+      "loss": 2.9783,
+      "step": 394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019629455909943715,
+      "loss": 3.2473,
+      "step": 395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019628517823639776,
+      "loss": 3.1384,
+      "step": 396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019627579737335835,
+      "loss": 3.1231,
+      "step": 397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019626641651031897,
+      "loss": 3.068,
+      "step": 398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019625703564727956,
+      "loss": 3.1763,
+      "step": 399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019624765478424015,
+      "loss": 3.1646,
+      "step": 400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019623827392120075,
+      "loss": 3.2737,
+      "step": 401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019622889305816136,
+      "loss": 2.936,
+      "step": 402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019621951219512195,
+      "loss": 2.807,
+      "step": 403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019621013133208257,
+      "loss": 3.0068,
+      "step": 404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019620075046904316,
+      "loss": 3.1226,
+      "step": 405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019619136960600378,
+      "loss": 3.1937,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019618198874296437,
+      "loss": 3.1578,
+      "step": 407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019617260787992496,
+      "loss": 3.2055,
+      "step": 408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019616322701688555,
+      "loss": 3.308,
+      "step": 409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019615384615384615,
+      "loss": 3.3338,
+      "step": 410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019614446529080676,
+      "loss": 3.1529,
+      "step": 411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019613508442776735,
+      "loss": 2.8154,
+      "step": 412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019612570356472797,
+      "loss": 3.1757,
+      "step": 413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019611632270168856,
+      "loss": 3.1299,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019610694183864918,
+      "loss": 3.1046,
+      "step": 415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019609756097560977,
+      "loss": 3.0446,
+      "step": 416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019608818011257036,
+      "loss": 3.0467,
+      "step": 417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019607879924953095,
+      "loss": 3.4481,
+      "step": 418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019606941838649157,
+      "loss": 3.0602,
+      "step": 419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019606003752345216,
+      "loss": 2.9465,
+      "step": 420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019605065666041275,
+      "loss": 2.7539,
+      "step": 421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019604127579737337,
+      "loss": 2.8985,
+      "step": 422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019603189493433396,
+      "loss": 3.0921,
+      "step": 423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019602251407129458,
+      "loss": 3.1786,
+      "step": 424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019601313320825517,
+      "loss": 2.9139,
+      "step": 425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019600375234521576,
+      "loss": 3.2591,
+      "step": 426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019599437148217635,
+      "loss": 3.3574,
+      "step": 427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019598499061913697,
+      "loss": 3.0397,
+      "step": 428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019597560975609756,
+      "loss": 2.8968,
+      "step": 429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019596622889305818,
+      "loss": 3.0519,
+      "step": 430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019595684803001877,
+      "loss": 3.2456,
+      "step": 431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001959474671669794,
+      "loss": 3.1925,
+      "step": 432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019593808630393998,
+      "loss": 3.0199,
+      "step": 433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019592870544090057,
+      "loss": 2.9992,
+      "step": 434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019591932457786116,
+      "loss": 2.9991,
+      "step": 435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019590994371482175,
+      "loss": 3.3517,
+      "step": 436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00019590056285178237,
+      "loss": 2.904,
+      "step": 437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019589118198874296,
+      "loss": 3.3845,
+      "step": 438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019588180112570358,
+      "loss": 2.9748,
+      "step": 439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019587242026266417,
+      "loss": 3.4354,
+      "step": 440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001958630393996248,
+      "loss": 3.1884,
+      "step": 441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019585365853658538,
+      "loss": 3.0647,
+      "step": 442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019584427767354597,
+      "loss": 3.0816,
+      "step": 443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019583489681050656,
+      "loss": 3.0807,
+      "step": 444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019582551594746718,
+      "loss": 3.3576,
+      "step": 445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019581613508442777,
+      "loss": 2.7145,
+      "step": 446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019580675422138836,
+      "loss": 3.2061,
+      "step": 447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019579737335834898,
+      "loss": 3.1307,
+      "step": 448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019578799249530957,
+      "loss": 3.107,
+      "step": 449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001957786116322702,
+      "loss": 3.2042,
+      "step": 450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019576923076923078,
+      "loss": 3.3162,
+      "step": 451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019575984990619137,
+      "loss": 3.0944,
+      "step": 452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019575046904315196,
+      "loss": 2.9056,
+      "step": 453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019574108818011258,
+      "loss": 3.1307,
+      "step": 454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019573170731707317,
+      "loss": 3.1115,
+      "step": 455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001957223264540338,
+      "loss": 2.9419,
+      "step": 456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019571294559099438,
+      "loss": 2.8983,
+      "step": 457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000195703564727955,
+      "loss": 3.1731,
+      "step": 458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001956941838649156,
+      "loss": 3.0275,
+      "step": 459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019568480300187618,
+      "loss": 3.1855,
+      "step": 460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019567542213883677,
+      "loss": 3.1892,
+      "step": 461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019566604127579736,
+      "loss": 3.1848,
+      "step": 462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019565666041275798,
+      "loss": 3.3083,
+      "step": 463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019564727954971857,
+      "loss": 3.3046,
+      "step": 464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001956378986866792,
+      "loss": 3.1709,
+      "step": 465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019562851782363978,
+      "loss": 2.8145,
+      "step": 466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001956191369606004,
+      "loss": 3.5426,
+      "step": 467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000195609756097561,
+      "loss": 3.1614,
+      "step": 468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001956003752345216,
+      "loss": 3.3498,
+      "step": 469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019559099437148217,
+      "loss": 3.0218,
+      "step": 470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001955816135084428,
+      "loss": 3.1899,
+      "step": 471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019557223264540338,
+      "loss": 3.2548,
+      "step": 472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019556285178236397,
+      "loss": 3.1656,
+      "step": 473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001955534709193246,
+      "loss": 3.2878,
+      "step": 474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019554409005628518,
+      "loss": 2.9428,
+      "step": 475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001955347091932458,
+      "loss": 2.8825,
+      "step": 476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001955253283302064,
+      "loss": 3.1964,
+      "step": 477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000195515947467167,
+      "loss": 3.1544,
+      "step": 478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019550656660412757,
+      "loss": 3.1118,
+      "step": 479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001954971857410882,
+      "loss": 2.9373,
+      "step": 480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019548780487804878,
+      "loss": 3.4792,
+      "step": 481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001954784240150094,
+      "loss": 2.9012,
+      "step": 482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019546904315197,
+      "loss": 3.0398,
+      "step": 483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001954596622889306,
+      "loss": 2.9935,
+      "step": 484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001954502814258912,
+      "loss": 3.136,
+      "step": 485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001954409005628518,
+      "loss": 2.957,
+      "step": 486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001954315196998124,
+      "loss": 3.2546,
+      "step": 487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019542213883677297,
+      "loss": 2.9323,
+      "step": 488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001954127579737336,
+      "loss": 3.2144,
+      "step": 489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019540337711069418,
+      "loss": 3.0524,
+      "step": 490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001953939962476548,
+      "loss": 3.0881,
+      "step": 491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001953846153846154,
+      "loss": 3.0603,
+      "step": 492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000195375234521576,
+      "loss": 2.7542,
+      "step": 493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001953658536585366,
+      "loss": 3.0851,
+      "step": 494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019535647279549722,
+      "loss": 3.0006,
+      "step": 495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001953470919324578,
+      "loss": 3.0807,
+      "step": 496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001953377110694184,
+      "loss": 3.0696,
+      "step": 497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000195328330206379,
+      "loss": 2.9916,
+      "step": 498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019531894934333958,
+      "loss": 3.075,
+      "step": 499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001953095684803002,
+      "loss": 3.1451,
+      "step": 500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001953001876172608,
+      "loss": 3.3378,
+      "step": 501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001952908067542214,
+      "loss": 3.2362,
+      "step": 502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000195281425891182,
+      "loss": 3.2573,
+      "step": 503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019527204502814262,
+      "loss": 3.2204,
+      "step": 504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001952626641651032,
+      "loss": 3.0927,
+      "step": 505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001952532833020638,
+      "loss": 3.1788,
+      "step": 506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001952439024390244,
+      "loss": 2.693,
+      "step": 507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000195234521575985,
+      "loss": 3.0395,
+      "step": 508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001952251407129456,
+      "loss": 3.0922,
+      "step": 509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019521575984990622,
+      "loss": 3.0552,
+      "step": 510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001952063789868668,
+      "loss": 2.8498,
+      "step": 511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001951969981238274,
+      "loss": 2.9657,
+      "step": 512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019518761726078801,
+      "loss": 3.0475,
+      "step": 513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001951782363977486,
+      "loss": 3.1197,
+      "step": 514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001951688555347092,
+      "loss": 3.0966,
+      "step": 515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001951594746716698,
+      "loss": 3.3986,
+      "step": 516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001951500938086304,
+      "loss": 3.2008,
+      "step": 517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000195140712945591,
+      "loss": 3.156,
+      "step": 518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019513133208255161,
+      "loss": 2.9226,
+      "step": 519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001951219512195122,
+      "loss": 2.9399,
+      "step": 520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019511257035647282,
+      "loss": 3.1263,
+      "step": 521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019510318949343341,
+      "loss": 3.0042,
+      "step": 522
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000195093808630394,
+      "loss": 3.3067,
+      "step": 523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001950844277673546,
+      "loss": 3.0069,
+      "step": 524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019507504690431521,
+      "loss": 3.0349,
+      "step": 525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001950656660412758,
+      "loss": 3.2025,
+      "step": 526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001950562851782364,
+      "loss": 2.7538,
+      "step": 527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019504690431519701,
+      "loss": 3.1622,
+      "step": 528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001950375234521576,
+      "loss": 3.1557,
+      "step": 529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019502814258911822,
+      "loss": 3.0088,
+      "step": 530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019501876172607881,
+      "loss": 3.2109,
+      "step": 531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001950093808630394,
+      "loss": 3.0408,
+      "step": 532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000195,
+      "loss": 3.1948,
+      "step": 533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019499061913696061,
+      "loss": 3.1765,
+      "step": 534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001949812382739212,
+      "loss": 3.0711,
+      "step": 535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019497185741088182,
+      "loss": 3.1334,
+      "step": 536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019496247654784241,
+      "loss": 3.018,
+      "step": 537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019495309568480303,
+      "loss": 2.7275,
+      "step": 538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019494371482176362,
+      "loss": 2.817,
+      "step": 539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019493433395872421,
+      "loss": 3.0515,
+      "step": 540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001949249530956848,
+      "loss": 3.1046,
+      "step": 541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001949155722326454,
+      "loss": 2.7649,
+      "step": 542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00019490619136960601,
+      "loss": 3.3219,
+      "step": 543
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001948968105065666,
+      "loss": 3.0036,
+      "step": 544
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019488742964352722,
+      "loss": 3.0524,
+      "step": 545
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019487804878048781,
+      "loss": 3.1458,
+      "step": 546
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019486866791744843,
+      "loss": 3.4298,
+      "step": 547
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019485928705440902,
+      "loss": 3.3886,
+      "step": 548
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019484990619136961,
+      "loss": 3.4971,
+      "step": 549
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001948405253283302,
+      "loss": 3.1215,
+      "step": 550
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019483114446529082,
+      "loss": 3.1412,
+      "step": 551
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019482176360225141,
+      "loss": 3.0919,
+      "step": 552
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000194812382739212,
+      "loss": 2.887,
+      "step": 553
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019480300187617262,
+      "loss": 3.3028,
+      "step": 554
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019479362101313321,
+      "loss": 3.2607,
+      "step": 555
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019478424015009383,
+      "loss": 3.0649,
+      "step": 556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019477485928705442,
+      "loss": 2.9031,
+      "step": 557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019476547842401501,
+      "loss": 3.2224,
+      "step": 558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001947560975609756,
+      "loss": 2.844,
+      "step": 559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019474671669793622,
+      "loss": 3.1729,
+      "step": 560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019473733583489681,
+      "loss": 3.0851,
+      "step": 561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019472795497185743,
+      "loss": 3.282,
+      "step": 562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019471857410881802,
+      "loss": 3.1163,
+      "step": 563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019470919324577864,
+      "loss": 3.2567,
+      "step": 564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019469981238273923,
+      "loss": 3.0017,
+      "step": 565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019469043151969982,
+      "loss": 2.8741,
+      "step": 566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019468105065666041,
+      "loss": 3.0884,
+      "step": 567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000194671669793621,
+      "loss": 2.8795,
+      "step": 568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019466228893058162,
+      "loss": 3.1967,
+      "step": 569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019465290806754221,
+      "loss": 2.8676,
+      "step": 570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019464352720450283,
+      "loss": 2.9411,
+      "step": 571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019463414634146342,
+      "loss": 3.2536,
+      "step": 572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019462476547842404,
+      "loss": 3.2045,
+      "step": 573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019461538461538463,
+      "loss": 3.1442,
+      "step": 574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019460600375234522,
+      "loss": 3.0194,
+      "step": 575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019459662288930581,
+      "loss": 2.9089,
+      "step": 576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019458724202626643,
+      "loss": 3.0816,
+      "step": 577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019457786116322702,
+      "loss": 3.0287,
+      "step": 578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001945684803001876,
+      "loss": 3.0447,
+      "step": 579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019455909943714823,
+      "loss": 3.1022,
+      "step": 580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019454971857410882,
+      "loss": 2.8802,
+      "step": 581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019454033771106944,
+      "loss": 3.1721,
+      "step": 582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019453095684803003,
+      "loss": 3.2243,
+      "step": 583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019452157598499062,
+      "loss": 2.9123,
+      "step": 584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0001945121951219512,
+      "loss": 3.1464,
+      "step": 585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019450281425891183,
+      "loss": 3.2424,
+      "step": 586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019449343339587242,
+      "loss": 3.0831,
+      "step": 587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019448405253283304,
+      "loss": 2.9059,
+      "step": 588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019447467166979363,
+      "loss": 3.0837,
+      "step": 589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019446529080675425,
+      "loss": 3.3349,
+      "step": 590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019445590994371484,
+      "loss": 2.6794,
+      "step": 591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019444652908067543,
+      "loss": 2.8404,
+      "step": 592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019443714821763602,
+      "loss": 3.4959,
+      "step": 593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001944277673545966,
+      "loss": 3.1384,
+      "step": 594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019441838649155723,
+      "loss": 3.1176,
+      "step": 595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019440900562851782,
+      "loss": 3.0553,
+      "step": 596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019439962476547844,
+      "loss": 3.2318,
+      "step": 597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019439024390243903,
+      "loss": 3.0702,
+      "step": 598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019438086303939965,
+      "loss": 2.9779,
+      "step": 599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019437148217636024,
+      "loss": 3.1416,
+      "step": 600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019436210131332083,
+      "loss": 2.8056,
+      "step": 601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019435272045028142,
+      "loss": 2.7396,
+      "step": 602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019434333958724204,
+      "loss": 2.7117,
+      "step": 603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019433395872420263,
+      "loss": 3.2221,
+      "step": 604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019432457786116322,
+      "loss": 3.0653,
+      "step": 605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019431519699812384,
+      "loss": 2.889,
+      "step": 606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019430581613508443,
+      "loss": 3.3049,
+      "step": 607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019429643527204505,
+      "loss": 3.3636,
+      "step": 608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019428705440900564,
+      "loss": 2.8255,
+      "step": 609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019427767354596623,
+      "loss": 2.8951,
+      "step": 610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019426829268292682,
+      "loss": 3.0836,
+      "step": 611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019425891181988744,
+      "loss": 2.8587,
+      "step": 612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019424953095684803,
+      "loss": 3.1714,
+      "step": 613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019424015009380865,
+      "loss": 3.3066,
+      "step": 614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019423076923076924,
+      "loss": 3.073,
+      "step": 615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019422138836772986,
+      "loss": 2.8586,
+      "step": 616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019421200750469045,
+      "loss": 2.9627,
+      "step": 617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019420262664165104,
+      "loss": 3.0333,
+      "step": 618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019419324577861163,
+      "loss": 2.9957,
+      "step": 619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019418386491557222,
+      "loss": 2.7639,
+      "step": 620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019417448405253284,
+      "loss": 3.0094,
+      "step": 621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019416510318949343,
+      "loss": 2.9303,
+      "step": 622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019415572232645405,
+      "loss": 2.7504,
+      "step": 623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019414634146341464,
+      "loss": 2.9521,
+      "step": 624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019413696060037526,
+      "loss": 3.0471,
+      "step": 625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019412757973733585,
+      "loss": 3.0855,
+      "step": 626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019411819887429644,
+      "loss": 3.1111,
+      "step": 627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019410881801125703,
+      "loss": 3.1357,
+      "step": 628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019409943714821765,
+      "loss": 3.1366,
+      "step": 629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019409005628517824,
+      "loss": 3.3789,
+      "step": 630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019408067542213883,
+      "loss": 3.2996,
+      "step": 631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019407129455909945,
+      "loss": 3.1689,
+      "step": 632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019406191369606004,
+      "loss": 2.8654,
+      "step": 633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019405253283302066,
+      "loss": 3.3052,
+      "step": 634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019404315196998125,
+      "loss": 3.1662,
+      "step": 635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019403377110694184,
+      "loss": 3.1173,
+      "step": 636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019402439024390243,
+      "loss": 2.8821,
+      "step": 637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019401500938086305,
+      "loss": 3.0943,
+      "step": 638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019400562851782364,
+      "loss": 3.0445,
+      "step": 639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019399624765478426,
+      "loss": 3.1412,
+      "step": 640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019398686679174485,
+      "loss": 3.0004,
+      "step": 641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019397748592870547,
+      "loss": 2.761,
+      "step": 642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019396810506566606,
+      "loss": 3.2268,
+      "step": 643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019395872420262665,
+      "loss": 3.1613,
+      "step": 644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019394934333958727,
+      "loss": 3.2458,
+      "step": 645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019393996247654783,
+      "loss": 3.2023,
+      "step": 646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019393058161350845,
+      "loss": 2.8217,
+      "step": 647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019392120075046904,
+      "loss": 3.0926,
+      "step": 648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019391181988742966,
+      "loss": 2.8252,
+      "step": 649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00019390243902439025,
+      "loss": 2.7624,
+      "step": 650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019389305816135087,
+      "loss": 2.9667,
+      "step": 651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019388367729831146,
+      "loss": 3.071,
+      "step": 652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019387429643527208,
+      "loss": 2.9323,
+      "step": 653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019386491557223267,
+      "loss": 3.0918,
+      "step": 654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019385553470919326,
+      "loss": 2.9776,
+      "step": 655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019384615384615385,
+      "loss": 2.8811,
+      "step": 656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019383677298311444,
+      "loss": 2.7976,
+      "step": 657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019382739212007506,
+      "loss": 3.0717,
+      "step": 658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019381801125703565,
+      "loss": 3.3336,
+      "step": 659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019380863039399627,
+      "loss": 2.6622,
+      "step": 660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019379924953095686,
+      "loss": 3.1295,
+      "step": 661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019378986866791747,
+      "loss": 2.9833,
+      "step": 662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019378048780487807,
+      "loss": 3.0008,
+      "step": 663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019377110694183866,
+      "loss": 3.1715,
+      "step": 664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019376172607879925,
+      "loss": 3.1821,
+      "step": 665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019375234521575987,
+      "loss": 3.0588,
+      "step": 666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019374296435272046,
+      "loss": 3.1356,
+      "step": 667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019373358348968107,
+      "loss": 3.2505,
+      "step": 668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019372420262664167,
+      "loss": 3.172,
+      "step": 669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019371482176360226,
+      "loss": 2.8879,
+      "step": 670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00019370544090056287,
+      "loss": 2.8009,
+      "step": 671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019369606003752347,
+      "loss": 3.1943,
+      "step": 672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019368667917448406,
+      "loss": 3.273,
+      "step": 673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019367729831144465,
+      "loss": 3.0476,
+      "step": 674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019366791744840527,
+      "loss": 3.1811,
+      "step": 675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019365853658536586,
+      "loss": 3.0571,
+      "step": 676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019364915572232647,
+      "loss": 2.8432,
+      "step": 677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019363977485928707,
+      "loss": 3.1828,
+      "step": 678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019363039399624768,
+      "loss": 3.0819,
+      "step": 679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019362101313320827,
+      "loss": 2.9993,
+      "step": 680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019361163227016887,
+      "loss": 3.1131,
+      "step": 681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019360225140712946,
+      "loss": 3.0949,
+      "step": 682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019359287054409005,
+      "loss": 2.9513,
+      "step": 683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019358348968105067,
+      "loss": 3.0277,
+      "step": 684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019357410881801126,
+      "loss": 3.1328,
+      "step": 685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019356472795497187,
+      "loss": 3.215,
+      "step": 686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019355534709193247,
+      "loss": 2.9753,
+      "step": 687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019354596622889308,
+      "loss": 3.233,
+      "step": 688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019353658536585367,
+      "loss": 2.9114,
+      "step": 689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019352720450281427,
+      "loss": 2.9451,
+      "step": 690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019351782363977486,
+      "loss": 2.8106,
+      "step": 691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019350844277673547,
+      "loss": 2.997,
+      "step": 692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019349906191369607,
+      "loss": 3.1396,
+      "step": 693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019348968105065668,
+      "loss": 3.1427,
+      "step": 694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019348030018761727,
+      "loss": 2.6956,
+      "step": 695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019347091932457787,
+      "loss": 3.0561,
+      "step": 696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019346153846153848,
+      "loss": 3.0072,
+      "step": 697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019345215759849907,
+      "loss": 3.2178,
+      "step": 698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019344277673545967,
+      "loss": 2.8966,
+      "step": 699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019343339587242026,
+      "loss": 2.8754,
+      "step": 700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019342401500938087,
+      "loss": 3.2778,
+      "step": 701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019341463414634146,
+      "loss": 2.6975,
+      "step": 702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019340525328330208,
+      "loss": 2.9741,
+      "step": 703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019339587242026267,
+      "loss": 2.9559,
+      "step": 704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001933864915572233,
+      "loss": 2.9405,
+      "step": 705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019337711069418388,
+      "loss": 3.112,
+      "step": 706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019336772983114447,
+      "loss": 3.1545,
+      "step": 707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019335834896810506,
+      "loss": 2.8339,
+      "step": 708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019334896810506566,
+      "loss": 3.4372,
+      "step": 709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019333958724202627,
+      "loss": 2.9161,
+      "step": 710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019333020637898686,
+      "loss": 2.927,
+      "step": 711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019332082551594748,
+      "loss": 3.0165,
+      "step": 712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019331144465290807,
+      "loss": 2.8245,
+      "step": 713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001933020637898687,
+      "loss": 3.1716,
+      "step": 714
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019329268292682928,
+      "loss": 2.9381,
+      "step": 715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019328330206378987,
+      "loss": 3.1958,
+      "step": 716
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019327392120075046,
+      "loss": 3.0518,
+      "step": 717
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019326454033771108,
+      "loss": 2.7724,
+      "step": 718
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019325515947467167,
+      "loss": 3.0531,
+      "step": 719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001932457786116323,
+      "loss": 2.9706,
+      "step": 720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019323639774859288,
+      "loss": 3.3402,
+      "step": 721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001932270168855535,
+      "loss": 2.9965,
+      "step": 722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001932176360225141,
+      "loss": 3.2885,
+      "step": 723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019320825515947468,
+      "loss": 2.8856,
+      "step": 724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019319887429643527,
+      "loss": 3.0961,
+      "step": 725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019318949343339586,
+      "loss": 2.6331,
+      "step": 726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019318011257035648,
+      "loss": 3.1912,
+      "step": 727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019317073170731707,
+      "loss": 3.2978,
+      "step": 728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001931613508442777,
+      "loss": 3.2238,
+      "step": 729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019315196998123828,
+      "loss": 2.8345,
+      "step": 730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001931425891181989,
+      "loss": 2.9237,
+      "step": 731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001931332082551595,
+      "loss": 2.8534,
+      "step": 732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019312382739212008,
+      "loss": 3.2027,
+      "step": 733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00019311444652908067,
+      "loss": 3.0346,
+      "step": 734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001931050656660413,
+      "loss": 3.1548,
+      "step": 735
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019309568480300188,
+      "loss": 3.1619,
+      "step": 736
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019308630393996247,
+      "loss": 3.0942,
+      "step": 737
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001930769230769231,
+      "loss": 2.9231,
+      "step": 738
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019306754221388368,
+      "loss": 2.8364,
+      "step": 739
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001930581613508443,
+      "loss": 2.9348,
+      "step": 740
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001930487804878049,
+      "loss": 3.0292,
+      "step": 741
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019303939962476548,
+      "loss": 2.9509,
+      "step": 742
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019303001876172607,
+      "loss": 2.9107,
+      "step": 743
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001930206378986867,
+      "loss": 2.9693,
+      "step": 744
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019301125703564728,
+      "loss": 3.0529,
+      "step": 745
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001930018761726079,
+      "loss": 3.4629,
+      "step": 746
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001929924953095685,
+      "loss": 2.9979,
+      "step": 747
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001929831144465291,
+      "loss": 3.2524,
+      "step": 748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001929737335834897,
+      "loss": 3.2609,
+      "step": 749
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001929643527204503,
+      "loss": 3.4833,
+      "step": 750
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019295497185741088,
+      "loss": 3.0933,
+      "step": 751
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019294559099437147,
+      "loss": 3.2914,
+      "step": 752
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001929362101313321,
+      "loss": 3.0512,
+      "step": 753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00019292682926829268,
+      "loss": 2.7888,
+      "step": 754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001929174484052533,
+      "loss": 3.2527,
+      "step": 755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001929080675422139,
+      "loss": 3.348,
+      "step": 756
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001928986866791745,
+      "loss": 2.8773,
+      "step": 757
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001928893058161351,
+      "loss": 3.1064,
+      "step": 758
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001928799249530957,
+      "loss": 3.3385,
+      "step": 759
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019287054409005628,
+      "loss": 2.7625,
+      "step": 760
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001928611632270169,
+      "loss": 2.9355,
+      "step": 761
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001928517823639775,
+      "loss": 2.5504,
+      "step": 762
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019284240150093808,
+      "loss": 3.0138,
+      "step": 763
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001928330206378987,
+      "loss": 3.0443,
+      "step": 764
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001928236397748593,
+      "loss": 3.1898,
+      "step": 765
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001928142589118199,
+      "loss": 3.1554,
+      "step": 766
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001928048780487805,
+      "loss": 2.7318,
+      "step": 767
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001927954971857411,
+      "loss": 3.2098,
+      "step": 768
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019278611632270168,
+      "loss": 2.9541,
+      "step": 769
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001927767354596623,
+      "loss": 2.6701,
+      "step": 770
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001927673545966229,
+      "loss": 2.9679,
+      "step": 771
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001927579737335835,
+      "loss": 3.1301,
+      "step": 772
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001927485928705441,
+      "loss": 2.787,
+      "step": 773
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019273921200750472,
+      "loss": 2.8549,
+      "step": 774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001927298311444653,
+      "loss": 2.9394,
+      "step": 775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001927204502814259,
+      "loss": 2.8855,
+      "step": 776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001927110694183865,
+      "loss": 3.0112,
+      "step": 777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019270168855534708,
+      "loss": 2.9168,
+      "step": 778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926923076923077,
+      "loss": 2.9643,
+      "step": 779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926829268292683,
+      "loss": 3.1243,
+      "step": 780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926735459662289,
+      "loss": 3.0134,
+      "step": 781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926641651031895,
+      "loss": 2.7943,
+      "step": 782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019265478424015012,
+      "loss": 3.2171,
+      "step": 783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926454033771107,
+      "loss": 3.2113,
+      "step": 784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926360225140713,
+      "loss": 3.0357,
+      "step": 785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926266416510319,
+      "loss": 3.101,
+      "step": 786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926172607879925,
+      "loss": 2.9796,
+      "step": 787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001926078799249531,
+      "loss": 3.0863,
+      "step": 788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001925984990619137,
+      "loss": 2.7836,
+      "step": 789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001925891181988743,
+      "loss": 3.3233,
+      "step": 790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001925797373358349,
+      "loss": 2.8138,
+      "step": 791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019257035647279552,
+      "loss": 3.0427,
+      "step": 792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001925609756097561,
+      "loss": 2.9022,
+      "step": 793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001925515947467167,
+      "loss": 2.9302,
+      "step": 794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001925422138836773,
+      "loss": 3.172,
+      "step": 795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001925328330206379,
+      "loss": 3.1596,
+      "step": 796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001925234521575985,
+      "loss": 2.7658,
+      "step": 797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019251407129455912,
+      "loss": 2.9452,
+      "step": 798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001925046904315197,
+      "loss": 3.2564,
+      "step": 799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019249530956848033,
+      "loss": 2.627,
+      "step": 800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019248592870544092,
+      "loss": 3.0656,
+      "step": 801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001924765478424015,
+      "loss": 2.9418,
+      "step": 802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001924671669793621,
+      "loss": 2.9137,
+      "step": 803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001924577861163227,
+      "loss": 3.1118,
+      "step": 804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001924484052532833,
+      "loss": 3.1386,
+      "step": 805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001924390243902439,
+      "loss": 3.1434,
+      "step": 806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019242964352720452,
+      "loss": 2.9494,
+      "step": 807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001924202626641651,
+      "loss": 3.1859,
+      "step": 808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019241088180112573,
+      "loss": 3.261,
+      "step": 809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019240150093808632,
+      "loss": 3.0718,
+      "step": 810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019239212007504693,
+      "loss": 2.7864,
+      "step": 811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019238273921200753,
+      "loss": 2.9417,
+      "step": 812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019237335834896812,
+      "loss": 2.886,
+      "step": 813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001923639774859287,
+      "loss": 3.2682,
+      "step": 814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001923545966228893,
+      "loss": 3.0338,
+      "step": 815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019234521575984992,
+      "loss": 3.0418,
+      "step": 816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001923358348968105,
+      "loss": 3.187,
+      "step": 817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019232645403377113,
+      "loss": 2.7942,
+      "step": 818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019231707317073172,
+      "loss": 2.8255,
+      "step": 819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 2.99,
+      "step": 820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019229831144465293,
+      "loss": 3.1627,
+      "step": 821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019228893058161352,
+      "loss": 2.9989,
+      "step": 822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001922795497185741,
+      "loss": 2.8494,
+      "step": 823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019227016885553473,
+      "loss": 2.9975,
+      "step": 824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019226078799249532,
+      "loss": 3.1517,
+      "step": 825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019225140712945593,
+      "loss": 3.002,
+      "step": 826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019224202626641653,
+      "loss": 2.9515,
+      "step": 827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019223264540337712,
+      "loss": 3.0847,
+      "step": 828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019222326454033773,
+      "loss": 2.9212,
+      "step": 829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019221388367729833,
+      "loss": 2.843,
+      "step": 830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019220450281425892,
+      "loss": 3.1478,
+      "step": 831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001921951219512195,
+      "loss": 3.1806,
+      "step": 832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019218574108818013,
+      "loss": 3.222,
+      "step": 833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019217636022514072,
+      "loss": 2.958,
+      "step": 834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019216697936210133,
+      "loss": 3.2037,
+      "step": 835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019215759849906193,
+      "loss": 2.8794,
+      "step": 836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019214821763602254,
+      "loss": 2.9278,
+      "step": 837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019213883677298313,
+      "loss": 2.9774,
+      "step": 838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019212945590994373,
+      "loss": 3.0807,
+      "step": 839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019212007504690432,
+      "loss": 2.8425,
+      "step": 840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001921106941838649,
+      "loss": 2.6458,
+      "step": 841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00019210131332082553,
+      "loss": 2.923,
+      "step": 842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019209193245778612,
+      "loss": 3.1217,
+      "step": 843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019208255159474673,
+      "loss": 2.9616,
+      "step": 844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019207317073170733,
+      "loss": 2.922,
+      "step": 845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019206378986866794,
+      "loss": 2.6615,
+      "step": 846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019205440900562853,
+      "loss": 3.1939,
+      "step": 847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019204502814258912,
+      "loss": 2.9317,
+      "step": 848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019203564727954972,
+      "loss": 3.2067,
+      "step": 849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019202626641651033,
+      "loss": 3.0489,
+      "step": 850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019201688555347092,
+      "loss": 2.9301,
+      "step": 851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019200750469043154,
+      "loss": 2.797,
+      "step": 852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019199812382739213,
+      "loss": 2.9671,
+      "step": 853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019198874296435272,
+      "loss": 3.0867,
+      "step": 854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019197936210131334,
+      "loss": 3.0829,
+      "step": 855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019196998123827393,
+      "loss": 3.0679,
+      "step": 856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019196060037523452,
+      "loss": 2.731,
+      "step": 857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019195121951219512,
+      "loss": 3.0005,
+      "step": 858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019194183864915573,
+      "loss": 3.2767,
+      "step": 859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019193245778611632,
+      "loss": 3.047,
+      "step": 860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019192307692307694,
+      "loss": 3.0944,
+      "step": 861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019191369606003753,
+      "loss": 2.944,
+      "step": 862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00019190431519699815,
+      "loss": 3.1411,
+      "step": 863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019189493433395874,
+      "loss": 3.1891,
+      "step": 864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019188555347091933,
+      "loss": 2.9038,
+      "step": 865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019187617260787992,
+      "loss": 2.8565,
+      "step": 866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019186679174484052,
+      "loss": 3.0884,
+      "step": 867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019185741088180113,
+      "loss": 2.8723,
+      "step": 868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019184803001876172,
+      "loss": 3.125,
+      "step": 869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019183864915572234,
+      "loss": 3.1713,
+      "step": 870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019182926829268293,
+      "loss": 2.9393,
+      "step": 871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019181988742964355,
+      "loss": 2.7862,
+      "step": 872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019181050656660414,
+      "loss": 3.1157,
+      "step": 873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019180112570356473,
+      "loss": 3.1183,
+      "step": 874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019179174484052532,
+      "loss": 3.0889,
+      "step": 875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019178236397748594,
+      "loss": 2.8773,
+      "step": 876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019177298311444653,
+      "loss": 3.0163,
+      "step": 877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019176360225140715,
+      "loss": 2.9783,
+      "step": 878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019175422138836774,
+      "loss": 3.0055,
+      "step": 879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019174484052532833,
+      "loss": 2.8967,
+      "step": 880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019173545966228895,
+      "loss": 3.0717,
+      "step": 881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019172607879924954,
+      "loss": 2.7889,
+      "step": 882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019171669793621013,
+      "loss": 2.9562,
+      "step": 883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019170731707317072,
+      "loss": 3.2933,
+      "step": 884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019169793621013134,
+      "loss": 2.9097,
+      "step": 885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019168855534709193,
+      "loss": 3.0579,
+      "step": 886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019167917448405255,
+      "loss": 2.9988,
+      "step": 887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019166979362101314,
+      "loss": 3.0666,
+      "step": 888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019166041275797376,
+      "loss": 3.52,
+      "step": 889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019165103189493435,
+      "loss": 2.8938,
+      "step": 890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019164165103189494,
+      "loss": 3.3815,
+      "step": 891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019163227016885553,
+      "loss": 2.9731,
+      "step": 892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019162288930581612,
+      "loss": 2.7938,
+      "step": 893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019161350844277674,
+      "loss": 2.8898,
+      "step": 894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019160412757973733,
+      "loss": 2.812,
+      "step": 895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019159474671669795,
+      "loss": 3.1399,
+      "step": 896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019158536585365854,
+      "loss": 2.8865,
+      "step": 897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019157598499061916,
+      "loss": 2.8734,
+      "step": 898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019156660412757975,
+      "loss": 2.6575,
+      "step": 899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019155722326454034,
+      "loss": 3.0334,
+      "step": 900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019154784240150093,
+      "loss": 2.9951,
+      "step": 901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019153846153846155,
+      "loss": 2.8947,
+      "step": 902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019152908067542214,
+      "loss": 2.7301,
+      "step": 903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019151969981238276,
+      "loss": 2.8416,
+      "step": 904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019151031894934335,
+      "loss": 2.7407,
+      "step": 905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019150093808630394,
+      "loss": 3.077,
+      "step": 906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019149155722326456,
+      "loss": 2.7963,
+      "step": 907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019148217636022515,
+      "loss": 2.6337,
+      "step": 908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019147279549718574,
+      "loss": 2.8098,
+      "step": 909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019146341463414633,
+      "loss": 2.9401,
+      "step": 910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019145403377110695,
+      "loss": 2.8184,
+      "step": 911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019144465290806754,
+      "loss": 2.8331,
+      "step": 912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019143527204502816,
+      "loss": 2.8344,
+      "step": 913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019142589118198875,
+      "loss": 3.3407,
+      "step": 914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019141651031894937,
+      "loss": 3.1281,
+      "step": 915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019140712945590996,
+      "loss": 3.2389,
+      "step": 916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019139774859287055,
+      "loss": 3.2404,
+      "step": 917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019138836772983114,
+      "loss": 2.7799,
+      "step": 918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019137898686679173,
+      "loss": 2.9161,
+      "step": 919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019136960600375235,
+      "loss": 3.0085,
+      "step": 920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019136022514071294,
+      "loss": 3.0726,
+      "step": 921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019135084427767356,
+      "loss": 2.8624,
+      "step": 922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019134146341463415,
+      "loss": 2.8307,
+      "step": 923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019133208255159477,
+      "loss": 2.7275,
+      "step": 924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019132270168855536,
+      "loss": 2.7972,
+      "step": 925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019131332082551595,
+      "loss": 3.0677,
+      "step": 926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00019130393996247654,
+      "loss": 2.977,
+      "step": 927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019129455909943716,
+      "loss": 2.893,
+      "step": 928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019128517823639775,
+      "loss": 3.0839,
+      "step": 929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019127579737335837,
+      "loss": 2.8416,
+      "step": 930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019126641651031896,
+      "loss": 2.8005,
+      "step": 931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019125703564727958,
+      "loss": 3.0253,
+      "step": 932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019124765478424017,
+      "loss": 2.9855,
+      "step": 933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019123827392120076,
+      "loss": 2.8752,
+      "step": 934
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019122889305816135,
+      "loss": 2.7181,
+      "step": 935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019121951219512194,
+      "loss": 2.9488,
+      "step": 936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019121013133208256,
+      "loss": 2.9229,
+      "step": 937
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019120075046904315,
+      "loss": 2.9734,
+      "step": 938
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019119136960600377,
+      "loss": 2.9686,
+      "step": 939
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019118198874296436,
+      "loss": 2.9439,
+      "step": 940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019117260787992498,
+      "loss": 2.9639,
+      "step": 941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019116322701688557,
+      "loss": 3.0734,
+      "step": 942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019115384615384616,
+      "loss": 3.2403,
+      "step": 943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019114446529080675,
+      "loss": 3.0726,
+      "step": 944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019113508442776737,
+      "loss": 3.103,
+      "step": 945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019112570356472796,
+      "loss": 2.8717,
+      "step": 946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019111632270168855,
+      "loss": 2.6882,
+      "step": 947
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00019110694183864917,
+      "loss": 2.9938,
+      "step": 948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019109756097560976,
+      "loss": 2.7618,
+      "step": 949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019108818011257038,
+      "loss": 3.1384,
+      "step": 950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019107879924953097,
+      "loss": 3.1859,
+      "step": 951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019106941838649156,
+      "loss": 2.9939,
+      "step": 952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019106003752345215,
+      "loss": 2.984,
+      "step": 953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019105065666041277,
+      "loss": 2.8694,
+      "step": 954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019104127579737336,
+      "loss": 3.1533,
+      "step": 955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019103189493433398,
+      "loss": 2.8467,
+      "step": 956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019102251407129457,
+      "loss": 2.5387,
+      "step": 957
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019101313320825519,
+      "loss": 3.1522,
+      "step": 958
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019100375234521578,
+      "loss": 3.0106,
+      "step": 959
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019099437148217637,
+      "loss": 3.0664,
+      "step": 960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019098499061913696,
+      "loss": 3.2813,
+      "step": 961
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019097560975609755,
+      "loss": 3.1021,
+      "step": 962
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019096622889305817,
+      "loss": 3.1599,
+      "step": 963
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019095684803001876,
+      "loss": 2.667,
+      "step": 964
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019094746716697938,
+      "loss": 3.0522,
+      "step": 965
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019093808630393997,
+      "loss": 3.0809,
+      "step": 966
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019092870544090059,
+      "loss": 3.1418,
+      "step": 967
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019091932457786118,
+      "loss": 2.7677,
+      "step": 968
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001909099437148218,
+      "loss": 2.9938,
+      "step": 969
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019090056285178236,
+      "loss": 3.1482,
+      "step": 970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019089118198874298,
+      "loss": 3.0108,
+      "step": 971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019088180112570357,
+      "loss": 2.7699,
+      "step": 972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019087242026266416,
+      "loss": 3.0861,
+      "step": 973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019086303939962478,
+      "loss": 2.867,
+      "step": 974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019085365853658537,
+      "loss": 2.7553,
+      "step": 975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019084427767354599,
+      "loss": 2.7359,
+      "step": 976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019083489681050658,
+      "loss": 3.1728,
+      "step": 977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001908255159474672,
+      "loss": 3.0198,
+      "step": 978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019081613508442779,
+      "loss": 3.1183,
+      "step": 979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019080675422138838,
+      "loss": 2.6916,
+      "step": 980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019079737335834897,
+      "loss": 3.2226,
+      "step": 981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019078799249530959,
+      "loss": 3.0847,
+      "step": 982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019077861163227018,
+      "loss": 2.9142,
+      "step": 983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001907692307692308,
+      "loss": 2.748,
+      "step": 984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019075984990619139,
+      "loss": 2.9454,
+      "step": 985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019075046904315198,
+      "loss": 3.0681,
+      "step": 986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001907410881801126,
+      "loss": 3.0545,
+      "step": 987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019073170731707319,
+      "loss": 2.8098,
+      "step": 988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019072232645403378,
+      "loss": 3.0755,
+      "step": 989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019071294559099437,
+      "loss": 3.0104,
+      "step": 990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019070356472795499,
+      "loss": 2.9962,
+      "step": 991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019069418386491558,
+      "loss": 2.7987,
+      "step": 992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001906848030018762,
+      "loss": 2.8979,
+      "step": 993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019067542213883678,
+      "loss": 3.2301,
+      "step": 994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001906660412757974,
+      "loss": 3.2264,
+      "step": 995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000190656660412758,
+      "loss": 2.9909,
+      "step": 996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019064727954971858,
+      "loss": 3.3543,
+      "step": 997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019063789868667918,
+      "loss": 2.7175,
+      "step": 998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019062851782363977,
+      "loss": 3.0089,
+      "step": 999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019061913696060038,
+      "loss": 3.0494,
+      "step": 1000
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 2.797795534133911,
+      "eval_runtime": 67.0815,
+      "eval_samples_per_second": 28.249,
+      "eval_steps_per_second": 3.533,
+      "step": 1000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019060975609756098,
+      "loss": 3.0516,
+      "step": 1001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001906003752345216,
+      "loss": 3.0629,
+      "step": 1002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019059099437148218,
+      "loss": 2.7196,
+      "step": 1003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001905816135084428,
+      "loss": 3.0818,
+      "step": 1004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001905722326454034,
+      "loss": 2.8809,
+      "step": 1005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019056285178236398,
+      "loss": 2.9791,
+      "step": 1006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019055347091932458,
+      "loss": 3.0223,
+      "step": 1007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001905440900562852,
+      "loss": 3.1681,
+      "step": 1008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019053470919324578,
+      "loss": 2.8116,
+      "step": 1009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001905253283302064,
+      "loss": 2.887,
+      "step": 1010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000190515947467167,
+      "loss": 2.8283,
+      "step": 1011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019050656660412758,
+      "loss": 2.8591,
+      "step": 1012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001904971857410882,
+      "loss": 3.1793,
+      "step": 1013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001904878048780488,
+      "loss": 2.9944,
+      "step": 1014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019047842401500938,
+      "loss": 2.914,
+      "step": 1015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019046904315196998,
+      "loss": 3.0477,
+      "step": 1016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001904596622889306,
+      "loss": 2.9382,
+      "step": 1017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019045028142589118,
+      "loss": 3.3881,
+      "step": 1018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001904409005628518,
+      "loss": 3.0613,
+      "step": 1019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001904315196998124,
+      "loss": 2.991,
+      "step": 1020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000190422138836773,
+      "loss": 2.8087,
+      "step": 1021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001904127579737336,
+      "loss": 2.656,
+      "step": 1022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001904033771106942,
+      "loss": 3.0734,
+      "step": 1023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019039399624765478,
+      "loss": 2.9263,
+      "step": 1024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019038461538461538,
+      "loss": 3.054,
+      "step": 1025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000190375234521576,
+      "loss": 2.9687,
+      "step": 1026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019036585365853658,
+      "loss": 3.0115,
+      "step": 1027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001903564727954972,
+      "loss": 2.8092,
+      "step": 1028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001903470919324578,
+      "loss": 2.9145,
+      "step": 1029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001903377110694184,
+      "loss": 2.8169,
+      "step": 1030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000190328330206379,
+      "loss": 3.0537,
+      "step": 1031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001903189493433396,
+      "loss": 2.8126,
+      "step": 1032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00019030956848030018,
+      "loss": 3.141,
+      "step": 1033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001903001876172608,
+      "loss": 2.8645,
+      "step": 1034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001902908067542214,
+      "loss": 3.2169,
+      "step": 1035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000190281425891182,
+      "loss": 2.5111,
+      "step": 1036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001902720450281426,
+      "loss": 2.6918,
+      "step": 1037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001902626641651032,
+      "loss": 2.729,
+      "step": 1038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001902532833020638,
+      "loss": 2.8083,
+      "step": 1039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001902439024390244,
+      "loss": 3.2349,
+      "step": 1040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000190234521575985,
+      "loss": 3.1935,
+      "step": 1041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019022514071294558,
+      "loss": 3.0334,
+      "step": 1042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001902157598499062,
+      "loss": 2.9925,
+      "step": 1043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001902063789868668,
+      "loss": 2.9766,
+      "step": 1044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001901969981238274,
+      "loss": 3.0165,
+      "step": 1045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000190187617260788,
+      "loss": 2.8067,
+      "step": 1046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019017823639774862,
+      "loss": 2.8697,
+      "step": 1047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001901688555347092,
+      "loss": 3.0561,
+      "step": 1048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001901594746716698,
+      "loss": 2.8741,
+      "step": 1049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001901500938086304,
+      "loss": 2.9487,
+      "step": 1050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00019014071294559098,
+      "loss": 3.0221,
+      "step": 1051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001901313320825516,
+      "loss": 3.1204,
+      "step": 1052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001901219512195122,
+      "loss": 3.1525,
+      "step": 1053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001901125703564728,
+      "loss": 2.9717,
+      "step": 1054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001901031894934334,
+      "loss": 2.5908,
+      "step": 1055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019009380863039402,
+      "loss": 3.0736,
+      "step": 1056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001900844277673546,
+      "loss": 2.7042,
+      "step": 1057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001900750469043152,
+      "loss": 2.797,
+      "step": 1058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001900656660412758,
+      "loss": 2.8776,
+      "step": 1059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001900562851782364,
+      "loss": 2.8561,
+      "step": 1060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000190046904315197,
+      "loss": 2.7824,
+      "step": 1061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019003752345215762,
+      "loss": 3.1061,
+      "step": 1062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001900281425891182,
+      "loss": 2.7346,
+      "step": 1063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001900187617260788,
+      "loss": 2.8759,
+      "step": 1064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019000938086303942,
+      "loss": 2.8238,
+      "step": 1065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019,
+      "loss": 3.078,
+      "step": 1066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001899906191369606,
+      "loss": 2.8403,
+      "step": 1067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001899812382739212,
+      "loss": 2.856,
+      "step": 1068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001899718574108818,
+      "loss": 2.8512,
+      "step": 1069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001899624765478424,
+      "loss": 2.8427,
+      "step": 1070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018995309568480302,
+      "loss": 3.1399,
+      "step": 1071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001899437148217636,
+      "loss": 2.8563,
+      "step": 1072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018993433395872423,
+      "loss": 3.1593,
+      "step": 1073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00018992495309568482,
+      "loss": 2.6344,
+      "step": 1074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0001899155722326454,
+      "loss": 2.99,
+      "step": 1075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000189906191369606,
+      "loss": 2.8432,
+      "step": 1076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001898968105065666,
+      "loss": 2.9653,
+      "step": 1077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001898874296435272,
+      "loss": 2.9792,
+      "step": 1078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001898780487804878,
+      "loss": 3.0286,
+      "step": 1079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018986866791744842,
+      "loss": 2.9487,
+      "step": 1080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000189859287054409,
+      "loss": 2.9279,
+      "step": 1081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018984990619136963,
+      "loss": 2.7981,
+      "step": 1082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018984052532833022,
+      "loss": 2.7137,
+      "step": 1083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001898311444652908,
+      "loss": 2.8545,
+      "step": 1084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001898217636022514,
+      "loss": 2.779,
+      "step": 1085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018981238273921202,
+      "loss": 2.7369,
+      "step": 1086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001898030018761726,
+      "loss": 2.8675,
+      "step": 1087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018979362101313323,
+      "loss": 2.9355,
+      "step": 1088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018978424015009382,
+      "loss": 2.8985,
+      "step": 1089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001897748592870544,
+      "loss": 2.7334,
+      "step": 1090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018976547842401503,
+      "loss": 3.0879,
+      "step": 1091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018975609756097562,
+      "loss": 2.9334,
+      "step": 1092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001897467166979362,
+      "loss": 2.7232,
+      "step": 1093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001897373358348968,
+      "loss": 2.4966,
+      "step": 1094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018972795497185742,
+      "loss": 3.0184,
+      "step": 1095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000189718574108818,
+      "loss": 3.0692,
+      "step": 1096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00018970919324577863,
+      "loss": 2.8241,
+      "step": 1097
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018969981238273922,
+      "loss": 2.8678,
+      "step": 1098
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018969043151969984,
+      "loss": 2.7833,
+      "step": 1099
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018968105065666043,
+      "loss": 2.7936,
+      "step": 1100
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018967166979362102,
+      "loss": 2.843,
+      "step": 1101
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001896622889305816,
+      "loss": 3.0425,
+      "step": 1102
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001896529080675422,
+      "loss": 2.6507,
+      "step": 1103
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018964352720450282,
+      "loss": 2.9896,
+      "step": 1104
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001896341463414634,
+      "loss": 2.9657,
+      "step": 1105
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018962476547842403,
+      "loss": 2.963,
+      "step": 1106
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018961538461538462,
+      "loss": 2.9941,
+      "step": 1107
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018960600375234524,
+      "loss": 2.9096,
+      "step": 1108
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018959662288930583,
+      "loss": 2.5706,
+      "step": 1109
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018958724202626642,
+      "loss": 2.683,
+      "step": 1110
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000189577861163227,
+      "loss": 2.662,
+      "step": 1111
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018956848030018763,
+      "loss": 2.797,
+      "step": 1112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018955909943714822,
+      "loss": 2.9618,
+      "step": 1113
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018954971857410884,
+      "loss": 3.1193,
+      "step": 1114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018954033771106943,
+      "loss": 3.0313,
+      "step": 1115
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018953095684803002,
+      "loss": 3.1727,
+      "step": 1116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018952157598499064,
+      "loss": 3.0579,
+      "step": 1117
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018951219512195123,
+      "loss": 2.9388,
+      "step": 1118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00018950281425891182,
+      "loss": 3.2291,
+      "step": 1119
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0001894934333958724,
+      "loss": 3.0498,
+      "step": 1120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018948405253283303,
+      "loss": 2.9807,
+      "step": 1121
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018947467166979362,
+      "loss": 2.7563,
+      "step": 1122
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018946529080675424,
+      "loss": 3.1314,
+      "step": 1123
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018945590994371483,
+      "loss": 3.4262,
+      "step": 1124
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018944652908067545,
+      "loss": 2.9919,
+      "step": 1125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018943714821763604,
+      "loss": 3.062,
+      "step": 1126
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018942776735459665,
+      "loss": 2.8518,
+      "step": 1127
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018941838649155722,
+      "loss": 3.0331,
+      "step": 1128
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018940900562851784,
+      "loss": 2.8717,
+      "step": 1129
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018939962476547843,
+      "loss": 2.7795,
+      "step": 1130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018939024390243902,
+      "loss": 3.1104,
+      "step": 1131
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018938086303939964,
+      "loss": 3.1577,
+      "step": 1132
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018937148217636023,
+      "loss": 3.2939,
+      "step": 1133
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018936210131332085,
+      "loss": 3.0018,
+      "step": 1134
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018935272045028144,
+      "loss": 3.0387,
+      "step": 1135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018934333958724205,
+      "loss": 2.9844,
+      "step": 1136
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018933395872420262,
+      "loss": 3.1031,
+      "step": 1137
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018932457786116324,
+      "loss": 3.0584,
+      "step": 1138
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018931519699812383,
+      "loss": 3.1007,
+      "step": 1139
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00018930581613508444,
+      "loss": 2.8978,
+      "step": 1140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018929643527204504,
+      "loss": 3.0263,
+      "step": 1141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018928705440900565,
+      "loss": 2.9969,
+      "step": 1142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018927767354596624,
+      "loss": 2.8073,
+      "step": 1143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018926829268292684,
+      "loss": 2.9601,
+      "step": 1144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018925891181988745,
+      "loss": 2.9874,
+      "step": 1145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018924953095684802,
+      "loss": 2.6502,
+      "step": 1146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018924015009380864,
+      "loss": 2.8289,
+      "step": 1147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018923076923076923,
+      "loss": 3.147,
+      "step": 1148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018922138836772984,
+      "loss": 2.7786,
+      "step": 1149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018921200750469044,
+      "loss": 2.7922,
+      "step": 1150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018920262664165105,
+      "loss": 2.9701,
+      "step": 1151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018919324577861164,
+      "loss": 2.815,
+      "step": 1152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018918386491557226,
+      "loss": 3.0574,
+      "step": 1153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018917448405253285,
+      "loss": 3.0454,
+      "step": 1154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018916510318949344,
+      "loss": 2.7891,
+      "step": 1155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018915572232645404,
+      "loss": 2.946,
+      "step": 1156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018914634146341463,
+      "loss": 3.3582,
+      "step": 1157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018913696060037524,
+      "loss": 3.1874,
+      "step": 1158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018912757973733584,
+      "loss": 3.4542,
+      "step": 1159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018911819887429645,
+      "loss": 3.1993,
+      "step": 1160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00018910881801125704,
+      "loss": 3.1083,
+      "step": 1161
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018909943714821766,
+      "loss": 3.2561,
+      "step": 1162
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018909005628517825,
+      "loss": 2.9495,
+      "step": 1163
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018908067542213884,
+      "loss": 3.0246,
+      "step": 1164
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018907129455909944,
+      "loss": 3.1864,
+      "step": 1165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018906191369606005,
+      "loss": 2.8162,
+      "step": 1166
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018905253283302064,
+      "loss": 2.8399,
+      "step": 1167
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018904315196998126,
+      "loss": 2.7636,
+      "step": 1168
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018903377110694185,
+      "loss": 2.9087,
+      "step": 1169
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018902439024390244,
+      "loss": 2.7955,
+      "step": 1170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018901500938086306,
+      "loss": 3.1134,
+      "step": 1171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018900562851782365,
+      "loss": 2.8721,
+      "step": 1172
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018899624765478424,
+      "loss": 2.8918,
+      "step": 1173
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018898686679174484,
+      "loss": 3.0079,
+      "step": 1174
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018897748592870545,
+      "loss": 2.5914,
+      "step": 1175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018896810506566604,
+      "loss": 2.8461,
+      "step": 1176
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018895872420262666,
+      "loss": 2.9849,
+      "step": 1177
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018894934333958725,
+      "loss": 2.8994,
+      "step": 1178
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018893996247654787,
+      "loss": 3.0483,
+      "step": 1179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018893058161350846,
+      "loss": 2.8253,
+      "step": 1180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018892120075046905,
+      "loss": 3.0557,
+      "step": 1181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018891181988742964,
+      "loss": 2.763,
+      "step": 1182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00018890243902439023,
+      "loss": 2.5843,
+      "step": 1183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018889305816135085,
+      "loss": 2.7823,
+      "step": 1184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018888367729831144,
+      "loss": 2.9862,
+      "step": 1185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018887429643527206,
+      "loss": 3.1134,
+      "step": 1186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018886491557223265,
+      "loss": 3.038,
+      "step": 1187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018885553470919327,
+      "loss": 2.8177,
+      "step": 1188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018884615384615386,
+      "loss": 2.8838,
+      "step": 1189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018883677298311445,
+      "loss": 2.7877,
+      "step": 1190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018882739212007504,
+      "loss": 3.0919,
+      "step": 1191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018881801125703566,
+      "loss": 2.8563,
+      "step": 1192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018880863039399625,
+      "loss": 2.9013,
+      "step": 1193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018879924953095687,
+      "loss": 2.6097,
+      "step": 1194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018878986866791746,
+      "loss": 2.9709,
+      "step": 1195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018878048780487805,
+      "loss": 2.7985,
+      "step": 1196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018877110694183867,
+      "loss": 2.7513,
+      "step": 1197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018876172607879926,
+      "loss": 2.8793,
+      "step": 1198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018875234521575985,
+      "loss": 2.7973,
+      "step": 1199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018874296435272044,
+      "loss": 3.0095,
+      "step": 1200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018873358348968106,
+      "loss": 2.7649,
+      "step": 1201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018872420262664165,
+      "loss": 2.8767,
+      "step": 1202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018871482176360227,
+      "loss": 3.0734,
+      "step": 1203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00018870544090056286,
+      "loss": 2.7219,
+      "step": 1204
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018869606003752348,
+      "loss": 2.9628,
+      "step": 1205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018868667917448407,
+      "loss": 3.097,
+      "step": 1206
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018867729831144466,
+      "loss": 2.9445,
+      "step": 1207
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018866791744840525,
+      "loss": 3.3434,
+      "step": 1208
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018865853658536584,
+      "loss": 2.947,
+      "step": 1209
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018864915572232646,
+      "loss": 3.2183,
+      "step": 1210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018863977485928705,
+      "loss": 2.6945,
+      "step": 1211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018863039399624767,
+      "loss": 2.818,
+      "step": 1212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018862101313320826,
+      "loss": 2.7418,
+      "step": 1213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018861163227016888,
+      "loss": 2.8909,
+      "step": 1214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018860225140712947,
+      "loss": 2.8927,
+      "step": 1215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018859287054409006,
+      "loss": 3.1397,
+      "step": 1216
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018858348968105065,
+      "loss": 2.8839,
+      "step": 1217
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018857410881801127,
+      "loss": 3.2013,
+      "step": 1218
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018856472795497186,
+      "loss": 2.8895,
+      "step": 1219
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018855534709193248,
+      "loss": 2.8568,
+      "step": 1220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018854596622889307,
+      "loss": 3.0233,
+      "step": 1221
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018853658536585366,
+      "loss": 2.9315,
+      "step": 1222
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018852720450281428,
+      "loss": 2.7675,
+      "step": 1223
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018851782363977487,
+      "loss": 2.865,
+      "step": 1224
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00018850844277673546,
+      "loss": 2.7999,
+      "step": 1225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018849906191369605,
+      "loss": 2.6047,
+      "step": 1226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018848968105065667,
+      "loss": 2.9389,
+      "step": 1227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018848030018761726,
+      "loss": 2.9934,
+      "step": 1228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018847091932457788,
+      "loss": 3.0862,
+      "step": 1229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018846153846153847,
+      "loss": 3.1609,
+      "step": 1230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001884521575984991,
+      "loss": 3.0754,
+      "step": 1231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018844277673545968,
+      "loss": 2.806,
+      "step": 1232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018843339587242027,
+      "loss": 3.1886,
+      "step": 1233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018842401500938086,
+      "loss": 2.9813,
+      "step": 1234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018841463414634145,
+      "loss": 2.9387,
+      "step": 1235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018840525328330207,
+      "loss": 2.624,
+      "step": 1236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018839587242026266,
+      "loss": 2.9747,
+      "step": 1237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018838649155722328,
+      "loss": 2.943,
+      "step": 1238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018837711069418387,
+      "loss": 2.8371,
+      "step": 1239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001883677298311445,
+      "loss": 2.9701,
+      "step": 1240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018835834896810508,
+      "loss": 2.9511,
+      "step": 1241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018834896810506567,
+      "loss": 2.7958,
+      "step": 1242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018833958724202626,
+      "loss": 3.0597,
+      "step": 1243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018833020637898688,
+      "loss": 2.7018,
+      "step": 1244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018832082551594747,
+      "loss": 2.9355,
+      "step": 1245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0001883114446529081,
+      "loss": 3.0433,
+      "step": 1246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00018830206378986868,
+      "loss": 2.7486,
+      "step": 1247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018829268292682927,
+      "loss": 2.9636,
+      "step": 1248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001882833020637899,
+      "loss": 3.2746,
+      "step": 1249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018827392120075048,
+      "loss": 2.9227,
+      "step": 1250
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018826454033771107,
+      "loss": 2.9261,
+      "step": 1251
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018825515947467166,
+      "loss": 2.7238,
+      "step": 1252
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018824577861163228,
+      "loss": 2.9302,
+      "step": 1253
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018823639774859287,
+      "loss": 3.0401,
+      "step": 1254
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001882270168855535,
+      "loss": 2.8064,
+      "step": 1255
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018821763602251408,
+      "loss": 3.0621,
+      "step": 1256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001882082551594747,
+      "loss": 3.0279,
+      "step": 1257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001881988742964353,
+      "loss": 3.2441,
+      "step": 1258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018818949343339588,
+      "loss": 3.2302,
+      "step": 1259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018818011257035647,
+      "loss": 3.1026,
+      "step": 1260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018817073170731706,
+      "loss": 3.0969,
+      "step": 1261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018816135084427768,
+      "loss": 3.1322,
+      "step": 1262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018815196998123827,
+      "loss": 3.2997,
+      "step": 1263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001881425891181989,
+      "loss": 2.8704,
+      "step": 1264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018813320825515948,
+      "loss": 2.9565,
+      "step": 1265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001881238273921201,
+      "loss": 2.6934,
+      "step": 1266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0001881144465290807,
+      "loss": 2.8841,
+      "step": 1267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00018810506566604128,
+      "loss": 2.9493,
+      "step": 1268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018809568480300187,
+      "loss": 3.1974,
+      "step": 1269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001880863039399625,
+      "loss": 2.9102,
+      "step": 1270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018807692307692308,
+      "loss": 2.8595,
+      "step": 1271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001880675422138837,
+      "loss": 2.8671,
+      "step": 1272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001880581613508443,
+      "loss": 2.8771,
+      "step": 1273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018804878048780488,
+      "loss": 2.7144,
+      "step": 1274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001880393996247655,
+      "loss": 2.752,
+      "step": 1275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001880300187617261,
+      "loss": 3.0382,
+      "step": 1276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018802063789868668,
+      "loss": 2.7202,
+      "step": 1277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018801125703564727,
+      "loss": 2.9063,
+      "step": 1278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001880018761726079,
+      "loss": 2.9604,
+      "step": 1279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018799249530956848,
+      "loss": 2.9846,
+      "step": 1280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001879831144465291,
+      "loss": 2.9984,
+      "step": 1281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001879737335834897,
+      "loss": 2.8147,
+      "step": 1282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001879643527204503,
+      "loss": 2.7045,
+      "step": 1283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001879549718574109,
+      "loss": 3.1211,
+      "step": 1284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018794559099437151,
+      "loss": 3.0123,
+      "step": 1285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018793621013133208,
+      "loss": 3.0403,
+      "step": 1286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018792682926829267,
+      "loss": 2.7382,
+      "step": 1287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0001879174484052533,
+      "loss": 2.7656,
+      "step": 1288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00018790806754221388,
+      "loss": 2.6217,
+      "step": 1289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001878986866791745,
+      "loss": 2.7959,
+      "step": 1290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001878893058161351,
+      "loss": 2.7583,
+      "step": 1291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001878799249530957,
+      "loss": 3.1735,
+      "step": 1292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001878705440900563,
+      "loss": 2.9687,
+      "step": 1293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018786116322701691,
+      "loss": 2.9381,
+      "step": 1294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018785178236397748,
+      "loss": 3.2917,
+      "step": 1295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001878424015009381,
+      "loss": 3.2729,
+      "step": 1296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001878330206378987,
+      "loss": 3.1061,
+      "step": 1297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001878236397748593,
+      "loss": 3.1827,
+      "step": 1298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001878142589118199,
+      "loss": 3.0686,
+      "step": 1299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001878048780487805,
+      "loss": 2.8846,
+      "step": 1300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001877954971857411,
+      "loss": 3.1197,
+      "step": 1301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001877861163227017,
+      "loss": 2.7983,
+      "step": 1302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018777673545966231,
+      "loss": 2.9215,
+      "step": 1303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018776735459662288,
+      "loss": 2.8533,
+      "step": 1304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001877579737335835,
+      "loss": 2.8784,
+      "step": 1305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018774859287054409,
+      "loss": 2.9417,
+      "step": 1306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001877392120075047,
+      "loss": 2.9764,
+      "step": 1307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001877298311444653,
+      "loss": 2.9166,
+      "step": 1308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001877204502814259,
+      "loss": 2.6936,
+      "step": 1309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0001877110694183865,
+      "loss": 2.7701,
+      "step": 1310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00018770168855534712,
+      "loss": 2.7917,
+      "step": 1311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001876923076923077,
+      "loss": 2.8857,
+      "step": 1312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018768292682926828,
+      "loss": 3.3227,
+      "step": 1313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001876735459662289,
+      "loss": 3.1256,
+      "step": 1314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018766416510318949,
+      "loss": 3.0305,
+      "step": 1315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001876547842401501,
+      "loss": 2.8719,
+      "step": 1316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001876454033771107,
+      "loss": 3.1429,
+      "step": 1317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001876360225140713,
+      "loss": 2.7836,
+      "step": 1318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001876266416510319,
+      "loss": 3.2413,
+      "step": 1319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018761726078799252,
+      "loss": 2.9082,
+      "step": 1320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001876078799249531,
+      "loss": 3.0704,
+      "step": 1321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001875984990619137,
+      "loss": 2.8799,
+      "step": 1322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001875891181988743,
+      "loss": 2.877,
+      "step": 1323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001875797373358349,
+      "loss": 2.833,
+      "step": 1324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001875703564727955,
+      "loss": 2.6646,
+      "step": 1325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001875609756097561,
+      "loss": 2.6501,
+      "step": 1326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001875515947467167,
+      "loss": 2.8265,
+      "step": 1327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001875422138836773,
+      "loss": 2.7275,
+      "step": 1328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00018753283302063792,
+      "loss": 2.8068,
+      "step": 1329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001875234521575985,
+      "loss": 2.8233,
+      "step": 1330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001875140712945591,
+      "loss": 2.5086,
+      "step": 1331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0001875046904315197,
+      "loss": 2.7185,
+      "step": 1332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001874953095684803,
+      "loss": 3.0033,
+      "step": 1333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001874859287054409,
+      "loss": 3.0376,
+      "step": 1334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018747654784240152,
+      "loss": 2.9012,
+      "step": 1335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001874671669793621,
+      "loss": 2.8212,
+      "step": 1336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018745778611632273,
+      "loss": 2.9189,
+      "step": 1337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018744840525328332,
+      "loss": 2.7766,
+      "step": 1338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001874390243902439,
+      "loss": 3.0552,
+      "step": 1339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001874296435272045,
+      "loss": 2.7469,
+      "step": 1340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001874202626641651,
+      "loss": 2.5709,
+      "step": 1341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001874108818011257,
+      "loss": 3.2473,
+      "step": 1342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001874015009380863,
+      "loss": 2.7506,
+      "step": 1343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018739212007504692,
+      "loss": 2.951,
+      "step": 1344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001873827392120075,
+      "loss": 2.9115,
+      "step": 1345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018737335834896813,
+      "loss": 3.0037,
+      "step": 1346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018736397748592872,
+      "loss": 2.8785,
+      "step": 1347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001873545966228893,
+      "loss": 2.7113,
+      "step": 1348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001873452157598499,
+      "loss": 3.0196,
+      "step": 1349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018733583489681052,
+      "loss": 2.8288,
+      "step": 1350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0001873264540337711,
+      "loss": 2.6476,
+      "step": 1351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018731707317073173,
+      "loss": 3.0263,
+      "step": 1352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00018730769230769232,
+      "loss": 2.6766,
+      "step": 1353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001872983114446529,
+      "loss": 3.0795,
+      "step": 1354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018728893058161353,
+      "loss": 2.7218,
+      "step": 1355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018727954971857412,
+      "loss": 2.8281,
+      "step": 1356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001872701688555347,
+      "loss": 2.9299,
+      "step": 1357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001872607879924953,
+      "loss": 2.9559,
+      "step": 1358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018725140712945592,
+      "loss": 3.1992,
+      "step": 1359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001872420262664165,
+      "loss": 2.8449,
+      "step": 1360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018723264540337713,
+      "loss": 3.1997,
+      "step": 1361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018722326454033772,
+      "loss": 3.1366,
+      "step": 1362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018721388367729834,
+      "loss": 3.0872,
+      "step": 1363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018720450281425893,
+      "loss": 2.9053,
+      "step": 1364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018719512195121952,
+      "loss": 3.0984,
+      "step": 1365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001871857410881801,
+      "loss": 2.7316,
+      "step": 1366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001871763602251407,
+      "loss": 2.9641,
+      "step": 1367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018716697936210132,
+      "loss": 3.1006,
+      "step": 1368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001871575984990619,
+      "loss": 2.7485,
+      "step": 1369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018714821763602253,
+      "loss": 3.0745,
+      "step": 1370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018713883677298312,
+      "loss": 2.9016,
+      "step": 1371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018712945590994374,
+      "loss": 2.9066,
+      "step": 1372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018712007504690433,
+      "loss": 2.9887,
+      "step": 1373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00018711069418386492,
+      "loss": 2.8577,
+      "step": 1374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0001871013133208255,
+      "loss": 2.8506,
+      "step": 1375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018709193245778613,
+      "loss": 2.7382,
+      "step": 1376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018708255159474672,
+      "loss": 2.8665,
+      "step": 1377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018707317073170734,
+      "loss": 2.7513,
+      "step": 1378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018706378986866793,
+      "loss": 2.9802,
+      "step": 1379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018705440900562852,
+      "loss": 2.6868,
+      "step": 1380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018704502814258914,
+      "loss": 2.4748,
+      "step": 1381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018703564727954973,
+      "loss": 2.8546,
+      "step": 1382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018702626641651032,
+      "loss": 2.9691,
+      "step": 1383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001870168855534709,
+      "loss": 3.3587,
+      "step": 1384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018700750469043153,
+      "loss": 3.1344,
+      "step": 1385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018699812382739212,
+      "loss": 3.0243,
+      "step": 1386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018698874296435274,
+      "loss": 2.8926,
+      "step": 1387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018697936210131333,
+      "loss": 2.9979,
+      "step": 1388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018696998123827395,
+      "loss": 3.033,
+      "step": 1389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018696060037523454,
+      "loss": 2.727,
+      "step": 1390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018695121951219513,
+      "loss": 3.195,
+      "step": 1391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018694183864915572,
+      "loss": 2.6585,
+      "step": 1392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001869324577861163,
+      "loss": 3.0783,
+      "step": 1393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018692307692307693,
+      "loss": 2.8382,
+      "step": 1394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018691369606003752,
+      "loss": 2.9467,
+      "step": 1395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00018690431519699814,
+      "loss": 2.9077,
+      "step": 1396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018689493433395873,
+      "loss": 3.1421,
+      "step": 1397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018688555347091935,
+      "loss": 2.7004,
+      "step": 1398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018687617260787994,
+      "loss": 2.7572,
+      "step": 1399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018686679174484053,
+      "loss": 2.7348,
+      "step": 1400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018685741088180112,
+      "loss": 2.7845,
+      "step": 1401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018684803001876174,
+      "loss": 2.6842,
+      "step": 1402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018683864915572233,
+      "loss": 3.2182,
+      "step": 1403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018682926829268295,
+      "loss": 2.6571,
+      "step": 1404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018681988742964354,
+      "loss": 3.1415,
+      "step": 1405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018681050656660413,
+      "loss": 2.6553,
+      "step": 1406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018680112570356475,
+      "loss": 2.8962,
+      "step": 1407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018679174484052534,
+      "loss": 2.7893,
+      "step": 1408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018678236397748593,
+      "loss": 2.728,
+      "step": 1409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018677298311444652,
+      "loss": 2.7376,
+      "step": 1410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018676360225140714,
+      "loss": 2.7359,
+      "step": 1411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018675422138836773,
+      "loss": 2.9868,
+      "step": 1412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018674484052532835,
+      "loss": 2.9534,
+      "step": 1413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018673545966228894,
+      "loss": 2.9655,
+      "step": 1414
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018672607879924956,
+      "loss": 3.0693,
+      "step": 1415
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018671669793621015,
+      "loss": 2.9329,
+      "step": 1416
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00018670731707317074,
+      "loss": 2.9242,
+      "step": 1417
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018669793621013133,
+      "loss": 2.8209,
+      "step": 1418
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018668855534709192,
+      "loss": 3.1824,
+      "step": 1419
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018667917448405254,
+      "loss": 2.8627,
+      "step": 1420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018666979362101313,
+      "loss": 2.7314,
+      "step": 1421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018666041275797375,
+      "loss": 2.795,
+      "step": 1422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018665103189493434,
+      "loss": 2.9818,
+      "step": 1423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018664165103189496,
+      "loss": 2.927,
+      "step": 1424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018663227016885555,
+      "loss": 2.8601,
+      "step": 1425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018662288930581614,
+      "loss": 2.4888,
+      "step": 1426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018661350844277673,
+      "loss": 2.7629,
+      "step": 1427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018660412757973735,
+      "loss": 2.6426,
+      "step": 1428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018659474671669794,
+      "loss": 3.197,
+      "step": 1429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018658536585365856,
+      "loss": 2.9239,
+      "step": 1430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018657598499061915,
+      "loss": 3.0096,
+      "step": 1431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018656660412757974,
+      "loss": 2.8395,
+      "step": 1432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018655722326454036,
+      "loss": 2.8805,
+      "step": 1433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018654784240150095,
+      "loss": 2.7824,
+      "step": 1434
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018653846153846154,
+      "loss": 3.2347,
+      "step": 1435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018652908067542213,
+      "loss": 2.698,
+      "step": 1436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018651969981238275,
+      "loss": 2.9045,
+      "step": 1437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018651031894934334,
+      "loss": 2.7172,
+      "step": 1438
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00018650093808630396,
+      "loss": 2.6245,
+      "step": 1439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018649155722326455,
+      "loss": 2.8719,
+      "step": 1440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018648217636022516,
+      "loss": 3.0515,
+      "step": 1441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018647279549718576,
+      "loss": 2.6147,
+      "step": 1442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018646341463414637,
+      "loss": 2.7264,
+      "step": 1443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018645403377110694,
+      "loss": 2.6915,
+      "step": 1444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018644465290806753,
+      "loss": 3.0793,
+      "step": 1445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018643527204502815,
+      "loss": 3.176,
+      "step": 1446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018642589118198874,
+      "loss": 2.8601,
+      "step": 1447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018641651031894936,
+      "loss": 3.0713,
+      "step": 1448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018640712945590995,
+      "loss": 2.7661,
+      "step": 1449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018639774859287056,
+      "loss": 2.8459,
+      "step": 1450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018638836772983116,
+      "loss": 2.5575,
+      "step": 1451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018637898686679177,
+      "loss": 2.5517,
+      "step": 1452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018636960600375234,
+      "loss": 2.7987,
+      "step": 1453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018636022514071296,
+      "loss": 2.6433,
+      "step": 1454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018635084427767355,
+      "loss": 2.5005,
+      "step": 1455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018634146341463416,
+      "loss": 2.9089,
+      "step": 1456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018633208255159476,
+      "loss": 2.9833,
+      "step": 1457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018632270168855535,
+      "loss": 2.8197,
+      "step": 1458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018631332082551596,
+      "loss": 2.8977,
+      "step": 1459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00018630393996247656,
+      "loss": 2.7701,
+      "step": 1460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018629455909943717,
+      "loss": 2.8706,
+      "step": 1461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018628517823639774,
+      "loss": 2.8456,
+      "step": 1462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018627579737335836,
+      "loss": 2.8821,
+      "step": 1463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018626641651031895,
+      "loss": 2.8787,
+      "step": 1464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018625703564727956,
+      "loss": 2.7068,
+      "step": 1465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018624765478424016,
+      "loss": 2.9281,
+      "step": 1466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018623827392120077,
+      "loss": 2.8469,
+      "step": 1467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018622889305816136,
+      "loss": 3.1916,
+      "step": 1468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018621951219512198,
+      "loss": 2.8798,
+      "step": 1469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018621013133208257,
+      "loss": 3.1782,
+      "step": 1470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018620075046904314,
+      "loss": 2.771,
+      "step": 1471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018619136960600376,
+      "loss": 3.1787,
+      "step": 1472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018618198874296435,
+      "loss": 2.897,
+      "step": 1473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018617260787992496,
+      "loss": 2.5676,
+      "step": 1474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018616322701688555,
+      "loss": 2.9853,
+      "step": 1475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018615384615384617,
+      "loss": 2.6582,
+      "step": 1476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018614446529080676,
+      "loss": 3.0366,
+      "step": 1477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018613508442776738,
+      "loss": 2.8494,
+      "step": 1478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018612570356472797,
+      "loss": 3.2195,
+      "step": 1479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018611632270168856,
+      "loss": 3.2296,
+      "step": 1480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00018610694183864915,
+      "loss": 3.0079,
+      "step": 1481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018609756097560977,
+      "loss": 3.0518,
+      "step": 1482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018608818011257036,
+      "loss": 2.9177,
+      "step": 1483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018607879924953095,
+      "loss": 3.0765,
+      "step": 1484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018606941838649157,
+      "loss": 2.6234,
+      "step": 1485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018606003752345216,
+      "loss": 2.7151,
+      "step": 1486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018605065666041278,
+      "loss": 2.7752,
+      "step": 1487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018604127579737337,
+      "loss": 2.7981,
+      "step": 1488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018603189493433396,
+      "loss": 2.8021,
+      "step": 1489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018602251407129455,
+      "loss": 2.773,
+      "step": 1490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018601313320825517,
+      "loss": 2.9147,
+      "step": 1491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018600375234521576,
+      "loss": 2.7007,
+      "step": 1492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018599437148217638,
+      "loss": 2.6256,
+      "step": 1493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018598499061913697,
+      "loss": 2.953,
+      "step": 1494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0001859756097560976,
+      "loss": 2.8985,
+      "step": 1495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018596622889305818,
+      "loss": 3.1013,
+      "step": 1496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018595684803001877,
+      "loss": 2.9821,
+      "step": 1497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018594746716697936,
+      "loss": 2.6058,
+      "step": 1498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018593808630393995,
+      "loss": 2.9501,
+      "step": 1499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018592870544090057,
+      "loss": 2.8685,
+      "step": 1500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018591932457786116,
+      "loss": 2.727,
+      "step": 1501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018590994371482178,
+      "loss": 2.8728,
+      "step": 1502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00018590056285178237,
+      "loss": 2.8457,
+      "step": 1503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000185891181988743,
+      "loss": 2.8086,
+      "step": 1504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018588180112570358,
+      "loss": 2.9864,
+      "step": 1505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018587242026266417,
+      "loss": 2.9177,
+      "step": 1506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018586303939962476,
+      "loss": 2.9362,
+      "step": 1507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018585365853658538,
+      "loss": 2.9298,
+      "step": 1508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018584427767354597,
+      "loss": 3.0195,
+      "step": 1509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018583489681050656,
+      "loss": 2.6902,
+      "step": 1510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018582551594746718,
+      "loss": 2.9572,
+      "step": 1511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018581613508442777,
+      "loss": 2.5902,
+      "step": 1512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001858067542213884,
+      "loss": 2.841,
+      "step": 1513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018579737335834898,
+      "loss": 3.1276,
+      "step": 1514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018578799249530957,
+      "loss": 2.8303,
+      "step": 1515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018577861163227016,
+      "loss": 2.9499,
+      "step": 1516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018576923076923078,
+      "loss": 2.9374,
+      "step": 1517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018575984990619137,
+      "loss": 2.9618,
+      "step": 1518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000185750469043152,
+      "loss": 2.9831,
+      "step": 1519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018574108818011258,
+      "loss": 2.8791,
+      "step": 1520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001857317073170732,
+      "loss": 3.151,
+      "step": 1521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001857223264540338,
+      "loss": 3.3349,
+      "step": 1522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018571294559099438,
+      "loss": 2.8303,
+      "step": 1523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00018570356472795497,
+      "loss": 2.9659,
+      "step": 1524
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018569418386491556,
+      "loss": 2.8096,
+      "step": 1525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018568480300187618,
+      "loss": 2.7828,
+      "step": 1526
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018567542213883677,
+      "loss": 2.67,
+      "step": 1527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001856660412757974,
+      "loss": 3.1676,
+      "step": 1528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018565666041275798,
+      "loss": 2.7553,
+      "step": 1529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001856472795497186,
+      "loss": 3.0531,
+      "step": 1530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001856378986866792,
+      "loss": 2.7057,
+      "step": 1531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018562851782363978,
+      "loss": 2.6544,
+      "step": 1532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018561913696060037,
+      "loss": 2.9622,
+      "step": 1533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000185609756097561,
+      "loss": 2.5786,
+      "step": 1534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018560037523452158,
+      "loss": 2.9004,
+      "step": 1535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018559099437148217,
+      "loss": 2.9145,
+      "step": 1536
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001855816135084428,
+      "loss": 3.2831,
+      "step": 1537
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018557223264540338,
+      "loss": 3.0169,
+      "step": 1538
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000185562851782364,
+      "loss": 3.0812,
+      "step": 1539
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001855534709193246,
+      "loss": 2.6127,
+      "step": 1540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018554409005628518,
+      "loss": 2.8412,
+      "step": 1541
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018553470919324577,
+      "loss": 2.6746,
+      "step": 1542
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001855253283302064,
+      "loss": 2.7347,
+      "step": 1543
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018551594746716698,
+      "loss": 2.9054,
+      "step": 1544
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001855065666041276,
+      "loss": 3.0103,
+      "step": 1545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001854971857410882,
+      "loss": 2.7654,
+      "step": 1546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001854878048780488,
+      "loss": 2.9644,
+      "step": 1547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001854784240150094,
+      "loss": 2.8823,
+      "step": 1548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018546904315197,
+      "loss": 3.0898,
+      "step": 1549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018545966228893058,
+      "loss": 2.781,
+      "step": 1550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018545028142589117,
+      "loss": 2.9024,
+      "step": 1551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001854409005628518,
+      "loss": 2.9958,
+      "step": 1552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018543151969981238,
+      "loss": 2.836,
+      "step": 1553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000185422138836773,
+      "loss": 3.0403,
+      "step": 1554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001854127579737336,
+      "loss": 2.7727,
+      "step": 1555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001854033771106942,
+      "loss": 3.0754,
+      "step": 1556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853939962476548,
+      "loss": 2.9421,
+      "step": 1557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853846153846154,
+      "loss": 2.9188,
+      "step": 1558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018537523452157598,
+      "loss": 2.8823,
+      "step": 1559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853658536585366,
+      "loss": 2.9923,
+      "step": 1560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853564727954972,
+      "loss": 2.8128,
+      "step": 1561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853470919324578,
+      "loss": 2.8034,
+      "step": 1562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853377110694184,
+      "loss": 2.9517,
+      "step": 1563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000185328330206379,
+      "loss": 2.961,
+      "step": 1564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853189493433396,
+      "loss": 2.6559,
+      "step": 1565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853095684803002,
+      "loss": 3.155,
+      "step": 1566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001853001876172608,
+      "loss": 2.6124,
+      "step": 1567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018529080675422138,
+      "loss": 3.1206,
+      "step": 1568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000185281425891182,
+      "loss": 2.8558,
+      "step": 1569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001852720450281426,
+      "loss": 2.8484,
+      "step": 1570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001852626641651032,
+      "loss": 2.6551,
+      "step": 1571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001852532833020638,
+      "loss": 2.9422,
+      "step": 1572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018524390243902442,
+      "loss": 2.8707,
+      "step": 1573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000185234521575985,
+      "loss": 3.075,
+      "step": 1574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001852251407129456,
+      "loss": 2.8242,
+      "step": 1575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001852157598499062,
+      "loss": 2.8684,
+      "step": 1576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018520637898686678,
+      "loss": 2.3384,
+      "step": 1577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001851969981238274,
+      "loss": 3.2875,
+      "step": 1578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000185187617260788,
+      "loss": 2.721,
+      "step": 1579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001851782363977486,
+      "loss": 2.7614,
+      "step": 1580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001851688555347092,
+      "loss": 3.0968,
+      "step": 1581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018515947467166982,
+      "loss": 3.0964,
+      "step": 1582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001851500938086304,
+      "loss": 2.4267,
+      "step": 1583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000185140712945591,
+      "loss": 2.871,
+      "step": 1584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001851313320825516,
+      "loss": 2.7951,
+      "step": 1585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001851219512195122,
+      "loss": 2.985,
+      "step": 1586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001851125703564728,
+      "loss": 2.8939,
+      "step": 1587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00018510318949343342,
+      "loss": 2.5799,
+      "step": 1588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000185093808630394,
+      "loss": 3.1105,
+      "step": 1589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001850844277673546,
+      "loss": 2.8819,
+      "step": 1590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018507504690431522,
+      "loss": 3.2191,
+      "step": 1591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001850656660412758,
+      "loss": 2.6002,
+      "step": 1592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001850562851782364,
+      "loss": 3.022,
+      "step": 1593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000185046904315197,
+      "loss": 2.8149,
+      "step": 1594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001850375234521576,
+      "loss": 2.767,
+      "step": 1595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001850281425891182,
+      "loss": 2.693,
+      "step": 1596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018501876172607882,
+      "loss": 2.9098,
+      "step": 1597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001850093808630394,
+      "loss": 2.8111,
+      "step": 1598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018500000000000002,
+      "loss": 2.8678,
+      "step": 1599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018499061913696062,
+      "loss": 2.9554,
+      "step": 1600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018498123827392123,
+      "loss": 2.9378,
+      "step": 1601
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001849718574108818,
+      "loss": 3.0315,
+      "step": 1602
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001849624765478424,
+      "loss": 2.8218,
+      "step": 1603
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000184953095684803,
+      "loss": 2.8377,
+      "step": 1604
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001849437148217636,
+      "loss": 3.0952,
+      "step": 1605
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018493433395872422,
+      "loss": 3.0533,
+      "step": 1606
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001849249530956848,
+      "loss": 2.9794,
+      "step": 1607
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018491557223264542,
+      "loss": 2.9608,
+      "step": 1608
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00018490619136960602,
+      "loss": 2.6585,
+      "step": 1609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018489681050656663,
+      "loss": 2.9339,
+      "step": 1610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001848874296435272,
+      "loss": 2.6162,
+      "step": 1611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018487804878048782,
+      "loss": 2.8276,
+      "step": 1612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001848686679174484,
+      "loss": 2.897,
+      "step": 1613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018485928705440902,
+      "loss": 3.0184,
+      "step": 1614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018484990619136962,
+      "loss": 2.9839,
+      "step": 1615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001848405253283302,
+      "loss": 2.922,
+      "step": 1616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018483114446529082,
+      "loss": 2.7377,
+      "step": 1617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018482176360225141,
+      "loss": 3.0839,
+      "step": 1618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018481238273921203,
+      "loss": 2.6509,
+      "step": 1619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001848030018761726,
+      "loss": 2.7313,
+      "step": 1620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018479362101313321,
+      "loss": 3.0233,
+      "step": 1621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001847842401500938,
+      "loss": 3.0138,
+      "step": 1622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018477485928705442,
+      "loss": 2.7654,
+      "step": 1623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018476547842401501,
+      "loss": 2.7642,
+      "step": 1624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018475609756097563,
+      "loss": 2.6059,
+      "step": 1625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018474671669793622,
+      "loss": 3.1302,
+      "step": 1626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018473733583489684,
+      "loss": 2.79,
+      "step": 1627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018472795497185743,
+      "loss": 2.8884,
+      "step": 1628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000184718574108818,
+      "loss": 2.8845,
+      "step": 1629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00018470919324577861,
+      "loss": 2.9649,
+      "step": 1630
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001846998123827392,
+      "loss": 2.5794,
+      "step": 1631
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018469043151969982,
+      "loss": 2.6996,
+      "step": 1632
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018468105065666041,
+      "loss": 3.215,
+      "step": 1633
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018467166979362103,
+      "loss": 3.0152,
+      "step": 1634
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018466228893058162,
+      "loss": 3.0556,
+      "step": 1635
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018465290806754224,
+      "loss": 2.9617,
+      "step": 1636
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018464352720450283,
+      "loss": 2.9317,
+      "step": 1637
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018463414634146342,
+      "loss": 2.742,
+      "step": 1638
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018462476547842401,
+      "loss": 3.0147,
+      "step": 1639
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 2.593,
+      "step": 1640
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018460600375234522,
+      "loss": 2.9859,
+      "step": 1641
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018459662288930581,
+      "loss": 2.7813,
+      "step": 1642
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018458724202626643,
+      "loss": 3.2439,
+      "step": 1643
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018457786116322702,
+      "loss": 3.0595,
+      "step": 1644
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018456848030018764,
+      "loss": 2.9818,
+      "step": 1645
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018455909943714823,
+      "loss": 2.9326,
+      "step": 1646
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018454971857410882,
+      "loss": 2.9995,
+      "step": 1647
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018454033771106941,
+      "loss": 2.5597,
+      "step": 1648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018453095684803003,
+      "loss": 2.8799,
+      "step": 1649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018452157598499062,
+      "loss": 2.8,
+      "step": 1650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018451219512195124,
+      "loss": 2.9127,
+      "step": 1651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018450281425891183,
+      "loss": 2.9202,
+      "step": 1652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018449343339587245,
+      "loss": 3.2213,
+      "step": 1653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018448405253283304,
+      "loss": 3.1159,
+      "step": 1654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018447467166979363,
+      "loss": 3.0792,
+      "step": 1655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018446529080675422,
+      "loss": 2.8201,
+      "step": 1656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018445590994371481,
+      "loss": 2.9863,
+      "step": 1657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018444652908067543,
+      "loss": 2.958,
+      "step": 1658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018443714821763602,
+      "loss": 2.9253,
+      "step": 1659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018442776735459664,
+      "loss": 3.1714,
+      "step": 1660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018441838649155723,
+      "loss": 3.2348,
+      "step": 1661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018440900562851785,
+      "loss": 2.9827,
+      "step": 1662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018439962476547844,
+      "loss": 2.7422,
+      "step": 1663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018439024390243903,
+      "loss": 2.8679,
+      "step": 1664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018438086303939962,
+      "loss": 2.8244,
+      "step": 1665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018437148217636024,
+      "loss": 2.7063,
+      "step": 1666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018436210131332083,
+      "loss": 3.0114,
+      "step": 1667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018435272045028142,
+      "loss": 2.884,
+      "step": 1668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018434333958724204,
+      "loss": 2.8999,
+      "step": 1669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018433395872420263,
+      "loss": 3.1188,
+      "step": 1670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018432457786116325,
+      "loss": 3.1033,
+      "step": 1671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018431519699812384,
+      "loss": 2.8498,
+      "step": 1672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018430581613508443,
+      "loss": 2.9387,
+      "step": 1673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018429643527204502,
+      "loss": 2.7833,
+      "step": 1674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018428705440900564,
+      "loss": 2.8318,
+      "step": 1675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018427767354596623,
+      "loss": 3.056,
+      "step": 1676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018426829268292685,
+      "loss": 2.9634,
+      "step": 1677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018425891181988744,
+      "loss": 3.0997,
+      "step": 1678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018424953095684806,
+      "loss": 3.0116,
+      "step": 1679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018424015009380865,
+      "loss": 2.6349,
+      "step": 1680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018423076923076924,
+      "loss": 2.909,
+      "step": 1681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018422138836772983,
+      "loss": 2.7181,
+      "step": 1682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018421200750469042,
+      "loss": 2.9177,
+      "step": 1683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018420262664165104,
+      "loss": 2.6897,
+      "step": 1684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018419324577861163,
+      "loss": 3.0603,
+      "step": 1685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018418386491557225,
+      "loss": 3.0362,
+      "step": 1686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018417448405253284,
+      "loss": 2.9174,
+      "step": 1687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018416510318949346,
+      "loss": 3.0829,
+      "step": 1688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018415572232645405,
+      "loss": 2.7685,
+      "step": 1689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018414634146341464,
+      "loss": 3.2717,
+      "step": 1690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018413696060037523,
+      "loss": 3.2392,
+      "step": 1691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018412757973733585,
+      "loss": 3.2888,
+      "step": 1692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018411819887429644,
+      "loss": 2.8266,
+      "step": 1693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00018410881801125703,
+      "loss": 2.5733,
+      "step": 1694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018409943714821765,
+      "loss": 2.7988,
+      "step": 1695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018409005628517824,
+      "loss": 2.9963,
+      "step": 1696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018408067542213886,
+      "loss": 2.9509,
+      "step": 1697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018407129455909945,
+      "loss": 3.0109,
+      "step": 1698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018406191369606004,
+      "loss": 2.8723,
+      "step": 1699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018405253283302063,
+      "loss": 2.4383,
+      "step": 1700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018404315196998125,
+      "loss": 2.7537,
+      "step": 1701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018403377110694184,
+      "loss": 2.794,
+      "step": 1702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018402439024390246,
+      "loss": 3.2093,
+      "step": 1703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018401500938086305,
+      "loss": 2.8129,
+      "step": 1704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018400562851782367,
+      "loss": 2.7393,
+      "step": 1705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018399624765478426,
+      "loss": 2.7815,
+      "step": 1706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018398686679174485,
+      "loss": 3.1864,
+      "step": 1707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018397748592870544,
+      "loss": 2.5515,
+      "step": 1708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018396810506566603,
+      "loss": 2.8603,
+      "step": 1709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018395872420262665,
+      "loss": 2.794,
+      "step": 1710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018394934333958724,
+      "loss": 2.7788,
+      "step": 1711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018393996247654786,
+      "loss": 2.8612,
+      "step": 1712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018393058161350845,
+      "loss": 2.5982,
+      "step": 1713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018392120075046907,
+      "loss": 2.7856,
+      "step": 1714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018391181988742966,
+      "loss": 2.9311,
+      "step": 1715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00018390243902439025,
+      "loss": 2.743,
+      "step": 1716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018389305816135084,
+      "loss": 2.9165,
+      "step": 1717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018388367729831146,
+      "loss": 2.9608,
+      "step": 1718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018387429643527205,
+      "loss": 2.8786,
+      "step": 1719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018386491557223264,
+      "loss": 3.1915,
+      "step": 1720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018385553470919326,
+      "loss": 3.1929,
+      "step": 1721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018384615384615385,
+      "loss": 3.152,
+      "step": 1722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018383677298311447,
+      "loss": 3.1318,
+      "step": 1723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018382739212007506,
+      "loss": 2.6454,
+      "step": 1724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018381801125703565,
+      "loss": 3.072,
+      "step": 1725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018380863039399624,
+      "loss": 2.2775,
+      "step": 1726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018379924953095686,
+      "loss": 3.1813,
+      "step": 1727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018378986866791745,
+      "loss": 2.8993,
+      "step": 1728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018378048780487807,
+      "loss": 2.9879,
+      "step": 1729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018377110694183866,
+      "loss": 2.7038,
+      "step": 1730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018376172607879928,
+      "loss": 2.7754,
+      "step": 1731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018375234521575987,
+      "loss": 2.7424,
+      "step": 1732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018374296435272046,
+      "loss": 2.605,
+      "step": 1733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018373358348968105,
+      "loss": 2.8621,
+      "step": 1734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018372420262664164,
+      "loss": 2.6127,
+      "step": 1735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018371482176360226,
+      "loss": 2.5798,
+      "step": 1736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018370544090056285,
+      "loss": 2.9664,
+      "step": 1737
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018369606003752347,
+      "loss": 2.8939,
+      "step": 1738
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018368667917448406,
+      "loss": 2.8719,
+      "step": 1739
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018367729831144468,
+      "loss": 3.0741,
+      "step": 1740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018366791744840527,
+      "loss": 2.9367,
+      "step": 1741
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018365853658536586,
+      "loss": 2.918,
+      "step": 1742
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018364915572232645,
+      "loss": 2.5897,
+      "step": 1743
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018363977485928707,
+      "loss": 2.9517,
+      "step": 1744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018363039399624766,
+      "loss": 2.7966,
+      "step": 1745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018362101313320828,
+      "loss": 2.9301,
+      "step": 1746
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018361163227016887,
+      "loss": 2.6587,
+      "step": 1747
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018360225140712946,
+      "loss": 2.77,
+      "step": 1748
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018359287054409008,
+      "loss": 2.9881,
+      "step": 1749
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018358348968105067,
+      "loss": 3.301,
+      "step": 1750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018357410881801126,
+      "loss": 2.8145,
+      "step": 1751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018356472795497185,
+      "loss": 3.0236,
+      "step": 1752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018355534709193247,
+      "loss": 2.5794,
+      "step": 1753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018354596622889306,
+      "loss": 3.3046,
+      "step": 1754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018353658536585368,
+      "loss": 2.6756,
+      "step": 1755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018352720450281427,
+      "loss": 2.7848,
+      "step": 1756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018351782363977488,
+      "loss": 2.8192,
+      "step": 1757
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018350844277673548,
+      "loss": 3.0384,
+      "step": 1758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001834990619136961,
+      "loss": 2.2596,
+      "step": 1759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018348968105065666,
+      "loss": 3.0687,
+      "step": 1760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018348030018761725,
+      "loss": 2.4826,
+      "step": 1761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018347091932457787,
+      "loss": 2.8485,
+      "step": 1762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018346153846153846,
+      "loss": 2.5195,
+      "step": 1763
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018345215759849907,
+      "loss": 3.0351,
+      "step": 1764
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018344277673545967,
+      "loss": 2.8601,
+      "step": 1765
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018343339587242028,
+      "loss": 2.706,
+      "step": 1766
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018342401500938087,
+      "loss": 3.0651,
+      "step": 1767
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001834146341463415,
+      "loss": 2.9071,
+      "step": 1768
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018340525328330206,
+      "loss": 2.8805,
+      "step": 1769
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018339587242026267,
+      "loss": 3.0039,
+      "step": 1770
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018338649155722327,
+      "loss": 2.6323,
+      "step": 1771
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018337711069418388,
+      "loss": 2.6026,
+      "step": 1772
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018336772983114447,
+      "loss": 2.7546,
+      "step": 1773
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018335834896810507,
+      "loss": 2.9124,
+      "step": 1774
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018334896810506568,
+      "loss": 3.0439,
+      "step": 1775
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018333958724202627,
+      "loss": 3.0376,
+      "step": 1776
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.0001833302063789869,
+      "loss": 2.7798,
+      "step": 1777
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018332082551594746,
+      "loss": 2.8865,
+      "step": 1778
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018331144465290807,
+      "loss": 3.2011,
+      "step": 1779
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018330206378986867,
+      "loss": 2.6914,
+      "step": 1780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018329268292682928,
+      "loss": 2.933,
+      "step": 1781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018328330206378987,
+      "loss": 2.8726,
+      "step": 1782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001832739212007505,
+      "loss": 3.2444,
+      "step": 1783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018326454033771108,
+      "loss": 2.7718,
+      "step": 1784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001832551594746717,
+      "loss": 2.3999,
+      "step": 1785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001832457786116323,
+      "loss": 3.0905,
+      "step": 1786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018323639774859286,
+      "loss": 2.8078,
+      "step": 1787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018322701688555347,
+      "loss": 2.6837,
+      "step": 1788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018321763602251407,
+      "loss": 2.7896,
+      "step": 1789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018320825515947468,
+      "loss": 3.2679,
+      "step": 1790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018319887429643527,
+      "loss": 2.7889,
+      "step": 1791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001831894934333959,
+      "loss": 2.8375,
+      "step": 1792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018318011257035648,
+      "loss": 2.8094,
+      "step": 1793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001831707317073171,
+      "loss": 3.071,
+      "step": 1794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001831613508442777,
+      "loss": 2.7771,
+      "step": 1795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018315196998123828,
+      "loss": 2.7684,
+      "step": 1796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018314258911819887,
+      "loss": 2.8529,
+      "step": 1797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001831332082551595,
+      "loss": 2.7161,
+      "step": 1798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018312382739212008,
+      "loss": 2.6015,
+      "step": 1799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.00018311444652908067,
+      "loss": 2.8625,
+      "step": 1800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 0.0001831050656660413,
+      "loss": 2.8502,
+      "step": 1801
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018309568480300188,
+      "loss": 2.7181,
+      "step": 1802
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001830863039399625,
+      "loss": 2.5004,
+      "step": 1803
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001830769230769231,
+      "loss": 3.1622,
+      "step": 1804
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018306754221388368,
+      "loss": 3.1595,
+      "step": 1805
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018305816135084427,
+      "loss": 2.9788,
+      "step": 1806
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001830487804878049,
+      "loss": 3.0773,
+      "step": 1807
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018303939962476548,
+      "loss": 2.7064,
+      "step": 1808
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001830300187617261,
+      "loss": 2.8956,
+      "step": 1809
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001830206378986867,
+      "loss": 2.4675,
+      "step": 1810
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001830112570356473,
+      "loss": 2.8514,
+      "step": 1811
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001830018761726079,
+      "loss": 2.8973,
+      "step": 1812
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001829924953095685,
+      "loss": 2.809,
+      "step": 1813
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018298311444652908,
+      "loss": 2.7547,
+      "step": 1814
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018297373358348967,
+      "loss": 2.6748,
+      "step": 1815
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001829643527204503,
+      "loss": 2.7911,
+      "step": 1816
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.00018295497185741088,
+      "loss": 3.0645,
+      "step": 1817
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001829455909943715,
+      "loss": 2.9165,
+      "step": 1818
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001829362101313321,
+      "loss": 2.9608,
+      "step": 1819
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001829268292682927,
+      "loss": 2.5631,
+      "step": 1820
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001829174484052533,
+      "loss": 2.8396,
+      "step": 1821
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001829080675422139,
+      "loss": 2.7184,
+      "step": 1822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018289868667917448,
+      "loss": 2.9166,
+      "step": 1823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001828893058161351,
+      "loss": 2.9088,
+      "step": 1824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001828799249530957,
+      "loss": 2.7836,
+      "step": 1825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018287054409005628,
+      "loss": 2.7678,
+      "step": 1826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001828611632270169,
+      "loss": 2.9346,
+      "step": 1827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001828517823639775,
+      "loss": 3.1084,
+      "step": 1828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001828424015009381,
+      "loss": 2.9251,
+      "step": 1829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001828330206378987,
+      "loss": 2.6791,
+      "step": 1830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001828236397748593,
+      "loss": 2.638,
+      "step": 1831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018281425891181988,
+      "loss": 3.0665,
+      "step": 1832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001828048780487805,
+      "loss": 2.8775,
+      "step": 1833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001827954971857411,
+      "loss": 2.742,
+      "step": 1834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001827861163227017,
+      "loss": 3.0903,
+      "step": 1835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001827767354596623,
+      "loss": 3.1162,
+      "step": 1836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018276735459662292,
+      "loss": 3.3869,
+      "step": 1837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001827579737335835,
+      "loss": 2.7017,
+      "step": 1838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001827485928705441,
+      "loss": 2.8956,
+      "step": 1839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001827392120075047,
+      "loss": 2.8746,
+      "step": 1840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018272983114446528,
+      "loss": 3.0796,
+      "step": 1841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001827204502814259,
+      "loss": 2.8755,
+      "step": 1842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001827110694183865,
+      "loss": 3.0471,
+      "step": 1843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.0001827016885553471,
+      "loss": 2.7191,
+      "step": 1844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001826923076923077,
+      "loss": 2.9682,
+      "step": 1845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018268292682926832,
+      "loss": 2.432,
+      "step": 1846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001826735459662289,
+      "loss": 2.9755,
+      "step": 1847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001826641651031895,
+      "loss": 3.028,
+      "step": 1848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001826547842401501,
+      "loss": 2.4453,
+      "step": 1849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001826454033771107,
+      "loss": 2.555,
+      "step": 1850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001826360225140713,
+      "loss": 2.9537,
+      "step": 1851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001826266416510319,
+      "loss": 3.0339,
+      "step": 1852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001826172607879925,
+      "loss": 2.7351,
+      "step": 1853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001826078799249531,
+      "loss": 2.6051,
+      "step": 1854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018259849906191372,
+      "loss": 2.7443,
+      "step": 1855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001825891181988743,
+      "loss": 2.4246,
+      "step": 1856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001825797373358349,
+      "loss": 2.9251,
+      "step": 1857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001825703564727955,
+      "loss": 3.1866,
+      "step": 1858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001825609756097561,
+      "loss": 3.0589,
+      "step": 1859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001825515947467167,
+      "loss": 2.536,
+      "step": 1860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018254221388367732,
+      "loss": 2.7218,
+      "step": 1861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001825328330206379,
+      "loss": 2.9613,
+      "step": 1862
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018252345215759853,
+      "loss": 2.8657,
+      "step": 1863
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018251407129455912,
+      "loss": 2.6259,
+      "step": 1864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.0001825046904315197,
+      "loss": 2.84,
+      "step": 1865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001824953095684803,
+      "loss": 2.8231,
+      "step": 1866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001824859287054409,
+      "loss": 2.8461,
+      "step": 1867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001824765478424015,
+      "loss": 2.9286,
+      "step": 1868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001824671669793621,
+      "loss": 3.1003,
+      "step": 1869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018245778611632272,
+      "loss": 2.7907,
+      "step": 1870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001824484052532833,
+      "loss": 2.815,
+      "step": 1871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018243902439024393,
+      "loss": 3.1225,
+      "step": 1872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018242964352720452,
+      "loss": 2.6833,
+      "step": 1873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001824202626641651,
+      "loss": 2.9465,
+      "step": 1874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001824108818011257,
+      "loss": 2.6586,
+      "step": 1875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018240150093808632,
+      "loss": 2.7114,
+      "step": 1876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001823921200750469,
+      "loss": 2.6707,
+      "step": 1877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001823827392120075,
+      "loss": 2.6667,
+      "step": 1878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018237335834896812,
+      "loss": 2.9026,
+      "step": 1879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001823639774859287,
+      "loss": 2.9729,
+      "step": 1880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018235459662288933,
+      "loss": 2.9119,
+      "step": 1881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018234521575984992,
+      "loss": 2.6845,
+      "step": 1882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001823358348968105,
+      "loss": 2.7928,
+      "step": 1883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001823264540337711,
+      "loss": 3.0361,
+      "step": 1884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.00018231707317073172,
+      "loss": 2.783,
+      "step": 1885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001823076923076923,
+      "loss": 2.8251,
+      "step": 1886
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018229831144465293,
+      "loss": 2.7769,
+      "step": 1887
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018228893058161352,
+      "loss": 2.8489,
+      "step": 1888
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018227954971857414,
+      "loss": 2.8711,
+      "step": 1889
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018227016885553473,
+      "loss": 2.7403,
+      "step": 1890
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018226078799249532,
+      "loss": 2.7495,
+      "step": 1891
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001822514071294559,
+      "loss": 2.6978,
+      "step": 1892
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001822420262664165,
+      "loss": 2.7882,
+      "step": 1893
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018223264540337712,
+      "loss": 2.7201,
+      "step": 1894
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001822232645403377,
+      "loss": 2.8478,
+      "step": 1895
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018221388367729833,
+      "loss": 2.8614,
+      "step": 1896
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018220450281425892,
+      "loss": 2.8845,
+      "step": 1897
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018219512195121954,
+      "loss": 2.7343,
+      "step": 1898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018218574108818013,
+      "loss": 2.9011,
+      "step": 1899
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018217636022514072,
+      "loss": 2.5713,
+      "step": 1900
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001821669793621013,
+      "loss": 2.7717,
+      "step": 1901
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018215759849906193,
+      "loss": 2.8689,
+      "step": 1902
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018214821763602252,
+      "loss": 2.9959,
+      "step": 1903
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.0001821388367729831,
+      "loss": 2.9722,
+      "step": 1904
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018212945590994373,
+      "loss": 2.578,
+      "step": 1905
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018212007504690432,
+      "loss": 3.0496,
+      "step": 1906
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018211069418386494,
+      "loss": 2.8482,
+      "step": 1907
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 0.00018210131332082553,
+      "loss": 2.7444,
+      "step": 1908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018209193245778612,
+      "loss": 2.7719,
+      "step": 1909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001820825515947467,
+      "loss": 2.8789,
+      "step": 1910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018207317073170733,
+      "loss": 2.8998,
+      "step": 1911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018206378986866792,
+      "loss": 2.6075,
+      "step": 1912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018205440900562853,
+      "loss": 2.9643,
+      "step": 1913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018204502814258913,
+      "loss": 2.7652,
+      "step": 1914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018203564727954974,
+      "loss": 2.9642,
+      "step": 1915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018202626641651033,
+      "loss": 2.959,
+      "step": 1916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018201688555347093,
+      "loss": 2.9101,
+      "step": 1917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018200750469043152,
+      "loss": 2.758,
+      "step": 1918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001819981238273921,
+      "loss": 2.8619,
+      "step": 1919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018198874296435273,
+      "loss": 2.9178,
+      "step": 1920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018197936210131332,
+      "loss": 2.6792,
+      "step": 1921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018196998123827393,
+      "loss": 2.6642,
+      "step": 1922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018196060037523453,
+      "loss": 2.8613,
+      "step": 1923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018195121951219514,
+      "loss": 2.7323,
+      "step": 1924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018194183864915573,
+      "loss": 2.6666,
+      "step": 1925
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018193245778611635,
+      "loss": 3.2057,
+      "step": 1926
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018192307692307692,
+      "loss": 2.6243,
+      "step": 1927
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018191369606003753,
+      "loss": 2.8498,
+      "step": 1928
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018190431519699813,
+      "loss": 3.2915,
+      "step": 1929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018189493433395872,
+      "loss": 2.9927,
+      "step": 1930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018188555347091933,
+      "loss": 2.5857,
+      "step": 1931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018187617260787993,
+      "loss": 2.8016,
+      "step": 1932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018186679174484054,
+      "loss": 2.5689,
+      "step": 1933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018185741088180113,
+      "loss": 3.0665,
+      "step": 1934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018184803001876175,
+      "loss": 2.8676,
+      "step": 1935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018183864915572232,
+      "loss": 2.6794,
+      "step": 1936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018182926829268293,
+      "loss": 2.839,
+      "step": 1937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018181988742964353,
+      "loss": 2.9924,
+      "step": 1938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018181050656660414,
+      "loss": 2.7864,
+      "step": 1939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018180112570356473,
+      "loss": 2.8443,
+      "step": 1940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018179174484052535,
+      "loss": 2.9152,
+      "step": 1941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018178236397748594,
+      "loss": 2.9899,
+      "step": 1942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018177298311444653,
+      "loss": 2.7245,
+      "step": 1943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018176360225140715,
+      "loss": 2.9388,
+      "step": 1944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018175422138836772,
+      "loss": 2.5637,
+      "step": 1945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018174484052532833,
+      "loss": 2.6825,
+      "step": 1946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018173545966228893,
+      "loss": 2.9115,
+      "step": 1947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018172607879924954,
+      "loss": 3.024,
+      "step": 1948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018171669793621013,
+      "loss": 2.9283,
+      "step": 1949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018170731707317075,
+      "loss": 2.9329,
+      "step": 1950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018169793621013134,
+      "loss": 2.8442,
+      "step": 1951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018168855534709196,
+      "loss": 2.9023,
+      "step": 1952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018167917448405255,
+      "loss": 2.9923,
+      "step": 1953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018166979362101314,
+      "loss": 2.9518,
+      "step": 1954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018166041275797373,
+      "loss": 2.9906,
+      "step": 1955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018165103189493435,
+      "loss": 2.8905,
+      "step": 1956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018164165103189494,
+      "loss": 2.8345,
+      "step": 1957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018163227016885553,
+      "loss": 2.6764,
+      "step": 1958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018162288930581615,
+      "loss": 2.7594,
+      "step": 1959
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018161350844277674,
+      "loss": 2.4669,
+      "step": 1960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018160412757973736,
+      "loss": 2.7891,
+      "step": 1961
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018159474671669795,
+      "loss": 2.8279,
+      "step": 1962
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018158536585365854,
+      "loss": 2.6635,
+      "step": 1963
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018157598499061913,
+      "loss": 2.7711,
+      "step": 1964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018156660412757975,
+      "loss": 2.9716,
+      "step": 1965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018155722326454034,
+      "loss": 2.6543,
+      "step": 1966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018154784240150096,
+      "loss": 2.7851,
+      "step": 1967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018153846153846155,
+      "loss": 2.8808,
+      "step": 1968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018152908067542217,
+      "loss": 2.8756,
+      "step": 1969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018151969981238276,
+      "loss": 2.7807,
+      "step": 1970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018151031894934335,
+      "loss": 2.7533,
+      "step": 1971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.00018150093808630394,
+      "loss": 2.9246,
+      "step": 1972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018149155722326453,
+      "loss": 3.2096,
+      "step": 1973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018148217636022515,
+      "loss": 2.9447,
+      "step": 1974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018147279549718574,
+      "loss": 2.7182,
+      "step": 1975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018146341463414636,
+      "loss": 2.9984,
+      "step": 1976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018145403377110695,
+      "loss": 2.9325,
+      "step": 1977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018144465290806757,
+      "loss": 2.8673,
+      "step": 1978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018143527204502816,
+      "loss": 2.9853,
+      "step": 1979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018142589118198875,
+      "loss": 2.8299,
+      "step": 1980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018141651031894934,
+      "loss": 2.766,
+      "step": 1981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018140712945590996,
+      "loss": 2.9221,
+      "step": 1982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018139774859287055,
+      "loss": 2.6051,
+      "step": 1983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018138836772983114,
+      "loss": 2.891,
+      "step": 1984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018137898686679176,
+      "loss": 2.7577,
+      "step": 1985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018136960600375235,
+      "loss": 3.0337,
+      "step": 1986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018136022514071297,
+      "loss": 2.5834,
+      "step": 1987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018135084427767356,
+      "loss": 2.7975,
+      "step": 1988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018134146341463415,
+      "loss": 2.7107,
+      "step": 1989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018133208255159474,
+      "loss": 2.6702,
+      "step": 1990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018132270168855536,
+      "loss": 2.6747,
+      "step": 1991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018131332082551595,
+      "loss": 3.0054,
+      "step": 1992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 0.00018130393996247657,
+      "loss": 2.5432,
+      "step": 1993
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018129455909943716,
+      "loss": 2.784,
+      "step": 1994
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018128517823639778,
+      "loss": 2.8278,
+      "step": 1995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018127579737335837,
+      "loss": 3.0657,
+      "step": 1996
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018126641651031896,
+      "loss": 2.989,
+      "step": 1997
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018125703564727955,
+      "loss": 3.002,
+      "step": 1998
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018124765478424014,
+      "loss": 2.7064,
+      "step": 1999
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018123827392120076,
+      "loss": 2.7278,
+      "step": 2000
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 2.7116825580596924,
+      "eval_runtime": 67.319,
+      "eval_samples_per_second": 28.15,
+      "eval_steps_per_second": 3.521,
+      "step": 2000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018122889305816135,
+      "loss": 2.7719,
+      "step": 2001
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018121951219512197,
+      "loss": 2.6463,
+      "step": 2002
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018121013133208256,
+      "loss": 2.7603,
+      "step": 2003
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018120075046904318,
+      "loss": 2.4644,
+      "step": 2004
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018119136960600377,
+      "loss": 2.6732,
+      "step": 2005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018118198874296436,
+      "loss": 2.8337,
+      "step": 2006
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018117260787992495,
+      "loss": 2.6149,
+      "step": 2007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018116322701688557,
+      "loss": 3.1207,
+      "step": 2008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018115384615384616,
+      "loss": 2.9406,
+      "step": 2009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018114446529080675,
+      "loss": 3.1662,
+      "step": 2010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018113508442776737,
+      "loss": 3.016,
+      "step": 2011
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018112570356472796,
+      "loss": 2.7439,
+      "step": 2012
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018111632270168858,
+      "loss": 2.9536,
+      "step": 2013
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018110694183864917,
+      "loss": 2.7411,
+      "step": 2014
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018109756097560976,
+      "loss": 2.5186,
+      "step": 2015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018108818011257035,
+      "loss": 2.8088,
+      "step": 2016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018107879924953097,
+      "loss": 2.6952,
+      "step": 2017
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018106941838649156,
+      "loss": 2.8547,
+      "step": 2018
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018106003752345218,
+      "loss": 3.0916,
+      "step": 2019
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018105065666041277,
+      "loss": 2.9534,
+      "step": 2020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001810412757973734,
+      "loss": 2.9232,
+      "step": 2021
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018103189493433398,
+      "loss": 2.6499,
+      "step": 2022
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018102251407129457,
+      "loss": 3.0511,
+      "step": 2023
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018101313320825516,
+      "loss": 2.6385,
+      "step": 2024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018100375234521575,
+      "loss": 2.4445,
+      "step": 2025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018099437148217637,
+      "loss": 2.6721,
+      "step": 2026
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018098499061913696,
+      "loss": 2.5676,
+      "step": 2027
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018097560975609758,
+      "loss": 2.969,
+      "step": 2028
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018096622889305817,
+      "loss": 2.764,
+      "step": 2029
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.0001809568480300188,
+      "loss": 2.9537,
+      "step": 2030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018094746716697938,
+      "loss": 2.9344,
+      "step": 2031
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018093808630393997,
+      "loss": 3.2573,
+      "step": 2032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018092870544090056,
+      "loss": 2.8277,
+      "step": 2033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018091932457786118,
+      "loss": 2.8219,
+      "step": 2034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018090994371482177,
+      "loss": 2.9622,
+      "step": 2035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018090056285178236,
+      "loss": 2.893,
+      "step": 2036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018089118198874298,
+      "loss": 2.7425,
+      "step": 2037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018088180112570357,
+      "loss": 2.9115,
+      "step": 2038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018087242026266419,
+      "loss": 2.8748,
+      "step": 2039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018086303939962478,
+      "loss": 2.8726,
+      "step": 2040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018085365853658537,
+      "loss": 3.3089,
+      "step": 2041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018084427767354596,
+      "loss": 3.0834,
+      "step": 2042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018083489681050658,
+      "loss": 2.8246,
+      "step": 2043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018082551594746717,
+      "loss": 2.9791,
+      "step": 2044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018081613508442779,
+      "loss": 2.8184,
+      "step": 2045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018080675422138838,
+      "loss": 2.9574,
+      "step": 2046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.000180797373358349,
+      "loss": 2.7365,
+      "step": 2047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018078799249530959,
+      "loss": 2.4522,
+      "step": 2048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018077861163227018,
+      "loss": 2.9125,
+      "step": 2049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018076923076923077,
+      "loss": 3.0157,
+      "step": 2050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018075984990619136,
+      "loss": 2.7907,
+      "step": 2051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018075046904315198,
+      "loss": 2.7907,
+      "step": 2052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018074108818011257,
+      "loss": 2.5705,
+      "step": 2053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018073170731707319,
+      "loss": 2.8346,
+      "step": 2054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018072232645403378,
+      "loss": 2.6216,
+      "step": 2055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001807129455909944,
+      "loss": 2.7373,
+      "step": 2056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.00018070356472795499,
+      "loss": 2.8679,
+      "step": 2057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018069418386491558,
+      "loss": 2.874,
+      "step": 2058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018068480300187617,
+      "loss": 3.0608,
+      "step": 2059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018067542213883679,
+      "loss": 2.9365,
+      "step": 2060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018066604127579738,
+      "loss": 2.5939,
+      "step": 2061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018065666041275797,
+      "loss": 2.7799,
+      "step": 2062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018064727954971859,
+      "loss": 2.8396,
+      "step": 2063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018063789868667918,
+      "loss": 2.8069,
+      "step": 2064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001806285178236398,
+      "loss": 2.7662,
+      "step": 2065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018061913696060039,
+      "loss": 2.8847,
+      "step": 2066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018060975609756098,
+      "loss": 2.5483,
+      "step": 2067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018060037523452157,
+      "loss": 2.7495,
+      "step": 2068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018059099437148219,
+      "loss": 2.7695,
+      "step": 2069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018058161350844278,
+      "loss": 2.8498,
+      "step": 2070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001805722326454034,
+      "loss": 2.9298,
+      "step": 2071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018056285178236399,
+      "loss": 3.0694,
+      "step": 2072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001805534709193246,
+      "loss": 3.1446,
+      "step": 2073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.0001805440900562852,
+      "loss": 2.8135,
+      "step": 2074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018053470919324579,
+      "loss": 3.1689,
+      "step": 2075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018052532833020638,
+      "loss": 2.9113,
+      "step": 2076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018051594746716697,
+      "loss": 2.8758,
+      "step": 2077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018050656660412759,
+      "loss": 3.0526,
+      "step": 2078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018049718574108818,
+      "loss": 3.0787,
+      "step": 2079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001804878048780488,
+      "loss": 2.8595,
+      "step": 2080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018047842401500939,
+      "loss": 2.9246,
+      "step": 2081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018046904315197,
+      "loss": 2.7735,
+      "step": 2082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001804596622889306,
+      "loss": 3.0584,
+      "step": 2083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001804502814258912,
+      "loss": 2.8884,
+      "step": 2084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018044090056285178,
+      "loss": 2.63,
+      "step": 2085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001804315196998124,
+      "loss": 2.6118,
+      "step": 2086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018042213883677299,
+      "loss": 3.0056,
+      "step": 2087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018041275797373358,
+      "loss": 2.7897,
+      "step": 2088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001804033771106942,
+      "loss": 2.82,
+      "step": 2089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018039399624765479,
+      "loss": 3.0259,
+      "step": 2090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001803846153846154,
+      "loss": 3.0458,
+      "step": 2091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000180375234521576,
+      "loss": 2.8735,
+      "step": 2092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001803658536585366,
+      "loss": 2.7535,
+      "step": 2093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018035647279549718,
+      "loss": 2.8523,
+      "step": 2094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001803470919324578,
+      "loss": 2.8311,
+      "step": 2095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.00018033771106941839,
+      "loss": 2.8017,
+      "step": 2096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.000180328330206379,
+      "loss": 2.6928,
+      "step": 2097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001803189493433396,
+      "loss": 3.0117,
+      "step": 2098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001803095684803002,
+      "loss": 3.1401,
+      "step": 2099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 0.0001803001876172608,
+      "loss": 3.0336,
+      "step": 2100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001802908067542214,
+      "loss": 2.8823,
+      "step": 2101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000180281425891182,
+      "loss": 2.905,
+      "step": 2102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018027204502814258,
+      "loss": 2.8724,
+      "step": 2103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001802626641651032,
+      "loss": 2.784,
+      "step": 2104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018025328330206378,
+      "loss": 2.8795,
+      "step": 2105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001802439024390244,
+      "loss": 2.5775,
+      "step": 2106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000180234521575985,
+      "loss": 3.1288,
+      "step": 2107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001802251407129456,
+      "loss": 3.1266,
+      "step": 2108
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001802157598499062,
+      "loss": 2.8435,
+      "step": 2109
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018020637898686682,
+      "loss": 2.6499,
+      "step": 2110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001801969981238274,
+      "loss": 3.0629,
+      "step": 2111
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000180187617260788,
+      "loss": 2.9173,
+      "step": 2112
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001801782363977486,
+      "loss": 2.7266,
+      "step": 2113
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018016885553470918,
+      "loss": 3.1065,
+      "step": 2114
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001801594746716698,
+      "loss": 2.7066,
+      "step": 2115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001801500938086304,
+      "loss": 2.758,
+      "step": 2116
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.000180140712945591,
+      "loss": 2.8674,
+      "step": 2117
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001801313320825516,
+      "loss": 2.6068,
+      "step": 2118
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018012195121951222,
+      "loss": 2.659,
+      "step": 2119
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001801125703564728,
+      "loss": 2.9637,
+      "step": 2120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.0001801031894934334,
+      "loss": 2.8606,
+      "step": 2121
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000180093808630394,
+      "loss": 2.6876,
+      "step": 2122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001800844277673546,
+      "loss": 2.9147,
+      "step": 2123
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001800750469043152,
+      "loss": 2.5767,
+      "step": 2124
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018006566604127582,
+      "loss": 2.8325,
+      "step": 2125
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001800562851782364,
+      "loss": 2.6036,
+      "step": 2126
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000180046904315197,
+      "loss": 3.0429,
+      "step": 2127
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018003752345215762,
+      "loss": 2.8601,
+      "step": 2128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001800281425891182,
+      "loss": 2.5428,
+      "step": 2129
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001800187617260788,
+      "loss": 2.9293,
+      "step": 2130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001800093808630394,
+      "loss": 2.7349,
+      "step": 2131
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00018,
+      "loss": 2.7829,
+      "step": 2132
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001799906191369606,
+      "loss": 2.8135,
+      "step": 2133
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017998123827392122,
+      "loss": 2.5587,
+      "step": 2134
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001799718574108818,
+      "loss": 2.9458,
+      "step": 2135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017996247654784243,
+      "loss": 2.4672,
+      "step": 2136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.00017995309568480302,
+      "loss": 2.8533,
+      "step": 2137
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001799437148217636,
+      "loss": 2.9956,
+      "step": 2138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001799343339587242,
+      "loss": 2.5347,
+      "step": 2139
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001799249530956848,
+      "loss": 3.0223,
+      "step": 2140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001799155722326454,
+      "loss": 2.9175,
+      "step": 2141
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.000179906191369606,
+      "loss": 2.7915,
+      "step": 2142
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017989681050656662,
+      "loss": 2.8964,
+      "step": 2143
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001798874296435272,
+      "loss": 2.9431,
+      "step": 2144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017987804878048783,
+      "loss": 2.817,
+      "step": 2145
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017986866791744842,
+      "loss": 2.784,
+      "step": 2146
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000179859287054409,
+      "loss": 2.6926,
+      "step": 2147
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001798499061913696,
+      "loss": 2.6331,
+      "step": 2148
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017984052532833022,
+      "loss": 2.5328,
+      "step": 2149
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001798311444652908,
+      "loss": 2.6487,
+      "step": 2150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017982176360225143,
+      "loss": 2.8806,
+      "step": 2151
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017981238273921202,
+      "loss": 2.8583,
+      "step": 2152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017980300187617264,
+      "loss": 2.5216,
+      "step": 2153
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017979362101313323,
+      "loss": 2.7769,
+      "step": 2154
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017978424015009382,
+      "loss": 2.6909,
+      "step": 2155
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001797748592870544,
+      "loss": 2.3549,
+      "step": 2156
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.000179765478424015,
+      "loss": 2.8306,
+      "step": 2157
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017975609756097562,
+      "loss": 2.8059,
+      "step": 2158
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.0001797467166979362,
+      "loss": 2.854,
+      "step": 2159
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017973733583489683,
+      "loss": 2.5748,
+      "step": 2160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017972795497185742,
+      "loss": 2.9529,
+      "step": 2161
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017971857410881804,
+      "loss": 2.7524,
+      "step": 2162
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00017970919324577863,
+      "loss": 2.8373,
+      "step": 2163
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017969981238273922,
+      "loss": 2.9041,
+      "step": 2164
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001796904315196998,
+      "loss": 2.9315,
+      "step": 2165
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017968105065666043,
+      "loss": 2.265,
+      "step": 2166
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017967166979362102,
+      "loss": 3.0426,
+      "step": 2167
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001796622889305816,
+      "loss": 2.6348,
+      "step": 2168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017965290806754223,
+      "loss": 2.9794,
+      "step": 2169
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017964352720450282,
+      "loss": 2.8104,
+      "step": 2170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017963414634146344,
+      "loss": 2.8258,
+      "step": 2171
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017962476547842403,
+      "loss": 2.6364,
+      "step": 2172
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017961538461538462,
+      "loss": 2.7155,
+      "step": 2173
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001796060037523452,
+      "loss": 2.7435,
+      "step": 2174
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017959662288930583,
+      "loss": 2.6017,
+      "step": 2175
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017958724202626642,
+      "loss": 2.5141,
+      "step": 2176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017957786116322704,
+      "loss": 2.5678,
+      "step": 2177
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017956848030018763,
+      "loss": 2.7168,
+      "step": 2178
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017955909943714825,
+      "loss": 2.8386,
+      "step": 2179
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017954971857410884,
+      "loss": 2.7928,
+      "step": 2180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017954033771106943,
+      "loss": 2.879,
+      "step": 2181
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017953095684803002,
+      "loss": 2.8149,
+      "step": 2182
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.0001795215759849906,
+      "loss": 2.8083,
+      "step": 2183
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017951219512195123,
+      "loss": 2.8136,
+      "step": 2184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 0.00017950281425891182,
+      "loss": 3.0287,
+      "step": 2185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017949343339587244,
+      "loss": 3.1003,
+      "step": 2186
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017948405253283303,
+      "loss": 2.7045,
+      "step": 2187
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017947467166979365,
+      "loss": 2.9984,
+      "step": 2188
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017946529080675424,
+      "loss": 2.529,
+      "step": 2189
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017945590994371483,
+      "loss": 2.7672,
+      "step": 2190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017944652908067542,
+      "loss": 2.6593,
+      "step": 2191
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017943714821763604,
+      "loss": 2.7052,
+      "step": 2192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017942776735459663,
+      "loss": 2.7894,
+      "step": 2193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017941838649155722,
+      "loss": 2.5259,
+      "step": 2194
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017940900562851784,
+      "loss": 2.8039,
+      "step": 2195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017939962476547843,
+      "loss": 2.7087,
+      "step": 2196
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017939024390243905,
+      "loss": 2.8537,
+      "step": 2197
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017938086303939964,
+      "loss": 2.7277,
+      "step": 2198
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017937148217636023,
+      "loss": 2.6104,
+      "step": 2199
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017936210131332082,
+      "loss": 2.5824,
+      "step": 2200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017935272045028144,
+      "loss": 2.7771,
+      "step": 2201
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017934333958724203,
+      "loss": 2.5709,
+      "step": 2202
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017933395872420265,
+      "loss": 2.9065,
+      "step": 2203
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017932457786116324,
+      "loss": 2.924,
+      "step": 2204
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017931519699812385,
+      "loss": 3.0831,
+      "step": 2205
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00017930581613508445,
+      "loss": 3.1193,
+      "step": 2206
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017929643527204504,
+      "loss": 2.8884,
+      "step": 2207
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017928705440900563,
+      "loss": 2.6995,
+      "step": 2208
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017927767354596622,
+      "loss": 2.7862,
+      "step": 2209
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017926829268292684,
+      "loss": 2.753,
+      "step": 2210
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017925891181988743,
+      "loss": 2.8082,
+      "step": 2211
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017924953095684805,
+      "loss": 2.6336,
+      "step": 2212
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017924015009380864,
+      "loss": 2.7258,
+      "step": 2213
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017923076923076925,
+      "loss": 2.794,
+      "step": 2214
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017922138836772985,
+      "loss": 2.6973,
+      "step": 2215
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017921200750469044,
+      "loss": 2.5784,
+      "step": 2216
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017920262664165103,
+      "loss": 3.1536,
+      "step": 2217
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017919324577861165,
+      "loss": 2.6238,
+      "step": 2218
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017918386491557224,
+      "loss": 2.9044,
+      "step": 2219
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017917448405253283,
+      "loss": 2.9836,
+      "step": 2220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017916510318949345,
+      "loss": 2.6194,
+      "step": 2221
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017915572232645404,
+      "loss": 2.6541,
+      "step": 2222
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017914634146341465,
+      "loss": 2.7568,
+      "step": 2223
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017913696060037525,
+      "loss": 2.861,
+      "step": 2224
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017912757973733584,
+      "loss": 2.892,
+      "step": 2225
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017911819887429643,
+      "loss": 2.6649,
+      "step": 2226
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00017910881801125705,
+      "loss": 2.6406,
+      "step": 2227
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017909943714821764,
+      "loss": 2.8542,
+      "step": 2228
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017909005628517825,
+      "loss": 2.6369,
+      "step": 2229
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017908067542213885,
+      "loss": 2.5957,
+      "step": 2230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017907129455909946,
+      "loss": 2.6583,
+      "step": 2231
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017906191369606005,
+      "loss": 2.6652,
+      "step": 2232
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017905253283302065,
+      "loss": 2.7401,
+      "step": 2233
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017904315196998124,
+      "loss": 2.78,
+      "step": 2234
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017903377110694183,
+      "loss": 2.9435,
+      "step": 2235
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017902439024390245,
+      "loss": 2.9519,
+      "step": 2236
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017901500938086304,
+      "loss": 2.8006,
+      "step": 2237
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017900562851782365,
+      "loss": 2.7361,
+      "step": 2238
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017899624765478425,
+      "loss": 3.0748,
+      "step": 2239
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017898686679174486,
+      "loss": 3.0642,
+      "step": 2240
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017897748592870545,
+      "loss": 2.5829,
+      "step": 2241
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017896810506566605,
+      "loss": 2.7141,
+      "step": 2242
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017895872420262664,
+      "loss": 2.8577,
+      "step": 2243
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017894934333958725,
+      "loss": 2.3303,
+      "step": 2244
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017893996247654784,
+      "loss": 2.7983,
+      "step": 2245
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017893058161350844,
+      "loss": 2.832,
+      "step": 2246
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017892120075046905,
+      "loss": 2.4911,
+      "step": 2247
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017891181988742964,
+      "loss": 3.0301,
+      "step": 2248
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.00017890243902439026,
+      "loss": 2.4567,
+      "step": 2249
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017889305816135085,
+      "loss": 2.7321,
+      "step": 2250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017888367729831147,
+      "loss": 2.8546,
+      "step": 2251
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017887429643527204,
+      "loss": 2.664,
+      "step": 2252
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017886491557223265,
+      "loss": 2.5241,
+      "step": 2253
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017885553470919324,
+      "loss": 2.5283,
+      "step": 2254
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017884615384615386,
+      "loss": 2.8066,
+      "step": 2255
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017883677298311445,
+      "loss": 2.6937,
+      "step": 2256
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017882739212007507,
+      "loss": 3.0494,
+      "step": 2257
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017881801125703566,
+      "loss": 2.7791,
+      "step": 2258
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017880863039399625,
+      "loss": 2.485,
+      "step": 2259
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017879924953095687,
+      "loss": 2.8198,
+      "step": 2260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017878986866791744,
+      "loss": 2.644,
+      "step": 2261
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017878048780487805,
+      "loss": 2.3981,
+      "step": 2262
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017877110694183864,
+      "loss": 2.7598,
+      "step": 2263
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017876172607879926,
+      "loss": 2.4037,
+      "step": 2264
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017875234521575985,
+      "loss": 2.7524,
+      "step": 2265
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017874296435272047,
+      "loss": 2.6362,
+      "step": 2266
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017873358348968106,
+      "loss": 2.7655,
+      "step": 2267
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017872420262664168,
+      "loss": 2.4043,
+      "step": 2268
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017871482176360227,
+      "loss": 2.4793,
+      "step": 2269
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00017870544090056286,
+      "loss": 3.0624,
+      "step": 2270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017869606003752345,
+      "loss": 2.8728,
+      "step": 2271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017868667917448404,
+      "loss": 2.7734,
+      "step": 2272
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017867729831144466,
+      "loss": 2.7265,
+      "step": 2273
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017866791744840525,
+      "loss": 2.9121,
+      "step": 2274
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017865853658536587,
+      "loss": 2.7132,
+      "step": 2275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017864915572232646,
+      "loss": 2.7305,
+      "step": 2276
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017863977485928708,
+      "loss": 2.6877,
+      "step": 2277
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017863039399624767,
+      "loss": 2.6331,
+      "step": 2278
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017862101313320826,
+      "loss": 2.7625,
+      "step": 2279
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017861163227016885,
+      "loss": 2.371,
+      "step": 2280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017860225140712947,
+      "loss": 2.8966,
+      "step": 2281
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017859287054409006,
+      "loss": 2.6959,
+      "step": 2282
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017858348968105068,
+      "loss": 2.5004,
+      "step": 2283
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017857410881801127,
+      "loss": 2.2999,
+      "step": 2284
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017856472795497186,
+      "loss": 2.6924,
+      "step": 2285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017855534709193248,
+      "loss": 2.6768,
+      "step": 2286
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017854596622889307,
+      "loss": 2.8324,
+      "step": 2287
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017853658536585366,
+      "loss": 2.738,
+      "step": 2288
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017852720450281425,
+      "loss": 2.8706,
+      "step": 2289
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017851782363977487,
+      "loss": 2.8017,
+      "step": 2290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 0.00017850844277673546,
+      "loss": 2.6967,
+      "step": 2291
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017849906191369608,
+      "loss": 2.7913,
+      "step": 2292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017848968105065667,
+      "loss": 2.6873,
+      "step": 2293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001784803001876173,
+      "loss": 2.6698,
+      "step": 2294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017847091932457788,
+      "loss": 2.5835,
+      "step": 2295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017846153846153847,
+      "loss": 2.9892,
+      "step": 2296
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017845215759849906,
+      "loss": 2.8762,
+      "step": 2297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017844277673545965,
+      "loss": 2.6872,
+      "step": 2298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017843339587242027,
+      "loss": 2.5452,
+      "step": 2299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017842401500938086,
+      "loss": 2.8198,
+      "step": 2300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017841463414634148,
+      "loss": 2.5954,
+      "step": 2301
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017840525328330207,
+      "loss": 2.5404,
+      "step": 2302
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001783958724202627,
+      "loss": 3.0363,
+      "step": 2303
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017838649155722328,
+      "loss": 2.7067,
+      "step": 2304
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017837711069418387,
+      "loss": 2.6618,
+      "step": 2305
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017836772983114446,
+      "loss": 2.7046,
+      "step": 2306
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017835834896810508,
+      "loss": 2.7796,
+      "step": 2307
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017834896810506567,
+      "loss": 3.1561,
+      "step": 2308
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001783395872420263,
+      "loss": 2.569,
+      "step": 2309
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017833020637898688,
+      "loss": 2.7567,
+      "step": 2310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017832082551594747,
+      "loss": 2.6724,
+      "step": 2311
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.0001783114446529081,
+      "loss": 2.6,
+      "step": 2312
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00017830206378986868,
+      "loss": 2.7089,
+      "step": 2313
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017829268292682927,
+      "loss": 2.8945,
+      "step": 2314
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017828330206378986,
+      "loss": 3.0929,
+      "step": 2315
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017827392120075048,
+      "loss": 2.7681,
+      "step": 2316
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017826454033771107,
+      "loss": 2.5842,
+      "step": 2317
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001782551594746717,
+      "loss": 2.6671,
+      "step": 2318
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017824577861163228,
+      "loss": 2.636,
+      "step": 2319
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001782363977485929,
+      "loss": 2.9599,
+      "step": 2320
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001782270168855535,
+      "loss": 2.6924,
+      "step": 2321
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017821763602251408,
+      "loss": 2.5855,
+      "step": 2322
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017820825515947467,
+      "loss": 2.9112,
+      "step": 2323
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017819887429643526,
+      "loss": 2.8697,
+      "step": 2324
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017818949343339588,
+      "loss": 2.8988,
+      "step": 2325
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017818011257035647,
+      "loss": 2.5526,
+      "step": 2326
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001781707317073171,
+      "loss": 3.1579,
+      "step": 2327
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017816135084427768,
+      "loss": 2.7543,
+      "step": 2328
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001781519699812383,
+      "loss": 2.7762,
+      "step": 2329
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001781425891181989,
+      "loss": 2.7674,
+      "step": 2330
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017813320825515948,
+      "loss": 3.0779,
+      "step": 2331
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017812382739212007,
+      "loss": 2.8276,
+      "step": 2332
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001781144465290807,
+      "loss": 2.6991,
+      "step": 2333
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.00017810506566604128,
+      "loss": 2.6501,
+      "step": 2334
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001780956848030019,
+      "loss": 2.879,
+      "step": 2335
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001780863039399625,
+      "loss": 2.4199,
+      "step": 2336
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017807692307692308,
+      "loss": 2.7083,
+      "step": 2337
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001780675422138837,
+      "loss": 3.0296,
+      "step": 2338
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001780581613508443,
+      "loss": 3.1255,
+      "step": 2339
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017804878048780488,
+      "loss": 2.8598,
+      "step": 2340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017803939962476547,
+      "loss": 2.7143,
+      "step": 2341
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001780300187617261,
+      "loss": 3.0119,
+      "step": 2342
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017802063789868668,
+      "loss": 2.7188,
+      "step": 2343
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001780112570356473,
+      "loss": 2.8554,
+      "step": 2344
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001780018761726079,
+      "loss": 2.6639,
+      "step": 2345
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001779924953095685,
+      "loss": 2.7565,
+      "step": 2346
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001779831144465291,
+      "loss": 2.7593,
+      "step": 2347
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001779737335834897,
+      "loss": 2.8004,
+      "step": 2348
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017796435272045028,
+      "loss": 2.9194,
+      "step": 2349
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017795497185741087,
+      "loss": 2.6366,
+      "step": 2350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001779455909943715,
+      "loss": 2.8047,
+      "step": 2351
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00017793621013133208,
+      "loss": 2.9868,
+      "step": 2352
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001779268292682927,
+      "loss": 2.5417,
+      "step": 2353
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001779174484052533,
+      "loss": 2.7013,
+      "step": 2354
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.0001779080675422139,
+      "loss": 2.7854,
+      "step": 2355
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001778986866791745,
+      "loss": 3.1739,
+      "step": 2356
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001778893058161351,
+      "loss": 2.7248,
+      "step": 2357
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00017787992495309568,
+      "loss": 2.8745,
+      "step": 2358
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001778705440900563,
+      "loss": 2.9649,
+      "step": 2359
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001778611632270169,
+      "loss": 2.5285,
+      "step": 2360
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001778517823639775,
+      "loss": 2.8288,
+      "step": 2361
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001778424015009381,
+      "loss": 2.5702,
+      "step": 2362
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00017783302063789871,
+      "loss": 2.7737,
+      "step": 2363
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001778236397748593,
+      "loss": 2.9733,
+      "step": 2364
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001778142589118199,
+      "loss": 2.9549,
+      "step": 2365
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001778048780487805,
+      "loss": 2.3753,
+      "step": 2366
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00017779549718574108,
+      "loss": 2.7128,
+      "step": 2367
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001777861163227017,
+      "loss": 2.7607,
+      "step": 2368
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001777767354596623,
+      "loss": 2.6074,
+      "step": 2369
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001777673545966229,
+      "loss": 2.7784,
+      "step": 2370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001777579737335835,
+      "loss": 2.7712,
+      "step": 2371
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.00017774859287054411,
+      "loss": 2.7641,
+      "step": 2372
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001777392120075047,
+      "loss": 2.7658,
+      "step": 2373
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001777298311444653,
+      "loss": 2.6893,
+      "step": 2374
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001777204502814259,
+      "loss": 2.7332,
+      "step": 2375
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001777110694183865,
+      "loss": 2.5995,
+      "step": 2376
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 0.0001777016885553471,
+      "loss": 2.5986,
+      "step": 2377
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001776923076923077,
+      "loss": 3.1733,
+      "step": 2378
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001776829268292683,
+      "loss": 2.8951,
+      "step": 2379
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001776735459662289,
+      "loss": 2.7755,
+      "step": 2380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00017766416510318951,
+      "loss": 2.6839,
+      "step": 2381
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001776547842401501,
+      "loss": 2.9711,
+      "step": 2382
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001776454033771107,
+      "loss": 2.7745,
+      "step": 2383
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001776360225140713,
+      "loss": 2.9178,
+      "step": 2384
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001776266416510319,
+      "loss": 2.9323,
+      "step": 2385
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001776172607879925,
+      "loss": 3.0357,
+      "step": 2386
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00017760787992495311,
+      "loss": 3.0192,
+      "step": 2387
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001775984990619137,
+      "loss": 2.7253,
+      "step": 2388
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00017758911819887432,
+      "loss": 2.9468,
+      "step": 2389
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00017757973733583491,
+      "loss": 2.6735,
+      "step": 2390
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001775703564727955,
+      "loss": 2.7393,
+      "step": 2391
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001775609756097561,
+      "loss": 2.8566,
+      "step": 2392
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001775515947467167,
+      "loss": 2.5949,
+      "step": 2393
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001775422138836773,
+      "loss": 2.7605,
+      "step": 2394
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001775328330206379,
+      "loss": 3.0126,
+      "step": 2395
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00017752345215759851,
+      "loss": 2.704,
+      "step": 2396
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.0001775140712945591,
+      "loss": 2.8355,
+      "step": 2397
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00017750469043151972,
+      "loss": 2.7209,
+      "step": 2398
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00017749530956848031,
+      "loss": 2.6583,
+      "step": 2399
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001774859287054409,
+      "loss": 2.9157,
+      "step": 2400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001774765478424015,
+      "loss": 2.9289,
+      "step": 2401
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00017746716697936211,
+      "loss": 2.4973,
+      "step": 2402
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001774577861163227,
+      "loss": 2.806,
+      "step": 2403
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001774484052532833,
+      "loss": 2.623,
+      "step": 2404
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00017743902439024391,
+      "loss": 2.6999,
+      "step": 2405
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001774296435272045,
+      "loss": 2.9798,
+      "step": 2406
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00017742026266416512,
+      "loss": 2.6563,
+      "step": 2407
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00017741088180112571,
+      "loss": 2.684,
+      "step": 2408
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001774015009380863,
+      "loss": 2.7437,
+      "step": 2409
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001773921200750469,
+      "loss": 2.9457,
+      "step": 2410
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001773827392120075,
+      "loss": 2.7264,
+      "step": 2411
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001773733583489681,
+      "loss": 2.9595,
+      "step": 2412
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00017736397748592872,
+      "loss": 2.5764,
+      "step": 2413
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001773545966228893,
+      "loss": 2.987,
+      "step": 2414
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00017734521575984993,
+      "loss": 2.6527,
+      "step": 2415
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00017733583489681052,
+      "loss": 2.7822,
+      "step": 2416
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001773264540337711,
+      "loss": 2.5965,
+      "step": 2417
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001773170731707317,
+      "loss": 2.6675,
+      "step": 2418
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.0001773076923076923,
+      "loss": 2.7947,
+      "step": 2419
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001772983114446529,
+      "loss": 3.0805,
+      "step": 2420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001772889305816135,
+      "loss": 2.6191,
+      "step": 2421
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017727954971857412,
+      "loss": 2.559,
+      "step": 2422
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001772701688555347,
+      "loss": 2.619,
+      "step": 2423
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017726078799249533,
+      "loss": 2.8194,
+      "step": 2424
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017725140712945592,
+      "loss": 2.7315,
+      "step": 2425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017724202626641654,
+      "loss": 2.8533,
+      "step": 2426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017723264540337713,
+      "loss": 2.6403,
+      "step": 2427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017722326454033772,
+      "loss": 2.861,
+      "step": 2428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001772138836772983,
+      "loss": 2.7439,
+      "step": 2429
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001772045028142589,
+      "loss": 2.9462,
+      "step": 2430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017719512195121952,
+      "loss": 3.0037,
+      "step": 2431
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001771857410881801,
+      "loss": 3.061,
+      "step": 2432
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017717636022514073,
+      "loss": 2.6754,
+      "step": 2433
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017716697936210132,
+      "loss": 2.5314,
+      "step": 2434
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017715759849906194,
+      "loss": 3.2069,
+      "step": 2435
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017714821763602253,
+      "loss": 2.4438,
+      "step": 2436
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017713883677298312,
+      "loss": 2.6984,
+      "step": 2437
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001771294559099437,
+      "loss": 3.0627,
+      "step": 2438
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017712007504690433,
+      "loss": 2.7072,
+      "step": 2439
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017711069418386492,
+      "loss": 2.7657,
+      "step": 2440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.00017710131332082554,
+      "loss": 2.9524,
+      "step": 2441
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017709193245778613,
+      "loss": 2.6442,
+      "step": 2442
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017708255159474672,
+      "loss": 2.8492,
+      "step": 2443
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017707317073170734,
+      "loss": 2.5956,
+      "step": 2444
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017706378986866793,
+      "loss": 2.7097,
+      "step": 2445
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017705440900562852,
+      "loss": 2.7155,
+      "step": 2446
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001770450281425891,
+      "loss": 2.5806,
+      "step": 2447
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017703564727954973,
+      "loss": 2.6176,
+      "step": 2448
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017702626641651032,
+      "loss": 2.9836,
+      "step": 2449
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017701688555347094,
+      "loss": 2.5357,
+      "step": 2450
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017700750469043153,
+      "loss": 2.6963,
+      "step": 2451
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017699812382739215,
+      "loss": 2.8154,
+      "step": 2452
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017698874296435274,
+      "loss": 2.6037,
+      "step": 2453
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017697936210131333,
+      "loss": 2.4801,
+      "step": 2454
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017696998123827392,
+      "loss": 2.8905,
+      "step": 2455
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.0001769606003752345,
+      "loss": 2.895,
+      "step": 2456
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017695121951219513,
+      "loss": 2.9593,
+      "step": 2457
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017694183864915572,
+      "loss": 2.4732,
+      "step": 2458
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017693245778611634,
+      "loss": 2.398,
+      "step": 2459
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017692307692307693,
+      "loss": 2.798,
+      "step": 2460
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017691369606003755,
+      "loss": 2.7656,
+      "step": 2461
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00017690431519699814,
+      "loss": 2.7412,
+      "step": 2462
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017689493433395873,
+      "loss": 2.7695,
+      "step": 2463
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017688555347091932,
+      "loss": 2.7461,
+      "step": 2464
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017687617260787994,
+      "loss": 2.7034,
+      "step": 2465
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017686679174484053,
+      "loss": 2.7636,
+      "step": 2466
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017685741088180115,
+      "loss": 2.7915,
+      "step": 2467
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017684803001876174,
+      "loss": 2.9637,
+      "step": 2468
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017683864915572233,
+      "loss": 3.0826,
+      "step": 2469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017682926829268295,
+      "loss": 2.6211,
+      "step": 2470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017681988742964354,
+      "loss": 2.4961,
+      "step": 2471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017681050656660413,
+      "loss": 2.735,
+      "step": 2472
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017680112570356472,
+      "loss": 2.6609,
+      "step": 2473
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017679174484052534,
+      "loss": 2.7821,
+      "step": 2474
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017678236397748593,
+      "loss": 2.8392,
+      "step": 2475
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017677298311444655,
+      "loss": 3.0648,
+      "step": 2476
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017676360225140714,
+      "loss": 2.7357,
+      "step": 2477
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017675422138836776,
+      "loss": 2.6379,
+      "step": 2478
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017674484052532835,
+      "loss": 2.6351,
+      "step": 2479
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017673545966228894,
+      "loss": 2.7124,
+      "step": 2480
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017672607879924953,
+      "loss": 2.9283,
+      "step": 2481
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017671669793621012,
+      "loss": 2.8275,
+      "step": 2482
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 0.00017670731707317074,
+      "loss": 2.6428,
+      "step": 2483
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017669793621013133,
+      "loss": 2.742,
+      "step": 2484
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017668855534709195,
+      "loss": 2.5269,
+      "step": 2485
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017667917448405254,
+      "loss": 2.6529,
+      "step": 2486
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017666979362101316,
+      "loss": 2.8457,
+      "step": 2487
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017666041275797375,
+      "loss": 2.4951,
+      "step": 2488
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017665103189493434,
+      "loss": 2.8297,
+      "step": 2489
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017664165103189493,
+      "loss": 2.8786,
+      "step": 2490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017663227016885555,
+      "loss": 2.6327,
+      "step": 2491
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017662288930581614,
+      "loss": 2.7587,
+      "step": 2492
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017661350844277676,
+      "loss": 2.4651,
+      "step": 2493
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017660412757973735,
+      "loss": 2.6157,
+      "step": 2494
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017659474671669794,
+      "loss": 2.8738,
+      "step": 2495
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017658536585365856,
+      "loss": 2.5844,
+      "step": 2496
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017657598499061915,
+      "loss": 2.9778,
+      "step": 2497
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017656660412757974,
+      "loss": 2.7451,
+      "step": 2498
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017655722326454033,
+      "loss": 2.5241,
+      "step": 2499
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017654784240150095,
+      "loss": 2.651,
+      "step": 2500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017653846153846154,
+      "loss": 2.7259,
+      "step": 2501
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017652908067542216,
+      "loss": 2.7125,
+      "step": 2502
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017651969981238275,
+      "loss": 2.8434,
+      "step": 2503
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017651031894934337,
+      "loss": 2.5683,
+      "step": 2504
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00017650093808630396,
+      "loss": 2.5857,
+      "step": 2505
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017649155722326455,
+      "loss": 2.4705,
+      "step": 2506
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017648217636022514,
+      "loss": 2.6284,
+      "step": 2507
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017647279549718573,
+      "loss": 2.6425,
+      "step": 2508
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017646341463414635,
+      "loss": 2.5557,
+      "step": 2509
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017645403377110694,
+      "loss": 2.778,
+      "step": 2510
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017644465290806756,
+      "loss": 2.8515,
+      "step": 2511
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017643527204502815,
+      "loss": 2.7194,
+      "step": 2512
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017642589118198877,
+      "loss": 2.7838,
+      "step": 2513
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017641651031894936,
+      "loss": 2.674,
+      "step": 2514
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017640712945590995,
+      "loss": 2.6284,
+      "step": 2515
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017639774859287054,
+      "loss": 2.7784,
+      "step": 2516
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017638836772983116,
+      "loss": 2.7481,
+      "step": 2517
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017637898686679175,
+      "loss": 2.3922,
+      "step": 2518
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017636960600375237,
+      "loss": 2.7599,
+      "step": 2519
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017636022514071296,
+      "loss": 2.7616,
+      "step": 2520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017635084427767355,
+      "loss": 2.5856,
+      "step": 2521
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017634146341463417,
+      "loss": 2.7235,
+      "step": 2522
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017633208255159476,
+      "loss": 2.7965,
+      "step": 2523
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017632270168855535,
+      "loss": 2.5772,
+      "step": 2524
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017631332082551594,
+      "loss": 2.9245,
+      "step": 2525
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00017630393996247656,
+      "loss": 2.6164,
+      "step": 2526
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017629455909943715,
+      "loss": 2.9337,
+      "step": 2527
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017628517823639777,
+      "loss": 2.9386,
+      "step": 2528
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017627579737335836,
+      "loss": 2.817,
+      "step": 2529
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017626641651031897,
+      "loss": 2.9091,
+      "step": 2530
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017625703564727957,
+      "loss": 2.5476,
+      "step": 2531
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017624765478424016,
+      "loss": 2.8553,
+      "step": 2532
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017623827392120075,
+      "loss": 2.5257,
+      "step": 2533
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017622889305816134,
+      "loss": 2.7594,
+      "step": 2534
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017621951219512196,
+      "loss": 2.7115,
+      "step": 2535
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017621013133208255,
+      "loss": 2.664,
+      "step": 2536
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017620075046904316,
+      "loss": 2.7755,
+      "step": 2537
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017619136960600376,
+      "loss": 2.7635,
+      "step": 2538
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017618198874296437,
+      "loss": 2.522,
+      "step": 2539
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017617260787992496,
+      "loss": 2.6564,
+      "step": 2540
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017616322701688556,
+      "loss": 2.5656,
+      "step": 2541
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017615384615384615,
+      "loss": 2.6601,
+      "step": 2542
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017614446529080676,
+      "loss": 2.7825,
+      "step": 2543
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017613508442776736,
+      "loss": 3.0149,
+      "step": 2544
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017612570356472797,
+      "loss": 2.9827,
+      "step": 2545
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017611632270168856,
+      "loss": 3.0486,
+      "step": 2546
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.00017610694183864916,
+      "loss": 2.6727,
+      "step": 2547
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017609756097560977,
+      "loss": 2.6107,
+      "step": 2548
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017608818011257036,
+      "loss": 3.0127,
+      "step": 2549
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017607879924953096,
+      "loss": 2.7042,
+      "step": 2550
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017606941838649155,
+      "loss": 3.0155,
+      "step": 2551
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017606003752345216,
+      "loss": 2.618,
+      "step": 2552
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017605065666041276,
+      "loss": 2.8955,
+      "step": 2553
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017604127579737337,
+      "loss": 2.8098,
+      "step": 2554
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017603189493433396,
+      "loss": 2.75,
+      "step": 2555
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017602251407129458,
+      "loss": 2.9045,
+      "step": 2556
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017601313320825517,
+      "loss": 2.4232,
+      "step": 2557
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017600375234521576,
+      "loss": 2.852,
+      "step": 2558
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017599437148217636,
+      "loss": 2.7031,
+      "step": 2559
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017598499061913695,
+      "loss": 2.6561,
+      "step": 2560
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017597560975609756,
+      "loss": 2.9355,
+      "step": 2561
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017596622889305816,
+      "loss": 2.8632,
+      "step": 2562
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017595684803001877,
+      "loss": 2.9418,
+      "step": 2563
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017594746716697936,
+      "loss": 2.9582,
+      "step": 2564
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017593808630393998,
+      "loss": 2.4873,
+      "step": 2565
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017592870544090057,
+      "loss": 2.7093,
+      "step": 2566
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017591932457786116,
+      "loss": 2.7906,
+      "step": 2567
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017590994371482176,
+      "loss": 2.6887,
+      "step": 2568
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 0.00017590056285178237,
+      "loss": 2.9915,
+      "step": 2569
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017589118198874296,
+      "loss": 2.7961,
+      "step": 2570
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017588180112570358,
+      "loss": 2.9589,
+      "step": 2571
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017587242026266417,
+      "loss": 2.9656,
+      "step": 2572
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001758630393996248,
+      "loss": 2.6804,
+      "step": 2573
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017585365853658538,
+      "loss": 2.365,
+      "step": 2574
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017584427767354597,
+      "loss": 2.6737,
+      "step": 2575
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017583489681050656,
+      "loss": 3.1079,
+      "step": 2576
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017582551594746716,
+      "loss": 2.691,
+      "step": 2577
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017581613508442777,
+      "loss": 2.8584,
+      "step": 2578
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017580675422138836,
+      "loss": 2.7991,
+      "step": 2579
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017579737335834898,
+      "loss": 2.8884,
+      "step": 2580
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017578799249530957,
+      "loss": 2.3977,
+      "step": 2581
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001757786116322702,
+      "loss": 2.6151,
+      "step": 2582
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017576923076923078,
+      "loss": 2.8109,
+      "step": 2583
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.0001757598499061914,
+      "loss": 2.7873,
+      "step": 2584
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017575046904315196,
+      "loss": 2.5677,
+      "step": 2585
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017574108818011258,
+      "loss": 2.8443,
+      "step": 2586
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017573170731707317,
+      "loss": 2.5472,
+      "step": 2587
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017572232645403376,
+      "loss": 2.6754,
+      "step": 2588
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017571294559099438,
+      "loss": 2.7304,
+      "step": 2589
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00017570356472795497,
+      "loss": 2.77,
+      "step": 2590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001756941838649156,
+      "loss": 2.3616,
+      "step": 2591
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017568480300187618,
+      "loss": 2.84,
+      "step": 2592
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001756754221388368,
+      "loss": 2.7774,
+      "step": 2593
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001756660412757974,
+      "loss": 2.8314,
+      "step": 2594
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017565666041275798,
+      "loss": 2.6921,
+      "step": 2595
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017564727954971857,
+      "loss": 2.6845,
+      "step": 2596
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001756378986866792,
+      "loss": 2.5985,
+      "step": 2597
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017562851782363978,
+      "loss": 2.647,
+      "step": 2598
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001756191369606004,
+      "loss": 2.6467,
+      "step": 2599
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000175609756097561,
+      "loss": 2.5385,
+      "step": 2600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017560037523452158,
+      "loss": 2.6171,
+      "step": 2601
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001755909943714822,
+      "loss": 2.878,
+      "step": 2602
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001755816135084428,
+      "loss": 2.9592,
+      "step": 2603
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017557223264540338,
+      "loss": 2.5698,
+      "step": 2604
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017556285178236397,
+      "loss": 2.4511,
+      "step": 2605
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001755534709193246,
+      "loss": 2.6617,
+      "step": 2606
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00017554409005628518,
+      "loss": 2.6556,
+      "step": 2607
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001755347091932458,
+      "loss": 2.7394,
+      "step": 2608
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001755253283302064,
+      "loss": 2.7747,
+      "step": 2609
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.000175515947467167,
+      "loss": 2.7034,
+      "step": 2610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.0001755065666041276,
+      "loss": 3.0033,
+      "step": 2611
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001754971857410882,
+      "loss": 2.5809,
+      "step": 2612
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017548780487804878,
+      "loss": 2.8828,
+      "step": 2613
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017547842401500937,
+      "loss": 2.6985,
+      "step": 2614
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017546904315197,
+      "loss": 2.422,
+      "step": 2615
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017545966228893058,
+      "loss": 2.7772,
+      "step": 2616
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001754502814258912,
+      "loss": 2.7201,
+      "step": 2617
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001754409005628518,
+      "loss": 2.6269,
+      "step": 2618
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001754315196998124,
+      "loss": 3.0177,
+      "step": 2619
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000175422138836773,
+      "loss": 2.9496,
+      "step": 2620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001754127579737336,
+      "loss": 2.7168,
+      "step": 2621
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017540337711069418,
+      "loss": 2.5694,
+      "step": 2622
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001753939962476548,
+      "loss": 2.7275,
+      "step": 2623
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001753846153846154,
+      "loss": 2.8089,
+      "step": 2624
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000175375234521576,
+      "loss": 2.5778,
+      "step": 2625
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001753658536585366,
+      "loss": 2.7088,
+      "step": 2626
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001753564727954972,
+      "loss": 2.671,
+      "step": 2627
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001753470919324578,
+      "loss": 2.7423,
+      "step": 2628
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001753377110694184,
+      "loss": 2.5767,
+      "step": 2629
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.000175328330206379,
+      "loss": 2.5133,
+      "step": 2630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00017531894934333958,
+      "loss": 2.7826,
+      "step": 2631
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001753095684803002,
+      "loss": 2.406,
+      "step": 2632
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.0001753001876172608,
+      "loss": 2.8637,
+      "step": 2633
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001752908067542214,
+      "loss": 2.977,
+      "step": 2634
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000175281425891182,
+      "loss": 2.7558,
+      "step": 2635
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017527204502814262,
+      "loss": 2.7519,
+      "step": 2636
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001752626641651032,
+      "loss": 2.4371,
+      "step": 2637
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001752532833020638,
+      "loss": 2.4316,
+      "step": 2638
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001752439024390244,
+      "loss": 2.8352,
+      "step": 2639
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017523452157598498,
+      "loss": 2.6653,
+      "step": 2640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001752251407129456,
+      "loss": 2.4015,
+      "step": 2641
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001752157598499062,
+      "loss": 2.7471,
+      "step": 2642
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001752063789868668,
+      "loss": 2.783,
+      "step": 2643
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001751969981238274,
+      "loss": 3.1554,
+      "step": 2644
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017518761726078802,
+      "loss": 2.4493,
+      "step": 2645
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001751782363977486,
+      "loss": 2.8377,
+      "step": 2646
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001751688555347092,
+      "loss": 2.4248,
+      "step": 2647
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001751594746716698,
+      "loss": 2.329,
+      "step": 2648
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001751500938086304,
+      "loss": 2.6732,
+      "step": 2649
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.000175140712945591,
+      "loss": 2.6792,
+      "step": 2650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017513133208255162,
+      "loss": 2.4037,
+      "step": 2651
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001751219512195122,
+      "loss": 2.8519,
+      "step": 2652
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001751125703564728,
+      "loss": 2.731,
+      "step": 2653
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.00017510318949343342,
+      "loss": 2.7575,
+      "step": 2654
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000175093808630394,
+      "loss": 2.8478,
+      "step": 2655
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001750844277673546,
+      "loss": 3.0658,
+      "step": 2656
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001750750469043152,
+      "loss": 2.6286,
+      "step": 2657
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001750656660412758,
+      "loss": 2.8308,
+      "step": 2658
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001750562851782364,
+      "loss": 2.4223,
+      "step": 2659
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017504690431519702,
+      "loss": 2.8435,
+      "step": 2660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001750375234521576,
+      "loss": 2.9686,
+      "step": 2661
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017502814258911823,
+      "loss": 2.571,
+      "step": 2662
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017501876172607882,
+      "loss": 2.7359,
+      "step": 2663
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001750093808630394,
+      "loss": 2.3338,
+      "step": 2664
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000175,
+      "loss": 2.6432,
+      "step": 2665
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001749906191369606,
+      "loss": 2.7189,
+      "step": 2666
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001749812382739212,
+      "loss": 2.9624,
+      "step": 2667
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001749718574108818,
+      "loss": 2.2556,
+      "step": 2668
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017496247654784242,
+      "loss": 2.7167,
+      "step": 2669
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.000174953095684803,
+      "loss": 2.8005,
+      "step": 2670
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017494371482176363,
+      "loss": 2.5642,
+      "step": 2671
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017493433395872422,
+      "loss": 2.8642,
+      "step": 2672
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001749249530956848,
+      "loss": 2.9146,
+      "step": 2673
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.0001749155722326454,
+      "loss": 2.8179,
+      "step": 2674
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 0.00017490619136960602,
+      "loss": 2.3617,
+      "step": 2675
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001748968105065666,
+      "loss": 2.5623,
+      "step": 2676
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017488742964352723,
+      "loss": 2.6273,
+      "step": 2677
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017487804878048782,
+      "loss": 2.9526,
+      "step": 2678
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001748686679174484,
+      "loss": 2.9335,
+      "step": 2679
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017485928705440902,
+      "loss": 2.9506,
+      "step": 2680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017484990619136962,
+      "loss": 2.7887,
+      "step": 2681
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001748405253283302,
+      "loss": 2.6214,
+      "step": 2682
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001748311444652908,
+      "loss": 2.5947,
+      "step": 2683
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017482176360225142,
+      "loss": 2.6309,
+      "step": 2684
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.000174812382739212,
+      "loss": 2.6745,
+      "step": 2685
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017480300187617262,
+      "loss": 2.7054,
+      "step": 2686
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017479362101313322,
+      "loss": 2.8561,
+      "step": 2687
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017478424015009383,
+      "loss": 2.5911,
+      "step": 2688
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017477485928705442,
+      "loss": 2.6642,
+      "step": 2689
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017476547842401502,
+      "loss": 2.5104,
+      "step": 2690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001747560975609756,
+      "loss": 2.4202,
+      "step": 2691
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001747467166979362,
+      "loss": 2.4022,
+      "step": 2692
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017473733583489682,
+      "loss": 3.1042,
+      "step": 2693
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.0001747279549718574,
+      "loss": 2.6442,
+      "step": 2694
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017471857410881802,
+      "loss": 2.8928,
+      "step": 2695
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00017470919324577862,
+      "loss": 2.4253,
+      "step": 2696
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017469981238273923,
+      "loss": 2.527,
+      "step": 2697
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017469043151969982,
+      "loss": 2.8034,
+      "step": 2698
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017468105065666042,
+      "loss": 3.017,
+      "step": 2699
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.000174671669793621,
+      "loss": 2.5462,
+      "step": 2700
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017466228893058162,
+      "loss": 2.7158,
+      "step": 2701
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017465290806754222,
+      "loss": 2.5705,
+      "step": 2702
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017464352720450283,
+      "loss": 2.8545,
+      "step": 2703
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017463414634146342,
+      "loss": 2.6984,
+      "step": 2704
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017462476547842402,
+      "loss": 2.7269,
+      "step": 2705
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017461538461538463,
+      "loss": 2.7389,
+      "step": 2706
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017460600375234522,
+      "loss": 2.4592,
+      "step": 2707
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017459662288930582,
+      "loss": 2.4404,
+      "step": 2708
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001745872420262664,
+      "loss": 2.6366,
+      "step": 2709
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017457786116322702,
+      "loss": 2.537,
+      "step": 2710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017456848030018762,
+      "loss": 2.5341,
+      "step": 2711
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017455909943714823,
+      "loss": 2.5857,
+      "step": 2712
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017454971857410882,
+      "loss": 2.8396,
+      "step": 2713
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017454033771106944,
+      "loss": 2.738,
+      "step": 2714
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017453095684803003,
+      "loss": 2.6693,
+      "step": 2715
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017452157598499062,
+      "loss": 2.5738,
+      "step": 2716
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.00017451219512195122,
+      "loss": 2.8307,
+      "step": 2717
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001745028142589118,
+      "loss": 2.8817,
+      "step": 2718
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017449343339587242,
+      "loss": 2.8649,
+      "step": 2719
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017448405253283302,
+      "loss": 2.5437,
+      "step": 2720
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017447467166979363,
+      "loss": 2.4885,
+      "step": 2721
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017446529080675422,
+      "loss": 2.8895,
+      "step": 2722
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017445590994371484,
+      "loss": 2.5804,
+      "step": 2723
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017444652908067543,
+      "loss": 2.532,
+      "step": 2724
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017443714821763602,
+      "loss": 2.75,
+      "step": 2725
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017442776735459661,
+      "loss": 2.5503,
+      "step": 2726
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017441838649155723,
+      "loss": 2.7162,
+      "step": 2727
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017440900562851782,
+      "loss": 2.6003,
+      "step": 2728
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017439962476547844,
+      "loss": 2.7587,
+      "step": 2729
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017439024390243903,
+      "loss": 2.8515,
+      "step": 2730
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017438086303939962,
+      "loss": 2.5379,
+      "step": 2731
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017437148217636024,
+      "loss": 2.7952,
+      "step": 2732
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017436210131332083,
+      "loss": 2.7908,
+      "step": 2733
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017435272045028142,
+      "loss": 2.806,
+      "step": 2734
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017434333958724201,
+      "loss": 2.5759,
+      "step": 2735
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017433395872420263,
+      "loss": 2.6878,
+      "step": 2736
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017432457786116322,
+      "loss": 2.7596,
+      "step": 2737
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017431519699812384,
+      "loss": 2.924,
+      "step": 2738
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00017430581613508443,
+      "loss": 2.8574,
+      "step": 2739
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017429643527204505,
+      "loss": 2.5343,
+      "step": 2740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017428705440900564,
+      "loss": 2.5614,
+      "step": 2741
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017427767354596626,
+      "loss": 2.7231,
+      "step": 2742
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017426829268292682,
+      "loss": 2.8226,
+      "step": 2743
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017425891181988741,
+      "loss": 2.9858,
+      "step": 2744
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017424953095684803,
+      "loss": 2.8521,
+      "step": 2745
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017424015009380862,
+      "loss": 2.8435,
+      "step": 2746
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017423076923076924,
+      "loss": 2.9071,
+      "step": 2747
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017422138836772983,
+      "loss": 3.0272,
+      "step": 2748
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017421200750469045,
+      "loss": 2.8263,
+      "step": 2749
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017420262664165104,
+      "loss": 2.6594,
+      "step": 2750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017419324577861166,
+      "loss": 2.5801,
+      "step": 2751
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017418386491557222,
+      "loss": 2.7521,
+      "step": 2752
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017417448405253284,
+      "loss": 2.7187,
+      "step": 2753
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017416510318949343,
+      "loss": 2.8916,
+      "step": 2754
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017415572232645405,
+      "loss": 2.6901,
+      "step": 2755
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017414634146341464,
+      "loss": 2.9323,
+      "step": 2756
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017413696060037523,
+      "loss": 2.907,
+      "step": 2757
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017412757973733585,
+      "loss": 2.6843,
+      "step": 2758
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017411819887429644,
+      "loss": 2.7675,
+      "step": 2759
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00017410881801125706,
+      "loss": 2.8587,
+      "step": 2760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017409943714821762,
+      "loss": 2.6754,
+      "step": 2761
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017409005628517824,
+      "loss": 2.63,
+      "step": 2762
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017408067542213883,
+      "loss": 2.868,
+      "step": 2763
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017407129455909945,
+      "loss": 2.768,
+      "step": 2764
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017406191369606004,
+      "loss": 2.5444,
+      "step": 2765
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017405253283302066,
+      "loss": 3.1147,
+      "step": 2766
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017404315196998125,
+      "loss": 2.8246,
+      "step": 2767
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017403377110694187,
+      "loss": 2.6623,
+      "step": 2768
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017402439024390246,
+      "loss": 2.881,
+      "step": 2769
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017401500938086305,
+      "loss": 2.6279,
+      "step": 2770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017400562851782364,
+      "loss": 2.5548,
+      "step": 2771
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017399624765478423,
+      "loss": 2.7583,
+      "step": 2772
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017398686679174485,
+      "loss": 2.7882,
+      "step": 2773
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017397748592870544,
+      "loss": 2.7563,
+      "step": 2774
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017396810506566606,
+      "loss": 2.6211,
+      "step": 2775
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017395872420262665,
+      "loss": 2.781,
+      "step": 2776
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017394934333958727,
+      "loss": 2.7469,
+      "step": 2777
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017393996247654786,
+      "loss": 2.692,
+      "step": 2778
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017393058161350845,
+      "loss": 2.9231,
+      "step": 2779
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017392120075046904,
+      "loss": 2.6219,
+      "step": 2780
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017391181988742966,
+      "loss": 2.7887,
+      "step": 2781
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 0.00017390243902439025,
+      "loss": 2.837,
+      "step": 2782
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017389305816135087,
+      "loss": 2.3381,
+      "step": 2783
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017388367729831146,
+      "loss": 2.8409,
+      "step": 2784
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017387429643527205,
+      "loss": 2.9969,
+      "step": 2785
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017386491557223267,
+      "loss": 2.58,
+      "step": 2786
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017385553470919326,
+      "loss": 2.6627,
+      "step": 2787
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017384615384615385,
+      "loss": 2.6169,
+      "step": 2788
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017383677298311444,
+      "loss": 2.6363,
+      "step": 2789
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017382739212007506,
+      "loss": 2.9219,
+      "step": 2790
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017381801125703565,
+      "loss": 2.6509,
+      "step": 2791
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017380863039399627,
+      "loss": 2.9314,
+      "step": 2792
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017379924953095686,
+      "loss": 2.7529,
+      "step": 2793
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017378986866791748,
+      "loss": 2.4649,
+      "step": 2794
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017378048780487807,
+      "loss": 2.7278,
+      "step": 2795
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017377110694183866,
+      "loss": 2.7685,
+      "step": 2796
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017376172607879925,
+      "loss": 2.7879,
+      "step": 2797
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017375234521575984,
+      "loss": 2.7789,
+      "step": 2798
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017374296435272046,
+      "loss": 2.8717,
+      "step": 2799
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017373358348968105,
+      "loss": 2.5379,
+      "step": 2800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017372420262664167,
+      "loss": 2.7643,
+      "step": 2801
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017371482176360226,
+      "loss": 2.3761,
+      "step": 2802
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00017370544090056288,
+      "loss": 2.549,
+      "step": 2803
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017369606003752347,
+      "loss": 2.5193,
+      "step": 2804
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017368667917448406,
+      "loss": 2.9015,
+      "step": 2805
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017367729831144465,
+      "loss": 2.8816,
+      "step": 2806
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017366791744840527,
+      "loss": 2.6155,
+      "step": 2807
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017365853658536586,
+      "loss": 2.5157,
+      "step": 2808
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017364915572232648,
+      "loss": 2.5676,
+      "step": 2809
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017363977485928707,
+      "loss": 2.8894,
+      "step": 2810
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017363039399624766,
+      "loss": 2.5436,
+      "step": 2811
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017362101313320828,
+      "loss": 2.3953,
+      "step": 2812
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017361163227016887,
+      "loss": 2.9903,
+      "step": 2813
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017360225140712946,
+      "loss": 3.123,
+      "step": 2814
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017359287054409005,
+      "loss": 2.9809,
+      "step": 2815
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017358348968105067,
+      "loss": 2.736,
+      "step": 2816
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017357410881801126,
+      "loss": 2.9917,
+      "step": 2817
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017356472795497188,
+      "loss": 2.8498,
+      "step": 2818
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017355534709193247,
+      "loss": 2.9127,
+      "step": 2819
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017354596622889309,
+      "loss": 2.6807,
+      "step": 2820
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017353658536585368,
+      "loss": 2.5719,
+      "step": 2821
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017352720450281427,
+      "loss": 2.8339,
+      "step": 2822
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017351782363977486,
+      "loss": 2.81,
+      "step": 2823
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.00017350844277673545,
+      "loss": 2.629,
+      "step": 2824
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017349906191369607,
+      "loss": 2.9081,
+      "step": 2825
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017348968105065666,
+      "loss": 2.8469,
+      "step": 2826
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017348030018761728,
+      "loss": 2.7125,
+      "step": 2827
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017347091932457787,
+      "loss": 2.8513,
+      "step": 2828
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017346153846153848,
+      "loss": 2.3669,
+      "step": 2829
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017345215759849908,
+      "loss": 2.7876,
+      "step": 2830
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017344277673545967,
+      "loss": 2.8502,
+      "step": 2831
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017343339587242026,
+      "loss": 2.7308,
+      "step": 2832
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017342401500938088,
+      "loss": 2.4411,
+      "step": 2833
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017341463414634147,
+      "loss": 2.7955,
+      "step": 2834
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017340525328330208,
+      "loss": 2.4387,
+      "step": 2835
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017339587242026268,
+      "loss": 2.9504,
+      "step": 2836
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017338649155722327,
+      "loss": 2.829,
+      "step": 2837
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017337711069418388,
+      "loss": 2.7204,
+      "step": 2838
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017336772983114448,
+      "loss": 2.3608,
+      "step": 2839
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017335834896810507,
+      "loss": 2.4626,
+      "step": 2840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017334896810506566,
+      "loss": 2.7792,
+      "step": 2841
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017333958724202628,
+      "loss": 2.7085,
+      "step": 2842
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017333020637898687,
+      "loss": 2.6101,
+      "step": 2843
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017332082551594748,
+      "loss": 2.884,
+      "step": 2844
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00017331144465290808,
+      "loss": 2.5912,
+      "step": 2845
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.0001733020637898687,
+      "loss": 2.5668,
+      "step": 2846
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017329268292682928,
+      "loss": 2.8099,
+      "step": 2847
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017328330206378988,
+      "loss": 2.4243,
+      "step": 2848
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017327392120075047,
+      "loss": 2.9681,
+      "step": 2849
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017326454033771106,
+      "loss": 2.7934,
+      "step": 2850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017325515947467168,
+      "loss": 2.7787,
+      "step": 2851
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017324577861163227,
+      "loss": 2.5907,
+      "step": 2852
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017323639774859288,
+      "loss": 2.7861,
+      "step": 2853
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017322701688555348,
+      "loss": 2.7653,
+      "step": 2854
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001732176360225141,
+      "loss": 2.7579,
+      "step": 2855
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017320825515947468,
+      "loss": 2.6271,
+      "step": 2856
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017319887429643528,
+      "loss": 2.2683,
+      "step": 2857
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017318949343339587,
+      "loss": 2.8784,
+      "step": 2858
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017318011257035648,
+      "loss": 2.9324,
+      "step": 2859
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017317073170731708,
+      "loss": 2.5774,
+      "step": 2860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001731613508442777,
+      "loss": 2.7155,
+      "step": 2861
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017315196998123828,
+      "loss": 3.0801,
+      "step": 2862
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017314258911819888,
+      "loss": 2.51,
+      "step": 2863
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.0001731332082551595,
+      "loss": 2.6469,
+      "step": 2864
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017312382739212008,
+      "loss": 2.7368,
+      "step": 2865
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017311444652908068,
+      "loss": 2.636,
+      "step": 2866
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 0.00017310506566604127,
+      "loss": 3.0948,
+      "step": 2867
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017309568480300188,
+      "loss": 2.8155,
+      "step": 2868
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017308630393996247,
+      "loss": 2.7643,
+      "step": 2869
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001730769230769231,
+      "loss": 2.5181,
+      "step": 2870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017306754221388368,
+      "loss": 2.5609,
+      "step": 2871
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001730581613508443,
+      "loss": 2.8349,
+      "step": 2872
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001730487804878049,
+      "loss": 2.5827,
+      "step": 2873
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017303939962476548,
+      "loss": 2.8193,
+      "step": 2874
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017303001876172607,
+      "loss": 2.3777,
+      "step": 2875
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017302063789868667,
+      "loss": 2.5061,
+      "step": 2876
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017301125703564728,
+      "loss": 2.7537,
+      "step": 2877
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017300187617260787,
+      "loss": 2.7167,
+      "step": 2878
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001729924953095685,
+      "loss": 2.8951,
+      "step": 2879
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017298311444652908,
+      "loss": 2.9841,
+      "step": 2880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001729737335834897,
+      "loss": 2.6912,
+      "step": 2881
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001729643527204503,
+      "loss": 2.8137,
+      "step": 2882
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017295497185741088,
+      "loss": 2.632,
+      "step": 2883
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017294559099437147,
+      "loss": 2.5795,
+      "step": 2884
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001729362101313321,
+      "loss": 2.5302,
+      "step": 2885
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00017292682926829268,
+      "loss": 2.3838,
+      "step": 2886
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001729174484052533,
+      "loss": 2.7179,
+      "step": 2887
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.0001729080675422139,
+      "loss": 2.6842,
+      "step": 2888
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017289868667917448,
+      "loss": 2.8273,
+      "step": 2889
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001728893058161351,
+      "loss": 2.9695,
+      "step": 2890
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001728799249530957,
+      "loss": 2.631,
+      "step": 2891
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017287054409005628,
+      "loss": 2.5815,
+      "step": 2892
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017286116322701687,
+      "loss": 2.7584,
+      "step": 2893
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001728517823639775,
+      "loss": 2.7597,
+      "step": 2894
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017284240150093808,
+      "loss": 2.6011,
+      "step": 2895
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001728330206378987,
+      "loss": 2.7372,
+      "step": 2896
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001728236397748593,
+      "loss": 2.7173,
+      "step": 2897
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001728142589118199,
+      "loss": 2.6619,
+      "step": 2898
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001728048780487805,
+      "loss": 2.6868,
+      "step": 2899
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017279549718574112,
+      "loss": 2.5498,
+      "step": 2900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017278611632270168,
+      "loss": 3.1252,
+      "step": 2901
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017277673545966227,
+      "loss": 2.9791,
+      "step": 2902
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001727673545966229,
+      "loss": 2.8496,
+      "step": 2903
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017275797373358348,
+      "loss": 2.7805,
+      "step": 2904
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001727485928705441,
+      "loss": 2.5874,
+      "step": 2905
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001727392120075047,
+      "loss": 2.657,
+      "step": 2906
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001727298311444653,
+      "loss": 2.886,
+      "step": 2907
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001727204502814259,
+      "loss": 2.8002,
+      "step": 2908
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017271106941838652,
+      "loss": 2.875,
+      "step": 2909
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.00017270168855534708,
+      "loss": 2.7533,
+      "step": 2910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001726923076923077,
+      "loss": 2.9956,
+      "step": 2911
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001726829268292683,
+      "loss": 2.8936,
+      "step": 2912
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001726735459662289,
+      "loss": 2.7823,
+      "step": 2913
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001726641651031895,
+      "loss": 2.8709,
+      "step": 2914
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001726547842401501,
+      "loss": 2.9329,
+      "step": 2915
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001726454033771107,
+      "loss": 2.4872,
+      "step": 2916
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001726360225140713,
+      "loss": 2.8178,
+      "step": 2917
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017262664165103192,
+      "loss": 2.6991,
+      "step": 2918
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017261726078799248,
+      "loss": 2.6051,
+      "step": 2919
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001726078799249531,
+      "loss": 2.835,
+      "step": 2920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001725984990619137,
+      "loss": 2.6392,
+      "step": 2921
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001725891181988743,
+      "loss": 2.6532,
+      "step": 2922
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001725797373358349,
+      "loss": 2.6429,
+      "step": 2923
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017257035647279552,
+      "loss": 3.1219,
+      "step": 2924
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001725609756097561,
+      "loss": 2.6075,
+      "step": 2925
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017255159474671673,
+      "loss": 2.9945,
+      "step": 2926
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017254221388367732,
+      "loss": 2.6863,
+      "step": 2927
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00017253283302063788,
+      "loss": 2.6071,
+      "step": 2928
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001725234521575985,
+      "loss": 2.9561,
+      "step": 2929
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001725140712945591,
+      "loss": 2.8819,
+      "step": 2930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.0001725046904315197,
+      "loss": 2.6875,
+      "step": 2931
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001724953095684803,
+      "loss": 2.6959,
+      "step": 2932
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017248592870544092,
+      "loss": 2.636,
+      "step": 2933
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001724765478424015,
+      "loss": 2.6776,
+      "step": 2934
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017246716697936213,
+      "loss": 2.5882,
+      "step": 2935
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017245778611632272,
+      "loss": 2.5588,
+      "step": 2936
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001724484052532833,
+      "loss": 2.9078,
+      "step": 2937
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001724390243902439,
+      "loss": 2.8146,
+      "step": 2938
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017242964352720452,
+      "loss": 2.6517,
+      "step": 2939
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001724202626641651,
+      "loss": 2.3733,
+      "step": 2940
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001724108818011257,
+      "loss": 2.4527,
+      "step": 2941
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017240150093808632,
+      "loss": 2.9069,
+      "step": 2942
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001723921200750469,
+      "loss": 2.8066,
+      "step": 2943
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017238273921200753,
+      "loss": 2.8338,
+      "step": 2944
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017237335834896812,
+      "loss": 2.6844,
+      "step": 2945
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001723639774859287,
+      "loss": 2.6926,
+      "step": 2946
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001723545966228893,
+      "loss": 2.9163,
+      "step": 2947
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017234521575984992,
+      "loss": 2.6944,
+      "step": 2948
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.0001723358348968105,
+      "loss": 2.886,
+      "step": 2949
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017232645403377113,
+      "loss": 2.4853,
+      "step": 2950
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017231707317073172,
+      "loss": 2.4897,
+      "step": 2951
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00017230769230769234,
+      "loss": 2.7466,
+      "step": 2952
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017229831144465293,
+      "loss": 2.5751,
+      "step": 2953
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017228893058161352,
+      "loss": 2.5717,
+      "step": 2954
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001722795497185741,
+      "loss": 2.7707,
+      "step": 2955
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001722701688555347,
+      "loss": 2.7309,
+      "step": 2956
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017226078799249532,
+      "loss": 2.3165,
+      "step": 2957
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001722514071294559,
+      "loss": 2.7589,
+      "step": 2958
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017224202626641653,
+      "loss": 2.6456,
+      "step": 2959
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017223264540337712,
+      "loss": 3.1477,
+      "step": 2960
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017222326454033774,
+      "loss": 2.6764,
+      "step": 2961
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017221388367729833,
+      "loss": 2.9063,
+      "step": 2962
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017220450281425892,
+      "loss": 2.6835,
+      "step": 2963
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001721951219512195,
+      "loss": 2.7365,
+      "step": 2964
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017218574108818013,
+      "loss": 2.5596,
+      "step": 2965
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017217636022514072,
+      "loss": 2.7755,
+      "step": 2966
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001721669793621013,
+      "loss": 2.6901,
+      "step": 2967
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017215759849906193,
+      "loss": 2.8908,
+      "step": 2968
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017214821763602252,
+      "loss": 2.7288,
+      "step": 2969
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017213883677298314,
+      "loss": 2.4949,
+      "step": 2970
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017212945590994373,
+      "loss": 2.4866,
+      "step": 2971
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017212007504690432,
+      "loss": 2.854,
+      "step": 2972
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.0001721106941838649,
+      "loss": 2.8079,
+      "step": 2973
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 0.00017210131332082553,
+      "loss": 3.0381,
+      "step": 2974
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017209193245778612,
+      "loss": 2.8287,
+      "step": 2975
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017208255159474674,
+      "loss": 2.6847,
+      "step": 2976
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017207317073170733,
+      "loss": 2.8799,
+      "step": 2977
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017206378986866794,
+      "loss": 2.9467,
+      "step": 2978
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017205440900562854,
+      "loss": 2.8992,
+      "step": 2979
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017204502814258913,
+      "loss": 2.9934,
+      "step": 2980
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017203564727954972,
+      "loss": 2.6477,
+      "step": 2981
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.0001720262664165103,
+      "loss": 2.8108,
+      "step": 2982
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017201688555347093,
+      "loss": 2.8022,
+      "step": 2983
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017200750469043152,
+      "loss": 2.6788,
+      "step": 2984
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017199812382739214,
+      "loss": 2.7339,
+      "step": 2985
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017198874296435273,
+      "loss": 2.6858,
+      "step": 2986
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017197936210131334,
+      "loss": 2.9278,
+      "step": 2987
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017196998123827394,
+      "loss": 3.0696,
+      "step": 2988
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017196060037523453,
+      "loss": 2.8182,
+      "step": 2989
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017195121951219512,
+      "loss": 2.5883,
+      "step": 2990
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017194183864915574,
+      "loss": 2.3694,
+      "step": 2991
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017193245778611633,
+      "loss": 2.5259,
+      "step": 2992
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017192307692307694,
+      "loss": 2.6945,
+      "step": 2993
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017191369606003754,
+      "loss": 2.5014,
+      "step": 2994
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00017190431519699813,
+      "loss": 2.6514,
+      "step": 2995
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017189493433395874,
+      "loss": 2.4,
+      "step": 2996
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017188555347091934,
+      "loss": 2.548,
+      "step": 2997
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017187617260787993,
+      "loss": 2.9236,
+      "step": 2998
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017186679174484052,
+      "loss": 2.6595,
+      "step": 2999
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017185741088180114,
+      "loss": 2.5533,
+      "step": 3000
+    },
+    {
+      "epoch": 1.41,
+      "eval_loss": 2.678913116455078,
+      "eval_runtime": 67.1091,
+      "eval_samples_per_second": 28.238,
+      "eval_steps_per_second": 3.532,
+      "step": 3000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017184803001876173,
+      "loss": 2.7793,
+      "step": 3001
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017183864915572234,
+      "loss": 2.5444,
+      "step": 3002
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017182926829268294,
+      "loss": 2.6881,
+      "step": 3003
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017181988742964355,
+      "loss": 3.0578,
+      "step": 3004
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017181050656660414,
+      "loss": 2.4098,
+      "step": 3005
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017180112570356474,
+      "loss": 2.8458,
+      "step": 3006
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017179174484052533,
+      "loss": 2.6158,
+      "step": 3007
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017178236397748592,
+      "loss": 2.8038,
+      "step": 3008
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017177298311444654,
+      "loss": 2.7593,
+      "step": 3009
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017176360225140713,
+      "loss": 2.7817,
+      "step": 3010
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017175422138836774,
+      "loss": 2.8711,
+      "step": 3011
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017174484052532834,
+      "loss": 2.7781,
+      "step": 3012
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017173545966228895,
+      "loss": 2.9707,
+      "step": 3013
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017172607879924954,
+      "loss": 2.6227,
+      "step": 3014
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017171669793621013,
+      "loss": 2.9498,
+      "step": 3015
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.00017170731707317073,
+      "loss": 2.8361,
+      "step": 3016
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017169793621013134,
+      "loss": 2.8943,
+      "step": 3017
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017168855534709193,
+      "loss": 2.5864,
+      "step": 3018
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017167917448405255,
+      "loss": 2.7841,
+      "step": 3019
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017166979362101314,
+      "loss": 2.5242,
+      "step": 3020
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017166041275797373,
+      "loss": 2.7418,
+      "step": 3021
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017165103189493435,
+      "loss": 2.7904,
+      "step": 3022
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017164165103189494,
+      "loss": 2.8021,
+      "step": 3023
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017163227016885553,
+      "loss": 2.937,
+      "step": 3024
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017162288930581613,
+      "loss": 2.7231,
+      "step": 3025
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017161350844277674,
+      "loss": 2.9367,
+      "step": 3026
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017160412757973733,
+      "loss": 2.4898,
+      "step": 3027
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017159474671669795,
+      "loss": 2.611,
+      "step": 3028
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017158536585365854,
+      "loss": 2.9048,
+      "step": 3029
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017157598499061916,
+      "loss": 3.0515,
+      "step": 3030
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017156660412757975,
+      "loss": 2.8316,
+      "step": 3031
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017155722326454034,
+      "loss": 2.4731,
+      "step": 3032
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017154784240150093,
+      "loss": 2.7237,
+      "step": 3033
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017153846153846153,
+      "loss": 2.5775,
+      "step": 3034
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017152908067542214,
+      "loss": 2.745,
+      "step": 3035
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017151969981238273,
+      "loss": 2.9841,
+      "step": 3036
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017151031894934335,
+      "loss": 3.002,
+      "step": 3037
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00017150093808630394,
+      "loss": 2.617,
+      "step": 3038
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017149155722326456,
+      "loss": 2.8369,
+      "step": 3039
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017148217636022515,
+      "loss": 2.853,
+      "step": 3040
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017147279549718574,
+      "loss": 2.3273,
+      "step": 3041
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017146341463414633,
+      "loss": 2.6838,
+      "step": 3042
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017145403377110695,
+      "loss": 2.538,
+      "step": 3043
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017144465290806754,
+      "loss": 2.6095,
+      "step": 3044
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017143527204502816,
+      "loss": 2.5863,
+      "step": 3045
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017142589118198875,
+      "loss": 2.8872,
+      "step": 3046
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017141651031894934,
+      "loss": 2.7005,
+      "step": 3047
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017140712945590996,
+      "loss": 2.5141,
+      "step": 3048
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017139774859287055,
+      "loss": 2.5289,
+      "step": 3049
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017138836772983114,
+      "loss": 2.6547,
+      "step": 3050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017137898686679173,
+      "loss": 2.6582,
+      "step": 3051
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017136960600375235,
+      "loss": 2.8159,
+      "step": 3052
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017136022514071294,
+      "loss": 2.8438,
+      "step": 3053
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017135084427767356,
+      "loss": 2.9122,
+      "step": 3054
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017134146341463415,
+      "loss": 2.6462,
+      "step": 3055
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017133208255159477,
+      "loss": 2.5949,
+      "step": 3056
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017132270168855536,
+      "loss": 2.8826,
+      "step": 3057
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017131332082551598,
+      "loss": 2.5963,
+      "step": 3058
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 0.00017130393996247654,
+      "loss": 2.4336,
+      "step": 3059
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017129455909943713,
+      "loss": 3.0232,
+      "step": 3060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017128517823639775,
+      "loss": 3.0003,
+      "step": 3061
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017127579737335834,
+      "loss": 2.715,
+      "step": 3062
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017126641651031896,
+      "loss": 3.0455,
+      "step": 3063
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017125703564727955,
+      "loss": 2.5397,
+      "step": 3064
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017124765478424017,
+      "loss": 2.5501,
+      "step": 3065
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017123827392120076,
+      "loss": 2.9175,
+      "step": 3066
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017122889305816138,
+      "loss": 2.6272,
+      "step": 3067
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017121951219512194,
+      "loss": 2.731,
+      "step": 3068
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017121013133208256,
+      "loss": 3.1438,
+      "step": 3069
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017120075046904315,
+      "loss": 2.8431,
+      "step": 3070
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017119136960600377,
+      "loss": 2.9129,
+      "step": 3071
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017118198874296436,
+      "loss": 2.6603,
+      "step": 3072
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017117260787992495,
+      "loss": 2.5675,
+      "step": 3073
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017116322701688557,
+      "loss": 2.5178,
+      "step": 3074
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017115384615384616,
+      "loss": 2.7317,
+      "step": 3075
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017114446529080678,
+      "loss": 2.6059,
+      "step": 3076
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017113508442776734,
+      "loss": 2.7325,
+      "step": 3077
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017112570356472796,
+      "loss": 2.7664,
+      "step": 3078
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017111632270168855,
+      "loss": 2.7549,
+      "step": 3079
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00017110694183864917,
+      "loss": 2.8152,
+      "step": 3080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017109756097560976,
+      "loss": 2.6663,
+      "step": 3081
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017108818011257038,
+      "loss": 2.6214,
+      "step": 3082
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017107879924953097,
+      "loss": 2.8116,
+      "step": 3083
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001710694183864916,
+      "loss": 2.8892,
+      "step": 3084
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017106003752345218,
+      "loss": 2.9481,
+      "step": 3085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017105065666041274,
+      "loss": 2.558,
+      "step": 3086
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017104127579737336,
+      "loss": 2.8382,
+      "step": 3087
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017103189493433395,
+      "loss": 2.7296,
+      "step": 3088
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017102251407129457,
+      "loss": 2.8102,
+      "step": 3089
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017101313320825516,
+      "loss": 2.855,
+      "step": 3090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017100375234521578,
+      "loss": 2.6151,
+      "step": 3091
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017099437148217637,
+      "loss": 2.8158,
+      "step": 3092
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.000170984990619137,
+      "loss": 2.9331,
+      "step": 3093
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017097560975609758,
+      "loss": 2.8462,
+      "step": 3094
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017096622889305817,
+      "loss": 2.6012,
+      "step": 3095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017095684803001876,
+      "loss": 2.6562,
+      "step": 3096
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017094746716697938,
+      "loss": 2.9534,
+      "step": 3097
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017093808630393997,
+      "loss": 3.2202,
+      "step": 3098
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017092870544090056,
+      "loss": 2.7568,
+      "step": 3099
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017091932457786118,
+      "loss": 2.8939,
+      "step": 3100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.00017090994371482177,
+      "loss": 2.9353,
+      "step": 3101
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001709005628517824,
+      "loss": 2.7958,
+      "step": 3102
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017089118198874298,
+      "loss": 2.31,
+      "step": 3103
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017088180112570357,
+      "loss": 2.5659,
+      "step": 3104
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017087242026266416,
+      "loss": 2.7299,
+      "step": 3105
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017086303939962478,
+      "loss": 2.7465,
+      "step": 3106
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017085365853658537,
+      "loss": 3.1164,
+      "step": 3107
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000170844277673546,
+      "loss": 2.6686,
+      "step": 3108
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017083489681050658,
+      "loss": 2.6841,
+      "step": 3109
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001708255159474672,
+      "loss": 2.776,
+      "step": 3110
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001708161350844278,
+      "loss": 2.7791,
+      "step": 3111
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017080675422138838,
+      "loss": 2.9564,
+      "step": 3112
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017079737335834897,
+      "loss": 2.3466,
+      "step": 3113
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017078799249530956,
+      "loss": 2.8656,
+      "step": 3114
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017077861163227018,
+      "loss": 2.9605,
+      "step": 3115
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017076923076923077,
+      "loss": 2.6853,
+      "step": 3116
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001707598499061914,
+      "loss": 2.4174,
+      "step": 3117
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017075046904315198,
+      "loss": 2.8485,
+      "step": 3118
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001707410881801126,
+      "loss": 2.7159,
+      "step": 3119
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.0001707317073170732,
+      "loss": 2.5873,
+      "step": 3120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017072232645403378,
+      "loss": 2.9645,
+      "step": 3121
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00017071294559099437,
+      "loss": 2.7442,
+      "step": 3122
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.000170703564727955,
+      "loss": 2.7018,
+      "step": 3123
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017069418386491558,
+      "loss": 2.9275,
+      "step": 3124
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017068480300187617,
+      "loss": 2.7435,
+      "step": 3125
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001706754221388368,
+      "loss": 2.8515,
+      "step": 3126
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017066604127579738,
+      "loss": 2.5158,
+      "step": 3127
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000170656660412758,
+      "loss": 2.7226,
+      "step": 3128
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001706472795497186,
+      "loss": 2.712,
+      "step": 3129
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017063789868667918,
+      "loss": 2.4729,
+      "step": 3130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017062851782363977,
+      "loss": 2.9757,
+      "step": 3131
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001706191369606004,
+      "loss": 2.7992,
+      "step": 3132
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017060975609756098,
+      "loss": 2.4064,
+      "step": 3133
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001706003752345216,
+      "loss": 2.6353,
+      "step": 3134
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001705909943714822,
+      "loss": 2.6061,
+      "step": 3135
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001705816135084428,
+      "loss": 2.9067,
+      "step": 3136
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.0001705722326454034,
+      "loss": 2.8703,
+      "step": 3137
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017056285178236399,
+      "loss": 2.9527,
+      "step": 3138
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017055347091932458,
+      "loss": 2.9619,
+      "step": 3139
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017054409005628517,
+      "loss": 2.6097,
+      "step": 3140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017053470919324579,
+      "loss": 2.6534,
+      "step": 3141
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017052532833020638,
+      "loss": 2.8959,
+      "step": 3142
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.000170515947467167,
+      "loss": 2.537,
+      "step": 3143
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00017050656660412759,
+      "loss": 3.1327,
+      "step": 3144
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001704971857410882,
+      "loss": 2.572,
+      "step": 3145
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001704878048780488,
+      "loss": 2.6131,
+      "step": 3146
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017047842401500939,
+      "loss": 2.5748,
+      "step": 3147
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017046904315196998,
+      "loss": 2.8521,
+      "step": 3148
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001704596622889306,
+      "loss": 2.5644,
+      "step": 3149
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017045028142589119,
+      "loss": 2.5617,
+      "step": 3150
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017044090056285178,
+      "loss": 2.7184,
+      "step": 3151
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001704315196998124,
+      "loss": 3.0365,
+      "step": 3152
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017042213883677299,
+      "loss": 2.635,
+      "step": 3153
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001704127579737336,
+      "loss": 2.9269,
+      "step": 3154
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001704033771106942,
+      "loss": 2.7477,
+      "step": 3155
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017039399624765479,
+      "loss": 2.8293,
+      "step": 3156
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017038461538461538,
+      "loss": 2.5961,
+      "step": 3157
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000170375234521576,
+      "loss": 2.3225,
+      "step": 3158
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017036585365853659,
+      "loss": 2.4833,
+      "step": 3159
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001703564727954972,
+      "loss": 2.6726,
+      "step": 3160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001703470919324578,
+      "loss": 2.6594,
+      "step": 3161
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001703377110694184,
+      "loss": 2.433,
+      "step": 3162
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.000170328330206379,
+      "loss": 2.701,
+      "step": 3163
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.0001703189493433396,
+      "loss": 2.4446,
+      "step": 3164
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017030956848030019,
+      "loss": 2.6611,
+      "step": 3165
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 0.00017030018761726078,
+      "loss": 2.6304,
+      "step": 3166
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001702908067542214,
+      "loss": 2.832,
+      "step": 3167
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017028142589118199,
+      "loss": 2.7743,
+      "step": 3168
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001702720450281426,
+      "loss": 2.4161,
+      "step": 3169
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001702626641651032,
+      "loss": 2.6128,
+      "step": 3170
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001702532833020638,
+      "loss": 2.7019,
+      "step": 3171
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001702439024390244,
+      "loss": 2.6002,
+      "step": 3172
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000170234521575985,
+      "loss": 3.1005,
+      "step": 3173
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017022514071294559,
+      "loss": 2.7425,
+      "step": 3174
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001702157598499062,
+      "loss": 2.7707,
+      "step": 3175
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001702063789868668,
+      "loss": 2.6075,
+      "step": 3176
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017019699812382739,
+      "loss": 2.6512,
+      "step": 3177
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.000170187617260788,
+      "loss": 3.0125,
+      "step": 3178
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001701782363977486,
+      "loss": 2.6634,
+      "step": 3179
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001701688555347092,
+      "loss": 2.6797,
+      "step": 3180
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001701594746716698,
+      "loss": 2.5175,
+      "step": 3181
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001701500938086304,
+      "loss": 3.0146,
+      "step": 3182
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00017014071294559099,
+      "loss": 2.3367,
+      "step": 3183
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001701313320825516,
+      "loss": 2.5247,
+      "step": 3184
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001701219512195122,
+      "loss": 2.7726,
+      "step": 3185
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001701125703564728,
+      "loss": 2.6658,
+      "step": 3186
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.0001701031894934334,
+      "loss": 2.355,
+      "step": 3187
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017009380863039402,
+      "loss": 2.5854,
+      "step": 3188
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001700844277673546,
+      "loss": 2.7396,
+      "step": 3189
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001700750469043152,
+      "loss": 2.5596,
+      "step": 3190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001700656660412758,
+      "loss": 3.0633,
+      "step": 3191
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017005628517823639,
+      "loss": 2.9406,
+      "step": 3192
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000170046904315197,
+      "loss": 2.5022,
+      "step": 3193
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001700375234521576,
+      "loss": 2.4979,
+      "step": 3194
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001700281425891182,
+      "loss": 2.6408,
+      "step": 3195
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001700187617260788,
+      "loss": 2.806,
+      "step": 3196
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017000938086303942,
+      "loss": 2.5097,
+      "step": 3197
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00017,
+      "loss": 2.6465,
+      "step": 3198
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001699906191369606,
+      "loss": 2.5482,
+      "step": 3199
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001699812382739212,
+      "loss": 2.5076,
+      "step": 3200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001699718574108818,
+      "loss": 2.8311,
+      "step": 3201
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001699624765478424,
+      "loss": 2.7215,
+      "step": 3202
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016995309568480302,
+      "loss": 2.7869,
+      "step": 3203
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001699437148217636,
+      "loss": 2.7288,
+      "step": 3204
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001699343339587242,
+      "loss": 2.8003,
+      "step": 3205
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00016992495309568482,
+      "loss": 2.8761,
+      "step": 3206
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.0001699155722326454,
+      "loss": 2.9263,
+      "step": 3207
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.000169906191369606,
+      "loss": 2.9429,
+      "step": 3208
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001698968105065666,
+      "loss": 2.7524,
+      "step": 3209
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001698874296435272,
+      "loss": 2.6806,
+      "step": 3210
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001698780487804878,
+      "loss": 2.6796,
+      "step": 3211
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016986866791744842,
+      "loss": 2.7909,
+      "step": 3212
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000169859287054409,
+      "loss": 2.7843,
+      "step": 3213
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016984990619136963,
+      "loss": 2.7157,
+      "step": 3214
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016984052532833022,
+      "loss": 2.8797,
+      "step": 3215
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016983114446529084,
+      "loss": 2.6407,
+      "step": 3216
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001698217636022514,
+      "loss": 2.7255,
+      "step": 3217
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000169812382739212,
+      "loss": 2.9456,
+      "step": 3218
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001698030018761726,
+      "loss": 2.6126,
+      "step": 3219
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001697936210131332,
+      "loss": 2.8366,
+      "step": 3220
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016978424015009382,
+      "loss": 2.8647,
+      "step": 3221
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001697748592870544,
+      "loss": 2.7379,
+      "step": 3222
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016976547842401503,
+      "loss": 3.2212,
+      "step": 3223
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016975609756097562,
+      "loss": 2.9804,
+      "step": 3224
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016974671669793624,
+      "loss": 2.8379,
+      "step": 3225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.0001697373358348968,
+      "loss": 2.613,
+      "step": 3226
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016972795497185742,
+      "loss": 2.558,
+      "step": 3227
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.000169718574108818,
+      "loss": 2.7772,
+      "step": 3228
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00016970919324577863,
+      "loss": 2.764,
+      "step": 3229
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016969981238273922,
+      "loss": 2.7303,
+      "step": 3230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001696904315196998,
+      "loss": 2.8115,
+      "step": 3231
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016968105065666043,
+      "loss": 2.8527,
+      "step": 3232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016967166979362102,
+      "loss": 2.5904,
+      "step": 3233
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016966228893058164,
+      "loss": 3.1062,
+      "step": 3234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001696529080675422,
+      "loss": 3.127,
+      "step": 3235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016964352720450282,
+      "loss": 2.4824,
+      "step": 3236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001696341463414634,
+      "loss": 2.8217,
+      "step": 3237
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016962476547842403,
+      "loss": 2.7807,
+      "step": 3238
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016961538461538462,
+      "loss": 2.8385,
+      "step": 3239
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016960600375234524,
+      "loss": 2.9902,
+      "step": 3240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016959662288930583,
+      "loss": 2.7922,
+      "step": 3241
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016958724202626645,
+      "loss": 2.5714,
+      "step": 3242
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016957786116322704,
+      "loss": 2.5704,
+      "step": 3243
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001695684803001876,
+      "loss": 2.6555,
+      "step": 3244
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016955909943714822,
+      "loss": 2.6278,
+      "step": 3245
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.0001695497185741088,
+      "loss": 2.7731,
+      "step": 3246
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016954033771106943,
+      "loss": 2.7912,
+      "step": 3247
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016953095684803002,
+      "loss": 2.8352,
+      "step": 3248
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016952157598499064,
+      "loss": 2.5883,
+      "step": 3249
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016951219512195123,
+      "loss": 2.5324,
+      "step": 3250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 0.00016950281425891185,
+      "loss": 3.0073,
+      "step": 3251
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016949343339587244,
+      "loss": 2.5758,
+      "step": 3252
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016948405253283303,
+      "loss": 2.9171,
+      "step": 3253
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016947467166979362,
+      "loss": 2.8146,
+      "step": 3254
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016946529080675424,
+      "loss": 2.6297,
+      "step": 3255
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016945590994371483,
+      "loss": 2.9102,
+      "step": 3256
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016944652908067542,
+      "loss": 2.7415,
+      "step": 3257
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016943714821763604,
+      "loss": 2.6939,
+      "step": 3258
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016942776735459663,
+      "loss": 2.698,
+      "step": 3259
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016941838649155725,
+      "loss": 2.5542,
+      "step": 3260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016940900562851784,
+      "loss": 2.5004,
+      "step": 3261
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016939962476547843,
+      "loss": 2.5766,
+      "step": 3262
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016939024390243902,
+      "loss": 2.3663,
+      "step": 3263
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016938086303939964,
+      "loss": 2.6894,
+      "step": 3264
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016937148217636023,
+      "loss": 2.7024,
+      "step": 3265
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016936210131332085,
+      "loss": 2.5683,
+      "step": 3266
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016935272045028144,
+      "loss": 2.3667,
+      "step": 3267
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016934333958724206,
+      "loss": 2.827,
+      "step": 3268
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016933395872420265,
+      "loss": 2.9044,
+      "step": 3269
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016932457786116324,
+      "loss": 2.5643,
+      "step": 3270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016931519699812383,
+      "loss": 2.7466,
+      "step": 3271
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00016930581613508442,
+      "loss": 2.7633,
+      "step": 3272
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016929643527204504,
+      "loss": 2.5183,
+      "step": 3273
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016928705440900563,
+      "loss": 2.5294,
+      "step": 3274
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016927767354596625,
+      "loss": 2.4856,
+      "step": 3275
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016926829268292684,
+      "loss": 2.7504,
+      "step": 3276
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016925891181988746,
+      "loss": 2.6776,
+      "step": 3277
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016924953095684805,
+      "loss": 2.5978,
+      "step": 3278
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016924015009380864,
+      "loss": 2.8175,
+      "step": 3279
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016923076923076923,
+      "loss": 2.8462,
+      "step": 3280
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016922138836772985,
+      "loss": 2.5172,
+      "step": 3281
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016921200750469044,
+      "loss": 2.3179,
+      "step": 3282
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016920262664165103,
+      "loss": 2.966,
+      "step": 3283
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016919324577861165,
+      "loss": 2.6992,
+      "step": 3284
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016918386491557224,
+      "loss": 2.5844,
+      "step": 3285
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016917448405253286,
+      "loss": 2.3403,
+      "step": 3286
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016916510318949345,
+      "loss": 2.7933,
+      "step": 3287
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016915572232645404,
+      "loss": 2.7571,
+      "step": 3288
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016914634146341463,
+      "loss": 2.8698,
+      "step": 3289
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016913696060037525,
+      "loss": 2.7853,
+      "step": 3290
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016912757973733584,
+      "loss": 2.5434,
+      "step": 3291
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016911819887429646,
+      "loss": 2.6172,
+      "step": 3292
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.00016910881801125705,
+      "loss": 2.4759,
+      "step": 3293
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016909943714821766,
+      "loss": 2.7916,
+      "step": 3294
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016909005628517826,
+      "loss": 2.6394,
+      "step": 3295
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016908067542213885,
+      "loss": 2.8407,
+      "step": 3296
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016907129455909944,
+      "loss": 2.6835,
+      "step": 3297
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016906191369606003,
+      "loss": 2.5883,
+      "step": 3298
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016905253283302065,
+      "loss": 2.7615,
+      "step": 3299
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016904315196998124,
+      "loss": 2.7157,
+      "step": 3300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016903377110694186,
+      "loss": 2.798,
+      "step": 3301
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016902439024390245,
+      "loss": 2.9943,
+      "step": 3302
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016901500938086306,
+      "loss": 2.674,
+      "step": 3303
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016900562851782366,
+      "loss": 2.8328,
+      "step": 3304
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016899624765478425,
+      "loss": 3.0931,
+      "step": 3305
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016898686679174484,
+      "loss": 2.8572,
+      "step": 3306
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016897748592870545,
+      "loss": 2.6116,
+      "step": 3307
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016896810506566605,
+      "loss": 2.5913,
+      "step": 3308
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016895872420262664,
+      "loss": 2.5108,
+      "step": 3309
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016894934333958725,
+      "loss": 2.6342,
+      "step": 3310
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016893996247654785,
+      "loss": 2.5162,
+      "step": 3311
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016893058161350846,
+      "loss": 2.6829,
+      "step": 3312
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016892120075046905,
+      "loss": 2.7835,
+      "step": 3313
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016891181988742965,
+      "loss": 2.8051,
+      "step": 3314
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00016890243902439024,
+      "loss": 2.7964,
+      "step": 3315
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016889305816135085,
+      "loss": 2.6301,
+      "step": 3316
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016888367729831145,
+      "loss": 2.6519,
+      "step": 3317
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016887429643527206,
+      "loss": 2.7591,
+      "step": 3318
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016886491557223265,
+      "loss": 2.606,
+      "step": 3319
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016885553470919327,
+      "loss": 2.8554,
+      "step": 3320
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016884615384615386,
+      "loss": 2.6425,
+      "step": 3321
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016883677298311445,
+      "loss": 2.8551,
+      "step": 3322
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016882739212007505,
+      "loss": 2.9051,
+      "step": 3323
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016881801125703564,
+      "loss": 2.6826,
+      "step": 3324
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016880863039399625,
+      "loss": 2.7913,
+      "step": 3325
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016879924953095685,
+      "loss": 2.7807,
+      "step": 3326
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016878986866791746,
+      "loss": 2.9735,
+      "step": 3327
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016878048780487805,
+      "loss": 2.7541,
+      "step": 3328
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016877110694183867,
+      "loss": 2.5714,
+      "step": 3329
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016876172607879926,
+      "loss": 3.1943,
+      "step": 3330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016875234521575985,
+      "loss": 2.6285,
+      "step": 3331
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016874296435272045,
+      "loss": 3.1035,
+      "step": 3332
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016873358348968106,
+      "loss": 3.0659,
+      "step": 3333
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016872420262664165,
+      "loss": 2.4533,
+      "step": 3334
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016871482176360225,
+      "loss": 2.7046,
+      "step": 3335
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00016870544090056286,
+      "loss": 2.8618,
+      "step": 3336
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016869606003752345,
+      "loss": 2.8188,
+      "step": 3337
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016868667917448407,
+      "loss": 2.7445,
+      "step": 3338
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016867729831144466,
+      "loss": 2.6032,
+      "step": 3339
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016866791744840525,
+      "loss": 2.6678,
+      "step": 3340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016865853658536585,
+      "loss": 2.6009,
+      "step": 3341
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016864915572232646,
+      "loss": 3.0442,
+      "step": 3342
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016863977485928705,
+      "loss": 2.6416,
+      "step": 3343
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016863039399624767,
+      "loss": 2.6542,
+      "step": 3344
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016862101313320826,
+      "loss": 2.6393,
+      "step": 3345
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016861163227016888,
+      "loss": 2.6088,
+      "step": 3346
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016860225140712947,
+      "loss": 2.7496,
+      "step": 3347
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016859287054409006,
+      "loss": 2.799,
+      "step": 3348
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016858348968105065,
+      "loss": 2.7073,
+      "step": 3349
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016857410881801124,
+      "loss": 2.7115,
+      "step": 3350
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016856472795497186,
+      "loss": 2.8898,
+      "step": 3351
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016855534709193245,
+      "loss": 2.7607,
+      "step": 3352
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016854596622889307,
+      "loss": 2.8326,
+      "step": 3353
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016853658536585366,
+      "loss": 2.8091,
+      "step": 3354
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016852720450281428,
+      "loss": 2.81,
+      "step": 3355
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016851782363977487,
+      "loss": 2.8279,
+      "step": 3356
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 0.00016850844277673546,
+      "loss": 2.9022,
+      "step": 3357
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016849906191369605,
+      "loss": 2.8199,
+      "step": 3358
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016848968105065667,
+      "loss": 2.6785,
+      "step": 3359
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016848030018761726,
+      "loss": 2.6314,
+      "step": 3360
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016847091932457785,
+      "loss": 2.6521,
+      "step": 3361
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016846153846153847,
+      "loss": 2.5509,
+      "step": 3362
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016845215759849906,
+      "loss": 2.5298,
+      "step": 3363
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016844277673545968,
+      "loss": 2.3979,
+      "step": 3364
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016843339587242027,
+      "loss": 2.5198,
+      "step": 3365
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016842401500938086,
+      "loss": 2.8004,
+      "step": 3366
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016841463414634145,
+      "loss": 2.2819,
+      "step": 3367
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016840525328330207,
+      "loss": 2.6506,
+      "step": 3368
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016839587242026266,
+      "loss": 2.9658,
+      "step": 3369
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016838649155722328,
+      "loss": 2.4935,
+      "step": 3370
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016837711069418387,
+      "loss": 2.5095,
+      "step": 3371
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.0001683677298311445,
+      "loss": 2.7531,
+      "step": 3372
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016835834896810508,
+      "loss": 2.6532,
+      "step": 3373
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016834896810506567,
+      "loss": 2.5548,
+      "step": 3374
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016833958724202626,
+      "loss": 2.7996,
+      "step": 3375
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016833020637898685,
+      "loss": 2.9221,
+      "step": 3376
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016832082551594747,
+      "loss": 2.9284,
+      "step": 3377
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016831144465290806,
+      "loss": 2.5078,
+      "step": 3378
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00016830206378986868,
+      "loss": 2.87,
+      "step": 3379
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016829268292682927,
+      "loss": 2.8679,
+      "step": 3380
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001682833020637899,
+      "loss": 2.8953,
+      "step": 3381
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016827392120075048,
+      "loss": 2.356,
+      "step": 3382
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001682645403377111,
+      "loss": 2.7207,
+      "step": 3383
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016825515947467166,
+      "loss": 2.9945,
+      "step": 3384
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016824577861163228,
+      "loss": 2.7043,
+      "step": 3385
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016823639774859287,
+      "loss": 2.7283,
+      "step": 3386
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001682270168855535,
+      "loss": 2.5132,
+      "step": 3387
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016821763602251408,
+      "loss": 2.6678,
+      "step": 3388
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016820825515947467,
+      "loss": 2.6594,
+      "step": 3389
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001681988742964353,
+      "loss": 2.707,
+      "step": 3390
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016818949343339588,
+      "loss": 2.7029,
+      "step": 3391
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001681801125703565,
+      "loss": 2.6919,
+      "step": 3392
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016817073170731706,
+      "loss": 2.7511,
+      "step": 3393
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016816135084427768,
+      "loss": 2.6862,
+      "step": 3394
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016815196998123827,
+      "loss": 2.7062,
+      "step": 3395
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001681425891181989,
+      "loss": 2.4869,
+      "step": 3396
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.00016813320825515948,
+      "loss": 2.8051,
+      "step": 3397
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001681238273921201,
+      "loss": 2.8148,
+      "step": 3398
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001681144465290807,
+      "loss": 3.1255,
+      "step": 3399
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001681050656660413,
+      "loss": 2.382,
+      "step": 3400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001680956848030019,
+      "loss": 2.9561,
+      "step": 3401
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016808630393996246,
+      "loss": 2.7397,
+      "step": 3402
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016807692307692308,
+      "loss": 2.6852,
+      "step": 3403
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016806754221388367,
+      "loss": 2.6811,
+      "step": 3404
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001680581613508443,
+      "loss": 2.9113,
+      "step": 3405
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016804878048780488,
+      "loss": 2.7597,
+      "step": 3406
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001680393996247655,
+      "loss": 2.6523,
+      "step": 3407
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001680300187617261,
+      "loss": 2.875,
+      "step": 3408
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001680206378986867,
+      "loss": 2.6835,
+      "step": 3409
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001680112570356473,
+      "loss": 2.6317,
+      "step": 3410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001680018761726079,
+      "loss": 2.722,
+      "step": 3411
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016799249530956848,
+      "loss": 2.7763,
+      "step": 3412
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001679831144465291,
+      "loss": 2.6403,
+      "step": 3413
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001679737335834897,
+      "loss": 2.1954,
+      "step": 3414
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016796435272045028,
+      "loss": 2.8,
+      "step": 3415
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001679549718574109,
+      "loss": 2.5908,
+      "step": 3416
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001679455909943715,
+      "loss": 2.8332,
+      "step": 3417
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001679362101313321,
+      "loss": 2.8429,
+      "step": 3418
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001679268292682927,
+      "loss": 2.614,
+      "step": 3419
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.0001679174484052533,
+      "loss": 2.7432,
+      "step": 3420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00016790806754221388,
+      "loss": 2.4398,
+      "step": 3421
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001678986866791745,
+      "loss": 2.8095,
+      "step": 3422
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001678893058161351,
+      "loss": 2.8508,
+      "step": 3423
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001678799249530957,
+      "loss": 2.6751,
+      "step": 3424
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001678705440900563,
+      "loss": 2.9275,
+      "step": 3425
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00016786116322701692,
+      "loss": 2.4208,
+      "step": 3426
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001678517823639775,
+      "loss": 2.4663,
+      "step": 3427
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001678424015009381,
+      "loss": 3.0583,
+      "step": 3428
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001678330206378987,
+      "loss": 2.7259,
+      "step": 3429
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00016782363977485928,
+      "loss": 2.7235,
+      "step": 3430
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001678142589118199,
+      "loss": 2.8606,
+      "step": 3431
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001678048780487805,
+      "loss": 2.825,
+      "step": 3432
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001677954971857411,
+      "loss": 2.5113,
+      "step": 3433
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001677861163227017,
+      "loss": 2.782,
+      "step": 3434
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.00016777673545966232,
+      "loss": 2.5811,
+      "step": 3435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001677673545966229,
+      "loss": 2.6539,
+      "step": 3436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001677579737335835,
+      "loss": 2.8292,
+      "step": 3437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001677485928705441,
+      "loss": 2.6189,
+      "step": 3438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001677392120075047,
+      "loss": 2.8311,
+      "step": 3439
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001677298311444653,
+      "loss": 2.5585,
+      "step": 3440
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001677204502814259,
+      "loss": 2.5255,
+      "step": 3441
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001677110694183865,
+      "loss": 2.5389,
+      "step": 3442
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 0.0001677016885553471,
+      "loss": 2.5739,
+      "step": 3443
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016769230769230772,
+      "loss": 2.6993,
+      "step": 3444
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001676829268292683,
+      "loss": 2.9356,
+      "step": 3445
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001676735459662289,
+      "loss": 2.7434,
+      "step": 3446
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001676641651031895,
+      "loss": 2.726,
+      "step": 3447
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001676547842401501,
+      "loss": 2.5775,
+      "step": 3448
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001676454033771107,
+      "loss": 2.9956,
+      "step": 3449
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016763602251407132,
+      "loss": 2.3254,
+      "step": 3450
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001676266416510319,
+      "loss": 2.8907,
+      "step": 3451
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016761726078799252,
+      "loss": 2.519,
+      "step": 3452
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016760787992495311,
+      "loss": 2.4202,
+      "step": 3453
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001675984990619137,
+      "loss": 2.7241,
+      "step": 3454
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001675891181988743,
+      "loss": 2.6497,
+      "step": 3455
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001675797373358349,
+      "loss": 2.6708,
+      "step": 3456
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001675703564727955,
+      "loss": 2.7781,
+      "step": 3457
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001675609756097561,
+      "loss": 3.025,
+      "step": 3458
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016755159474671671,
+      "loss": 2.7392,
+      "step": 3459
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001675422138836773,
+      "loss": 2.8282,
+      "step": 3460
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016753283302063792,
+      "loss": 2.6872,
+      "step": 3461
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00016752345215759851,
+      "loss": 2.7759,
+      "step": 3462
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001675140712945591,
+      "loss": 2.5507,
+      "step": 3463
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.0001675046904315197,
+      "loss": 2.6322,
+      "step": 3464
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016749530956848031,
+      "loss": 2.8307,
+      "step": 3465
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001674859287054409,
+      "loss": 2.7539,
+      "step": 3466
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001674765478424015,
+      "loss": 2.7712,
+      "step": 3467
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016746716697936211,
+      "loss": 2.301,
+      "step": 3468
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001674577861163227,
+      "loss": 2.4611,
+      "step": 3469
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016744840525328332,
+      "loss": 3.0594,
+      "step": 3470
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016743902439024391,
+      "loss": 2.445,
+      "step": 3471
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001674296435272045,
+      "loss": 2.3474,
+      "step": 3472
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001674202626641651,
+      "loss": 2.8957,
+      "step": 3473
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016741088180112571,
+      "loss": 2.505,
+      "step": 3474
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001674015009380863,
+      "loss": 2.8272,
+      "step": 3475
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016739212007504692,
+      "loss": 2.4458,
+      "step": 3476
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016738273921200751,
+      "loss": 2.6717,
+      "step": 3477
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016737335834896813,
+      "loss": 2.5137,
+      "step": 3478
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016736397748592872,
+      "loss": 2.5359,
+      "step": 3479
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016735459662288931,
+      "loss": 2.8222,
+      "step": 3480
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001673452157598499,
+      "loss": 2.9168,
+      "step": 3481
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001673358348968105,
+      "loss": 2.6595,
+      "step": 3482
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016732645403377111,
+      "loss": 2.6091,
+      "step": 3483
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001673170731707317,
+      "loss": 2.8862,
+      "step": 3484
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.00016730769230769232,
+      "loss": 2.658,
+      "step": 3485
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016729831144465291,
+      "loss": 2.6674,
+      "step": 3486
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016728893058161353,
+      "loss": 2.6555,
+      "step": 3487
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016727954971857412,
+      "loss": 2.6065,
+      "step": 3488
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016727016885553471,
+      "loss": 2.7141,
+      "step": 3489
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001672607879924953,
+      "loss": 2.8294,
+      "step": 3490
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016725140712945592,
+      "loss": 2.7744,
+      "step": 3491
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016724202626641651,
+      "loss": 2.7444,
+      "step": 3492
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001672326454033771,
+      "loss": 2.7435,
+      "step": 3493
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016722326454033772,
+      "loss": 2.7414,
+      "step": 3494
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016721388367729831,
+      "loss": 2.2332,
+      "step": 3495
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016720450281425893,
+      "loss": 2.6,
+      "step": 3496
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016719512195121952,
+      "loss": 2.5542,
+      "step": 3497
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016718574108818011,
+      "loss": 2.6883,
+      "step": 3498
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.0001671763602251407,
+      "loss": 2.5196,
+      "step": 3499
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016716697936210132,
+      "loss": 2.6634,
+      "step": 3500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016715759849906191,
+      "loss": 2.8055,
+      "step": 3501
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016714821763602253,
+      "loss": 2.849,
+      "step": 3502
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016713883677298312,
+      "loss": 2.3239,
+      "step": 3503
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016712945590994374,
+      "loss": 3.1481,
+      "step": 3504
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016712007504690433,
+      "loss": 2.8351,
+      "step": 3505
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016711069418386492,
+      "loss": 2.3995,
+      "step": 3506
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00016710131332082551,
+      "loss": 2.813,
+      "step": 3507
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001670919324577861,
+      "loss": 2.8204,
+      "step": 3508
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016708255159474672,
+      "loss": 2.6752,
+      "step": 3509
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016707317073170731,
+      "loss": 2.815,
+      "step": 3510
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016706378986866793,
+      "loss": 2.6306,
+      "step": 3511
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016705440900562852,
+      "loss": 3.0369,
+      "step": 3512
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016704502814258914,
+      "loss": 2.5752,
+      "step": 3513
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016703564727954973,
+      "loss": 2.6842,
+      "step": 3514
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016702626641651032,
+      "loss": 2.7679,
+      "step": 3515
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001670168855534709,
+      "loss": 3.0093,
+      "step": 3516
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016700750469043153,
+      "loss": 2.5871,
+      "step": 3517
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016699812382739212,
+      "loss": 2.2161,
+      "step": 3518
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001669887429643527,
+      "loss": 2.7879,
+      "step": 3519
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016697936210131333,
+      "loss": 2.7762,
+      "step": 3520
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016696998123827392,
+      "loss": 2.7995,
+      "step": 3521
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016696060037523454,
+      "loss": 2.7007,
+      "step": 3522
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016695121951219513,
+      "loss": 2.8814,
+      "step": 3523
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016694183864915572,
+      "loss": 2.769,
+      "step": 3524
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.0001669324577861163,
+      "loss": 2.86,
+      "step": 3525
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016692307692307693,
+      "loss": 2.592,
+      "step": 3526
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016691369606003752,
+      "loss": 2.8128,
+      "step": 3527
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00016690431519699814,
+      "loss": 2.5144,
+      "step": 3528
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016689493433395873,
+      "loss": 2.7165,
+      "step": 3529
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016688555347091935,
+      "loss": 2.4918,
+      "step": 3530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016687617260787994,
+      "loss": 2.7291,
+      "step": 3531
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016686679174484053,
+      "loss": 2.6888,
+      "step": 3532
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016685741088180112,
+      "loss": 2.8757,
+      "step": 3533
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.0001668480300187617,
+      "loss": 2.6241,
+      "step": 3534
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016683864915572233,
+      "loss": 2.6709,
+      "step": 3535
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016682926829268292,
+      "loss": 2.612,
+      "step": 3536
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016681988742964354,
+      "loss": 2.4131,
+      "step": 3537
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016681050656660413,
+      "loss": 2.3536,
+      "step": 3538
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016680112570356475,
+      "loss": 2.5888,
+      "step": 3539
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016679174484052534,
+      "loss": 2.7973,
+      "step": 3540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016678236397748596,
+      "loss": 2.6931,
+      "step": 3541
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016677298311444652,
+      "loss": 3.1156,
+      "step": 3542
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016676360225140714,
+      "loss": 2.6293,
+      "step": 3543
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016675422138836773,
+      "loss": 2.4903,
+      "step": 3544
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016674484052532832,
+      "loss": 2.6814,
+      "step": 3545
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016673545966228894,
+      "loss": 2.5745,
+      "step": 3546
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016672607879924953,
+      "loss": 2.8226,
+      "step": 3547
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016671669793621015,
+      "loss": 2.6872,
+      "step": 3548
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 0.00016670731707317074,
+      "loss": 2.8657,
+      "step": 3549
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016669793621013136,
+      "loss": 2.8954,
+      "step": 3550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016668855534709192,
+      "loss": 3.0231,
+      "step": 3551
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016667917448405254,
+      "loss": 2.6934,
+      "step": 3552
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016666979362101313,
+      "loss": 2.4226,
+      "step": 3553
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016666041275797375,
+      "loss": 2.7355,
+      "step": 3554
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016665103189493434,
+      "loss": 2.4497,
+      "step": 3555
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016664165103189496,
+      "loss": 2.8212,
+      "step": 3556
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016663227016885555,
+      "loss": 2.8701,
+      "step": 3557
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016662288930581614,
+      "loss": 2.6691,
+      "step": 3558
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016661350844277676,
+      "loss": 2.5678,
+      "step": 3559
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016660412757973732,
+      "loss": 2.6383,
+      "step": 3560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016659474671669794,
+      "loss": 2.5562,
+      "step": 3561
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016658536585365853,
+      "loss": 2.7795,
+      "step": 3562
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016657598499061915,
+      "loss": 2.6331,
+      "step": 3563
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016656660412757974,
+      "loss": 2.5397,
+      "step": 3564
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016655722326454036,
+      "loss": 3.0401,
+      "step": 3565
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016654784240150095,
+      "loss": 2.7769,
+      "step": 3566
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016653846153846157,
+      "loss": 2.8563,
+      "step": 3567
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016652908067542216,
+      "loss": 2.9863,
+      "step": 3568
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016651969981238275,
+      "loss": 2.6444,
+      "step": 3569
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016651031894934334,
+      "loss": 2.8229,
+      "step": 3570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00016650093808630393,
+      "loss": 2.8416,
+      "step": 3571
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016649155722326455,
+      "loss": 2.3178,
+      "step": 3572
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016648217636022514,
+      "loss": 2.7612,
+      "step": 3573
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016647279549718576,
+      "loss": 2.6651,
+      "step": 3574
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016646341463414635,
+      "loss": 2.4703,
+      "step": 3575
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016645403377110697,
+      "loss": 2.8069,
+      "step": 3576
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016644465290806756,
+      "loss": 2.5286,
+      "step": 3577
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016643527204502815,
+      "loss": 2.5153,
+      "step": 3578
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016642589118198874,
+      "loss": 2.6734,
+      "step": 3579
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016641651031894936,
+      "loss": 2.7182,
+      "step": 3580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016640712945590995,
+      "loss": 2.7736,
+      "step": 3581
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016639774859287057,
+      "loss": 2.5623,
+      "step": 3582
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016638836772983116,
+      "loss": 2.5353,
+      "step": 3583
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016637898686679175,
+      "loss": 2.4283,
+      "step": 3584
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016636960600375237,
+      "loss": 2.8571,
+      "step": 3585
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016636022514071296,
+      "loss": 2.4918,
+      "step": 3586
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016635084427767355,
+      "loss": 2.5916,
+      "step": 3587
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016634146341463414,
+      "loss": 2.7472,
+      "step": 3588
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016633208255159476,
+      "loss": 2.8516,
+      "step": 3589
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016632270168855535,
+      "loss": 2.6794,
+      "step": 3590
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016631332082551597,
+      "loss": 2.7153,
+      "step": 3591
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00016630393996247656,
+      "loss": 2.724,
+      "step": 3592
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016629455909943718,
+      "loss": 2.7349,
+      "step": 3593
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016628517823639777,
+      "loss": 2.4864,
+      "step": 3594
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016627579737335836,
+      "loss": 2.6805,
+      "step": 3595
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016626641651031895,
+      "loss": 2.5554,
+      "step": 3596
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016625703564727957,
+      "loss": 2.717,
+      "step": 3597
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016624765478424016,
+      "loss": 2.843,
+      "step": 3598
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016623827392120075,
+      "loss": 2.9979,
+      "step": 3599
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016622889305816137,
+      "loss": 2.6712,
+      "step": 3600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016621951219512196,
+      "loss": 2.7554,
+      "step": 3601
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016621013133208257,
+      "loss": 2.9465,
+      "step": 3602
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016620075046904317,
+      "loss": 2.967,
+      "step": 3603
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016619136960600376,
+      "loss": 3.1262,
+      "step": 3604
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016618198874296435,
+      "loss": 3.0653,
+      "step": 3605
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016617260787992497,
+      "loss": 2.4918,
+      "step": 3606
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016616322701688556,
+      "loss": 2.7288,
+      "step": 3607
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016615384615384617,
+      "loss": 2.4881,
+      "step": 3608
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016614446529080677,
+      "loss": 2.6464,
+      "step": 3609
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016613508442776738,
+      "loss": 2.4582,
+      "step": 3610
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016612570356472797,
+      "loss": 2.6081,
+      "step": 3611
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016611632270168857,
+      "loss": 2.7038,
+      "step": 3612
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00016610694183864916,
+      "loss": 2.9198,
+      "step": 3613
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016609756097560975,
+      "loss": 2.7181,
+      "step": 3614
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016608818011257037,
+      "loss": 2.7473,
+      "step": 3615
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016607879924953096,
+      "loss": 2.5305,
+      "step": 3616
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016606941838649157,
+      "loss": 2.4432,
+      "step": 3617
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016606003752345217,
+      "loss": 2.6834,
+      "step": 3618
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016605065666041278,
+      "loss": 2.7929,
+      "step": 3619
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016604127579737337,
+      "loss": 2.6369,
+      "step": 3620
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016603189493433397,
+      "loss": 2.5849,
+      "step": 3621
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016602251407129456,
+      "loss": 2.6525,
+      "step": 3622
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016601313320825517,
+      "loss": 2.7342,
+      "step": 3623
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016600375234521577,
+      "loss": 2.5908,
+      "step": 3624
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016599437148217636,
+      "loss": 2.6766,
+      "step": 3625
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016598499061913697,
+      "loss": 2.6271,
+      "step": 3626
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016597560975609757,
+      "loss": 2.6041,
+      "step": 3627
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016596622889305818,
+      "loss": 2.7395,
+      "step": 3628
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016595684803001877,
+      "loss": 2.4884,
+      "step": 3629
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016594746716697937,
+      "loss": 2.8614,
+      "step": 3630
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016593808630393996,
+      "loss": 2.7721,
+      "step": 3631
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016592870544090057,
+      "loss": 2.6257,
+      "step": 3632
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016591932457786117,
+      "loss": 2.4934,
+      "step": 3633
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016590994371482178,
+      "loss": 2.5681,
+      "step": 3634
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 0.00016590056285178237,
+      "loss": 2.6614,
+      "step": 3635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.000165891181988743,
+      "loss": 2.4028,
+      "step": 3636
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016588180112570358,
+      "loss": 2.4959,
+      "step": 3637
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016587242026266417,
+      "loss": 2.6023,
+      "step": 3638
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016586303939962477,
+      "loss": 2.7371,
+      "step": 3639
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016585365853658536,
+      "loss": 2.7496,
+      "step": 3640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016584427767354597,
+      "loss": 2.9512,
+      "step": 3641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016583489681050656,
+      "loss": 2.5671,
+      "step": 3642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016582551594746718,
+      "loss": 2.9815,
+      "step": 3643
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016581613508442777,
+      "loss": 2.6417,
+      "step": 3644
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001658067542213884,
+      "loss": 2.6227,
+      "step": 3645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016579737335834898,
+      "loss": 2.6634,
+      "step": 3646
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016578799249530957,
+      "loss": 2.6733,
+      "step": 3647
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016577861163227016,
+      "loss": 2.7896,
+      "step": 3648
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016576923076923078,
+      "loss": 2.6974,
+      "step": 3649
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016575984990619137,
+      "loss": 2.4413,
+      "step": 3650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016575046904315196,
+      "loss": 2.8933,
+      "step": 3651
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016574108818011258,
+      "loss": 2.7779,
+      "step": 3652
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016573170731707317,
+      "loss": 2.6106,
+      "step": 3653
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.0001657223264540338,
+      "loss": 2.8009,
+      "step": 3654
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016571294559099438,
+      "loss": 2.4406,
+      "step": 3655
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00016570356472795497,
+      "loss": 2.4898,
+      "step": 3656
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016569418386491556,
+      "loss": 2.8223,
+      "step": 3657
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016568480300187618,
+      "loss": 2.6238,
+      "step": 3658
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016567542213883677,
+      "loss": 2.834,
+      "step": 3659
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001656660412757974,
+      "loss": 2.5655,
+      "step": 3660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016565666041275798,
+      "loss": 2.7513,
+      "step": 3661
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001656472795497186,
+      "loss": 2.6505,
+      "step": 3662
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001656378986866792,
+      "loss": 2.6942,
+      "step": 3663
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016562851782363978,
+      "loss": 3.2004,
+      "step": 3664
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016561913696060037,
+      "loss": 2.6122,
+      "step": 3665
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016560975609756096,
+      "loss": 2.8187,
+      "step": 3666
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016560037523452158,
+      "loss": 2.6933,
+      "step": 3667
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016559099437148217,
+      "loss": 2.8311,
+      "step": 3668
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001655816135084428,
+      "loss": 3.0775,
+      "step": 3669
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016557223264540338,
+      "loss": 2.6729,
+      "step": 3670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.000165562851782364,
+      "loss": 2.9599,
+      "step": 3671
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001655534709193246,
+      "loss": 2.8546,
+      "step": 3672
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016554409005628518,
+      "loss": 2.81,
+      "step": 3673
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016553470919324577,
+      "loss": 2.7291,
+      "step": 3674
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001655253283302064,
+      "loss": 2.5488,
+      "step": 3675
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016551594746716698,
+      "loss": 3.0312,
+      "step": 3676
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.00016550656660412757,
+      "loss": 2.6262,
+      "step": 3677
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001654971857410882,
+      "loss": 2.4772,
+      "step": 3678
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016548780487804878,
+      "loss": 2.7413,
+      "step": 3679
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001654784240150094,
+      "loss": 2.8314,
+      "step": 3680
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016546904315197,
+      "loss": 2.3644,
+      "step": 3681
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016545966228893058,
+      "loss": 2.7166,
+      "step": 3682
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016545028142589117,
+      "loss": 2.6347,
+      "step": 3683
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001654409005628518,
+      "loss": 2.9215,
+      "step": 3684
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016543151969981238,
+      "loss": 2.7879,
+      "step": 3685
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000165422138836773,
+      "loss": 2.7713,
+      "step": 3686
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001654127579737336,
+      "loss": 2.7753,
+      "step": 3687
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001654033771106942,
+      "loss": 2.7984,
+      "step": 3688
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001653939962476548,
+      "loss": 2.578,
+      "step": 3689
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001653846153846154,
+      "loss": 2.794,
+      "step": 3690
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016537523452157598,
+      "loss": 2.7517,
+      "step": 3691
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016536585365853657,
+      "loss": 2.4917,
+      "step": 3692
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001653564727954972,
+      "loss": 2.819,
+      "step": 3693
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016534709193245778,
+      "loss": 2.5459,
+      "step": 3694
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001653377110694184,
+      "loss": 2.8051,
+      "step": 3695
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.000165328330206379,
+      "loss": 2.4982,
+      "step": 3696
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001653189493433396,
+      "loss": 2.4646,
+      "step": 3697
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.0001653095684803002,
+      "loss": 2.6896,
+      "step": 3698
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00016530018761726082,
+      "loss": 2.4124,
+      "step": 3699
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016529080675422138,
+      "loss": 2.7742,
+      "step": 3700
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000165281425891182,
+      "loss": 2.789,
+      "step": 3701
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001652720450281426,
+      "loss": 2.8089,
+      "step": 3702
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016526266416510318,
+      "loss": 2.4877,
+      "step": 3703
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001652532833020638,
+      "loss": 2.7532,
+      "step": 3704
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001652439024390244,
+      "loss": 3.0726,
+      "step": 3705
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000165234521575985,
+      "loss": 2.5745,
+      "step": 3706
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001652251407129456,
+      "loss": 2.7739,
+      "step": 3707
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016521575984990622,
+      "loss": 2.5675,
+      "step": 3708
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016520637898686678,
+      "loss": 2.8877,
+      "step": 3709
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001651969981238274,
+      "loss": 2.8046,
+      "step": 3710
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000165187617260788,
+      "loss": 2.8576,
+      "step": 3711
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001651782363977486,
+      "loss": 3.0635,
+      "step": 3712
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001651688555347092,
+      "loss": 2.6875,
+      "step": 3713
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016515947467166982,
+      "loss": 2.7224,
+      "step": 3714
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001651500938086304,
+      "loss": 2.7178,
+      "step": 3715
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.000165140712945591,
+      "loss": 2.6802,
+      "step": 3716
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016513133208255162,
+      "loss": 2.8185,
+      "step": 3717
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00016512195121951218,
+      "loss": 2.6961,
+      "step": 3718
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001651125703564728,
+      "loss": 2.3949,
+      "step": 3719
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.0001651031894934334,
+      "loss": 2.8023,
+      "step": 3720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000165093808630394,
+      "loss": 2.97,
+      "step": 3721
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001650844277673546,
+      "loss": 2.7585,
+      "step": 3722
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016507504690431522,
+      "loss": 2.6248,
+      "step": 3723
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001650656660412758,
+      "loss": 2.6644,
+      "step": 3724
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016505628517823643,
+      "loss": 2.9187,
+      "step": 3725
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016504690431519702,
+      "loss": 2.7612,
+      "step": 3726
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001650375234521576,
+      "loss": 2.8203,
+      "step": 3727
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001650281425891182,
+      "loss": 2.6651,
+      "step": 3728
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001650187617260788,
+      "loss": 2.6125,
+      "step": 3729
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001650093808630394,
+      "loss": 2.6723,
+      "step": 3730
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000165,
+      "loss": 2.7942,
+      "step": 3731
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016499061913696062,
+      "loss": 2.7991,
+      "step": 3732
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001649812382739212,
+      "loss": 2.7659,
+      "step": 3733
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016497185741088183,
+      "loss": 2.6925,
+      "step": 3734
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016496247654784242,
+      "loss": 2.7697,
+      "step": 3735
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.000164953095684803,
+      "loss": 2.6782,
+      "step": 3736
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001649437148217636,
+      "loss": 2.8691,
+      "step": 3737
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016493433395872422,
+      "loss": 2.6385,
+      "step": 3738
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.0001649249530956848,
+      "loss": 2.8277,
+      "step": 3739
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016491557223264543,
+      "loss": 2.7602,
+      "step": 3740
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 0.00016490619136960602,
+      "loss": 2.9801,
+      "step": 3741
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001648968105065666,
+      "loss": 2.6385,
+      "step": 3742
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016488742964352723,
+      "loss": 2.6444,
+      "step": 3743
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016487804878048782,
+      "loss": 2.7908,
+      "step": 3744
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001648686679174484,
+      "loss": 2.8018,
+      "step": 3745
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.000164859287054409,
+      "loss": 2.5374,
+      "step": 3746
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016484990619136962,
+      "loss": 2.5676,
+      "step": 3747
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001648405253283302,
+      "loss": 2.7501,
+      "step": 3748
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016483114446529083,
+      "loss": 2.7526,
+      "step": 3749
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016482176360225142,
+      "loss": 2.6123,
+      "step": 3750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016481238273921203,
+      "loss": 2.5922,
+      "step": 3751
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016480300187617263,
+      "loss": 2.7866,
+      "step": 3752
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016479362101313322,
+      "loss": 2.3925,
+      "step": 3753
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001647842401500938,
+      "loss": 2.3067,
+      "step": 3754
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001647748592870544,
+      "loss": 2.61,
+      "step": 3755
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016476547842401502,
+      "loss": 2.5844,
+      "step": 3756
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001647560975609756,
+      "loss": 2.3769,
+      "step": 3757
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016474671669793623,
+      "loss": 2.7116,
+      "step": 3758
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016473733583489682,
+      "loss": 2.606,
+      "step": 3759
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016472795497185743,
+      "loss": 2.7236,
+      "step": 3760
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016471857410881803,
+      "loss": 2.521,
+      "step": 3761
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.00016470919324577862,
+      "loss": 2.6682,
+      "step": 3762
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001646998123827392,
+      "loss": 2.922,
+      "step": 3763
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016469043151969983,
+      "loss": 2.6432,
+      "step": 3764
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016468105065666042,
+      "loss": 2.5367,
+      "step": 3765
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016467166979362103,
+      "loss": 2.8795,
+      "step": 3766
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016466228893058163,
+      "loss": 2.7843,
+      "step": 3767
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016465290806754222,
+      "loss": 2.7666,
+      "step": 3768
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016464352720450283,
+      "loss": 2.7495,
+      "step": 3769
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016463414634146343,
+      "loss": 2.843,
+      "step": 3770
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016462476547842402,
+      "loss": 2.523,
+      "step": 3771
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.0001646153846153846,
+      "loss": 2.7772,
+      "step": 3772
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016460600375234523,
+      "loss": 2.9553,
+      "step": 3773
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016459662288930582,
+      "loss": 2.6155,
+      "step": 3774
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016458724202626643,
+      "loss": 2.9009,
+      "step": 3775
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016457786116322703,
+      "loss": 2.4408,
+      "step": 3776
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016456848030018764,
+      "loss": 2.6791,
+      "step": 3777
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016455909943714823,
+      "loss": 2.5636,
+      "step": 3778
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016454971857410883,
+      "loss": 2.6279,
+      "step": 3779
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016454033771106942,
+      "loss": 2.9617,
+      "step": 3780
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016453095684803,
+      "loss": 2.9085,
+      "step": 3781
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016452157598499063,
+      "loss": 2.582,
+      "step": 3782
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016451219512195122,
+      "loss": 2.8953,
+      "step": 3783
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00016450281425891183,
+      "loss": 2.7591,
+      "step": 3784
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016449343339587242,
+      "loss": 2.733,
+      "step": 3785
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016448405253283304,
+      "loss": 2.7194,
+      "step": 3786
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016447467166979363,
+      "loss": 2.7526,
+      "step": 3787
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016446529080675422,
+      "loss": 2.7229,
+      "step": 3788
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016445590994371482,
+      "loss": 2.858,
+      "step": 3789
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016444652908067543,
+      "loss": 2.9767,
+      "step": 3790
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016443714821763602,
+      "loss": 2.5479,
+      "step": 3791
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016442776735459664,
+      "loss": 2.7288,
+      "step": 3792
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016441838649155723,
+      "loss": 2.772,
+      "step": 3793
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016440900562851785,
+      "loss": 2.6496,
+      "step": 3794
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016439962476547844,
+      "loss": 2.8835,
+      "step": 3795
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016439024390243903,
+      "loss": 2.4379,
+      "step": 3796
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016438086303939962,
+      "loss": 2.674,
+      "step": 3797
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016437148217636022,
+      "loss": 2.7436,
+      "step": 3798
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016436210131332083,
+      "loss": 2.6231,
+      "step": 3799
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016435272045028142,
+      "loss": 2.4736,
+      "step": 3800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016434333958724204,
+      "loss": 2.7012,
+      "step": 3801
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016433395872420263,
+      "loss": 2.1999,
+      "step": 3802
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016432457786116325,
+      "loss": 2.7806,
+      "step": 3803
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016431519699812384,
+      "loss": 2.6268,
+      "step": 3804
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00016430581613508443,
+      "loss": 2.692,
+      "step": 3805
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016429643527204502,
+      "loss": 2.7885,
+      "step": 3806
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016428705440900564,
+      "loss": 2.632,
+      "step": 3807
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016427767354596623,
+      "loss": 2.9089,
+      "step": 3808
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016426829268292682,
+      "loss": 2.529,
+      "step": 3809
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016425891181988744,
+      "loss": 2.3595,
+      "step": 3810
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016424953095684803,
+      "loss": 2.6006,
+      "step": 3811
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016424015009380865,
+      "loss": 2.7347,
+      "step": 3812
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016423076923076924,
+      "loss": 2.4719,
+      "step": 3813
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016422138836772983,
+      "loss": 2.3066,
+      "step": 3814
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016421200750469042,
+      "loss": 2.6575,
+      "step": 3815
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016420262664165104,
+      "loss": 2.5434,
+      "step": 3816
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016419324577861163,
+      "loss": 2.8603,
+      "step": 3817
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016418386491557225,
+      "loss": 2.5461,
+      "step": 3818
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016417448405253284,
+      "loss": 2.6002,
+      "step": 3819
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016416510318949346,
+      "loss": 2.5482,
+      "step": 3820
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016415572232645405,
+      "loss": 2.7098,
+      "step": 3821
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016414634146341464,
+      "loss": 2.6235,
+      "step": 3822
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016413696060037523,
+      "loss": 2.5645,
+      "step": 3823
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016412757973733582,
+      "loss": 2.3281,
+      "step": 3824
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016411819887429644,
+      "loss": 2.5481,
+      "step": 3825
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00016410881801125703,
+      "loss": 2.4779,
+      "step": 3826
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016409943714821765,
+      "loss": 2.5234,
+      "step": 3827
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016409005628517824,
+      "loss": 2.7182,
+      "step": 3828
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016408067542213886,
+      "loss": 2.8809,
+      "step": 3829
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016407129455909945,
+      "loss": 3.0274,
+      "step": 3830
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016406191369606004,
+      "loss": 2.6978,
+      "step": 3831
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016405253283302063,
+      "loss": 2.6275,
+      "step": 3832
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016404315196998125,
+      "loss": 2.5059,
+      "step": 3833
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016403377110694184,
+      "loss": 2.5586,
+      "step": 3834
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016402439024390243,
+      "loss": 2.8948,
+      "step": 3835
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016401500938086305,
+      "loss": 2.6775,
+      "step": 3836
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016400562851782364,
+      "loss": 2.4372,
+      "step": 3837
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016399624765478426,
+      "loss": 2.4766,
+      "step": 3838
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016398686679174485,
+      "loss": 2.7171,
+      "step": 3839
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016397748592870544,
+      "loss": 2.6498,
+      "step": 3840
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016396810506566603,
+      "loss": 2.5522,
+      "step": 3841
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016395872420262665,
+      "loss": 2.8424,
+      "step": 3842
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016394934333958724,
+      "loss": 2.447,
+      "step": 3843
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016393996247654786,
+      "loss": 2.7143,
+      "step": 3844
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016393058161350845,
+      "loss": 2.8272,
+      "step": 3845
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016392120075046907,
+      "loss": 2.7187,
+      "step": 3846
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016391181988742966,
+      "loss": 2.6451,
+      "step": 3847
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 0.00016390243902439025,
+      "loss": 2.6959,
+      "step": 3848
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016389305816135084,
+      "loss": 2.7407,
+      "step": 3849
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016388367729831143,
+      "loss": 2.9298,
+      "step": 3850
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016387429643527205,
+      "loss": 2.733,
+      "step": 3851
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016386491557223264,
+      "loss": 2.6959,
+      "step": 3852
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016385553470919326,
+      "loss": 2.7665,
+      "step": 3853
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016384615384615385,
+      "loss": 2.5855,
+      "step": 3854
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016383677298311447,
+      "loss": 2.9251,
+      "step": 3855
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016382739212007506,
+      "loss": 2.6159,
+      "step": 3856
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016381801125703565,
+      "loss": 2.8888,
+      "step": 3857
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016380863039399624,
+      "loss": 2.7719,
+      "step": 3858
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016379924953095686,
+      "loss": 2.6531,
+      "step": 3859
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016378986866791745,
+      "loss": 2.722,
+      "step": 3860
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016378048780487804,
+      "loss": 2.7346,
+      "step": 3861
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016377110694183866,
+      "loss": 2.7796,
+      "step": 3862
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016376172607879925,
+      "loss": 2.7277,
+      "step": 3863
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016375234521575987,
+      "loss": 2.9068,
+      "step": 3864
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016374296435272046,
+      "loss": 2.7927,
+      "step": 3865
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016373358348968108,
+      "loss": 2.7386,
+      "step": 3866
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016372420262664164,
+      "loss": 2.6535,
+      "step": 3867
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016371482176360226,
+      "loss": 2.7735,
+      "step": 3868
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.00016370544090056285,
+      "loss": 2.6627,
+      "step": 3869
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016369606003752347,
+      "loss": 2.4872,
+      "step": 3870
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016368667917448406,
+      "loss": 2.7223,
+      "step": 3871
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016367729831144468,
+      "loss": 2.626,
+      "step": 3872
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016366791744840527,
+      "loss": 2.5736,
+      "step": 3873
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016365853658536586,
+      "loss": 2.7547,
+      "step": 3874
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016364915572232648,
+      "loss": 2.8237,
+      "step": 3875
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016363977485928704,
+      "loss": 2.5855,
+      "step": 3876
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016363039399624766,
+      "loss": 2.5106,
+      "step": 3877
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016362101313320825,
+      "loss": 2.6381,
+      "step": 3878
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016361163227016887,
+      "loss": 2.4953,
+      "step": 3879
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016360225140712946,
+      "loss": 2.847,
+      "step": 3880
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016359287054409008,
+      "loss": 2.8576,
+      "step": 3881
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016358348968105067,
+      "loss": 2.489,
+      "step": 3882
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016357410881801129,
+      "loss": 2.6957,
+      "step": 3883
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016356472795497188,
+      "loss": 2.6545,
+      "step": 3884
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016355534709193247,
+      "loss": 2.6495,
+      "step": 3885
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016354596622889306,
+      "loss": 2.6407,
+      "step": 3886
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016353658536585365,
+      "loss": 3.183,
+      "step": 3887
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016352720450281427,
+      "loss": 2.7352,
+      "step": 3888
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016351782363977486,
+      "loss": 2.445,
+      "step": 3889
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00016350844277673548,
+      "loss": 2.6849,
+      "step": 3890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016349906191369607,
+      "loss": 2.8733,
+      "step": 3891
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016348968105065669,
+      "loss": 3.0096,
+      "step": 3892
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016348030018761728,
+      "loss": 2.767,
+      "step": 3893
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016347091932457787,
+      "loss": 2.803,
+      "step": 3894
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016346153846153846,
+      "loss": 2.5881,
+      "step": 3895
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016345215759849908,
+      "loss": 2.6437,
+      "step": 3896
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016344277673545967,
+      "loss": 2.6072,
+      "step": 3897
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016343339587242029,
+      "loss": 2.6393,
+      "step": 3898
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016342401500938088,
+      "loss": 2.8644,
+      "step": 3899
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016341463414634147,
+      "loss": 2.644,
+      "step": 3900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016340525328330209,
+      "loss": 2.7478,
+      "step": 3901
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016339587242026268,
+      "loss": 2.5842,
+      "step": 3902
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016338649155722327,
+      "loss": 2.9765,
+      "step": 3903
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016337711069418386,
+      "loss": 2.5882,
+      "step": 3904
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016336772983114448,
+      "loss": 2.2671,
+      "step": 3905
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016335834896810507,
+      "loss": 2.7694,
+      "step": 3906
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016334896810506569,
+      "loss": 2.7318,
+      "step": 3907
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016333958724202628,
+      "loss": 2.5364,
+      "step": 3908
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.0001633302063789869,
+      "loss": 2.9061,
+      "step": 3909
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016332082551594749,
+      "loss": 2.5551,
+      "step": 3910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016331144465290808,
+      "loss": 2.6453,
+      "step": 3911
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00016330206378986867,
+      "loss": 2.7721,
+      "step": 3912
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016329268292682926,
+      "loss": 2.855,
+      "step": 3913
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016328330206378988,
+      "loss": 2.484,
+      "step": 3914
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016327392120075047,
+      "loss": 2.8358,
+      "step": 3915
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016326454033771109,
+      "loss": 2.6166,
+      "step": 3916
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016325515947467168,
+      "loss": 3.0429,
+      "step": 3917
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001632457786116323,
+      "loss": 2.4368,
+      "step": 3918
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016323639774859289,
+      "loss": 2.6952,
+      "step": 3919
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016322701688555348,
+      "loss": 2.8315,
+      "step": 3920
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016321763602251407,
+      "loss": 2.7682,
+      "step": 3921
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016320825515947469,
+      "loss": 2.8226,
+      "step": 3922
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016319887429643528,
+      "loss": 2.4935,
+      "step": 3923
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001631894934333959,
+      "loss": 2.6736,
+      "step": 3924
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016318011257035649,
+      "loss": 2.6287,
+      "step": 3925
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016317073170731708,
+      "loss": 2.9065,
+      "step": 3926
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001631613508442777,
+      "loss": 2.414,
+      "step": 3927
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016315196998123829,
+      "loss": 2.5613,
+      "step": 3928
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016314258911819888,
+      "loss": 2.8349,
+      "step": 3929
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016313320825515947,
+      "loss": 2.771,
+      "step": 3930
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016312382739212008,
+      "loss": 2.7019,
+      "step": 3931
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.00016311444652908068,
+      "loss": 2.6171,
+      "step": 3932
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 0.0001631050656660413,
+      "loss": 2.7688,
+      "step": 3933
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016309568480300188,
+      "loss": 2.7053,
+      "step": 3934
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001630863039399625,
+      "loss": 2.5286,
+      "step": 3935
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001630769230769231,
+      "loss": 2.9101,
+      "step": 3936
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016306754221388368,
+      "loss": 2.4861,
+      "step": 3937
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016305816135084428,
+      "loss": 2.9975,
+      "step": 3938
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016304878048780487,
+      "loss": 2.5152,
+      "step": 3939
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016303939962476548,
+      "loss": 2.5568,
+      "step": 3940
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016303001876172608,
+      "loss": 2.856,
+      "step": 3941
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001630206378986867,
+      "loss": 2.5329,
+      "step": 3942
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016301125703564728,
+      "loss": 2.7329,
+      "step": 3943
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001630018761726079,
+      "loss": 2.5559,
+      "step": 3944
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001629924953095685,
+      "loss": 2.6716,
+      "step": 3945
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016298311444652908,
+      "loss": 2.5884,
+      "step": 3946
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016297373358348968,
+      "loss": 2.4214,
+      "step": 3947
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001629643527204503,
+      "loss": 2.934,
+      "step": 3948
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016295497185741088,
+      "loss": 2.7265,
+      "step": 3949
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001629455909943715,
+      "loss": 2.7268,
+      "step": 3950
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001629362101313321,
+      "loss": 2.4822,
+      "step": 3951
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.00016292682926829268,
+      "loss": 2.5917,
+      "step": 3952
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001629174484052533,
+      "loss": 2.8121,
+      "step": 3953
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001629080675422139,
+      "loss": 2.4811,
+      "step": 3954
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016289868667917448,
+      "loss": 2.5886,
+      "step": 3955
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016288930581613508,
+      "loss": 2.648,
+      "step": 3956
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001628799249530957,
+      "loss": 2.6128,
+      "step": 3957
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016287054409005628,
+      "loss": 2.5582,
+      "step": 3958
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001628611632270169,
+      "loss": 2.8893,
+      "step": 3959
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001628517823639775,
+      "loss": 2.6665,
+      "step": 3960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001628424015009381,
+      "loss": 2.6367,
+      "step": 3961
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001628330206378987,
+      "loss": 2.9561,
+      "step": 3962
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001628236397748593,
+      "loss": 2.7206,
+      "step": 3963
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016281425891181988,
+      "loss": 2.8841,
+      "step": 3964
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016280487804878048,
+      "loss": 2.8621,
+      "step": 3965
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001627954971857411,
+      "loss": 2.6488,
+      "step": 3966
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016278611632270168,
+      "loss": 2.7478,
+      "step": 3967
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001627767354596623,
+      "loss": 2.8935,
+      "step": 3968
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001627673545966229,
+      "loss": 2.4598,
+      "step": 3969
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001627579737335835,
+      "loss": 2.5319,
+      "step": 3970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001627485928705441,
+      "loss": 2.8588,
+      "step": 3971
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001627392120075047,
+      "loss": 2.7242,
+      "step": 3972
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00016272983114446528,
+      "loss": 2.6858,
+      "step": 3973
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001627204502814259,
+      "loss": 2.4375,
+      "step": 3974
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001627110694183865,
+      "loss": 2.6743,
+      "step": 3975
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.0001627016885553471,
+      "loss": 2.4818,
+      "step": 3976
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001626923076923077,
+      "loss": 2.6537,
+      "step": 3977
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001626829268292683,
+      "loss": 2.5392,
+      "step": 3978
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001626735459662289,
+      "loss": 2.5496,
+      "step": 3979
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001626641651031895,
+      "loss": 2.7516,
+      "step": 3980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001626547842401501,
+      "loss": 2.7859,
+      "step": 3981
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016264540337711068,
+      "loss": 2.8793,
+      "step": 3982
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001626360225140713,
+      "loss": 2.7584,
+      "step": 3983
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001626266416510319,
+      "loss": 2.6995,
+      "step": 3984
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001626172607879925,
+      "loss": 2.7091,
+      "step": 3985
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001626078799249531,
+      "loss": 2.5924,
+      "step": 3986
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016259849906191372,
+      "loss": 2.7412,
+      "step": 3987
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001625891181988743,
+      "loss": 2.6573,
+      "step": 3988
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001625797373358349,
+      "loss": 2.5042,
+      "step": 3989
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001625703564727955,
+      "loss": 2.7013,
+      "step": 3990
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016256097560975608,
+      "loss": 2.4827,
+      "step": 3991
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001625515947467167,
+      "loss": 2.6993,
+      "step": 3992
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001625422138836773,
+      "loss": 2.7377,
+      "step": 3993
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001625328330206379,
+      "loss": 2.5525,
+      "step": 3994
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001625234521575985,
+      "loss": 2.8293,
+      "step": 3995
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00016251407129455912,
+      "loss": 3.0844,
+      "step": 3996
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.0001625046904315197,
+      "loss": 2.7294,
+      "step": 3997
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001624953095684803,
+      "loss": 2.5592,
+      "step": 3998
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001624859287054409,
+      "loss": 2.8294,
+      "step": 3999
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001624765478424015,
+      "loss": 2.5714,
+      "step": 4000
+    },
+    {
+      "epoch": 1.88,
+      "eval_loss": 2.648087501525879,
+      "eval_runtime": 67.7429,
+      "eval_samples_per_second": 27.973,
+      "eval_steps_per_second": 3.499,
+      "step": 4000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001624671669793621,
+      "loss": 2.4386,
+      "step": 4001
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016245778611632272,
+      "loss": 2.6064,
+      "step": 4002
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001624484052532833,
+      "loss": 2.8086,
+      "step": 4003
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016243902439024393,
+      "loss": 2.8018,
+      "step": 4004
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016242964352720452,
+      "loss": 2.3774,
+      "step": 4005
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001624202626641651,
+      "loss": 2.6216,
+      "step": 4006
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001624108818011257,
+      "loss": 2.6106,
+      "step": 4007
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001624015009380863,
+      "loss": 3.013,
+      "step": 4008
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001623921200750469,
+      "loss": 2.4835,
+      "step": 4009
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001623827392120075,
+      "loss": 2.7192,
+      "step": 4010
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016237335834896812,
+      "loss": 2.8594,
+      "step": 4011
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001623639774859287,
+      "loss": 2.6929,
+      "step": 4012
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016235459662288933,
+      "loss": 2.5357,
+      "step": 4013
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016234521575984992,
+      "loss": 2.6699,
+      "step": 4014
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001623358348968105,
+      "loss": 2.8571,
+      "step": 4015
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001623264540337711,
+      "loss": 2.6172,
+      "step": 4016
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00016231707317073172,
+      "loss": 2.6455,
+      "step": 4017
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.0001623076923076923,
+      "loss": 2.7639,
+      "step": 4018
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001622983114446529,
+      "loss": 2.6935,
+      "step": 4019
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016228893058161352,
+      "loss": 2.8153,
+      "step": 4020
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001622795497185741,
+      "loss": 2.5122,
+      "step": 4021
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016227016885553473,
+      "loss": 2.4674,
+      "step": 4022
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016226078799249532,
+      "loss": 2.8248,
+      "step": 4023
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001622514071294559,
+      "loss": 2.6232,
+      "step": 4024
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001622420262664165,
+      "loss": 2.5176,
+      "step": 4025
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016223264540337712,
+      "loss": 2.8115,
+      "step": 4026
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001622232645403377,
+      "loss": 2.4642,
+      "step": 4027
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016221388367729833,
+      "loss": 2.7615,
+      "step": 4028
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016220450281425892,
+      "loss": 2.7773,
+      "step": 4029
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016219512195121954,
+      "loss": 2.5017,
+      "step": 4030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016218574108818013,
+      "loss": 2.6202,
+      "step": 4031
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016217636022514072,
+      "loss": 2.84,
+      "step": 4032
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001621669793621013,
+      "loss": 2.6514,
+      "step": 4033
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001621575984990619,
+      "loss": 2.5212,
+      "step": 4034
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016214821763602252,
+      "loss": 2.6134,
+      "step": 4035
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.0001621388367729831,
+      "loss": 2.4773,
+      "step": 4036
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016212945590994373,
+      "loss": 2.3506,
+      "step": 4037
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016212007504690432,
+      "loss": 2.6849,
+      "step": 4038
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016211069418386494,
+      "loss": 3.1789,
+      "step": 4039
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 0.00016210131332082553,
+      "loss": 2.6594,
+      "step": 4040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016209193245778615,
+      "loss": 2.693,
+      "step": 4041
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016208255159474674,
+      "loss": 2.5682,
+      "step": 4042
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016207317073170733,
+      "loss": 2.6793,
+      "step": 4043
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016206378986866792,
+      "loss": 2.89,
+      "step": 4044
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001620544090056285,
+      "loss": 2.8315,
+      "step": 4045
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016204502814258913,
+      "loss": 2.5325,
+      "step": 4046
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016203564727954972,
+      "loss": 2.7888,
+      "step": 4047
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016202626641651034,
+      "loss": 2.8641,
+      "step": 4048
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016201688555347093,
+      "loss": 2.6568,
+      "step": 4049
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016200750469043155,
+      "loss": 2.5029,
+      "step": 4050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016199812382739214,
+      "loss": 2.7383,
+      "step": 4051
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016198874296435273,
+      "loss": 2.239,
+      "step": 4052
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016197936210131332,
+      "loss": 2.5704,
+      "step": 4053
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016196998123827394,
+      "loss": 2.5428,
+      "step": 4054
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016196060037523453,
+      "loss": 2.7653,
+      "step": 4055
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016195121951219515,
+      "loss": 2.9441,
+      "step": 4056
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016194183864915574,
+      "loss": 2.7144,
+      "step": 4057
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016193245778611633,
+      "loss": 2.3977,
+      "step": 4058
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016192307692307695,
+      "loss": 2.6155,
+      "step": 4059
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016191369606003754,
+      "loss": 2.8586,
+      "step": 4060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.00016190431519699813,
+      "loss": 2.5208,
+      "step": 4061
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016189493433395872,
+      "loss": 2.4907,
+      "step": 4062
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016188555347091934,
+      "loss": 2.7306,
+      "step": 4063
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016187617260787993,
+      "loss": 2.7567,
+      "step": 4064
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016186679174484055,
+      "loss": 2.6058,
+      "step": 4065
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016185741088180114,
+      "loss": 2.8476,
+      "step": 4066
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016184803001876175,
+      "loss": 2.6281,
+      "step": 4067
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016183864915572235,
+      "loss": 2.5717,
+      "step": 4068
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016182926829268294,
+      "loss": 2.9905,
+      "step": 4069
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016181988742964353,
+      "loss": 2.9127,
+      "step": 4070
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016181050656660412,
+      "loss": 2.7387,
+      "step": 4071
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016180112570356474,
+      "loss": 2.5266,
+      "step": 4072
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016179174484052533,
+      "loss": 2.8156,
+      "step": 4073
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016178236397748595,
+      "loss": 2.9501,
+      "step": 4074
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016177298311444654,
+      "loss": 2.544,
+      "step": 4075
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016176360225140715,
+      "loss": 2.6328,
+      "step": 4076
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016175422138836774,
+      "loss": 2.7864,
+      "step": 4077
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016174484052532834,
+      "loss": 2.6574,
+      "step": 4078
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016173545966228893,
+      "loss": 2.7127,
+      "step": 4079
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016172607879924954,
+      "loss": 2.8013,
+      "step": 4080
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016171669793621014,
+      "loss": 2.6867,
+      "step": 4081
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00016170731707317075,
+      "loss": 2.4273,
+      "step": 4082
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016169793621013134,
+      "loss": 2.9141,
+      "step": 4083
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016168855534709194,
+      "loss": 2.6691,
+      "step": 4084
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016167917448405255,
+      "loss": 2.5201,
+      "step": 4085
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016166979362101314,
+      "loss": 2.5761,
+      "step": 4086
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016166041275797374,
+      "loss": 2.6404,
+      "step": 4087
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016165103189493433,
+      "loss": 2.4609,
+      "step": 4088
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016164165103189494,
+      "loss": 2.6874,
+      "step": 4089
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016163227016885554,
+      "loss": 2.9403,
+      "step": 4090
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016162288930581615,
+      "loss": 2.5388,
+      "step": 4091
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016161350844277674,
+      "loss": 2.671,
+      "step": 4092
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016160412757973736,
+      "loss": 2.6181,
+      "step": 4093
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016159474671669795,
+      "loss": 3.225,
+      "step": 4094
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016158536585365854,
+      "loss": 2.6136,
+      "step": 4095
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016157598499061914,
+      "loss": 2.747,
+      "step": 4096
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016156660412757973,
+      "loss": 2.9777,
+      "step": 4097
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016155722326454034,
+      "loss": 2.8182,
+      "step": 4098
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016154784240150094,
+      "loss": 2.4788,
+      "step": 4099
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016153846153846155,
+      "loss": 2.4631,
+      "step": 4100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016152908067542214,
+      "loss": 2.381,
+      "step": 4101
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016151969981238276,
+      "loss": 2.7491,
+      "step": 4102
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016151031894934335,
+      "loss": 2.7003,
+      "step": 4103
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00016150093808630394,
+      "loss": 2.8756,
+      "step": 4104
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016149155722326454,
+      "loss": 2.4964,
+      "step": 4105
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016148217636022515,
+      "loss": 2.6896,
+      "step": 4106
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016147279549718574,
+      "loss": 2.5936,
+      "step": 4107
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016146341463414636,
+      "loss": 2.857,
+      "step": 4108
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016145403377110695,
+      "loss": 2.7877,
+      "step": 4109
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016144465290806754,
+      "loss": 2.8112,
+      "step": 4110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016143527204502816,
+      "loss": 2.4985,
+      "step": 4111
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016142589118198875,
+      "loss": 2.3477,
+      "step": 4112
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016141651031894934,
+      "loss": 2.7147,
+      "step": 4113
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016140712945590994,
+      "loss": 2.2425,
+      "step": 4114
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016139774859287055,
+      "loss": 2.4512,
+      "step": 4115
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016138836772983114,
+      "loss": 2.9064,
+      "step": 4116
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016137898686679176,
+      "loss": 2.423,
+      "step": 4117
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016136960600375235,
+      "loss": 2.6373,
+      "step": 4118
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016136022514071297,
+      "loss": 2.6161,
+      "step": 4119
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016135084427767356,
+      "loss": 2.4443,
+      "step": 4120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016134146341463415,
+      "loss": 2.43,
+      "step": 4121
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016133208255159474,
+      "loss": 2.7103,
+      "step": 4122
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016132270168855533,
+      "loss": 2.3951,
+      "step": 4123
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016131332082551595,
+      "loss": 3.059,
+      "step": 4124
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 0.00016130393996247654,
+      "loss": 2.5631,
+      "step": 4125
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016129455909943716,
+      "loss": 2.4294,
+      "step": 4126
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016128517823639775,
+      "loss": 2.5061,
+      "step": 4127
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016127579737335837,
+      "loss": 2.817,
+      "step": 4128
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016126641651031896,
+      "loss": 2.7667,
+      "step": 4129
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016125703564727955,
+      "loss": 3.0027,
+      "step": 4130
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016124765478424014,
+      "loss": 2.8919,
+      "step": 4131
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016123827392120076,
+      "loss": 2.7175,
+      "step": 4132
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016122889305816135,
+      "loss": 2.9163,
+      "step": 4133
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016121951219512197,
+      "loss": 2.8002,
+      "step": 4134
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016121013133208256,
+      "loss": 2.6096,
+      "step": 4135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016120075046904315,
+      "loss": 2.435,
+      "step": 4136
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016119136960600377,
+      "loss": 2.7852,
+      "step": 4137
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016118198874296436,
+      "loss": 2.5853,
+      "step": 4138
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016117260787992495,
+      "loss": 2.6613,
+      "step": 4139
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016116322701688554,
+      "loss": 2.5206,
+      "step": 4140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016115384615384616,
+      "loss": 2.642,
+      "step": 4141
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016114446529080675,
+      "loss": 2.7993,
+      "step": 4142
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016113508442776737,
+      "loss": 2.5837,
+      "step": 4143
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016112570356472796,
+      "loss": 2.7013,
+      "step": 4144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016111632270168858,
+      "loss": 2.5499,
+      "step": 4145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.00016110694183864917,
+      "loss": 2.7058,
+      "step": 4146
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016109756097560976,
+      "loss": 2.5037,
+      "step": 4147
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016108818011257035,
+      "loss": 2.9363,
+      "step": 4148
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016107879924953094,
+      "loss": 2.5336,
+      "step": 4149
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016106941838649156,
+      "loss": 2.8657,
+      "step": 4150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016106003752345215,
+      "loss": 2.8976,
+      "step": 4151
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016105065666041277,
+      "loss": 2.6973,
+      "step": 4152
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016104127579737336,
+      "loss": 2.401,
+      "step": 4153
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016103189493433398,
+      "loss": 2.5991,
+      "step": 4154
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016102251407129457,
+      "loss": 2.5995,
+      "step": 4155
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016101313320825516,
+      "loss": 2.7195,
+      "step": 4156
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016100375234521575,
+      "loss": 2.7846,
+      "step": 4157
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016099437148217637,
+      "loss": 2.4251,
+      "step": 4158
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016098499061913696,
+      "loss": 2.6358,
+      "step": 4159
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016097560975609758,
+      "loss": 2.5617,
+      "step": 4160
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016096622889305817,
+      "loss": 2.4611,
+      "step": 4161
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016095684803001876,
+      "loss": 2.3117,
+      "step": 4162
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016094746716697938,
+      "loss": 2.6469,
+      "step": 4163
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016093808630393997,
+      "loss": 2.4819,
+      "step": 4164
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016092870544090056,
+      "loss": 2.7945,
+      "step": 4165
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016091932457786115,
+      "loss": 2.5133,
+      "step": 4166
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016090994371482177,
+      "loss": 2.4385,
+      "step": 4167
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00016090056285178236,
+      "loss": 3.0925,
+      "step": 4168
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016089118198874298,
+      "loss": 2.5047,
+      "step": 4169
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016088180112570357,
+      "loss": 2.4428,
+      "step": 4170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001608724202626642,
+      "loss": 2.7364,
+      "step": 4171
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016086303939962478,
+      "loss": 2.7336,
+      "step": 4172
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016085365853658537,
+      "loss": 2.8493,
+      "step": 4173
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016084427767354596,
+      "loss": 2.7381,
+      "step": 4174
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016083489681050655,
+      "loss": 2.4805,
+      "step": 4175
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016082551594746717,
+      "loss": 2.9908,
+      "step": 4176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016081613508442776,
+      "loss": 2.5088,
+      "step": 4177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016080675422138838,
+      "loss": 3.0171,
+      "step": 4178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016079737335834897,
+      "loss": 2.5843,
+      "step": 4179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001607879924953096,
+      "loss": 2.9156,
+      "step": 4180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016077861163227018,
+      "loss": 2.4578,
+      "step": 4181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016076923076923077,
+      "loss": 3.0387,
+      "step": 4182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016075984990619136,
+      "loss": 2.6059,
+      "step": 4183
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016075046904315198,
+      "loss": 2.4996,
+      "step": 4184
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016074108818011257,
+      "loss": 2.758,
+      "step": 4185
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.0001607317073170732,
+      "loss": 2.667,
+      "step": 4186
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016072232645403378,
+      "loss": 2.8222,
+      "step": 4187
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00016071294559099437,
+      "loss": 2.6463,
+      "step": 4188
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.000160703564727955,
+      "loss": 2.8315,
+      "step": 4189
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016069418386491558,
+      "loss": 2.4974,
+      "step": 4190
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016068480300187617,
+      "loss": 2.5474,
+      "step": 4191
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016067542213883676,
+      "loss": 2.606,
+      "step": 4192
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016066604127579738,
+      "loss": 2.9196,
+      "step": 4193
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016065666041275797,
+      "loss": 2.5604,
+      "step": 4194
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001606472795497186,
+      "loss": 2.5176,
+      "step": 4195
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016063789868667918,
+      "loss": 2.5784,
+      "step": 4196
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001606285178236398,
+      "loss": 2.5692,
+      "step": 4197
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001606191369606004,
+      "loss": 2.4823,
+      "step": 4198
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000160609756097561,
+      "loss": 2.6247,
+      "step": 4199
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016060037523452157,
+      "loss": 2.4799,
+      "step": 4200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016059099437148216,
+      "loss": 2.5021,
+      "step": 4201
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016058161350844278,
+      "loss": 2.9064,
+      "step": 4202
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016057223264540337,
+      "loss": 2.7843,
+      "step": 4203
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000160562851782364,
+      "loss": 2.7897,
+      "step": 4204
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00016055347091932458,
+      "loss": 2.7682,
+      "step": 4205
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001605440900562852,
+      "loss": 2.5505,
+      "step": 4206
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001605347091932458,
+      "loss": 2.5553,
+      "step": 4207
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001605253283302064,
+      "loss": 2.7328,
+      "step": 4208
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.000160515947467167,
+      "loss": 2.6528,
+      "step": 4209
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.0001605065666041276,
+      "loss": 2.9207,
+      "step": 4210
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00016049718574108818,
+      "loss": 2.8282,
+      "step": 4211
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001604878048780488,
+      "loss": 2.5891,
+      "step": 4212
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001604784240150094,
+      "loss": 2.5119,
+      "step": 4213
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00016046904315197,
+      "loss": 2.6426,
+      "step": 4214
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001604596622889306,
+      "loss": 2.5548,
+      "step": 4215
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001604502814258912,
+      "loss": 2.6827,
+      "step": 4216
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001604409005628518,
+      "loss": 2.3476,
+      "step": 4217
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001604315196998124,
+      "loss": 2.5322,
+      "step": 4218
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000160422138836773,
+      "loss": 2.695,
+      "step": 4219
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00016041275797373358,
+      "loss": 2.9052,
+      "step": 4220
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001604033771106942,
+      "loss": 2.6413,
+      "step": 4221
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001603939962476548,
+      "loss": 2.4719,
+      "step": 4222
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001603846153846154,
+      "loss": 2.86,
+      "step": 4223
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.000160375234521576,
+      "loss": 2.6855,
+      "step": 4224
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00016036585365853661,
+      "loss": 2.7983,
+      "step": 4225
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001603564727954972,
+      "loss": 2.7687,
+      "step": 4226
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001603470919324578,
+      "loss": 2.6046,
+      "step": 4227
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001603377110694184,
+      "loss": 2.9213,
+      "step": 4228
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.00016032833020637898,
+      "loss": 2.5634,
+      "step": 4229
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001603189493433396,
+      "loss": 2.5346,
+      "step": 4230
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001603095684803002,
+      "loss": 2.6682,
+      "step": 4231
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 0.0001603001876172608,
+      "loss": 2.4205,
+      "step": 4232
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001602908067542214,
+      "loss": 2.6869,
+      "step": 4233
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00016028142589118201,
+      "loss": 2.5726,
+      "step": 4234
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001602720450281426,
+      "loss": 2.8606,
+      "step": 4235
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001602626641651032,
+      "loss": 2.751,
+      "step": 4236
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001602532833020638,
+      "loss": 2.8042,
+      "step": 4237
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001602439024390244,
+      "loss": 2.6361,
+      "step": 4238
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000160234521575985,
+      "loss": 2.6214,
+      "step": 4239
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00016022514071294561,
+      "loss": 2.8811,
+      "step": 4240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001602157598499062,
+      "loss": 2.712,
+      "step": 4241
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001602063789868668,
+      "loss": 2.6892,
+      "step": 4242
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00016019699812382741,
+      "loss": 2.4703,
+      "step": 4243
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000160187617260788,
+      "loss": 2.7969,
+      "step": 4244
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001601782363977486,
+      "loss": 2.7607,
+      "step": 4245
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00016016885553470919,
+      "loss": 2.6963,
+      "step": 4246
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001601594746716698,
+      "loss": 2.7542,
+      "step": 4247
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001601500938086304,
+      "loss": 2.7268,
+      "step": 4248
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.000160140712945591,
+      "loss": 2.8696,
+      "step": 4249
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001601313320825516,
+      "loss": 2.3562,
+      "step": 4250
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.00016012195121951222,
+      "loss": 2.6008,
+      "step": 4251
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001601125703564728,
+      "loss": 2.3246,
+      "step": 4252
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001601031894934334,
+      "loss": 2.5882,
+      "step": 4253
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.000160093808630394,
+      "loss": 2.6346,
+      "step": 4254
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00016008442776735459,
+      "loss": 2.8709,
+      "step": 4255
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001600750469043152,
+      "loss": 2.4627,
+      "step": 4256
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001600656660412758,
+      "loss": 2.6251,
+      "step": 4257
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001600562851782364,
+      "loss": 2.7109,
+      "step": 4258
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.000160046904315197,
+      "loss": 2.5865,
+      "step": 4259
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00016003752345215762,
+      "loss": 2.5123,
+      "step": 4260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001600281425891182,
+      "loss": 2.5783,
+      "step": 4261
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001600187617260788,
+      "loss": 2.414,
+      "step": 4262
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001600093808630394,
+      "loss": 2.4989,
+      "step": 4263
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00016,
+      "loss": 2.6924,
+      "step": 4264
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001599906191369606,
+      "loss": 2.8841,
+      "step": 4265
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015998123827392122,
+      "loss": 2.3258,
+      "step": 4266
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001599718574108818,
+      "loss": 2.4356,
+      "step": 4267
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001599624765478424,
+      "loss": 2.803,
+      "step": 4268
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00015995309568480302,
+      "loss": 2.5616,
+      "step": 4269
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001599437148217636,
+      "loss": 2.7071,
+      "step": 4270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001599343339587242,
+      "loss": 2.8097,
+      "step": 4271
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001599249530956848,
+      "loss": 2.6138,
+      "step": 4272
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.0001599155722326454,
+      "loss": 2.4676,
+      "step": 4273
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.000159906191369606,
+      "loss": 2.6412,
+      "step": 4274
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015989681050656662,
+      "loss": 2.6454,
+      "step": 4275
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001598874296435272,
+      "loss": 2.5127,
+      "step": 4276
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015987804878048783,
+      "loss": 2.8622,
+      "step": 4277
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015986866791744842,
+      "loss": 2.7196,
+      "step": 4278
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000159859287054409,
+      "loss": 2.6336,
+      "step": 4279
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001598499061913696,
+      "loss": 2.5792,
+      "step": 4280
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001598405253283302,
+      "loss": 2.5589,
+      "step": 4281
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001598311444652908,
+      "loss": 2.7144,
+      "step": 4282
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001598217636022514,
+      "loss": 2.4687,
+      "step": 4283
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015981238273921202,
+      "loss": 2.6438,
+      "step": 4284
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001598030018761726,
+      "loss": 2.5238,
+      "step": 4285
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015979362101313323,
+      "loss": 2.6789,
+      "step": 4286
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015978424015009382,
+      "loss": 2.5925,
+      "step": 4287
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001597748592870544,
+      "loss": 2.6846,
+      "step": 4288
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000159765478424015,
+      "loss": 2.7189,
+      "step": 4289
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015975609756097562,
+      "loss": 2.5373,
+      "step": 4290
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.0001597467166979362,
+      "loss": 2.7961,
+      "step": 4291
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015973733583489683,
+      "loss": 2.496,
+      "step": 4292
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015972795497185742,
+      "loss": 2.4313,
+      "step": 4293
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.000159718574108818,
+      "loss": 2.4178,
+      "step": 4294
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00015970919324577863,
+      "loss": 2.3434,
+      "step": 4295
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015969981238273922,
+      "loss": 2.7571,
+      "step": 4296
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001596904315196998,
+      "loss": 2.1461,
+      "step": 4297
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001596810506566604,
+      "loss": 2.3914,
+      "step": 4298
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015967166979362102,
+      "loss": 2.4864,
+      "step": 4299
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001596622889305816,
+      "loss": 2.6362,
+      "step": 4300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015965290806754223,
+      "loss": 2.6122,
+      "step": 4301
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015964352720450282,
+      "loss": 2.2975,
+      "step": 4302
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015963414634146344,
+      "loss": 2.7408,
+      "step": 4303
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015962476547842403,
+      "loss": 2.3459,
+      "step": 4304
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015961538461538462,
+      "loss": 2.2357,
+      "step": 4305
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001596060037523452,
+      "loss": 2.8457,
+      "step": 4306
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001595966228893058,
+      "loss": 2.6238,
+      "step": 4307
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015958724202626642,
+      "loss": 2.5209,
+      "step": 4308
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.000159577861163227,
+      "loss": 2.5543,
+      "step": 4309
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015956848030018763,
+      "loss": 2.3184,
+      "step": 4310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015955909943714822,
+      "loss": 2.2894,
+      "step": 4311
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015954971857410884,
+      "loss": 2.3982,
+      "step": 4312
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015954033771106943,
+      "loss": 2.6022,
+      "step": 4313
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015953095684803002,
+      "loss": 2.571,
+      "step": 4314
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.0001595215759849906,
+      "loss": 2.7985,
+      "step": 4315
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015951219512195123,
+      "loss": 2.8214,
+      "step": 4316
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 0.00015950281425891182,
+      "loss": 2.6328,
+      "step": 4317
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015949343339587244,
+      "loss": 2.514,
+      "step": 4318
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015948405253283303,
+      "loss": 2.2955,
+      "step": 4319
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015947467166979362,
+      "loss": 2.5515,
+      "step": 4320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015946529080675424,
+      "loss": 2.5283,
+      "step": 4321
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015945590994371483,
+      "loss": 2.7762,
+      "step": 4322
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015944652908067542,
+      "loss": 2.2884,
+      "step": 4323
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000159437148217636,
+      "loss": 2.1808,
+      "step": 4324
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015942776735459663,
+      "loss": 2.3423,
+      "step": 4325
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015941838649155722,
+      "loss": 2.7525,
+      "step": 4326
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015940900562851784,
+      "loss": 2.737,
+      "step": 4327
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015939962476547843,
+      "loss": 2.7176,
+      "step": 4328
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015939024390243905,
+      "loss": 2.5015,
+      "step": 4329
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015938086303939964,
+      "loss": 2.7966,
+      "step": 4330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015937148217636023,
+      "loss": 2.504,
+      "step": 4331
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015936210131332082,
+      "loss": 2.6491,
+      "step": 4332
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.0001593527204502814,
+      "loss": 2.6621,
+      "step": 4333
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015934333958724203,
+      "loss": 2.3439,
+      "step": 4334
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015933395872420262,
+      "loss": 2.5198,
+      "step": 4335
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015932457786116324,
+      "loss": 2.4023,
+      "step": 4336
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015931519699812383,
+      "loss": 2.7751,
+      "step": 4337
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.00015930581613508445,
+      "loss": 2.5927,
+      "step": 4338
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015929643527204504,
+      "loss": 2.4802,
+      "step": 4339
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015928705440900563,
+      "loss": 2.6819,
+      "step": 4340
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015927767354596622,
+      "loss": 2.5051,
+      "step": 4341
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015926829268292684,
+      "loss": 2.4713,
+      "step": 4342
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015925891181988743,
+      "loss": 2.5482,
+      "step": 4343
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015924953095684805,
+      "loss": 2.6316,
+      "step": 4344
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015924015009380864,
+      "loss": 2.6454,
+      "step": 4345
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015923076923076923,
+      "loss": 2.7869,
+      "step": 4346
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015922138836772985,
+      "loss": 2.4611,
+      "step": 4347
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015921200750469044,
+      "loss": 2.4787,
+      "step": 4348
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015920262664165103,
+      "loss": 2.9897,
+      "step": 4349
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015919324577861162,
+      "loss": 2.958,
+      "step": 4350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015918386491557224,
+      "loss": 2.6433,
+      "step": 4351
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015917448405253283,
+      "loss": 2.3537,
+      "step": 4352
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015916510318949345,
+      "loss": 2.5942,
+      "step": 4353
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015915572232645404,
+      "loss": 2.5404,
+      "step": 4354
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015914634146341466,
+      "loss": 2.4124,
+      "step": 4355
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015913696060037525,
+      "loss": 2.5295,
+      "step": 4356
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015912757973733587,
+      "loss": 2.6119,
+      "step": 4357
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015911819887429643,
+      "loss": 2.4915,
+      "step": 4358
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00015910881801125702,
+      "loss": 2.6004,
+      "step": 4359
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015909943714821764,
+      "loss": 2.4624,
+      "step": 4360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015909005628517823,
+      "loss": 2.6588,
+      "step": 4361
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015908067542213885,
+      "loss": 2.8425,
+      "step": 4362
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015907129455909944,
+      "loss": 2.8369,
+      "step": 4363
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015906191369606006,
+      "loss": 2.694,
+      "step": 4364
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015905253283302065,
+      "loss": 2.8706,
+      "step": 4365
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015904315196998127,
+      "loss": 2.8241,
+      "step": 4366
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015903377110694183,
+      "loss": 2.6251,
+      "step": 4367
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015902439024390245,
+      "loss": 2.73,
+      "step": 4368
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015901500938086304,
+      "loss": 2.8176,
+      "step": 4369
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015900562851782366,
+      "loss": 2.7228,
+      "step": 4370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015899624765478425,
+      "loss": 2.7757,
+      "step": 4371
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015898686679174484,
+      "loss": 2.5456,
+      "step": 4372
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015897748592870546,
+      "loss": 2.5219,
+      "step": 4373
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015896810506566605,
+      "loss": 2.6633,
+      "step": 4374
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015895872420262666,
+      "loss": 2.834,
+      "step": 4375
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015894934333958726,
+      "loss": 2.5723,
+      "step": 4376
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015893996247654785,
+      "loss": 2.6135,
+      "step": 4377
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015893058161350844,
+      "loss": 2.3467,
+      "step": 4378
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015892120075046906,
+      "loss": 2.5899,
+      "step": 4379
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015891181988742965,
+      "loss": 2.638,
+      "step": 4380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00015890243902439026,
+      "loss": 2.5398,
+      "step": 4381
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015889305816135086,
+      "loss": 2.7,
+      "step": 4382
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015888367729831147,
+      "loss": 2.6375,
+      "step": 4383
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015887429643527206,
+      "loss": 2.9955,
+      "step": 4384
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015886491557223266,
+      "loss": 2.3147,
+      "step": 4385
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015885553470919325,
+      "loss": 2.5985,
+      "step": 4386
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015884615384615384,
+      "loss": 2.431,
+      "step": 4387
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015883677298311446,
+      "loss": 2.6375,
+      "step": 4388
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015882739212007505,
+      "loss": 2.9809,
+      "step": 4389
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015881801125703566,
+      "loss": 2.8009,
+      "step": 4390
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015880863039399626,
+      "loss": 2.3156,
+      "step": 4391
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015879924953095687,
+      "loss": 2.6264,
+      "step": 4392
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015878986866791746,
+      "loss": 2.4117,
+      "step": 4393
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015878048780487806,
+      "loss": 2.5986,
+      "step": 4394
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015877110694183865,
+      "loss": 2.6604,
+      "step": 4395
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015876172607879926,
+      "loss": 2.6349,
+      "step": 4396
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015875234521575986,
+      "loss": 2.5371,
+      "step": 4397
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015874296435272045,
+      "loss": 2.6603,
+      "step": 4398
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015873358348968106,
+      "loss": 2.6669,
+      "step": 4399
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015872420262664166,
+      "loss": 2.9338,
+      "step": 4400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015871482176360227,
+      "loss": 2.6395,
+      "step": 4401
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.00015870544090056286,
+      "loss": 2.8299,
+      "step": 4402
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015869606003752346,
+      "loss": 2.455,
+      "step": 4403
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015868667917448405,
+      "loss": 2.5819,
+      "step": 4404
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015867729831144466,
+      "loss": 2.7593,
+      "step": 4405
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015866791744840526,
+      "loss": 2.8726,
+      "step": 4406
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015865853658536587,
+      "loss": 2.8762,
+      "step": 4407
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015864915572232646,
+      "loss": 2.6853,
+      "step": 4408
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015863977485928708,
+      "loss": 2.7081,
+      "step": 4409
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015863039399624767,
+      "loss": 2.631,
+      "step": 4410
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015862101313320826,
+      "loss": 2.7838,
+      "step": 4411
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015861163227016885,
+      "loss": 2.5972,
+      "step": 4412
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015860225140712945,
+      "loss": 2.6454,
+      "step": 4413
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015859287054409006,
+      "loss": 2.7778,
+      "step": 4414
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015858348968105065,
+      "loss": 2.5076,
+      "step": 4415
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015857410881801127,
+      "loss": 2.6991,
+      "step": 4416
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015856472795497186,
+      "loss": 3.1005,
+      "step": 4417
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015855534709193248,
+      "loss": 2.5318,
+      "step": 4418
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015854596622889307,
+      "loss": 2.8442,
+      "step": 4419
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015853658536585366,
+      "loss": 2.4801,
+      "step": 4420
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015852720450281425,
+      "loss": 2.5809,
+      "step": 4421
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015851782363977487,
+      "loss": 2.6102,
+      "step": 4422
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 0.00015850844277673546,
+      "loss": 2.6515,
+      "step": 4423
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015849906191369608,
+      "loss": 2.5928,
+      "step": 4424
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015848968105065667,
+      "loss": 2.686,
+      "step": 4425
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015848030018761726,
+      "loss": 2.8485,
+      "step": 4426
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015847091932457788,
+      "loss": 2.5993,
+      "step": 4427
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015846153846153847,
+      "loss": 2.4292,
+      "step": 4428
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015845215759849906,
+      "loss": 2.7786,
+      "step": 4429
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015844277673545965,
+      "loss": 2.5398,
+      "step": 4430
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015843339587242027,
+      "loss": 2.475,
+      "step": 4431
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015842401500938086,
+      "loss": 2.591,
+      "step": 4432
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015841463414634148,
+      "loss": 2.619,
+      "step": 4433
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015840525328330207,
+      "loss": 2.49,
+      "step": 4434
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001583958724202627,
+      "loss": 2.4578,
+      "step": 4435
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015838649155722328,
+      "loss": 2.6424,
+      "step": 4436
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015837711069418387,
+      "loss": 2.4022,
+      "step": 4437
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015836772983114446,
+      "loss": 2.2913,
+      "step": 4438
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015835834896810505,
+      "loss": 2.6793,
+      "step": 4439
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015834896810506567,
+      "loss": 2.4993,
+      "step": 4440
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015833958724202626,
+      "loss": 2.6183,
+      "step": 4441
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015833020637898688,
+      "loss": 2.4326,
+      "step": 4442
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015832082551594747,
+      "loss": 2.4443,
+      "step": 4443
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.0001583114446529081,
+      "loss": 2.8405,
+      "step": 4444
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00015830206378986868,
+      "loss": 2.882,
+      "step": 4445
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015829268292682927,
+      "loss": 2.9185,
+      "step": 4446
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015828330206378986,
+      "loss": 2.5148,
+      "step": 4447
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015827392120075048,
+      "loss": 2.5304,
+      "step": 4448
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015826454033771107,
+      "loss": 2.6908,
+      "step": 4449
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001582551594746717,
+      "loss": 2.5274,
+      "step": 4450
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015824577861163228,
+      "loss": 2.4607,
+      "step": 4451
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015823639774859287,
+      "loss": 2.7682,
+      "step": 4452
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001582270168855535,
+      "loss": 2.582,
+      "step": 4453
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015821763602251408,
+      "loss": 2.5211,
+      "step": 4454
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015820825515947467,
+      "loss": 2.54,
+      "step": 4455
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015819887429643526,
+      "loss": 2.6795,
+      "step": 4456
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015818949343339588,
+      "loss": 2.5428,
+      "step": 4457
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015818011257035647,
+      "loss": 2.558,
+      "step": 4458
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001581707317073171,
+      "loss": 2.4914,
+      "step": 4459
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015816135084427768,
+      "loss": 2.45,
+      "step": 4460
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001581519699812383,
+      "loss": 2.3112,
+      "step": 4461
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.0001581425891181989,
+      "loss": 2.5382,
+      "step": 4462
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015813320825515948,
+      "loss": 2.8523,
+      "step": 4463
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015812382739212007,
+      "loss": 2.7175,
+      "step": 4464
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015811444652908066,
+      "loss": 2.5541,
+      "step": 4465
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00015810506566604128,
+      "loss": 2.6396,
+      "step": 4466
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015809568480300187,
+      "loss": 2.5392,
+      "step": 4467
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001580863039399625,
+      "loss": 2.4749,
+      "step": 4468
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015807692307692308,
+      "loss": 2.4131,
+      "step": 4469
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001580675422138837,
+      "loss": 2.5359,
+      "step": 4470
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001580581613508443,
+      "loss": 2.732,
+      "step": 4471
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015804878048780488,
+      "loss": 2.6314,
+      "step": 4472
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015803939962476547,
+      "loss": 2.3516,
+      "step": 4473
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001580300187617261,
+      "loss": 2.4623,
+      "step": 4474
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015802063789868668,
+      "loss": 2.4334,
+      "step": 4475
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001580112570356473,
+      "loss": 2.6486,
+      "step": 4476
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001580018761726079,
+      "loss": 2.7754,
+      "step": 4477
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015799249530956848,
+      "loss": 2.6093,
+      "step": 4478
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001579831144465291,
+      "loss": 2.517,
+      "step": 4479
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001579737335834897,
+      "loss": 2.5766,
+      "step": 4480
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015796435272045028,
+      "loss": 2.3294,
+      "step": 4481
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015795497185741087,
+      "loss": 2.4976,
+      "step": 4482
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001579455909943715,
+      "loss": 2.3736,
+      "step": 4483
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.00015793621013133208,
+      "loss": 2.9948,
+      "step": 4484
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001579268292682927,
+      "loss": 2.6622,
+      "step": 4485
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001579174484052533,
+      "loss": 2.5038,
+      "step": 4486
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.0001579080675422139,
+      "loss": 2.8637,
+      "step": 4487
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001578986866791745,
+      "loss": 2.7754,
+      "step": 4488
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001578893058161351,
+      "loss": 2.5832,
+      "step": 4489
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015787992495309568,
+      "loss": 2.5565,
+      "step": 4490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015787054409005627,
+      "loss": 2.3583,
+      "step": 4491
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001578611632270169,
+      "loss": 2.4977,
+      "step": 4492
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015785178236397748,
+      "loss": 2.6984,
+      "step": 4493
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001578424015009381,
+      "loss": 2.4549,
+      "step": 4494
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001578330206378987,
+      "loss": 3.0345,
+      "step": 4495
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001578236397748593,
+      "loss": 2.5691,
+      "step": 4496
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001578142589118199,
+      "loss": 2.3983,
+      "step": 4497
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001578048780487805,
+      "loss": 2.2813,
+      "step": 4498
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015779549718574108,
+      "loss": 2.5941,
+      "step": 4499
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001577861163227017,
+      "loss": 2.541,
+      "step": 4500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001577767354596623,
+      "loss": 2.6884,
+      "step": 4501
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001577673545966229,
+      "loss": 2.3077,
+      "step": 4502
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001577579737335835,
+      "loss": 2.5913,
+      "step": 4503
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001577485928705441,
+      "loss": 2.7315,
+      "step": 4504
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001577392120075047,
+      "loss": 2.8268,
+      "step": 4505
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001577298311444653,
+      "loss": 2.4511,
+      "step": 4506
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001577204502814259,
+      "loss": 2.1029,
+      "step": 4507
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.00015771106941838648,
+      "loss": 2.6093,
+      "step": 4508
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 0.0001577016885553471,
+      "loss": 2.1777,
+      "step": 4509
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001576923076923077,
+      "loss": 2.7264,
+      "step": 4510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001576829268292683,
+      "loss": 2.5728,
+      "step": 4511
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001576735459662289,
+      "loss": 2.4928,
+      "step": 4512
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015766416510318952,
+      "loss": 2.6295,
+      "step": 4513
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001576547842401501,
+      "loss": 2.3677,
+      "step": 4514
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015764540337711072,
+      "loss": 2.5688,
+      "step": 4515
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001576360225140713,
+      "loss": 2.5962,
+      "step": 4516
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015762664165103188,
+      "loss": 2.4424,
+      "step": 4517
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001576172607879925,
+      "loss": 2.457,
+      "step": 4518
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001576078799249531,
+      "loss": 2.5801,
+      "step": 4519
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001575984990619137,
+      "loss": 2.3588,
+      "step": 4520
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001575891181988743,
+      "loss": 2.4908,
+      "step": 4521
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015757973733583492,
+      "loss": 2.7426,
+      "step": 4522
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001575703564727955,
+      "loss": 2.4304,
+      "step": 4523
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015756097560975612,
+      "loss": 2.6162,
+      "step": 4524
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001575515947467167,
+      "loss": 2.6822,
+      "step": 4525
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001575422138836773,
+      "loss": 2.6371,
+      "step": 4526
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001575328330206379,
+      "loss": 2.7805,
+      "step": 4527
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.00015752345215759852,
+      "loss": 2.9804,
+      "step": 4528
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001575140712945591,
+      "loss": 2.4899,
+      "step": 4529
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001575046904315197,
+      "loss": 2.5062,
+      "step": 4530
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015749530956848032,
+      "loss": 2.8375,
+      "step": 4531
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001574859287054409,
+      "loss": 2.781,
+      "step": 4532
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015747654784240152,
+      "loss": 2.5738,
+      "step": 4533
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001574671669793621,
+      "loss": 2.6057,
+      "step": 4534
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001574577861163227,
+      "loss": 2.468,
+      "step": 4535
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001574484052532833,
+      "loss": 2.4755,
+      "step": 4536
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015743902439024392,
+      "loss": 2.4174,
+      "step": 4537
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001574296435272045,
+      "loss": 2.6015,
+      "step": 4538
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015742026266416512,
+      "loss": 2.4516,
+      "step": 4539
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015741088180112572,
+      "loss": 2.5687,
+      "step": 4540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015740150093808633,
+      "loss": 2.4381,
+      "step": 4541
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015739212007504692,
+      "loss": 2.326,
+      "step": 4542
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001573827392120075,
+      "loss": 2.6144,
+      "step": 4543
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001573733583489681,
+      "loss": 2.8568,
+      "step": 4544
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001573639774859287,
+      "loss": 2.7369,
+      "step": 4545
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015735459662288932,
+      "loss": 2.6062,
+      "step": 4546
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.0001573452157598499,
+      "loss": 2.7098,
+      "step": 4547
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015733583489681052,
+      "loss": 2.8925,
+      "step": 4548
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015732645403377112,
+      "loss": 2.8546,
+      "step": 4549
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015731707317073173,
+      "loss": 2.7105,
+      "step": 4550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00015730769230769232,
+      "loss": 2.4396,
+      "step": 4551
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015729831144465292,
+      "loss": 2.2911,
+      "step": 4552
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001572889305816135,
+      "loss": 2.5974,
+      "step": 4553
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015727954971857412,
+      "loss": 2.6138,
+      "step": 4554
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015727016885553472,
+      "loss": 2.8115,
+      "step": 4555
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001572607879924953,
+      "loss": 2.6296,
+      "step": 4556
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015725140712945592,
+      "loss": 2.701,
+      "step": 4557
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015724202626641651,
+      "loss": 2.6037,
+      "step": 4558
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015723264540337713,
+      "loss": 2.405,
+      "step": 4559
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015722326454033772,
+      "loss": 2.4049,
+      "step": 4560
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015721388367729831,
+      "loss": 2.7859,
+      "step": 4561
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001572045028142589,
+      "loss": 2.7079,
+      "step": 4562
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015719512195121952,
+      "loss": 2.762,
+      "step": 4563
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015718574108818011,
+      "loss": 2.4563,
+      "step": 4564
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015717636022514073,
+      "loss": 2.356,
+      "step": 4565
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015716697936210132,
+      "loss": 2.4473,
+      "step": 4566
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015715759849906194,
+      "loss": 2.4606,
+      "step": 4567
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015714821763602253,
+      "loss": 2.8029,
+      "step": 4568
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015713883677298312,
+      "loss": 2.7371,
+      "step": 4569
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015712945590994371,
+      "loss": 2.7657,
+      "step": 4570
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.0001571200750469043,
+      "loss": 2.6155,
+      "step": 4571
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015711069418386492,
+      "loss": 2.4182,
+      "step": 4572
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00015710131332082551,
+      "loss": 2.6032,
+      "step": 4573
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015709193245778613,
+      "loss": 2.7099,
+      "step": 4574
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015708255159474672,
+      "loss": 2.3518,
+      "step": 4575
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015707317073170734,
+      "loss": 2.4256,
+      "step": 4576
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015706378986866793,
+      "loss": 2.3353,
+      "step": 4577
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015705440900562852,
+      "loss": 2.396,
+      "step": 4578
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015704502814258911,
+      "loss": 2.5121,
+      "step": 4579
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015703564727954973,
+      "loss": 2.5774,
+      "step": 4580
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015702626641651032,
+      "loss": 2.5347,
+      "step": 4581
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015701688555347091,
+      "loss": 2.605,
+      "step": 4582
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015700750469043153,
+      "loss": 2.3884,
+      "step": 4583
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015699812382739212,
+      "loss": 2.7602,
+      "step": 4584
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015698874296435274,
+      "loss": 2.3384,
+      "step": 4585
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015697936210131333,
+      "loss": 2.6052,
+      "step": 4586
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015696998123827392,
+      "loss": 2.7431,
+      "step": 4587
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015696060037523451,
+      "loss": 2.5941,
+      "step": 4588
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015695121951219513,
+      "loss": 2.8989,
+      "step": 4589
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015694183864915572,
+      "loss": 2.6636,
+      "step": 4590
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015693245778611634,
+      "loss": 2.7466,
+      "step": 4591
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015692307692307693,
+      "loss": 2.4429,
+      "step": 4592
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015691369606003755,
+      "loss": 2.7664,
+      "step": 4593
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00015690431519699814,
+      "loss": 2.5882,
+      "step": 4594
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015689493433395873,
+      "loss": 2.6781,
+      "step": 4595
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015688555347091932,
+      "loss": 2.7669,
+      "step": 4596
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015687617260787991,
+      "loss": 2.4841,
+      "step": 4597
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015686679174484053,
+      "loss": 2.7594,
+      "step": 4598
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015685741088180112,
+      "loss": 2.6214,
+      "step": 4599
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015684803001876174,
+      "loss": 2.4713,
+      "step": 4600
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015683864915572233,
+      "loss": 2.3734,
+      "step": 4601
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015682926829268295,
+      "loss": 2.8466,
+      "step": 4602
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015681988742964354,
+      "loss": 2.5066,
+      "step": 4603
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015681050656660413,
+      "loss": 2.739,
+      "step": 4604
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015680112570356472,
+      "loss": 2.7431,
+      "step": 4605
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015679174484052534,
+      "loss": 2.5016,
+      "step": 4606
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015678236397748593,
+      "loss": 2.595,
+      "step": 4607
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015677298311444652,
+      "loss": 2.6689,
+      "step": 4608
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015676360225140714,
+      "loss": 2.2807,
+      "step": 4609
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015675422138836773,
+      "loss": 2.6073,
+      "step": 4610
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015674484052532835,
+      "loss": 2.703,
+      "step": 4611
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015673545966228894,
+      "loss": 2.4423,
+      "step": 4612
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015672607879924953,
+      "loss": 2.7458,
+      "step": 4613
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015671669793621012,
+      "loss": 2.5413,
+      "step": 4614
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 0.00015670731707317074,
+      "loss": 2.9935,
+      "step": 4615
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015669793621013133,
+      "loss": 2.8188,
+      "step": 4616
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015668855534709195,
+      "loss": 2.5313,
+      "step": 4617
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015667917448405254,
+      "loss": 2.6752,
+      "step": 4618
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015666979362101316,
+      "loss": 2.6092,
+      "step": 4619
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015666041275797375,
+      "loss": 2.5309,
+      "step": 4620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015665103189493434,
+      "loss": 2.2886,
+      "step": 4621
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015664165103189493,
+      "loss": 2.5226,
+      "step": 4622
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015663227016885552,
+      "loss": 2.4919,
+      "step": 4623
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015662288930581614,
+      "loss": 2.44,
+      "step": 4624
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015661350844277673,
+      "loss": 2.5225,
+      "step": 4625
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015660412757973735,
+      "loss": 2.3869,
+      "step": 4626
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015659474671669794,
+      "loss": 2.4686,
+      "step": 4627
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015658536585365856,
+      "loss": 2.7758,
+      "step": 4628
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015657598499061915,
+      "loss": 2.6064,
+      "step": 4629
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015656660412757974,
+      "loss": 2.6679,
+      "step": 4630
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015655722326454033,
+      "loss": 2.4254,
+      "step": 4631
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015654784240150095,
+      "loss": 2.2285,
+      "step": 4632
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015653846153846154,
+      "loss": 2.5671,
+      "step": 4633
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015652908067542216,
+      "loss": 2.8871,
+      "step": 4634
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015651969981238275,
+      "loss": 2.589,
+      "step": 4635
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015651031894934334,
+      "loss": 2.5392,
+      "step": 4636
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.00015650093808630396,
+      "loss": 2.8035,
+      "step": 4637
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015649155722326455,
+      "loss": 2.7223,
+      "step": 4638
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015648217636022514,
+      "loss": 2.5631,
+      "step": 4639
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015647279549718573,
+      "loss": 2.5549,
+      "step": 4640
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015646341463414635,
+      "loss": 2.3788,
+      "step": 4641
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015645403377110694,
+      "loss": 2.589,
+      "step": 4642
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015644465290806756,
+      "loss": 2.5018,
+      "step": 4643
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015643527204502815,
+      "loss": 2.7609,
+      "step": 4644
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015642589118198877,
+      "loss": 2.6805,
+      "step": 4645
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015641651031894936,
+      "loss": 2.4299,
+      "step": 4646
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015640712945590995,
+      "loss": 2.4477,
+      "step": 4647
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015639774859287054,
+      "loss": 2.468,
+      "step": 4648
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015638836772983113,
+      "loss": 2.8652,
+      "step": 4649
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015637898686679175,
+      "loss": 2.7143,
+      "step": 4650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015636960600375234,
+      "loss": 2.6036,
+      "step": 4651
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015636022514071296,
+      "loss": 2.5935,
+      "step": 4652
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015635084427767355,
+      "loss": 2.6033,
+      "step": 4653
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015634146341463417,
+      "loss": 2.3635,
+      "step": 4654
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015633208255159476,
+      "loss": 2.5079,
+      "step": 4655
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015632270168855535,
+      "loss": 2.523,
+      "step": 4656
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015631332082551594,
+      "loss": 2.6048,
+      "step": 4657
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00015630393996247656,
+      "loss": 2.3536,
+      "step": 4658
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015629455909943715,
+      "loss": 2.7908,
+      "step": 4659
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015628517823639777,
+      "loss": 2.384,
+      "step": 4660
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015627579737335836,
+      "loss": 2.8094,
+      "step": 4661
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015626641651031895,
+      "loss": 2.5696,
+      "step": 4662
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015625703564727957,
+      "loss": 2.4851,
+      "step": 4663
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015624765478424016,
+      "loss": 2.363,
+      "step": 4664
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015623827392120075,
+      "loss": 2.5613,
+      "step": 4665
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015622889305816134,
+      "loss": 2.5545,
+      "step": 4666
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015621951219512196,
+      "loss": 2.5806,
+      "step": 4667
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015621013133208255,
+      "loss": 2.509,
+      "step": 4668
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015620075046904317,
+      "loss": 2.821,
+      "step": 4669
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015619136960600376,
+      "loss": 2.4945,
+      "step": 4670
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015618198874296438,
+      "loss": 2.6719,
+      "step": 4671
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015617260787992497,
+      "loss": 2.644,
+      "step": 4672
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015616322701688558,
+      "loss": 2.6105,
+      "step": 4673
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015615384615384615,
+      "loss": 2.2277,
+      "step": 4674
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015614446529080674,
+      "loss": 2.8137,
+      "step": 4675
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015613508442776736,
+      "loss": 2.4648,
+      "step": 4676
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015612570356472795,
+      "loss": 2.71,
+      "step": 4677
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015611632270168857,
+      "loss": 2.5108,
+      "step": 4678
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00015610694183864916,
+      "loss": 2.5616,
+      "step": 4679
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015609756097560978,
+      "loss": 2.8568,
+      "step": 4680
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015608818011257037,
+      "loss": 2.7518,
+      "step": 4681
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015607879924953098,
+      "loss": 2.4356,
+      "step": 4682
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015606941838649155,
+      "loss": 2.7974,
+      "step": 4683
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015606003752345217,
+      "loss": 2.4115,
+      "step": 4684
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015605065666041276,
+      "loss": 2.5956,
+      "step": 4685
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015604127579737338,
+      "loss": 2.5632,
+      "step": 4686
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015603189493433397,
+      "loss": 2.4229,
+      "step": 4687
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015602251407129456,
+      "loss": 2.71,
+      "step": 4688
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015601313320825518,
+      "loss": 2.6684,
+      "step": 4689
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015600375234521577,
+      "loss": 2.5481,
+      "step": 4690
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015599437148217638,
+      "loss": 2.5983,
+      "step": 4691
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015598499061913695,
+      "loss": 2.6677,
+      "step": 4692
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015597560975609757,
+      "loss": 2.3774,
+      "step": 4693
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015596622889305816,
+      "loss": 2.6539,
+      "step": 4694
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015595684803001878,
+      "loss": 2.8216,
+      "step": 4695
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015594746716697937,
+      "loss": 2.6979,
+      "step": 4696
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015593808630393998,
+      "loss": 2.568,
+      "step": 4697
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015592870544090058,
+      "loss": 2.6946,
+      "step": 4698
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.0001559193245778612,
+      "loss": 2.8089,
+      "step": 4699
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015590994371482178,
+      "loss": 2.7449,
+      "step": 4700
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 0.00015590056285178235,
+      "loss": 2.3646,
+      "step": 4701
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015589118198874297,
+      "loss": 2.427,
+      "step": 4702
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015588180112570356,
+      "loss": 2.5397,
+      "step": 4703
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015587242026266417,
+      "loss": 2.4491,
+      "step": 4704
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015586303939962477,
+      "loss": 2.5012,
+      "step": 4705
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015585365853658538,
+      "loss": 2.5868,
+      "step": 4706
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015584427767354597,
+      "loss": 2.4577,
+      "step": 4707
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001558348968105066,
+      "loss": 2.9577,
+      "step": 4708
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015582551594746718,
+      "loss": 2.5232,
+      "step": 4709
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015581613508442777,
+      "loss": 2.7753,
+      "step": 4710
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015580675422138837,
+      "loss": 2.6917,
+      "step": 4711
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015579737335834898,
+      "loss": 2.456,
+      "step": 4712
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015578799249530957,
+      "loss": 2.8124,
+      "step": 4713
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015577861163227017,
+      "loss": 2.4999,
+      "step": 4714
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015576923076923078,
+      "loss": 2.4959,
+      "step": 4715
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015575984990619137,
+      "loss": 2.6472,
+      "step": 4716
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.000155750469043152,
+      "loss": 2.4436,
+      "step": 4717
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015574108818011258,
+      "loss": 2.5421,
+      "step": 4718
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015573170731707317,
+      "loss": 2.608,
+      "step": 4719
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015572232645403377,
+      "loss": 2.4529,
+      "step": 4720
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015571294559099438,
+      "loss": 2.5398,
+      "step": 4721
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.00015570356472795497,
+      "loss": 2.3635,
+      "step": 4722
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001556941838649156,
+      "loss": 2.7334,
+      "step": 4723
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015568480300187618,
+      "loss": 2.4792,
+      "step": 4724
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001556754221388368,
+      "loss": 2.4423,
+      "step": 4725
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001556660412757974,
+      "loss": 2.5252,
+      "step": 4726
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015565666041275798,
+      "loss": 2.4,
+      "step": 4727
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015564727954971857,
+      "loss": 2.6982,
+      "step": 4728
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015563789868667917,
+      "loss": 2.434,
+      "step": 4729
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015562851782363978,
+      "loss": 2.7046,
+      "step": 4730
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015561913696060037,
+      "loss": 2.6209,
+      "step": 4731
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.000155609756097561,
+      "loss": 2.7975,
+      "step": 4732
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015560037523452158,
+      "loss": 2.7338,
+      "step": 4733
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001555909943714822,
+      "loss": 2.5192,
+      "step": 4734
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001555816135084428,
+      "loss": 2.6204,
+      "step": 4735
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015557223264540338,
+      "loss": 2.3582,
+      "step": 4736
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015556285178236397,
+      "loss": 2.8805,
+      "step": 4737
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001555534709193246,
+      "loss": 2.5105,
+      "step": 4738
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015554409005628518,
+      "loss": 2.6388,
+      "step": 4739
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015553470919324577,
+      "loss": 2.712,
+      "step": 4740
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001555253283302064,
+      "loss": 2.5594,
+      "step": 4741
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00015551594746716698,
+      "loss": 2.5769,
+      "step": 4742
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.0001555065666041276,
+      "loss": 2.7724,
+      "step": 4743
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001554971857410882,
+      "loss": 2.5441,
+      "step": 4744
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00015548780487804878,
+      "loss": 2.7344,
+      "step": 4745
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00015547842401500937,
+      "loss": 2.3914,
+      "step": 4746
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00015546904315197,
+      "loss": 2.3502,
+      "step": 4747
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00015545966228893058,
+      "loss": 2.2865,
+      "step": 4748
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001554502814258912,
+      "loss": 2.4968,
+      "step": 4749
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001554409005628518,
+      "loss": 2.7232,
+      "step": 4750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001554315196998124,
+      "loss": 2.4349,
+      "step": 4751
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000155422138836773,
+      "loss": 2.6052,
+      "step": 4752
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001554127579737336,
+      "loss": 2.5086,
+      "step": 4753
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00015540337711069418,
+      "loss": 2.7502,
+      "step": 4754
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00015539399624765477,
+      "loss": 2.5342,
+      "step": 4755
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001553846153846154,
+      "loss": 2.535,
+      "step": 4756
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00015537523452157598,
+      "loss": 2.6982,
+      "step": 4757
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001553658536585366,
+      "loss": 2.7595,
+      "step": 4758
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001553564727954972,
+      "loss": 2.6816,
+      "step": 4759
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001553470919324578,
+      "loss": 2.5823,
+      "step": 4760
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001553377110694184,
+      "loss": 2.739,
+      "step": 4761
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.000155328330206379,
+      "loss": 2.5341,
+      "step": 4762
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00015531894934333958,
+      "loss": 2.7363,
+      "step": 4763
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001553095684803002,
+      "loss": 2.5168,
+      "step": 4764
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.0001553001876172608,
+      "loss": 2.4562,
+      "step": 4765
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00015529080675422138,
+      "loss": 2.865,
+      "step": 4766
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.000155281425891182,
+      "loss": 2.4407,
+      "step": 4767
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001552720450281426,
+      "loss": 2.2743,
+      "step": 4768
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001552626641651032,
+      "loss": 2.685,
+      "step": 4769
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001552532833020638,
+      "loss": 2.4946,
+      "step": 4770
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001552439024390244,
+      "loss": 2.6935,
+      "step": 4771
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00015523452157598498,
+      "loss": 2.5332,
+      "step": 4772
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001552251407129456,
+      "loss": 2.8331,
+      "step": 4773
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001552157598499062,
+      "loss": 2.5144,
+      "step": 4774
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001552063789868668,
+      "loss": 2.4781,
+      "step": 4775
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001551969981238274,
+      "loss": 2.8311,
+      "step": 4776
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00015518761726078802,
+      "loss": 2.5813,
+      "step": 4777
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001551782363977486,
+      "loss": 2.8364,
+      "step": 4778
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001551688555347092,
+      "loss": 2.6004,
+      "step": 4779
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001551594746716698,
+      "loss": 2.6317,
+      "step": 4780
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00015515009380863038,
+      "loss": 2.6714,
+      "step": 4781
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.000155140712945591,
+      "loss": 3.0313,
+      "step": 4782
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001551313320825516,
+      "loss": 2.5693,
+      "step": 4783
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001551219512195122,
+      "loss": 2.3874,
+      "step": 4784
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.0001551125703564728,
+      "loss": 2.8292,
+      "step": 4785
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00015510318949343342,
+      "loss": 2.5666,
+      "step": 4786
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000155093808630394,
+      "loss": 2.5758,
+      "step": 4787
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001550844277673546,
+      "loss": 2.4667,
+      "step": 4788
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001550750469043152,
+      "loss": 2.4272,
+      "step": 4789
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001550656660412758,
+      "loss": 2.7953,
+      "step": 4790
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001550562851782364,
+      "loss": 2.6911,
+      "step": 4791
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000155046904315197,
+      "loss": 2.6465,
+      "step": 4792
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001550375234521576,
+      "loss": 2.7449,
+      "step": 4793
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001550281425891182,
+      "loss": 2.4406,
+      "step": 4794
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00015501876172607882,
+      "loss": 2.4562,
+      "step": 4795
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001550093808630394,
+      "loss": 2.2677,
+      "step": 4796
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000155,
+      "loss": 2.723,
+      "step": 4797
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001549906191369606,
+      "loss": 2.537,
+      "step": 4798
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001549812382739212,
+      "loss": 2.7183,
+      "step": 4799
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001549718574108818,
+      "loss": 2.5496,
+      "step": 4800
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00015496247654784242,
+      "loss": 2.3643,
+      "step": 4801
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000154953095684803,
+      "loss": 2.4046,
+      "step": 4802
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00015494371482176363,
+      "loss": 2.5431,
+      "step": 4803
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.00015493433395872422,
+      "loss": 2.5782,
+      "step": 4804
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001549249530956848,
+      "loss": 2.4381,
+      "step": 4805
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.0001549155722326454,
+      "loss": 2.4244,
+      "step": 4806
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 0.000154906191369606,
+      "loss": 2.6344,
+      "step": 4807
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001548968105065666,
+      "loss": 2.6307,
+      "step": 4808
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001548874296435272,
+      "loss": 2.5604,
+      "step": 4809
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015487804878048782,
+      "loss": 2.6362,
+      "step": 4810
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001548686679174484,
+      "loss": 2.6185,
+      "step": 4811
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015485928705440903,
+      "loss": 2.7175,
+      "step": 4812
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015484990619136962,
+      "loss": 2.8736,
+      "step": 4813
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001548405253283302,
+      "loss": 2.7096,
+      "step": 4814
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001548311444652908,
+      "loss": 2.3862,
+      "step": 4815
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015482176360225142,
+      "loss": 2.707,
+      "step": 4816
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.000154812382739212,
+      "loss": 2.8595,
+      "step": 4817
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015480300187617263,
+      "loss": 2.3817,
+      "step": 4818
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015479362101313322,
+      "loss": 2.5008,
+      "step": 4819
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001547842401500938,
+      "loss": 2.5749,
+      "step": 4820
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015477485928705443,
+      "loss": 2.6903,
+      "step": 4821
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015476547842401502,
+      "loss": 2.7524,
+      "step": 4822
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001547560975609756,
+      "loss": 2.8333,
+      "step": 4823
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001547467166979362,
+      "loss": 2.5003,
+      "step": 4824
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015473733583489682,
+      "loss": 2.5043,
+      "step": 4825
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001547279549718574,
+      "loss": 2.4399,
+      "step": 4826
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015471857410881803,
+      "loss": 2.5996,
+      "step": 4827
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.00015470919324577862,
+      "loss": 2.7021,
+      "step": 4828
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015469981238273924,
+      "loss": 2.1381,
+      "step": 4829
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015469043151969983,
+      "loss": 2.7273,
+      "step": 4830
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015468105065666044,
+      "loss": 2.7093,
+      "step": 4831
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.000154671669793621,
+      "loss": 2.7245,
+      "step": 4832
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001546622889305816,
+      "loss": 2.636,
+      "step": 4833
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015465290806754222,
+      "loss": 2.5842,
+      "step": 4834
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001546435272045028,
+      "loss": 2.4467,
+      "step": 4835
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015463414634146343,
+      "loss": 2.4958,
+      "step": 4836
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015462476547842402,
+      "loss": 2.9019,
+      "step": 4837
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015461538461538464,
+      "loss": 2.477,
+      "step": 4838
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015460600375234523,
+      "loss": 2.7242,
+      "step": 4839
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015459662288930584,
+      "loss": 2.5058,
+      "step": 4840
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001545872420262664,
+      "loss": 2.3354,
+      "step": 4841
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015457786116322703,
+      "loss": 2.8194,
+      "step": 4842
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015456848030018762,
+      "loss": 2.4446,
+      "step": 4843
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015455909943714824,
+      "loss": 2.5466,
+      "step": 4844
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015454971857410883,
+      "loss": 2.7617,
+      "step": 4845
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015454033771106942,
+      "loss": 2.6805,
+      "step": 4846
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015453095684803003,
+      "loss": 2.5101,
+      "step": 4847
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015452157598499063,
+      "loss": 2.6279,
+      "step": 4848
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00015451219512195124,
+      "loss": 2.3681,
+      "step": 4849
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.0001545028142589118,
+      "loss": 2.6378,
+      "step": 4850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015449343339587243,
+      "loss": 2.7388,
+      "step": 4851
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015448405253283302,
+      "loss": 2.4724,
+      "step": 4852
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015447467166979363,
+      "loss": 2.6351,
+      "step": 4853
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015446529080675423,
+      "loss": 2.4582,
+      "step": 4854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015445590994371484,
+      "loss": 2.3912,
+      "step": 4855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015444652908067543,
+      "loss": 2.4146,
+      "step": 4856
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015443714821763605,
+      "loss": 2.6767,
+      "step": 4857
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015442776735459664,
+      "loss": 2.884,
+      "step": 4858
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.0001544183864915572,
+      "loss": 2.5058,
+      "step": 4859
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015440900562851783,
+      "loss": 2.797,
+      "step": 4860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015439962476547842,
+      "loss": 2.7933,
+      "step": 4861
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015439024390243903,
+      "loss": 2.3097,
+      "step": 4862
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015438086303939963,
+      "loss": 2.7184,
+      "step": 4863
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015437148217636024,
+      "loss": 2.8811,
+      "step": 4864
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015436210131332083,
+      "loss": 2.507,
+      "step": 4865
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015435272045028145,
+      "loss": 2.582,
+      "step": 4866
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015434333958724204,
+      "loss": 2.2897,
+      "step": 4867
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015433395872420263,
+      "loss": 2.4532,
+      "step": 4868
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015432457786116323,
+      "loss": 2.3394,
+      "step": 4869
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015431519699812384,
+      "loss": 2.5551,
+      "step": 4870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00015430581613508443,
+      "loss": 2.5727,
+      "step": 4871
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015429643527204503,
+      "loss": 2.5446,
+      "step": 4872
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015428705440900564,
+      "loss": 2.5391,
+      "step": 4873
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015427767354596623,
+      "loss": 2.8477,
+      "step": 4874
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015426829268292685,
+      "loss": 2.4357,
+      "step": 4875
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015425891181988744,
+      "loss": 2.4459,
+      "step": 4876
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015424953095684803,
+      "loss": 2.3209,
+      "step": 4877
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015424015009380863,
+      "loss": 2.4376,
+      "step": 4878
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015423076923076924,
+      "loss": 2.6161,
+      "step": 4879
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015422138836772983,
+      "loss": 2.5588,
+      "step": 4880
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015421200750469045,
+      "loss": 2.7722,
+      "step": 4881
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015420262664165104,
+      "loss": 2.5109,
+      "step": 4882
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015419324577861166,
+      "loss": 2.3767,
+      "step": 4883
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015418386491557225,
+      "loss": 2.5005,
+      "step": 4884
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015417448405253284,
+      "loss": 2.7071,
+      "step": 4885
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015416510318949343,
+      "loss": 2.4465,
+      "step": 4886
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015415572232645403,
+      "loss": 2.6411,
+      "step": 4887
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015414634146341464,
+      "loss": 2.6318,
+      "step": 4888
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015413696060037523,
+      "loss": 2.383,
+      "step": 4889
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015412757973733585,
+      "loss": 2.6139,
+      "step": 4890
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015411819887429644,
+      "loss": 2.5303,
+      "step": 4891
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.00015410881801125706,
+      "loss": 2.4132,
+      "step": 4892
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015409943714821765,
+      "loss": 2.5593,
+      "step": 4893
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015409005628517824,
+      "loss": 2.6888,
+      "step": 4894
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015408067542213883,
+      "loss": 2.9395,
+      "step": 4895
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015407129455909945,
+      "loss": 2.4411,
+      "step": 4896
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015406191369606004,
+      "loss": 2.784,
+      "step": 4897
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015405253283302063,
+      "loss": 2.6382,
+      "step": 4898
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015404315196998125,
+      "loss": 2.3597,
+      "step": 4899
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015403377110694184,
+      "loss": 2.4053,
+      "step": 4900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015402439024390246,
+      "loss": 2.6453,
+      "step": 4901
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015401500938086305,
+      "loss": 2.5599,
+      "step": 4902
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015400562851782364,
+      "loss": 2.7974,
+      "step": 4903
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015399624765478423,
+      "loss": 2.7938,
+      "step": 4904
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015398686679174485,
+      "loss": 2.4707,
+      "step": 4905
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015397748592870544,
+      "loss": 2.787,
+      "step": 4906
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015396810506566606,
+      "loss": 2.6118,
+      "step": 4907
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015395872420262665,
+      "loss": 2.8754,
+      "step": 4908
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015394934333958727,
+      "loss": 2.6239,
+      "step": 4909
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015393996247654786,
+      "loss": 2.4888,
+      "step": 4910
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015393058161350845,
+      "loss": 2.6802,
+      "step": 4911
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015392120075046904,
+      "loss": 2.5181,
+      "step": 4912
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015391181988742963,
+      "loss": 2.4085,
+      "step": 4913
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 0.00015390243902439025,
+      "loss": 2.6526,
+      "step": 4914
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015389305816135084,
+      "loss": 2.8999,
+      "step": 4915
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015388367729831146,
+      "loss": 2.6765,
+      "step": 4916
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015387429643527205,
+      "loss": 2.5339,
+      "step": 4917
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015386491557223267,
+      "loss": 2.6782,
+      "step": 4918
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015385553470919326,
+      "loss": 2.5436,
+      "step": 4919
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 2.4876,
+      "step": 4920
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015383677298311444,
+      "loss": 2.4286,
+      "step": 4921
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015382739212007506,
+      "loss": 2.5494,
+      "step": 4922
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015381801125703565,
+      "loss": 2.7036,
+      "step": 4923
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015380863039399624,
+      "loss": 2.8458,
+      "step": 4924
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015379924953095686,
+      "loss": 3.0224,
+      "step": 4925
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015378986866791745,
+      "loss": 2.6589,
+      "step": 4926
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015378048780487807,
+      "loss": 2.7869,
+      "step": 4927
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015377110694183866,
+      "loss": 2.5201,
+      "step": 4928
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015376172607879925,
+      "loss": 2.4817,
+      "step": 4929
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015375234521575984,
+      "loss": 2.452,
+      "step": 4930
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015374296435272046,
+      "loss": 2.7444,
+      "step": 4931
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015373358348968105,
+      "loss": 2.594,
+      "step": 4932
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015372420262664167,
+      "loss": 2.5866,
+      "step": 4933
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015371482176360226,
+      "loss": 2.3802,
+      "step": 4934
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00015370544090056288,
+      "loss": 2.4955,
+      "step": 4935
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015369606003752347,
+      "loss": 2.5638,
+      "step": 4936
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015368667917448406,
+      "loss": 3.1693,
+      "step": 4937
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015367729831144465,
+      "loss": 2.5334,
+      "step": 4938
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015366791744840524,
+      "loss": 2.6191,
+      "step": 4939
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015365853658536586,
+      "loss": 2.5611,
+      "step": 4940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015364915572232645,
+      "loss": 2.5906,
+      "step": 4941
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015363977485928707,
+      "loss": 2.6009,
+      "step": 4942
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015363039399624766,
+      "loss": 2.3454,
+      "step": 4943
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015362101313320828,
+      "loss": 2.7993,
+      "step": 4944
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015361163227016887,
+      "loss": 2.3538,
+      "step": 4945
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015360225140712946,
+      "loss": 2.7001,
+      "step": 4946
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015359287054409005,
+      "loss": 2.6139,
+      "step": 4947
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015358348968105067,
+      "loss": 2.6324,
+      "step": 4948
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015357410881801126,
+      "loss": 2.4727,
+      "step": 4949
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015356472795497185,
+      "loss": 2.5334,
+      "step": 4950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015355534709193247,
+      "loss": 2.7448,
+      "step": 4951
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015354596622889306,
+      "loss": 2.6224,
+      "step": 4952
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015353658536585368,
+      "loss": 2.566,
+      "step": 4953
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015352720450281427,
+      "loss": 2.8086,
+      "step": 4954
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015351782363977486,
+      "loss": 2.6812,
+      "step": 4955
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00015350844277673545,
+      "loss": 2.6766,
+      "step": 4956
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015349906191369607,
+      "loss": 2.6775,
+      "step": 4957
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015348968105065666,
+      "loss": 2.8635,
+      "step": 4958
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015348030018761728,
+      "loss": 2.8654,
+      "step": 4959
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015347091932457787,
+      "loss": 2.6547,
+      "step": 4960
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001534615384615385,
+      "loss": 2.8409,
+      "step": 4961
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015345215759849908,
+      "loss": 2.724,
+      "step": 4962
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015344277673545967,
+      "loss": 2.7394,
+      "step": 4963
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015343339587242026,
+      "loss": 2.6232,
+      "step": 4964
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015342401500938085,
+      "loss": 2.5548,
+      "step": 4965
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015341463414634147,
+      "loss": 2.5527,
+      "step": 4966
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015340525328330206,
+      "loss": 2.8705,
+      "step": 4967
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015339587242026268,
+      "loss": 2.4845,
+      "step": 4968
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015338649155722327,
+      "loss": 2.461,
+      "step": 4969
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.0001533771106941839,
+      "loss": 2.2811,
+      "step": 4970
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015336772983114448,
+      "loss": 2.5773,
+      "step": 4971
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015335834896810507,
+      "loss": 2.4992,
+      "step": 4972
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015334896810506566,
+      "loss": 2.7263,
+      "step": 4973
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015333958724202628,
+      "loss": 2.7368,
+      "step": 4974
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015333020637898687,
+      "loss": 2.5679,
+      "step": 4975
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015332082551594746,
+      "loss": 2.8406,
+      "step": 4976
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015331144465290808,
+      "loss": 2.6146,
+      "step": 4977
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00015330206378986867,
+      "loss": 2.9998,
+      "step": 4978
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015329268292682929,
+      "loss": 2.3668,
+      "step": 4979
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015328330206378988,
+      "loss": 2.7155,
+      "step": 4980
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015327392120075047,
+      "loss": 2.5115,
+      "step": 4981
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015326454033771106,
+      "loss": 2.6553,
+      "step": 4982
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015325515947467168,
+      "loss": 2.5712,
+      "step": 4983
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015324577861163227,
+      "loss": 2.4741,
+      "step": 4984
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015323639774859289,
+      "loss": 2.473,
+      "step": 4985
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015322701688555348,
+      "loss": 2.8188,
+      "step": 4986
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001532176360225141,
+      "loss": 2.6108,
+      "step": 4987
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015320825515947469,
+      "loss": 2.6271,
+      "step": 4988
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015319887429643528,
+      "loss": 2.6145,
+      "step": 4989
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015318949343339587,
+      "loss": 2.1351,
+      "step": 4990
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015318011257035646,
+      "loss": 2.6286,
+      "step": 4991
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015317073170731708,
+      "loss": 2.4518,
+      "step": 4992
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015316135084427767,
+      "loss": 2.739,
+      "step": 4993
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015315196998123829,
+      "loss": 2.9006,
+      "step": 4994
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015314258911819888,
+      "loss": 2.6811,
+      "step": 4995
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001531332082551595,
+      "loss": 2.6674,
+      "step": 4996
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015312382739212009,
+      "loss": 2.4357,
+      "step": 4997
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.0001531144465290807,
+      "loss": 2.5687,
+      "step": 4998
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 0.00015310506566604127,
+      "loss": 2.4391,
+      "step": 4999
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015309568480300189,
+      "loss": 2.5401,
+      "step": 5000
+    },
+    {
+      "epoch": 2.35,
+      "eval_loss": 2.6361289024353027,
+      "eval_runtime": 67.4934,
+      "eval_samples_per_second": 28.077,
+      "eval_steps_per_second": 3.511,
+      "step": 5000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015308630393996248,
+      "loss": 2.4259,
+      "step": 5001
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015307692307692307,
+      "loss": 2.9872,
+      "step": 5002
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015306754221388369,
+      "loss": 2.5688,
+      "step": 5003
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015305816135084428,
+      "loss": 2.7028,
+      "step": 5004
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001530487804878049,
+      "loss": 2.4351,
+      "step": 5005
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015303939962476549,
+      "loss": 2.7249,
+      "step": 5006
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001530300187617261,
+      "loss": 2.1647,
+      "step": 5007
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015302063789868667,
+      "loss": 2.7017,
+      "step": 5008
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015301125703564729,
+      "loss": 2.4981,
+      "step": 5009
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015300187617260788,
+      "loss": 2.5883,
+      "step": 5010
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001529924953095685,
+      "loss": 2.6359,
+      "step": 5011
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015298311444652909,
+      "loss": 2.2109,
+      "step": 5012
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001529737335834897,
+      "loss": 3.1574,
+      "step": 5013
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001529643527204503,
+      "loss": 2.6887,
+      "step": 5014
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015295497185741089,
+      "loss": 2.848,
+      "step": 5015
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001529455909943715,
+      "loss": 2.6861,
+      "step": 5016
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015293621013133207,
+      "loss": 2.4916,
+      "step": 5017
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015292682926829269,
+      "loss": 2.7704,
+      "step": 5018
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00015291744840525328,
+      "loss": 2.648,
+      "step": 5019
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.0001529080675422139,
+      "loss": 2.7629,
+      "step": 5020
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00015289868667917449,
+      "loss": 2.702,
+      "step": 5021
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001528893058161351,
+      "loss": 2.7109,
+      "step": 5022
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001528799249530957,
+      "loss": 2.9193,
+      "step": 5023
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001528705440900563,
+      "loss": 2.4296,
+      "step": 5024
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001528611632270169,
+      "loss": 2.573,
+      "step": 5025
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001528517823639775,
+      "loss": 2.8118,
+      "step": 5026
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00015284240150093809,
+      "loss": 2.3945,
+      "step": 5027
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001528330206378987,
+      "loss": 2.5737,
+      "step": 5028
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001528236397748593,
+      "loss": 2.8957,
+      "step": 5029
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00015281425891181989,
+      "loss": 2.6596,
+      "step": 5030
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001528048780487805,
+      "loss": 2.5252,
+      "step": 5031
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001527954971857411,
+      "loss": 2.7855,
+      "step": 5032
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001527861163227017,
+      "loss": 2.7261,
+      "step": 5033
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001527767354596623,
+      "loss": 2.4172,
+      "step": 5034
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001527673545966229,
+      "loss": 2.9609,
+      "step": 5035
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00015275797373358348,
+      "loss": 2.4358,
+      "step": 5036
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001527485928705441,
+      "loss": 2.3059,
+      "step": 5037
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001527392120075047,
+      "loss": 2.7714,
+      "step": 5038
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001527298311444653,
+      "loss": 2.7331,
+      "step": 5039
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001527204502814259,
+      "loss": 2.5882,
+      "step": 5040
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00015271106941838652,
+      "loss": 2.5151,
+      "step": 5041
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.0001527016885553471,
+      "loss": 2.8262,
+      "step": 5042
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001526923076923077,
+      "loss": 2.4196,
+      "step": 5043
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001526829268292683,
+      "loss": 2.5619,
+      "step": 5044
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00015267354596622888,
+      "loss": 2.7732,
+      "step": 5045
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001526641651031895,
+      "loss": 2.6842,
+      "step": 5046
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001526547842401501,
+      "loss": 2.4202,
+      "step": 5047
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001526454033771107,
+      "loss": 2.2937,
+      "step": 5048
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001526360225140713,
+      "loss": 2.5798,
+      "step": 5049
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00015262664165103192,
+      "loss": 2.3335,
+      "step": 5050
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001526172607879925,
+      "loss": 2.5682,
+      "step": 5051
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001526078799249531,
+      "loss": 2.5875,
+      "step": 5052
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001525984990619137,
+      "loss": 2.5321,
+      "step": 5053
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001525891181988743,
+      "loss": 2.8773,
+      "step": 5054
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001525797373358349,
+      "loss": 2.4142,
+      "step": 5055
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001525703564727955,
+      "loss": 2.6445,
+      "step": 5056
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001525609756097561,
+      "loss": 2.646,
+      "step": 5057
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001525515947467167,
+      "loss": 2.766,
+      "step": 5058
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00015254221388367732,
+      "loss": 2.8784,
+      "step": 5059
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001525328330206379,
+      "loss": 2.6556,
+      "step": 5060
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001525234521575985,
+      "loss": 2.7884,
+      "step": 5061
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001525140712945591,
+      "loss": 2.7249,
+      "step": 5062
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.0001525046904315197,
+      "loss": 2.7342,
+      "step": 5063
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001524953095684803,
+      "loss": 2.4705,
+      "step": 5064
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00015248592870544092,
+      "loss": 2.6515,
+      "step": 5065
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001524765478424015,
+      "loss": 2.5734,
+      "step": 5066
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00015246716697936213,
+      "loss": 2.612,
+      "step": 5067
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00015245778611632272,
+      "loss": 2.2915,
+      "step": 5068
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001524484052532833,
+      "loss": 2.3514,
+      "step": 5069
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001524390243902439,
+      "loss": 2.5863,
+      "step": 5070
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001524296435272045,
+      "loss": 2.6981,
+      "step": 5071
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001524202626641651,
+      "loss": 2.6653,
+      "step": 5072
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001524108818011257,
+      "loss": 2.5296,
+      "step": 5073
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00015240150093808632,
+      "loss": 2.8438,
+      "step": 5074
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001523921200750469,
+      "loss": 2.7266,
+      "step": 5075
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00015238273921200753,
+      "loss": 2.9262,
+      "step": 5076
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00015237335834896812,
+      "loss": 2.3913,
+      "step": 5077
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001523639774859287,
+      "loss": 2.6852,
+      "step": 5078
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001523545966228893,
+      "loss": 2.5731,
+      "step": 5079
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00015234521575984992,
+      "loss": 2.7292,
+      "step": 5080
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001523358348968105,
+      "loss": 2.6584,
+      "step": 5081
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001523264540337711,
+      "loss": 2.7514,
+      "step": 5082
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00015231707317073172,
+      "loss": 2.6789,
+      "step": 5083
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.0001523076923076923,
+      "loss": 2.4296,
+      "step": 5084
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015229831144465293,
+      "loss": 2.5094,
+      "step": 5085
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015228893058161352,
+      "loss": 2.3915,
+      "step": 5086
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001522795497185741,
+      "loss": 2.4911,
+      "step": 5087
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001522701688555347,
+      "loss": 2.7437,
+      "step": 5088
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015226078799249532,
+      "loss": 2.6946,
+      "step": 5089
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001522514071294559,
+      "loss": 2.529,
+      "step": 5090
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015224202626641653,
+      "loss": 2.7256,
+      "step": 5091
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015223264540337712,
+      "loss": 2.6789,
+      "step": 5092
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015222326454033774,
+      "loss": 2.4644,
+      "step": 5093
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015221388367729833,
+      "loss": 2.6795,
+      "step": 5094
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015220450281425892,
+      "loss": 2.4766,
+      "step": 5095
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001521951219512195,
+      "loss": 2.4644,
+      "step": 5096
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001521857410881801,
+      "loss": 2.8547,
+      "step": 5097
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015217636022514072,
+      "loss": 2.8652,
+      "step": 5098
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001521669793621013,
+      "loss": 2.9219,
+      "step": 5099
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015215759849906193,
+      "loss": 2.6638,
+      "step": 5100
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015214821763602252,
+      "loss": 2.727,
+      "step": 5101
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015213883677298314,
+      "loss": 2.6301,
+      "step": 5102
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015212945590994373,
+      "loss": 2.4834,
+      "step": 5103
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015212007504690432,
+      "loss": 2.8176,
+      "step": 5104
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.0001521106941838649,
+      "loss": 2.3939,
+      "step": 5105
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 0.00015210131332082553,
+      "loss": 2.5802,
+      "step": 5106
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015209193245778612,
+      "loss": 2.5652,
+      "step": 5107
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001520825515947467,
+      "loss": 2.5446,
+      "step": 5108
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015207317073170733,
+      "loss": 2.8032,
+      "step": 5109
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015206378986866792,
+      "loss": 2.6628,
+      "step": 5110
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015205440900562854,
+      "loss": 2.8373,
+      "step": 5111
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015204502814258913,
+      "loss": 2.5434,
+      "step": 5112
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015203564727954972,
+      "loss": 2.3724,
+      "step": 5113
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001520262664165103,
+      "loss": 2.5205,
+      "step": 5114
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015201688555347093,
+      "loss": 2.6989,
+      "step": 5115
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015200750469043152,
+      "loss": 2.4986,
+      "step": 5116
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015199812382739214,
+      "loss": 2.6909,
+      "step": 5117
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015198874296435273,
+      "loss": 2.2571,
+      "step": 5118
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015197936210131335,
+      "loss": 2.6504,
+      "step": 5119
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015196998123827394,
+      "loss": 2.5914,
+      "step": 5120
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015196060037523453,
+      "loss": 2.7675,
+      "step": 5121
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015195121951219512,
+      "loss": 2.6709,
+      "step": 5122
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.0001519418386491557,
+      "loss": 2.8893,
+      "step": 5123
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015193245778611633,
+      "loss": 2.5308,
+      "step": 5124
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015192307692307692,
+      "loss": 2.3618,
+      "step": 5125
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015191369606003754,
+      "loss": 2.489,
+      "step": 5126
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00015190431519699813,
+      "loss": 2.6542,
+      "step": 5127
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015189493433395875,
+      "loss": 2.6872,
+      "step": 5128
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015188555347091934,
+      "loss": 2.1639,
+      "step": 5129
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015187617260787993,
+      "loss": 2.7536,
+      "step": 5130
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015186679174484052,
+      "loss": 2.6956,
+      "step": 5131
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015185741088180114,
+      "loss": 2.632,
+      "step": 5132
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015184803001876173,
+      "loss": 2.7084,
+      "step": 5133
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015183864915572232,
+      "loss": 2.513,
+      "step": 5134
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015182926829268294,
+      "loss": 2.3915,
+      "step": 5135
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015181988742964353,
+      "loss": 2.3278,
+      "step": 5136
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015181050656660415,
+      "loss": 2.3803,
+      "step": 5137
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015180112570356474,
+      "loss": 2.6543,
+      "step": 5138
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015179174484052533,
+      "loss": 2.4342,
+      "step": 5139
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015178236397748592,
+      "loss": 2.408,
+      "step": 5140
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015177298311444654,
+      "loss": 2.4651,
+      "step": 5141
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015176360225140713,
+      "loss": 2.5997,
+      "step": 5142
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015175422138836775,
+      "loss": 2.7193,
+      "step": 5143
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015174484052532834,
+      "loss": 2.4984,
+      "step": 5144
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015173545966228895,
+      "loss": 2.7702,
+      "step": 5145
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015172607879924955,
+      "loss": 2.6559,
+      "step": 5146
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015171669793621014,
+      "loss": 2.7272,
+      "step": 5147
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00015170731707317073,
+      "loss": 2.2507,
+      "step": 5148
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015169793621013132,
+      "loss": 2.5681,
+      "step": 5149
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015168855534709194,
+      "loss": 2.509,
+      "step": 5150
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015167917448405253,
+      "loss": 2.4245,
+      "step": 5151
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015166979362101315,
+      "loss": 2.8352,
+      "step": 5152
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015166041275797374,
+      "loss": 2.4843,
+      "step": 5153
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015165103189493435,
+      "loss": 2.6425,
+      "step": 5154
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015164165103189495,
+      "loss": 2.9276,
+      "step": 5155
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015163227016885556,
+      "loss": 2.4568,
+      "step": 5156
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015162288930581613,
+      "loss": 2.6233,
+      "step": 5157
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015161350844277675,
+      "loss": 2.5837,
+      "step": 5158
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015160412757973734,
+      "loss": 2.6963,
+      "step": 5159
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015159474671669793,
+      "loss": 2.5446,
+      "step": 5160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015158536585365855,
+      "loss": 2.6815,
+      "step": 5161
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015157598499061914,
+      "loss": 2.6114,
+      "step": 5162
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015156660412757975,
+      "loss": 2.8934,
+      "step": 5163
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015155722326454035,
+      "loss": 2.6488,
+      "step": 5164
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015154784240150096,
+      "loss": 2.2903,
+      "step": 5165
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015153846153846153,
+      "loss": 2.4164,
+      "step": 5166
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015152908067542215,
+      "loss": 2.7351,
+      "step": 5167
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015151969981238274,
+      "loss": 2.8082,
+      "step": 5168
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015151031894934335,
+      "loss": 2.8303,
+      "step": 5169
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00015150093808630395,
+      "loss": 2.2486,
+      "step": 5170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015149155722326456,
+      "loss": 2.4717,
+      "step": 5171
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015148217636022515,
+      "loss": 2.5344,
+      "step": 5172
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015147279549718575,
+      "loss": 2.6054,
+      "step": 5173
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015146341463414636,
+      "loss": 2.4478,
+      "step": 5174
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015145403377110693,
+      "loss": 2.6541,
+      "step": 5175
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015144465290806755,
+      "loss": 2.7783,
+      "step": 5176
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015143527204502814,
+      "loss": 2.5975,
+      "step": 5177
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015142589118198875,
+      "loss": 2.7052,
+      "step": 5178
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015141651031894935,
+      "loss": 2.5909,
+      "step": 5179
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015140712945590996,
+      "loss": 2.7214,
+      "step": 5180
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015139774859287055,
+      "loss": 2.9073,
+      "step": 5181
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015138836772983117,
+      "loss": 2.5451,
+      "step": 5182
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015137898686679176,
+      "loss": 2.5056,
+      "step": 5183
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015136960600375235,
+      "loss": 2.5956,
+      "step": 5184
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015136022514071294,
+      "loss": 2.7651,
+      "step": 5185
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015135084427767354,
+      "loss": 2.5967,
+      "step": 5186
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015134146341463415,
+      "loss": 2.8269,
+      "step": 5187
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015133208255159474,
+      "loss": 2.4964,
+      "step": 5188
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015132270168855536,
+      "loss": 2.503,
+      "step": 5189
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015131332082551595,
+      "loss": 2.4273,
+      "step": 5190
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 0.00015130393996247657,
+      "loss": 2.6539,
+      "step": 5191
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015129455909943716,
+      "loss": 2.3555,
+      "step": 5192
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015128517823639775,
+      "loss": 2.7522,
+      "step": 5193
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015127579737335834,
+      "loss": 2.4705,
+      "step": 5194
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015126641651031896,
+      "loss": 2.5069,
+      "step": 5195
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015125703564727955,
+      "loss": 2.6354,
+      "step": 5196
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015124765478424017,
+      "loss": 2.4962,
+      "step": 5197
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015123827392120076,
+      "loss": 2.4286,
+      "step": 5198
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015122889305816135,
+      "loss": 2.5907,
+      "step": 5199
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015121951219512197,
+      "loss": 2.5526,
+      "step": 5200
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015121013133208256,
+      "loss": 2.3959,
+      "step": 5201
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015120075046904315,
+      "loss": 2.3606,
+      "step": 5202
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015119136960600374,
+      "loss": 2.5807,
+      "step": 5203
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015118198874296436,
+      "loss": 2.6295,
+      "step": 5204
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015117260787992495,
+      "loss": 2.3832,
+      "step": 5205
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015116322701688557,
+      "loss": 2.6847,
+      "step": 5206
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015115384615384616,
+      "loss": 2.6339,
+      "step": 5207
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015114446529080678,
+      "loss": 2.7686,
+      "step": 5208
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015113508442776737,
+      "loss": 2.6314,
+      "step": 5209
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015112570356472796,
+      "loss": 2.7167,
+      "step": 5210
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015111632270168855,
+      "loss": 2.273,
+      "step": 5211
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00015110694183864914,
+      "loss": 2.541,
+      "step": 5212
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015109756097560976,
+      "loss": 2.7938,
+      "step": 5213
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015108818011257035,
+      "loss": 2.7354,
+      "step": 5214
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015107879924953097,
+      "loss": 2.6191,
+      "step": 5215
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015106941838649156,
+      "loss": 2.4519,
+      "step": 5216
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015106003752345218,
+      "loss": 2.6186,
+      "step": 5217
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015105065666041277,
+      "loss": 2.4757,
+      "step": 5218
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015104127579737336,
+      "loss": 2.829,
+      "step": 5219
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015103189493433395,
+      "loss": 2.7351,
+      "step": 5220
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015102251407129457,
+      "loss": 2.605,
+      "step": 5221
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015101313320825516,
+      "loss": 2.2642,
+      "step": 5222
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015100375234521578,
+      "loss": 2.684,
+      "step": 5223
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015099437148217637,
+      "loss": 2.7195,
+      "step": 5224
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015098499061913696,
+      "loss": 2.6093,
+      "step": 5225
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015097560975609758,
+      "loss": 2.4932,
+      "step": 5226
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015096622889305817,
+      "loss": 2.5811,
+      "step": 5227
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015095684803001876,
+      "loss": 2.6347,
+      "step": 5228
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015094746716697935,
+      "loss": 2.5549,
+      "step": 5229
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015093808630393997,
+      "loss": 2.4195,
+      "step": 5230
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015092870544090056,
+      "loss": 2.6993,
+      "step": 5231
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015091932457786118,
+      "loss": 2.2518,
+      "step": 5232
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00015090994371482177,
+      "loss": 2.5439,
+      "step": 5233
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.0001509005628517824,
+      "loss": 2.7091,
+      "step": 5234
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015089118198874298,
+      "loss": 2.5034,
+      "step": 5235
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015088180112570357,
+      "loss": 2.9716,
+      "step": 5236
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015087242026266416,
+      "loss": 2.5558,
+      "step": 5237
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015086303939962478,
+      "loss": 2.6153,
+      "step": 5238
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015085365853658537,
+      "loss": 2.6353,
+      "step": 5239
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015084427767354596,
+      "loss": 2.8702,
+      "step": 5240
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015083489681050658,
+      "loss": 2.538,
+      "step": 5241
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015082551594746717,
+      "loss": 2.5177,
+      "step": 5242
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001508161350844278,
+      "loss": 2.4715,
+      "step": 5243
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015080675422138838,
+      "loss": 2.7138,
+      "step": 5244
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015079737335834897,
+      "loss": 2.8759,
+      "step": 5245
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015078799249530956,
+      "loss": 2.3693,
+      "step": 5246
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015077861163227018,
+      "loss": 2.3638,
+      "step": 5247
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015076923076923077,
+      "loss": 2.4229,
+      "step": 5248
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001507598499061914,
+      "loss": 2.743,
+      "step": 5249
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015075046904315198,
+      "loss": 2.8686,
+      "step": 5250
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001507410881801126,
+      "loss": 2.8831,
+      "step": 5251
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001507317073170732,
+      "loss": 2.8759,
+      "step": 5252
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015072232645403378,
+      "loss": 2.4513,
+      "step": 5253
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015071294559099437,
+      "loss": 2.7309,
+      "step": 5254
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.00015070356472795496,
+      "loss": 2.8675,
+      "step": 5255
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015069418386491558,
+      "loss": 2.91,
+      "step": 5256
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015068480300187617,
+      "loss": 2.4872,
+      "step": 5257
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001506754221388368,
+      "loss": 2.4991,
+      "step": 5258
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015066604127579738,
+      "loss": 2.6409,
+      "step": 5259
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.000150656660412758,
+      "loss": 2.5364,
+      "step": 5260
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001506472795497186,
+      "loss": 2.5563,
+      "step": 5261
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015063789868667918,
+      "loss": 2.6737,
+      "step": 5262
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015062851782363977,
+      "loss": 2.3836,
+      "step": 5263
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001506191369606004,
+      "loss": 2.4221,
+      "step": 5264
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015060975609756098,
+      "loss": 2.7824,
+      "step": 5265
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015060037523452157,
+      "loss": 2.4257,
+      "step": 5266
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001505909943714822,
+      "loss": 2.4769,
+      "step": 5267
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015058161350844278,
+      "loss": 2.5757,
+      "step": 5268
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001505722326454034,
+      "loss": 2.6884,
+      "step": 5269
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.000150562851782364,
+      "loss": 2.7117,
+      "step": 5270
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015055347091932458,
+      "loss": 2.4422,
+      "step": 5271
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015054409005628517,
+      "loss": 2.4482,
+      "step": 5272
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001505347091932458,
+      "loss": 2.5131,
+      "step": 5273
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00015052532833020638,
+      "loss": 2.5536,
+      "step": 5274
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.000150515947467167,
+      "loss": 2.3656,
+      "step": 5275
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.0001505065666041276,
+      "loss": 2.4307,
+      "step": 5276
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001504971857410882,
+      "loss": 2.9286,
+      "step": 5277
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001504878048780488,
+      "loss": 2.3818,
+      "step": 5278
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001504784240150094,
+      "loss": 2.7399,
+      "step": 5279
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015046904315196998,
+      "loss": 2.6697,
+      "step": 5280
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015045966228893057,
+      "loss": 2.096,
+      "step": 5281
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001504502814258912,
+      "loss": 2.6956,
+      "step": 5282
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015044090056285178,
+      "loss": 2.5372,
+      "step": 5283
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001504315196998124,
+      "loss": 2.3172,
+      "step": 5284
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000150422138836773,
+      "loss": 2.3611,
+      "step": 5285
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001504127579737336,
+      "loss": 2.4452,
+      "step": 5286
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001504033771106942,
+      "loss": 2.764,
+      "step": 5287
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001503939962476548,
+      "loss": 2.5081,
+      "step": 5288
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015038461538461538,
+      "loss": 2.9962,
+      "step": 5289
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000150375234521576,
+      "loss": 2.3971,
+      "step": 5290
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001503658536585366,
+      "loss": 3.0419,
+      "step": 5291
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015035647279549718,
+      "loss": 2.823,
+      "step": 5292
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001503470919324578,
+      "loss": 2.8495,
+      "step": 5293
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001503377110694184,
+      "loss": 2.6318,
+      "step": 5294
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.000150328330206379,
+      "loss": 3.0263,
+      "step": 5295
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001503189493433396,
+      "loss": 2.7609,
+      "step": 5296
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.0001503095684803002,
+      "loss": 2.753,
+      "step": 5297
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 0.00015030018761726078,
+      "loss": 2.4182,
+      "step": 5298
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001502908067542214,
+      "loss": 2.5426,
+      "step": 5299
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000150281425891182,
+      "loss": 2.7619,
+      "step": 5300
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001502720450281426,
+      "loss": 2.3209,
+      "step": 5301
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001502626641651032,
+      "loss": 2.5838,
+      "step": 5302
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015025328330206381,
+      "loss": 2.6806,
+      "step": 5303
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001502439024390244,
+      "loss": 2.7335,
+      "step": 5304
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000150234521575985,
+      "loss": 2.7712,
+      "step": 5305
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001502251407129456,
+      "loss": 2.5757,
+      "step": 5306
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015021575984990618,
+      "loss": 2.6638,
+      "step": 5307
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001502063789868668,
+      "loss": 2.5954,
+      "step": 5308
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001501969981238274,
+      "loss": 2.4274,
+      "step": 5309
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000150187617260788,
+      "loss": 2.5427,
+      "step": 5310
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001501782363977486,
+      "loss": 2.4656,
+      "step": 5311
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015016885553470921,
+      "loss": 2.488,
+      "step": 5312
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001501594746716698,
+      "loss": 2.4046,
+      "step": 5313
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00015015009380863042,
+      "loss": 2.7982,
+      "step": 5314
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.000150140712945591,
+      "loss": 2.3052,
+      "step": 5315
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001501313320825516,
+      "loss": 2.7036,
+      "step": 5316
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001501219512195122,
+      "loss": 2.4761,
+      "step": 5317
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001501125703564728,
+      "loss": 2.5089,
+      "step": 5318
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.0001501031894934334,
+      "loss": 2.6313,
+      "step": 5319
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000150093808630394,
+      "loss": 2.8152,
+      "step": 5320
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015008442776735461,
+      "loss": 2.3215,
+      "step": 5321
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001500750469043152,
+      "loss": 2.4189,
+      "step": 5322
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015006566604127582,
+      "loss": 2.4362,
+      "step": 5323
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001500562851782364,
+      "loss": 2.5925,
+      "step": 5324
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000150046904315197,
+      "loss": 2.4806,
+      "step": 5325
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001500375234521576,
+      "loss": 2.6432,
+      "step": 5326
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015002814258911821,
+      "loss": 2.6321,
+      "step": 5327
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001500187617260788,
+      "loss": 2.7601,
+      "step": 5328
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015000938086303942,
+      "loss": 2.3653,
+      "step": 5329
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 2.6552,
+      "step": 5330
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001499906191369606,
+      "loss": 2.3606,
+      "step": 5331
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00014998123827392122,
+      "loss": 2.4783,
+      "step": 5332
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001499718574108818,
+      "loss": 2.683,
+      "step": 5333
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001499624765478424,
+      "loss": 2.7166,
+      "step": 5334
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.000149953095684803,
+      "loss": 2.5805,
+      "step": 5335
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00014994371482176361,
+      "loss": 2.5738,
+      "step": 5336
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001499343339587242,
+      "loss": 2.6366,
+      "step": 5337
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00014992495309568482,
+      "loss": 2.4679,
+      "step": 5338
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00014991557223264541,
+      "loss": 2.7133,
+      "step": 5339
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.00014990619136960603,
+      "loss": 2.6045,
+      "step": 5340
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014989681050656662,
+      "loss": 2.3516,
+      "step": 5341
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014988742964352721,
+      "loss": 2.3814,
+      "step": 5342
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001498780487804878,
+      "loss": 2.7744,
+      "step": 5343
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001498686679174484,
+      "loss": 2.4929,
+      "step": 5344
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014985928705440901,
+      "loss": 2.2559,
+      "step": 5345
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001498499061913696,
+      "loss": 2.8385,
+      "step": 5346
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014984052532833022,
+      "loss": 2.6375,
+      "step": 5347
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014983114446529081,
+      "loss": 2.3876,
+      "step": 5348
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014982176360225143,
+      "loss": 2.7768,
+      "step": 5349
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014981238273921202,
+      "loss": 2.7688,
+      "step": 5350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001498030018761726,
+      "loss": 2.729,
+      "step": 5351
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001497936210131332,
+      "loss": 2.6501,
+      "step": 5352
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014978424015009382,
+      "loss": 2.3254,
+      "step": 5353
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001497748592870544,
+      "loss": 2.701,
+      "step": 5354
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014976547842401503,
+      "loss": 2.3968,
+      "step": 5355
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014975609756097562,
+      "loss": 2.6821,
+      "step": 5356
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001497467166979362,
+      "loss": 2.765,
+      "step": 5357
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014973733583489683,
+      "loss": 2.5287,
+      "step": 5358
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00014972795497185742,
+      "loss": 2.4932,
+      "step": 5359
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.000149718574108818,
+      "loss": 2.441,
+      "step": 5360
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.0001497091932457786,
+      "loss": 2.3633,
+      "step": 5361
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014969981238273922,
+      "loss": 2.7496,
+      "step": 5362
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001496904315196998,
+      "loss": 2.4906,
+      "step": 5363
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014968105065666043,
+      "loss": 2.623,
+      "step": 5364
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014967166979362102,
+      "loss": 2.686,
+      "step": 5365
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014966228893058164,
+      "loss": 2.4436,
+      "step": 5366
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014965290806754223,
+      "loss": 2.4499,
+      "step": 5367
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014964352720450282,
+      "loss": 2.5892,
+      "step": 5368
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001496341463414634,
+      "loss": 2.4727,
+      "step": 5369
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.000149624765478424,
+      "loss": 2.4689,
+      "step": 5370
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014961538461538462,
+      "loss": 2.5431,
+      "step": 5371
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001496060037523452,
+      "loss": 2.362,
+      "step": 5372
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014959662288930583,
+      "loss": 2.5801,
+      "step": 5373
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014958724202626642,
+      "loss": 2.4827,
+      "step": 5374
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014957786116322704,
+      "loss": 2.3344,
+      "step": 5375
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014956848030018763,
+      "loss": 2.471,
+      "step": 5376
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014955909943714822,
+      "loss": 2.2503,
+      "step": 5377
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.0001495497185741088,
+      "loss": 2.499,
+      "step": 5378
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014954033771106943,
+      "loss": 2.6366,
+      "step": 5379
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014953095684803002,
+      "loss": 2.4341,
+      "step": 5380
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014952157598499064,
+      "loss": 2.497,
+      "step": 5381
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014951219512195123,
+      "loss": 2.5917,
+      "step": 5382
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 0.00014950281425891182,
+      "loss": 2.4059,
+      "step": 5383
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014949343339587244,
+      "loss": 2.8979,
+      "step": 5384
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014948405253283303,
+      "loss": 2.2564,
+      "step": 5385
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014947467166979362,
+      "loss": 2.5257,
+      "step": 5386
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001494652908067542,
+      "loss": 2.4983,
+      "step": 5387
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014945590994371483,
+      "loss": 2.2996,
+      "step": 5388
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014944652908067542,
+      "loss": 2.4939,
+      "step": 5389
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014943714821763604,
+      "loss": 2.495,
+      "step": 5390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014942776735459663,
+      "loss": 2.7088,
+      "step": 5391
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014941838649155725,
+      "loss": 2.6786,
+      "step": 5392
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014940900562851784,
+      "loss": 2.5358,
+      "step": 5393
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014939962476547843,
+      "loss": 2.6234,
+      "step": 5394
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014939024390243902,
+      "loss": 2.6901,
+      "step": 5395
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.0001493808630393996,
+      "loss": 2.4782,
+      "step": 5396
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014937148217636023,
+      "loss": 2.4639,
+      "step": 5397
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014936210131332082,
+      "loss": 2.3621,
+      "step": 5398
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014935272045028144,
+      "loss": 2.6911,
+      "step": 5399
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014934333958724203,
+      "loss": 2.5832,
+      "step": 5400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014933395872420265,
+      "loss": 2.7799,
+      "step": 5401
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014932457786116324,
+      "loss": 2.582,
+      "step": 5402
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014931519699812383,
+      "loss": 2.4168,
+      "step": 5403
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00014930581613508442,
+      "loss": 2.6677,
+      "step": 5404
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014929643527204504,
+      "loss": 2.6312,
+      "step": 5405
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014928705440900563,
+      "loss": 2.6293,
+      "step": 5406
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014927767354596625,
+      "loss": 2.712,
+      "step": 5407
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014926829268292684,
+      "loss": 2.5526,
+      "step": 5408
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014925891181988743,
+      "loss": 2.4683,
+      "step": 5409
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014924953095684805,
+      "loss": 2.4413,
+      "step": 5410
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014924015009380864,
+      "loss": 2.4827,
+      "step": 5411
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014923076923076923,
+      "loss": 2.6804,
+      "step": 5412
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014922138836772982,
+      "loss": 2.4062,
+      "step": 5413
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014921200750469044,
+      "loss": 2.8775,
+      "step": 5414
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014920262664165103,
+      "loss": 3.0209,
+      "step": 5415
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014919324577861165,
+      "loss": 2.639,
+      "step": 5416
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014918386491557224,
+      "loss": 2.6257,
+      "step": 5417
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014917448405253286,
+      "loss": 2.5111,
+      "step": 5418
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014916510318949345,
+      "loss": 2.6924,
+      "step": 5419
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014915572232645404,
+      "loss": 2.6602,
+      "step": 5420
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014914634146341463,
+      "loss": 2.5175,
+      "step": 5421
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014913696060037522,
+      "loss": 2.217,
+      "step": 5422
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014912757973733584,
+      "loss": 2.5672,
+      "step": 5423
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014911819887429643,
+      "loss": 2.6266,
+      "step": 5424
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00014910881801125705,
+      "loss": 2.4881,
+      "step": 5425
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014909943714821764,
+      "loss": 2.4951,
+      "step": 5426
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014909005628517826,
+      "loss": 2.3763,
+      "step": 5427
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014908067542213885,
+      "loss": 2.3256,
+      "step": 5428
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014907129455909944,
+      "loss": 2.4033,
+      "step": 5429
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014906191369606003,
+      "loss": 2.3875,
+      "step": 5430
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014905253283302065,
+      "loss": 2.3082,
+      "step": 5431
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014904315196998124,
+      "loss": 2.6707,
+      "step": 5432
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014903377110694186,
+      "loss": 2.5776,
+      "step": 5433
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014902439024390245,
+      "loss": 2.5393,
+      "step": 5434
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014901500938086307,
+      "loss": 2.6069,
+      "step": 5435
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014900562851782366,
+      "loss": 2.6965,
+      "step": 5436
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014899624765478425,
+      "loss": 2.7179,
+      "step": 5437
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014898686679174484,
+      "loss": 2.6,
+      "step": 5438
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014897748592870543,
+      "loss": 2.7437,
+      "step": 5439
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014896810506566605,
+      "loss": 2.623,
+      "step": 5440
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014895872420262664,
+      "loss": 2.6514,
+      "step": 5441
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014894934333958726,
+      "loss": 2.536,
+      "step": 5442
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014893996247654785,
+      "loss": 2.3257,
+      "step": 5443
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014893058161350847,
+      "loss": 2.797,
+      "step": 5444
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014892120075046906,
+      "loss": 2.6749,
+      "step": 5445
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014891181988742965,
+      "loss": 2.7766,
+      "step": 5446
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.00014890243902439024,
+      "loss": 2.5222,
+      "step": 5447
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014889305816135086,
+      "loss": 2.8344,
+      "step": 5448
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014888367729831145,
+      "loss": 2.5577,
+      "step": 5449
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014887429643527204,
+      "loss": 2.7352,
+      "step": 5450
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014886491557223266,
+      "loss": 2.6076,
+      "step": 5451
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014885553470919325,
+      "loss": 2.6593,
+      "step": 5452
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014884615384615387,
+      "loss": 2.6069,
+      "step": 5453
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014883677298311446,
+      "loss": 2.7934,
+      "step": 5454
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014882739212007505,
+      "loss": 2.6354,
+      "step": 5455
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014881801125703564,
+      "loss": 2.8476,
+      "step": 5456
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014880863039399626,
+      "loss": 2.7161,
+      "step": 5457
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014879924953095685,
+      "loss": 2.7504,
+      "step": 5458
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014878986866791747,
+      "loss": 2.5737,
+      "step": 5459
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014878048780487806,
+      "loss": 2.6934,
+      "step": 5460
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014877110694183867,
+      "loss": 2.6013,
+      "step": 5461
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014876172607879927,
+      "loss": 2.6373,
+      "step": 5462
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014875234521575986,
+      "loss": 2.5049,
+      "step": 5463
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014874296435272045,
+      "loss": 2.476,
+      "step": 5464
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014873358348968104,
+      "loss": 2.8001,
+      "step": 5465
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014872420262664166,
+      "loss": 2.4771,
+      "step": 5466
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014871482176360225,
+      "loss": 2.7235,
+      "step": 5467
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00014870544090056287,
+      "loss": 2.2767,
+      "step": 5468
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014869606003752346,
+      "loss": 2.3464,
+      "step": 5469
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014868667917448407,
+      "loss": 2.4827,
+      "step": 5470
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014867729831144467,
+      "loss": 2.5983,
+      "step": 5471
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014866791744840526,
+      "loss": 2.3226,
+      "step": 5472
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014865853658536585,
+      "loss": 2.5741,
+      "step": 5473
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014864915572232646,
+      "loss": 2.3745,
+      "step": 5474
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014863977485928706,
+      "loss": 2.676,
+      "step": 5475
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014863039399624765,
+      "loss": 2.9855,
+      "step": 5476
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014862101313320826,
+      "loss": 2.4196,
+      "step": 5477
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014861163227016886,
+      "loss": 2.5919,
+      "step": 5478
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014860225140712947,
+      "loss": 2.3634,
+      "step": 5479
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014859287054409006,
+      "loss": 2.7145,
+      "step": 5480
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014858348968105068,
+      "loss": 2.2073,
+      "step": 5481
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014857410881801125,
+      "loss": 2.6568,
+      "step": 5482
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014856472795497186,
+      "loss": 2.5917,
+      "step": 5483
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014855534709193246,
+      "loss": 2.539,
+      "step": 5484
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014854596622889307,
+      "loss": 2.5749,
+      "step": 5485
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014853658536585366,
+      "loss": 2.7686,
+      "step": 5486
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014852720450281428,
+      "loss": 2.367,
+      "step": 5487
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014851782363977487,
+      "loss": 2.3488,
+      "step": 5488
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 0.00014850844277673546,
+      "loss": 2.4805,
+      "step": 5489
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014849906191369608,
+      "loss": 2.499,
+      "step": 5490
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014848968105065665,
+      "loss": 2.5603,
+      "step": 5491
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014848030018761726,
+      "loss": 2.7441,
+      "step": 5492
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014847091932457786,
+      "loss": 2.6168,
+      "step": 5493
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014846153846153847,
+      "loss": 2.6305,
+      "step": 5494
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014845215759849906,
+      "loss": 2.7855,
+      "step": 5495
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014844277673545968,
+      "loss": 2.6485,
+      "step": 5496
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014843339587242027,
+      "loss": 2.3539,
+      "step": 5497
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001484240150093809,
+      "loss": 2.4013,
+      "step": 5498
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014841463414634148,
+      "loss": 2.7364,
+      "step": 5499
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014840525328330207,
+      "loss": 2.6102,
+      "step": 5500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014839587242026266,
+      "loss": 2.6388,
+      "step": 5501
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014838649155722326,
+      "loss": 2.6721,
+      "step": 5502
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014837711069418387,
+      "loss": 2.69,
+      "step": 5503
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014836772983114446,
+      "loss": 2.658,
+      "step": 5504
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014835834896810508,
+      "loss": 2.4672,
+      "step": 5505
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014834896810506567,
+      "loss": 2.5412,
+      "step": 5506
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.0001483395872420263,
+      "loss": 2.7013,
+      "step": 5507
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014833020637898688,
+      "loss": 3.0254,
+      "step": 5508
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014832082551594747,
+      "loss": 2.5531,
+      "step": 5509
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014831144465290806,
+      "loss": 2.8643,
+      "step": 5510
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00014830206378986868,
+      "loss": 2.7692,
+      "step": 5511
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014829268292682927,
+      "loss": 2.523,
+      "step": 5512
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001482833020637899,
+      "loss": 2.6101,
+      "step": 5513
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014827392120075048,
+      "loss": 2.6769,
+      "step": 5514
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014826454033771107,
+      "loss": 2.798,
+      "step": 5515
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001482551594746717,
+      "loss": 2.8306,
+      "step": 5516
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014824577861163228,
+      "loss": 2.4207,
+      "step": 5517
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014823639774859287,
+      "loss": 2.577,
+      "step": 5518
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014822701688555346,
+      "loss": 2.5098,
+      "step": 5519
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014821763602251408,
+      "loss": 2.7943,
+      "step": 5520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014820825515947467,
+      "loss": 2.4916,
+      "step": 5521
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001481988742964353,
+      "loss": 2.5892,
+      "step": 5522
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014818949343339588,
+      "loss": 2.3692,
+      "step": 5523
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001481801125703565,
+      "loss": 2.5692,
+      "step": 5524
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001481707317073171,
+      "loss": 2.8397,
+      "step": 5525
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014816135084427768,
+      "loss": 2.7103,
+      "step": 5526
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014815196998123827,
+      "loss": 2.5131,
+      "step": 5527
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014814258911819886,
+      "loss": 2.1983,
+      "step": 5528
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014813320825515948,
+      "loss": 2.8312,
+      "step": 5529
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014812382739212007,
+      "loss": 2.4581,
+      "step": 5530
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.0001481144465290807,
+      "loss": 2.5414,
+      "step": 5531
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00014810506566604128,
+      "loss": 2.8816,
+      "step": 5532
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001480956848030019,
+      "loss": 2.5386,
+      "step": 5533
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001480863039399625,
+      "loss": 2.5165,
+      "step": 5534
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014807692307692308,
+      "loss": 2.4989,
+      "step": 5535
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014806754221388367,
+      "loss": 2.3006,
+      "step": 5536
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001480581613508443,
+      "loss": 2.8084,
+      "step": 5537
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014804878048780488,
+      "loss": 2.7442,
+      "step": 5538
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001480393996247655,
+      "loss": 2.6831,
+      "step": 5539
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001480300187617261,
+      "loss": 2.496,
+      "step": 5540
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014802063789868668,
+      "loss": 2.4392,
+      "step": 5541
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001480112570356473,
+      "loss": 2.8562,
+      "step": 5542
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001480018761726079,
+      "loss": 2.8082,
+      "step": 5543
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014799249530956848,
+      "loss": 2.5672,
+      "step": 5544
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014798311444652907,
+      "loss": 2.639,
+      "step": 5545
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001479737335834897,
+      "loss": 2.4795,
+      "step": 5546
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014796435272045028,
+      "loss": 2.445,
+      "step": 5547
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001479549718574109,
+      "loss": 2.6622,
+      "step": 5548
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001479455909943715,
+      "loss": 2.3704,
+      "step": 5549
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001479362101313321,
+      "loss": 2.581,
+      "step": 5550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001479268292682927,
+      "loss": 2.5247,
+      "step": 5551
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001479174484052533,
+      "loss": 2.6675,
+      "step": 5552
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.00014790806754221388,
+      "loss": 2.1829,
+      "step": 5553
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014789868667917447,
+      "loss": 2.4712,
+      "step": 5554
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001478893058161351,
+      "loss": 2.6583,
+      "step": 5555
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014787992495309568,
+      "loss": 2.7334,
+      "step": 5556
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001478705440900563,
+      "loss": 2.8836,
+      "step": 5557
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001478611632270169,
+      "loss": 2.7391,
+      "step": 5558
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001478517823639775,
+      "loss": 2.3048,
+      "step": 5559
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001478424015009381,
+      "loss": 2.4467,
+      "step": 5560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001478330206378987,
+      "loss": 2.7361,
+      "step": 5561
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014782363977485928,
+      "loss": 2.4626,
+      "step": 5562
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001478142589118199,
+      "loss": 2.7606,
+      "step": 5563
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001478048780487805,
+      "loss": 2.6709,
+      "step": 5564
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001477954971857411,
+      "loss": 2.4858,
+      "step": 5565
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001477861163227017,
+      "loss": 2.7279,
+      "step": 5566
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001477767354596623,
+      "loss": 2.4189,
+      "step": 5567
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001477673545966229,
+      "loss": 2.5888,
+      "step": 5568
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001477579737335835,
+      "loss": 2.6523,
+      "step": 5569
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001477485928705441,
+      "loss": 2.4517,
+      "step": 5570
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.00014773921200750468,
+      "loss": 2.6793,
+      "step": 5571
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001477298311444653,
+      "loss": 2.6261,
+      "step": 5572
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001477204502814259,
+      "loss": 2.8553,
+      "step": 5573
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001477110694183865,
+      "loss": 2.5795,
+      "step": 5574
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 0.0001477016885553471,
+      "loss": 2.4992,
+      "step": 5575
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014769230769230772,
+      "loss": 2.6063,
+      "step": 5576
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001476829268292683,
+      "loss": 2.2976,
+      "step": 5577
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001476735459662289,
+      "loss": 2.462,
+      "step": 5578
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001476641651031895,
+      "loss": 2.4274,
+      "step": 5579
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014765478424015008,
+      "loss": 2.6666,
+      "step": 5580
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001476454033771107,
+      "loss": 2.4807,
+      "step": 5581
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001476360225140713,
+      "loss": 2.3065,
+      "step": 5582
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001476266416510319,
+      "loss": 2.4225,
+      "step": 5583
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001476172607879925,
+      "loss": 2.6288,
+      "step": 5584
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014760787992495312,
+      "loss": 2.4106,
+      "step": 5585
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001475984990619137,
+      "loss": 2.5513,
+      "step": 5586
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001475891181988743,
+      "loss": 2.6885,
+      "step": 5587
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001475797373358349,
+      "loss": 2.5458,
+      "step": 5588
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001475703564727955,
+      "loss": 2.4399,
+      "step": 5589
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001475609756097561,
+      "loss": 2.4765,
+      "step": 5590
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014755159474671672,
+      "loss": 2.5308,
+      "step": 5591
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001475422138836773,
+      "loss": 2.6474,
+      "step": 5592
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001475328330206379,
+      "loss": 2.5635,
+      "step": 5593
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00014752345215759852,
+      "loss": 2.506,
+      "step": 5594
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001475140712945591,
+      "loss": 2.5679,
+      "step": 5595
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.0001475046904315197,
+      "loss": 2.6743,
+      "step": 5596
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001474953095684803,
+      "loss": 2.7538,
+      "step": 5597
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001474859287054409,
+      "loss": 2.7798,
+      "step": 5598
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001474765478424015,
+      "loss": 2.6568,
+      "step": 5599
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014746716697936212,
+      "loss": 2.5951,
+      "step": 5600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001474577861163227,
+      "loss": 2.6371,
+      "step": 5601
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014744840525328333,
+      "loss": 2.3936,
+      "step": 5602
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014743902439024392,
+      "loss": 2.8597,
+      "step": 5603
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001474296435272045,
+      "loss": 2.7813,
+      "step": 5604
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001474202626641651,
+      "loss": 2.6046,
+      "step": 5605
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001474108818011257,
+      "loss": 2.6205,
+      "step": 5606
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001474015009380863,
+      "loss": 2.6347,
+      "step": 5607
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001473921200750469,
+      "loss": 2.4963,
+      "step": 5608
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014738273921200752,
+      "loss": 2.5231,
+      "step": 5609
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001473733583489681,
+      "loss": 2.2082,
+      "step": 5610
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014736397748592873,
+      "loss": 3.0958,
+      "step": 5611
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014735459662288932,
+      "loss": 2.4344,
+      "step": 5612
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001473452157598499,
+      "loss": 2.4647,
+      "step": 5613
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001473358348968105,
+      "loss": 2.5542,
+      "step": 5614
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014732645403377112,
+      "loss": 2.4923,
+      "step": 5615
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.0001473170731707317,
+      "loss": 2.4284,
+      "step": 5616
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00014730769230769233,
+      "loss": 2.2539,
+      "step": 5617
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014729831144465292,
+      "loss": 2.3268,
+      "step": 5618
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001472889305816135,
+      "loss": 2.7626,
+      "step": 5619
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014727954971857412,
+      "loss": 2.7709,
+      "step": 5620
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014727016885553472,
+      "loss": 2.6709,
+      "step": 5621
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001472607879924953,
+      "loss": 2.6056,
+      "step": 5622
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001472514071294559,
+      "loss": 2.9191,
+      "step": 5623
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014724202626641652,
+      "loss": 2.839,
+      "step": 5624
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001472326454033771,
+      "loss": 2.3454,
+      "step": 5625
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014722326454033772,
+      "loss": 2.7592,
+      "step": 5626
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014721388367729832,
+      "loss": 2.4031,
+      "step": 5627
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014720450281425893,
+      "loss": 2.4965,
+      "step": 5628
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014719512195121952,
+      "loss": 2.7842,
+      "step": 5629
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014718574108818012,
+      "loss": 2.8073,
+      "step": 5630
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001471763602251407,
+      "loss": 2.4806,
+      "step": 5631
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001471669793621013,
+      "loss": 2.815,
+      "step": 5632
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014715759849906192,
+      "loss": 2.4133,
+      "step": 5633
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.0001471482176360225,
+      "loss": 2.7838,
+      "step": 5634
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014713883677298312,
+      "loss": 2.2586,
+      "step": 5635
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014712945590994372,
+      "loss": 2.6976,
+      "step": 5636
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014712007504690433,
+      "loss": 2.4223,
+      "step": 5637
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014711069418386492,
+      "loss": 2.6436,
+      "step": 5638
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00014710131332082552,
+      "loss": 2.7548,
+      "step": 5639
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001470919324577861,
+      "loss": 2.6862,
+      "step": 5640
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014708255159474672,
+      "loss": 2.4623,
+      "step": 5641
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014707317073170732,
+      "loss": 2.5155,
+      "step": 5642
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014706378986866793,
+      "loss": 2.6025,
+      "step": 5643
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014705440900562852,
+      "loss": 2.3544,
+      "step": 5644
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014704502814258914,
+      "loss": 2.7422,
+      "step": 5645
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014703564727954973,
+      "loss": 2.7538,
+      "step": 5646
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014702626641651032,
+      "loss": 2.4236,
+      "step": 5647
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014701688555347092,
+      "loss": 2.5859,
+      "step": 5648
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001470075046904315,
+      "loss": 2.3662,
+      "step": 5649
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014699812382739212,
+      "loss": 2.6782,
+      "step": 5650
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014698874296435272,
+      "loss": 2.7347,
+      "step": 5651
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014697936210131333,
+      "loss": 2.5046,
+      "step": 5652
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014696998123827392,
+      "loss": 2.5365,
+      "step": 5653
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014696060037523454,
+      "loss": 2.5489,
+      "step": 5654
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014695121951219513,
+      "loss": 2.8471,
+      "step": 5655
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014694183864915575,
+      "loss": 2.6457,
+      "step": 5656
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014693245778611634,
+      "loss": 2.5247,
+      "step": 5657
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014692307692307693,
+      "loss": 2.6209,
+      "step": 5658
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014691369606003752,
+      "loss": 2.4887,
+      "step": 5659
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.00014690431519699812,
+      "loss": 2.4374,
+      "step": 5660
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014689493433395873,
+      "loss": 2.5814,
+      "step": 5661
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014688555347091932,
+      "loss": 2.6983,
+      "step": 5662
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014687617260787994,
+      "loss": 2.5486,
+      "step": 5663
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014686679174484053,
+      "loss": 2.6802,
+      "step": 5664
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014685741088180115,
+      "loss": 2.4934,
+      "step": 5665
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014684803001876174,
+      "loss": 2.6044,
+      "step": 5666
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014683864915572233,
+      "loss": 2.5725,
+      "step": 5667
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014682926829268292,
+      "loss": 2.6507,
+      "step": 5668
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014681988742964354,
+      "loss": 2.7172,
+      "step": 5669
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014681050656660413,
+      "loss": 2.3882,
+      "step": 5670
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014680112570356475,
+      "loss": 2.2532,
+      "step": 5671
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014679174484052534,
+      "loss": 2.5491,
+      "step": 5672
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014678236397748593,
+      "loss": 2.5977,
+      "step": 5673
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014677298311444655,
+      "loss": 2.3924,
+      "step": 5674
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014676360225140714,
+      "loss": 2.6678,
+      "step": 5675
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014675422138836773,
+      "loss": 2.6964,
+      "step": 5676
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014674484052532832,
+      "loss": 2.4351,
+      "step": 5677
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014673545966228894,
+      "loss": 2.2036,
+      "step": 5678
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014672607879924953,
+      "loss": 2.601,
+      "step": 5679
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014671669793621015,
+      "loss": 2.6957,
+      "step": 5680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 0.00014670731707317074,
+      "loss": 2.3339,
+      "step": 5681
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014669793621013136,
+      "loss": 2.6553,
+      "step": 5682
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014668855534709195,
+      "loss": 2.6517,
+      "step": 5683
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014667917448405254,
+      "loss": 2.6328,
+      "step": 5684
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014666979362101313,
+      "loss": 2.6309,
+      "step": 5685
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014666041275797372,
+      "loss": 2.5072,
+      "step": 5686
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014665103189493434,
+      "loss": 2.5971,
+      "step": 5687
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014664165103189493,
+      "loss": 2.9054,
+      "step": 5688
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014663227016885555,
+      "loss": 2.4797,
+      "step": 5689
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014662288930581614,
+      "loss": 2.7757,
+      "step": 5690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014661350844277676,
+      "loss": 2.4673,
+      "step": 5691
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014660412757973735,
+      "loss": 2.4943,
+      "step": 5692
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014659474671669794,
+      "loss": 2.869,
+      "step": 5693
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014658536585365853,
+      "loss": 2.5962,
+      "step": 5694
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014657598499061915,
+      "loss": 2.6944,
+      "step": 5695
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014656660412757974,
+      "loss": 2.5045,
+      "step": 5696
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014655722326454036,
+      "loss": 2.625,
+      "step": 5697
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014654784240150095,
+      "loss": 2.8829,
+      "step": 5698
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014653846153846154,
+      "loss": 2.4949,
+      "step": 5699
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014652908067542216,
+      "loss": 2.3755,
+      "step": 5700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014651969981238275,
+      "loss": 2.6031,
+      "step": 5701
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014651031894934334,
+      "loss": 2.8191,
+      "step": 5702
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00014650093808630393,
+      "loss": 2.7115,
+      "step": 5703
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014649155722326455,
+      "loss": 2.6915,
+      "step": 5704
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014648217636022514,
+      "loss": 2.5409,
+      "step": 5705
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014647279549718576,
+      "loss": 2.7499,
+      "step": 5706
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014646341463414635,
+      "loss": 2.7328,
+      "step": 5707
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014645403377110697,
+      "loss": 2.8397,
+      "step": 5708
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014644465290806756,
+      "loss": 2.6772,
+      "step": 5709
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014643527204502815,
+      "loss": 2.566,
+      "step": 5710
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014642589118198874,
+      "loss": 2.7671,
+      "step": 5711
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014641651031894933,
+      "loss": 2.7891,
+      "step": 5712
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014640712945590995,
+      "loss": 2.5209,
+      "step": 5713
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014639774859287054,
+      "loss": 2.6671,
+      "step": 5714
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014638836772983116,
+      "loss": 2.6322,
+      "step": 5715
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014637898686679175,
+      "loss": 2.5267,
+      "step": 5716
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014636960600375237,
+      "loss": 2.6835,
+      "step": 5717
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014636022514071296,
+      "loss": 2.5425,
+      "step": 5718
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014635084427767355,
+      "loss": 2.7119,
+      "step": 5719
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014634146341463414,
+      "loss": 2.5677,
+      "step": 5720
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014633208255159476,
+      "loss": 2.6436,
+      "step": 5721
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014632270168855535,
+      "loss": 2.8173,
+      "step": 5722
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014631332082551597,
+      "loss": 2.7572,
+      "step": 5723
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00014630393996247656,
+      "loss": 2.5163,
+      "step": 5724
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014629455909943715,
+      "loss": 2.8146,
+      "step": 5725
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014628517823639777,
+      "loss": 2.8232,
+      "step": 5726
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014627579737335836,
+      "loss": 2.3598,
+      "step": 5727
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014626641651031895,
+      "loss": 2.6632,
+      "step": 5728
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014625703564727954,
+      "loss": 2.6403,
+      "step": 5729
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014624765478424016,
+      "loss": 2.5634,
+      "step": 5730
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014623827392120075,
+      "loss": 2.5165,
+      "step": 5731
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014622889305816137,
+      "loss": 2.5702,
+      "step": 5732
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014621951219512196,
+      "loss": 2.4775,
+      "step": 5733
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014621013133208258,
+      "loss": 2.712,
+      "step": 5734
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014620075046904317,
+      "loss": 2.2821,
+      "step": 5735
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014619136960600376,
+      "loss": 2.36,
+      "step": 5736
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014618198874296435,
+      "loss": 2.3792,
+      "step": 5737
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014617260787992494,
+      "loss": 2.9556,
+      "step": 5738
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014616322701688556,
+      "loss": 2.4737,
+      "step": 5739
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014615384615384615,
+      "loss": 2.5959,
+      "step": 5740
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014614446529080677,
+      "loss": 2.8865,
+      "step": 5741
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014613508442776736,
+      "loss": 2.8457,
+      "step": 5742
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014612570356472798,
+      "loss": 2.5426,
+      "step": 5743
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014611632270168857,
+      "loss": 2.4741,
+      "step": 5744
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.00014610694183864916,
+      "loss": 2.4021,
+      "step": 5745
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014609756097560975,
+      "loss": 2.7799,
+      "step": 5746
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014608818011257037,
+      "loss": 2.4877,
+      "step": 5747
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014607879924953096,
+      "loss": 2.5441,
+      "step": 5748
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014606941838649158,
+      "loss": 2.3972,
+      "step": 5749
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014606003752345217,
+      "loss": 2.7602,
+      "step": 5750
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014605065666041276,
+      "loss": 2.5926,
+      "step": 5751
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014604127579737338,
+      "loss": 2.5148,
+      "step": 5752
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014603189493433397,
+      "loss": 2.5641,
+      "step": 5753
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014602251407129456,
+      "loss": 2.5308,
+      "step": 5754
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014601313320825515,
+      "loss": 2.482,
+      "step": 5755
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014600375234521577,
+      "loss": 2.6494,
+      "step": 5756
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014599437148217636,
+      "loss": 2.7002,
+      "step": 5757
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014598499061913698,
+      "loss": 2.6751,
+      "step": 5758
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014597560975609757,
+      "loss": 2.455,
+      "step": 5759
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014596622889305819,
+      "loss": 2.7713,
+      "step": 5760
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014595684803001878,
+      "loss": 2.4384,
+      "step": 5761
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014594746716697937,
+      "loss": 2.6661,
+      "step": 5762
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014593808630393996,
+      "loss": 2.6304,
+      "step": 5763
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014592870544090055,
+      "loss": 2.4975,
+      "step": 5764
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014591932457786117,
+      "loss": 2.531,
+      "step": 5765
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014590994371482176,
+      "loss": 2.4611,
+      "step": 5766
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 0.00014590056285178238,
+      "loss": 2.6108,
+      "step": 5767
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014589118198874297,
+      "loss": 2.4476,
+      "step": 5768
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014588180112570358,
+      "loss": 2.6732,
+      "step": 5769
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014587242026266418,
+      "loss": 2.746,
+      "step": 5770
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014586303939962477,
+      "loss": 2.3879,
+      "step": 5771
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014585365853658536,
+      "loss": 2.7022,
+      "step": 5772
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014584427767354598,
+      "loss": 2.5203,
+      "step": 5773
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014583489681050657,
+      "loss": 2.5993,
+      "step": 5774
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014582551594746718,
+      "loss": 2.5291,
+      "step": 5775
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014581613508442778,
+      "loss": 2.5749,
+      "step": 5776
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014580675422138837,
+      "loss": 2.6918,
+      "step": 5777
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014579737335834898,
+      "loss": 2.6304,
+      "step": 5778
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014578799249530958,
+      "loss": 2.6043,
+      "step": 5779
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014577861163227017,
+      "loss": 2.5585,
+      "step": 5780
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014576923076923076,
+      "loss": 2.636,
+      "step": 5781
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014575984990619138,
+      "loss": 2.5181,
+      "step": 5782
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014575046904315197,
+      "loss": 2.5908,
+      "step": 5783
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014574108818011258,
+      "loss": 2.6029,
+      "step": 5784
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014573170731707318,
+      "loss": 2.5372,
+      "step": 5785
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.0001457223264540338,
+      "loss": 2.8799,
+      "step": 5786
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014571294559099438,
+      "loss": 2.834,
+      "step": 5787
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00014570356472795498,
+      "loss": 2.4862,
+      "step": 5788
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014569418386491557,
+      "loss": 2.4408,
+      "step": 5789
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014568480300187616,
+      "loss": 2.5442,
+      "step": 5790
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014567542213883678,
+      "loss": 2.8399,
+      "step": 5791
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014566604127579737,
+      "loss": 2.7285,
+      "step": 5792
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014565666041275798,
+      "loss": 2.4569,
+      "step": 5793
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014564727954971858,
+      "loss": 2.5661,
+      "step": 5794
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001456378986866792,
+      "loss": 2.699,
+      "step": 5795
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014562851782363978,
+      "loss": 2.3898,
+      "step": 5796
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014561913696060038,
+      "loss": 2.5699,
+      "step": 5797
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014560975609756097,
+      "loss": 2.5799,
+      "step": 5798
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014560037523452158,
+      "loss": 2.5925,
+      "step": 5799
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014559099437148218,
+      "loss": 2.8037,
+      "step": 5800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001455816135084428,
+      "loss": 2.3172,
+      "step": 5801
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014557223264540338,
+      "loss": 2.609,
+      "step": 5802
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014556285178236398,
+      "loss": 2.3338,
+      "step": 5803
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.0001455534709193246,
+      "loss": 2.5567,
+      "step": 5804
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014554409005628518,
+      "loss": 2.5905,
+      "step": 5805
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014553470919324578,
+      "loss": 2.6664,
+      "step": 5806
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014552532833020637,
+      "loss": 2.6862,
+      "step": 5807
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014551594746716698,
+      "loss": 2.541,
+      "step": 5808
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00014550656660412757,
+      "loss": 2.5794,
+      "step": 5809
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001454971857410882,
+      "loss": 2.7317,
+      "step": 5810
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00014548780487804878,
+      "loss": 2.4369,
+      "step": 5811
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001454784240150094,
+      "loss": 2.3562,
+      "step": 5812
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00014546904315197,
+      "loss": 2.4912,
+      "step": 5813
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001454596622889306,
+      "loss": 2.8363,
+      "step": 5814
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00014545028142589117,
+      "loss": 2.5952,
+      "step": 5815
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00014544090056285177,
+      "loss": 2.5098,
+      "step": 5816
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00014543151969981238,
+      "loss": 2.6769,
+      "step": 5817
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00014542213883677297,
+      "loss": 2.4358,
+      "step": 5818
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001454127579737336,
+      "loss": 2.6613,
+      "step": 5819
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00014540337711069418,
+      "loss": 2.5551,
+      "step": 5820
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001453939962476548,
+      "loss": 2.1326,
+      "step": 5821
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001453846153846154,
+      "loss": 2.8451,
+      "step": 5822
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000145375234521576,
+      "loss": 2.7511,
+      "step": 5823
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001453658536585366,
+      "loss": 2.6148,
+      "step": 5824
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001453564727954972,
+      "loss": 2.5936,
+      "step": 5825
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00014534709193245778,
+      "loss": 2.6565,
+      "step": 5826
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001453377110694184,
+      "loss": 2.538,
+      "step": 5827
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.000145328330206379,
+      "loss": 2.2381,
+      "step": 5828
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00014531894934333958,
+      "loss": 2.6353,
+      "step": 5829
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001453095684803002,
+      "loss": 2.6683,
+      "step": 5830
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.0001453001876172608,
+      "loss": 2.437,
+      "step": 5831
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001452908067542214,
+      "loss": 2.812,
+      "step": 5832
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000145281425891182,
+      "loss": 2.3567,
+      "step": 5833
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001452720450281426,
+      "loss": 2.5484,
+      "step": 5834
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00014526266416510318,
+      "loss": 2.6037,
+      "step": 5835
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001452532833020638,
+      "loss": 2.6508,
+      "step": 5836
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001452439024390244,
+      "loss": 2.5735,
+      "step": 5837
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000145234521575985,
+      "loss": 2.3878,
+      "step": 5838
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001452251407129456,
+      "loss": 2.768,
+      "step": 5839
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00014521575984990622,
+      "loss": 2.7782,
+      "step": 5840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001452063789868668,
+      "loss": 2.4607,
+      "step": 5841
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001451969981238274,
+      "loss": 2.7952,
+      "step": 5842
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000145187617260788,
+      "loss": 2.6503,
+      "step": 5843
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00014517823639774858,
+      "loss": 2.5714,
+      "step": 5844
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001451688555347092,
+      "loss": 2.7187,
+      "step": 5845
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001451594746716698,
+      "loss": 2.7041,
+      "step": 5846
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001451500938086304,
+      "loss": 2.7007,
+      "step": 5847
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.000145140712945591,
+      "loss": 2.4232,
+      "step": 5848
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00014513133208255162,
+      "loss": 2.5147,
+      "step": 5849
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001451219512195122,
+      "loss": 2.4844,
+      "step": 5850
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001451125703564728,
+      "loss": 2.863,
+      "step": 5851
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.0001451031894934334,
+      "loss": 2.3978,
+      "step": 5852
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.000145093808630394,
+      "loss": 2.5095,
+      "step": 5853
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001450844277673546,
+      "loss": 2.9659,
+      "step": 5854
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00014507504690431522,
+      "loss": 2.7434,
+      "step": 5855
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001450656660412758,
+      "loss": 2.6375,
+      "step": 5856
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001450562851782364,
+      "loss": 2.301,
+      "step": 5857
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00014504690431519702,
+      "loss": 2.549,
+      "step": 5858
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001450375234521576,
+      "loss": 2.4588,
+      "step": 5859
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001450281425891182,
+      "loss": 2.9834,
+      "step": 5860
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001450187617260788,
+      "loss": 2.7307,
+      "step": 5861
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001450093808630394,
+      "loss": 2.3139,
+      "step": 5862
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.000145,
+      "loss": 2.6645,
+      "step": 5863
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00014499061913696062,
+      "loss": 2.9094,
+      "step": 5864
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001449812382739212,
+      "loss": 2.3966,
+      "step": 5865
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00014497185741088183,
+      "loss": 2.4945,
+      "step": 5866
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00014496247654784242,
+      "loss": 2.5934,
+      "step": 5867
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.000144953095684803,
+      "loss": 2.4859,
+      "step": 5868
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001449437148217636,
+      "loss": 2.572,
+      "step": 5869
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001449343339587242,
+      "loss": 2.6936,
+      "step": 5870
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001449249530956848,
+      "loss": 2.4188,
+      "step": 5871
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.0001449155722326454,
+      "loss": 2.7737,
+      "step": 5872
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 0.00014490619136960602,
+      "loss": 2.4597,
+      "step": 5873
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001448968105065666,
+      "loss": 2.3749,
+      "step": 5874
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014488742964352723,
+      "loss": 2.4477,
+      "step": 5875
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014487804878048782,
+      "loss": 2.4372,
+      "step": 5876
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001448686679174484,
+      "loss": 2.4549,
+      "step": 5877
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000144859287054409,
+      "loss": 2.6109,
+      "step": 5878
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014484990619136962,
+      "loss": 2.5768,
+      "step": 5879
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001448405253283302,
+      "loss": 2.6921,
+      "step": 5880
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014483114446529083,
+      "loss": 2.7537,
+      "step": 5881
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014482176360225142,
+      "loss": 2.4477,
+      "step": 5882
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.000144812382739212,
+      "loss": 2.6309,
+      "step": 5883
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014480300187617263,
+      "loss": 2.5535,
+      "step": 5884
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014479362101313322,
+      "loss": 2.158,
+      "step": 5885
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001447842401500938,
+      "loss": 2.6021,
+      "step": 5886
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001447748592870544,
+      "loss": 2.3468,
+      "step": 5887
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014476547842401502,
+      "loss": 2.7722,
+      "step": 5888
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.0001447560975609756,
+      "loss": 2.5037,
+      "step": 5889
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014474671669793623,
+      "loss": 2.8823,
+      "step": 5890
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014473733583489682,
+      "loss": 2.628,
+      "step": 5891
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014472795497185744,
+      "loss": 2.4083,
+      "step": 5892
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014471857410881803,
+      "loss": 2.7351,
+      "step": 5893
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00014470919324577862,
+      "loss": 2.7604,
+      "step": 5894
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001446998123827392,
+      "loss": 2.549,
+      "step": 5895
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001446904315196998,
+      "loss": 2.8201,
+      "step": 5896
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014468105065666042,
+      "loss": 2.5681,
+      "step": 5897
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.000144671669793621,
+      "loss": 2.5869,
+      "step": 5898
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014466228893058163,
+      "loss": 2.5584,
+      "step": 5899
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014465290806754222,
+      "loss": 2.5536,
+      "step": 5900
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014464352720450284,
+      "loss": 2.3814,
+      "step": 5901
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014463414634146343,
+      "loss": 2.2768,
+      "step": 5902
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014462476547842402,
+      "loss": 2.6718,
+      "step": 5903
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001446153846153846,
+      "loss": 2.5268,
+      "step": 5904
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014460600375234523,
+      "loss": 2.6105,
+      "step": 5905
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014459662288930582,
+      "loss": 2.4472,
+      "step": 5906
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014458724202626644,
+      "loss": 2.6141,
+      "step": 5907
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014457786116322703,
+      "loss": 2.5864,
+      "step": 5908
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014456848030018762,
+      "loss": 2.6668,
+      "step": 5909
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014455909943714824,
+      "loss": 2.4536,
+      "step": 5910
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014454971857410883,
+      "loss": 2.6828,
+      "step": 5911
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014454033771106942,
+      "loss": 2.4935,
+      "step": 5912
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014453095684803,
+      "loss": 2.7946,
+      "step": 5913
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014452157598499063,
+      "loss": 2.6093,
+      "step": 5914
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014451219512195122,
+      "loss": 2.6248,
+      "step": 5915
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.00014450281425891184,
+      "loss": 2.5395,
+      "step": 5916
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014449343339587243,
+      "loss": 2.4609,
+      "step": 5917
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014448405253283304,
+      "loss": 2.6541,
+      "step": 5918
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014447467166979364,
+      "loss": 2.4377,
+      "step": 5919
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014446529080675423,
+      "loss": 2.3898,
+      "step": 5920
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014445590994371482,
+      "loss": 2.5393,
+      "step": 5921
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.0001444465290806754,
+      "loss": 2.6941,
+      "step": 5922
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014443714821763603,
+      "loss": 2.5286,
+      "step": 5923
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014442776735459662,
+      "loss": 2.6741,
+      "step": 5924
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014441838649155724,
+      "loss": 2.7141,
+      "step": 5925
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014440900562851783,
+      "loss": 2.8274,
+      "step": 5926
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014439962476547844,
+      "loss": 2.4932,
+      "step": 5927
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014439024390243904,
+      "loss": 2.36,
+      "step": 5928
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014438086303939963,
+      "loss": 3.0635,
+      "step": 5929
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014437148217636022,
+      "loss": 2.3774,
+      "step": 5930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014436210131332084,
+      "loss": 2.6429,
+      "step": 5931
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014435272045028143,
+      "loss": 2.4491,
+      "step": 5932
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014434333958724204,
+      "loss": 2.6821,
+      "step": 5933
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014433395872420264,
+      "loss": 2.7015,
+      "step": 5934
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014432457786116323,
+      "loss": 2.3668,
+      "step": 5935
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014431519699812384,
+      "loss": 2.4572,
+      "step": 5936
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00014430581613508444,
+      "loss": 2.4047,
+      "step": 5937
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014429643527204503,
+      "loss": 2.6207,
+      "step": 5938
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014428705440900562,
+      "loss": 2.7891,
+      "step": 5939
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014427767354596624,
+      "loss": 2.4392,
+      "step": 5940
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014426829268292683,
+      "loss": 2.5042,
+      "step": 5941
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014425891181988744,
+      "loss": 2.4251,
+      "step": 5942
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014424953095684804,
+      "loss": 2.5696,
+      "step": 5943
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014424015009380865,
+      "loss": 2.8172,
+      "step": 5944
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014423076923076924,
+      "loss": 2.7605,
+      "step": 5945
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014422138836772984,
+      "loss": 2.4624,
+      "step": 5946
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014421200750469043,
+      "loss": 2.6228,
+      "step": 5947
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014420262664165102,
+      "loss": 3.0227,
+      "step": 5948
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014419324577861164,
+      "loss": 2.8161,
+      "step": 5949
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014418386491557223,
+      "loss": 2.7392,
+      "step": 5950
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014417448405253284,
+      "loss": 2.609,
+      "step": 5951
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014416510318949344,
+      "loss": 2.5082,
+      "step": 5952
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014415572232645405,
+      "loss": 2.5522,
+      "step": 5953
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014414634146341464,
+      "loss": 2.7928,
+      "step": 5954
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014413696060037523,
+      "loss": 2.9763,
+      "step": 5955
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014412757973733583,
+      "loss": 2.5501,
+      "step": 5956
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014411819887429644,
+      "loss": 2.6155,
+      "step": 5957
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00014410881801125703,
+      "loss": 2.9322,
+      "step": 5958
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014409943714821765,
+      "loss": 2.7319,
+      "step": 5959
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014409005628517824,
+      "loss": 2.7483,
+      "step": 5960
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014408067542213883,
+      "loss": 2.8291,
+      "step": 5961
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014407129455909945,
+      "loss": 2.5968,
+      "step": 5962
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014406191369606004,
+      "loss": 2.4737,
+      "step": 5963
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014405253283302063,
+      "loss": 2.667,
+      "step": 5964
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014404315196998123,
+      "loss": 2.605,
+      "step": 5965
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014403377110694184,
+      "loss": 2.5506,
+      "step": 5966
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014402439024390243,
+      "loss": 2.6116,
+      "step": 5967
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014401500938086305,
+      "loss": 2.6989,
+      "step": 5968
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014400562851782364,
+      "loss": 2.8122,
+      "step": 5969
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014399624765478426,
+      "loss": 2.6631,
+      "step": 5970
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014398686679174485,
+      "loss": 2.213,
+      "step": 5971
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014397748592870547,
+      "loss": 2.1687,
+      "step": 5972
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014396810506566603,
+      "loss": 2.405,
+      "step": 5973
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014395872420262663,
+      "loss": 2.6102,
+      "step": 5974
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014394934333958724,
+      "loss": 2.5672,
+      "step": 5975
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014393996247654783,
+      "loss": 2.4904,
+      "step": 5976
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014393058161350845,
+      "loss": 2.6185,
+      "step": 5977
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014392120075046904,
+      "loss": 2.4347,
+      "step": 5978
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014391181988742966,
+      "loss": 2.629,
+      "step": 5979
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 0.00014390243902439025,
+      "loss": 2.7129,
+      "step": 5980
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014389305816135087,
+      "loss": 2.2635,
+      "step": 5981
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014388367729831143,
+      "loss": 2.898,
+      "step": 5982
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014387429643527205,
+      "loss": 2.7421,
+      "step": 5983
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014386491557223264,
+      "loss": 2.4147,
+      "step": 5984
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014385553470919326,
+      "loss": 2.6461,
+      "step": 5985
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014384615384615385,
+      "loss": 2.2413,
+      "step": 5986
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014383677298311444,
+      "loss": 2.635,
+      "step": 5987
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014382739212007506,
+      "loss": 2.5204,
+      "step": 5988
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014381801125703565,
+      "loss": 2.5139,
+      "step": 5989
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014380863039399627,
+      "loss": 2.6605,
+      "step": 5990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014379924953095686,
+      "loss": 2.4119,
+      "step": 5991
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014378986866791745,
+      "loss": 2.6751,
+      "step": 5992
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014378048780487804,
+      "loss": 2.5118,
+      "step": 5993
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014377110694183866,
+      "loss": 2.7012,
+      "step": 5994
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014376172607879925,
+      "loss": 2.5069,
+      "step": 5995
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014375234521575987,
+      "loss": 2.6769,
+      "step": 5996
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014374296435272046,
+      "loss": 2.44,
+      "step": 5997
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014373358348968108,
+      "loss": 2.4644,
+      "step": 5998
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014372420262664167,
+      "loss": 2.5583,
+      "step": 5999
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014371482176360226,
+      "loss": 2.5505,
+      "step": 6000
+    },
+    {
+      "epoch": 2.81,
+      "eval_loss": 2.6150174140930176,
+      "eval_runtime": 67.649,
+      "eval_samples_per_second": 28.012,
+      "eval_steps_per_second": 3.503,
+      "step": 6000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00014370544090056285,
+      "loss": 2.422,
+      "step": 6001
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014369606003752344,
+      "loss": 2.2897,
+      "step": 6002
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014368667917448406,
+      "loss": 2.9737,
+      "step": 6003
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014367729831144465,
+      "loss": 2.3454,
+      "step": 6004
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014366791744840527,
+      "loss": 2.485,
+      "step": 6005
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014365853658536586,
+      "loss": 2.4827,
+      "step": 6006
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014364915572232648,
+      "loss": 2.7594,
+      "step": 6007
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014363977485928707,
+      "loss": 2.5596,
+      "step": 6008
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014363039399624766,
+      "loss": 2.4475,
+      "step": 6009
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014362101313320825,
+      "loss": 2.314,
+      "step": 6010
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014361163227016887,
+      "loss": 2.6814,
+      "step": 6011
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014360225140712946,
+      "loss": 2.9425,
+      "step": 6012
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014359287054409005,
+      "loss": 2.6652,
+      "step": 6013
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014358348968105067,
+      "loss": 2.7187,
+      "step": 6014
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014357410881801126,
+      "loss": 2.5555,
+      "step": 6015
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014356472795497188,
+      "loss": 2.3975,
+      "step": 6016
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014355534709193247,
+      "loss": 2.49,
+      "step": 6017
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014354596622889306,
+      "loss": 2.6382,
+      "step": 6018
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014353658536585365,
+      "loss": 2.2276,
+      "step": 6019
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014352720450281427,
+      "loss": 2.5,
+      "step": 6020
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014351782363977486,
+      "loss": 2.2747,
+      "step": 6021
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.00014350844277673548,
+      "loss": 2.6395,
+      "step": 6022
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014349906191369607,
+      "loss": 2.5744,
+      "step": 6023
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001434896810506567,
+      "loss": 2.5374,
+      "step": 6024
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014348030018761728,
+      "loss": 2.8979,
+      "step": 6025
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014347091932457787,
+      "loss": 2.5441,
+      "step": 6026
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014346153846153846,
+      "loss": 2.3994,
+      "step": 6027
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014345215759849905,
+      "loss": 2.7594,
+      "step": 6028
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014344277673545967,
+      "loss": 2.375,
+      "step": 6029
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014343339587242026,
+      "loss": 2.5754,
+      "step": 6030
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014342401500938088,
+      "loss": 2.8555,
+      "step": 6031
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014341463414634147,
+      "loss": 2.5363,
+      "step": 6032
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001434052532833021,
+      "loss": 2.3111,
+      "step": 6033
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014339587242026268,
+      "loss": 2.5219,
+      "step": 6034
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014338649155722327,
+      "loss": 2.7915,
+      "step": 6035
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014337711069418386,
+      "loss": 2.5639,
+      "step": 6036
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014336772983114448,
+      "loss": 2.2505,
+      "step": 6037
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014335834896810507,
+      "loss": 2.7369,
+      "step": 6038
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014334896810506566,
+      "loss": 2.6651,
+      "step": 6039
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014333958724202628,
+      "loss": 2.4045,
+      "step": 6040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014333020637898687,
+      "loss": 2.755,
+      "step": 6041
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.0001433208255159475,
+      "loss": 2.5786,
+      "step": 6042
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014331144465290808,
+      "loss": 2.5211,
+      "step": 6043
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00014330206378986867,
+      "loss": 2.4617,
+      "step": 6044
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014329268292682926,
+      "loss": 2.6158,
+      "step": 6045
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014328330206378988,
+      "loss": 2.4753,
+      "step": 6046
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014327392120075047,
+      "loss": 2.704,
+      "step": 6047
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001432645403377111,
+      "loss": 2.6326,
+      "step": 6048
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014325515947467168,
+      "loss": 2.4892,
+      "step": 6049
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001432457786116323,
+      "loss": 2.7071,
+      "step": 6050
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001432363977485929,
+      "loss": 2.682,
+      "step": 6051
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014322701688555348,
+      "loss": 2.4754,
+      "step": 6052
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014321763602251407,
+      "loss": 2.5799,
+      "step": 6053
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014320825515947466,
+      "loss": 2.4962,
+      "step": 6054
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014319887429643528,
+      "loss": 2.4025,
+      "step": 6055
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014318949343339587,
+      "loss": 2.2363,
+      "step": 6056
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001431801125703565,
+      "loss": 2.2129,
+      "step": 6057
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014317073170731708,
+      "loss": 2.5992,
+      "step": 6058
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001431613508442777,
+      "loss": 2.3765,
+      "step": 6059
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001431519699812383,
+      "loss": 2.9112,
+      "step": 6060
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014314258911819888,
+      "loss": 2.5162,
+      "step": 6061
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014313320825515947,
+      "loss": 2.8259,
+      "step": 6062
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001431238273921201,
+      "loss": 2.4581,
+      "step": 6063
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.00014311444652908068,
+      "loss": 2.6709,
+      "step": 6064
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 0.0001431050656660413,
+      "loss": 2.834,
+      "step": 6065
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001430956848030019,
+      "loss": 2.7409,
+      "step": 6066
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00014308630393996248,
+      "loss": 2.9472,
+      "step": 6067
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001430769230769231,
+      "loss": 2.5763,
+      "step": 6068
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001430675422138837,
+      "loss": 2.8822,
+      "step": 6069
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00014305816135084428,
+      "loss": 2.6403,
+      "step": 6070
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00014304878048780487,
+      "loss": 2.6737,
+      "step": 6071
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001430393996247655,
+      "loss": 2.9331,
+      "step": 6072
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00014303001876172608,
+      "loss": 2.7016,
+      "step": 6073
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001430206378986867,
+      "loss": 2.6282,
+      "step": 6074
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001430112570356473,
+      "loss": 2.6785,
+      "step": 6075
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001430018761726079,
+      "loss": 2.4507,
+      "step": 6076
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001429924953095685,
+      "loss": 2.5217,
+      "step": 6077
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00014298311444652909,
+      "loss": 2.5578,
+      "step": 6078
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00014297373358348968,
+      "loss": 2.3695,
+      "step": 6079
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00014296435272045027,
+      "loss": 2.763,
+      "step": 6080
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00014295497185741089,
+      "loss": 2.6775,
+      "step": 6081
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00014294559099437148,
+      "loss": 2.6711,
+      "step": 6082
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001429362101313321,
+      "loss": 2.5005,
+      "step": 6083
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00014292682926829269,
+      "loss": 2.7785,
+      "step": 6084
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001429174484052533,
+      "loss": 2.8693,
+      "step": 6085
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.0001429080675422139,
+      "loss": 2.5875,
+      "step": 6086
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00014289868667917449,
+      "loss": 2.4457,
+      "step": 6087
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00014288930581613508,
+      "loss": 2.6728,
+      "step": 6088
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001428799249530957,
+      "loss": 2.4809,
+      "step": 6089
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00014287054409005629,
+      "loss": 2.3187,
+      "step": 6090
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001428611632270169,
+      "loss": 2.6427,
+      "step": 6091
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001428517823639775,
+      "loss": 2.5339,
+      "step": 6092
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00014284240150093809,
+      "loss": 2.6237,
+      "step": 6093
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001428330206378987,
+      "loss": 2.8806,
+      "step": 6094
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001428236397748593,
+      "loss": 2.4188,
+      "step": 6095
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00014281425891181989,
+      "loss": 2.5925,
+      "step": 6096
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00014280487804878048,
+      "loss": 2.5807,
+      "step": 6097
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001427954971857411,
+      "loss": 2.5904,
+      "step": 6098
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00014278611632270169,
+      "loss": 2.8711,
+      "step": 6099
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001427767354596623,
+      "loss": 2.3445,
+      "step": 6100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001427673545966229,
+      "loss": 2.5516,
+      "step": 6101
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001427579737335835,
+      "loss": 2.7599,
+      "step": 6102
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001427485928705441,
+      "loss": 2.79,
+      "step": 6103
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001427392120075047,
+      "loss": 2.6707,
+      "step": 6104
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00014272983114446529,
+      "loss": 2.5626,
+      "step": 6105
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00014272045028142588,
+      "loss": 2.3868,
+      "step": 6106
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001427110694183865,
+      "loss": 2.804,
+      "step": 6107
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.00014270168855534709,
+      "loss": 2.5302,
+      "step": 6108
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001426923076923077,
+      "loss": 2.6054,
+      "step": 6109
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001426829268292683,
+      "loss": 2.4006,
+      "step": 6110
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001426735459662289,
+      "loss": 2.4237,
+      "step": 6111
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001426641651031895,
+      "loss": 2.8051,
+      "step": 6112
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001426547842401501,
+      "loss": 2.3785,
+      "step": 6113
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00014264540337711069,
+      "loss": 2.5592,
+      "step": 6114
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001426360225140713,
+      "loss": 2.4957,
+      "step": 6115
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001426266416510319,
+      "loss": 2.6327,
+      "step": 6116
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001426172607879925,
+      "loss": 2.7182,
+      "step": 6117
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001426078799249531,
+      "loss": 2.4737,
+      "step": 6118
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001425984990619137,
+      "loss": 2.5229,
+      "step": 6119
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001425891181988743,
+      "loss": 2.3247,
+      "step": 6120
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001425797373358349,
+      "loss": 2.6035,
+      "step": 6121
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001425703564727955,
+      "loss": 2.5888,
+      "step": 6122
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00014256097560975609,
+      "loss": 2.6023,
+      "step": 6123
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001425515947467167,
+      "loss": 2.6995,
+      "step": 6124
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001425422138836773,
+      "loss": 2.5747,
+      "step": 6125
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001425328330206379,
+      "loss": 2.4997,
+      "step": 6126
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001425234521575985,
+      "loss": 2.8528,
+      "step": 6127
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00014251407129455912,
+      "loss": 2.5578,
+      "step": 6128
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.0001425046904315197,
+      "loss": 2.7808,
+      "step": 6129
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00014249530956848033,
+      "loss": 2.1775,
+      "step": 6130
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001424859287054409,
+      "loss": 2.5409,
+      "step": 6131
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00014247654784240149,
+      "loss": 2.5363,
+      "step": 6132
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001424671669793621,
+      "loss": 2.3011,
+      "step": 6133
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001424577861163227,
+      "loss": 2.488,
+      "step": 6134
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001424484052532833,
+      "loss": 2.5191,
+      "step": 6135
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001424390243902439,
+      "loss": 2.6244,
+      "step": 6136
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00014242964352720452,
+      "loss": 2.4557,
+      "step": 6137
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001424202626641651,
+      "loss": 2.5641,
+      "step": 6138
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00014241088180112573,
+      "loss": 2.2994,
+      "step": 6139
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001424015009380863,
+      "loss": 2.6158,
+      "step": 6140
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001423921200750469,
+      "loss": 2.6962,
+      "step": 6141
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001423827392120075,
+      "loss": 2.3078,
+      "step": 6142
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00014237335834896812,
+      "loss": 2.4164,
+      "step": 6143
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001423639774859287,
+      "loss": 2.4559,
+      "step": 6144
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001423545966228893,
+      "loss": 2.6605,
+      "step": 6145
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00014234521575984992,
+      "loss": 2.8834,
+      "step": 6146
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001423358348968105,
+      "loss": 2.8453,
+      "step": 6147
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00014232645403377113,
+      "loss": 2.3986,
+      "step": 6148
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001423170731707317,
+      "loss": 2.3829,
+      "step": 6149
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.0001423076923076923,
+      "loss": 2.6511,
+      "step": 6150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001422983114446529,
+      "loss": 2.6069,
+      "step": 6151
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014228893058161352,
+      "loss": 2.5163,
+      "step": 6152
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001422795497185741,
+      "loss": 2.4905,
+      "step": 6153
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014227016885553473,
+      "loss": 2.5999,
+      "step": 6154
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014226078799249532,
+      "loss": 2.6444,
+      "step": 6155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014225140712945594,
+      "loss": 2.4382,
+      "step": 6156
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014224202626641653,
+      "loss": 2.8553,
+      "step": 6157
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001422326454033771,
+      "loss": 2.475,
+      "step": 6158
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001422232645403377,
+      "loss": 2.4843,
+      "step": 6159
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001422138836772983,
+      "loss": 2.6909,
+      "step": 6160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014220450281425892,
+      "loss": 2.7933,
+      "step": 6161
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001421951219512195,
+      "loss": 2.7961,
+      "step": 6162
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014218574108818013,
+      "loss": 2.2126,
+      "step": 6163
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014217636022514072,
+      "loss": 2.7481,
+      "step": 6164
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014216697936210134,
+      "loss": 2.4117,
+      "step": 6165
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014215759849906193,
+      "loss": 2.4404,
+      "step": 6166
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014214821763602252,
+      "loss": 2.3766,
+      "step": 6167
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001421388367729831,
+      "loss": 2.6153,
+      "step": 6168
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014212945590994373,
+      "loss": 2.6886,
+      "step": 6169
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014212007504690432,
+      "loss": 2.4473,
+      "step": 6170
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.0001421106941838649,
+      "loss": 2.3683,
+      "step": 6171
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 0.00014210131332082553,
+      "loss": 2.6929,
+      "step": 6172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014209193245778612,
+      "loss": 2.6452,
+      "step": 6173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014208255159474674,
+      "loss": 2.7155,
+      "step": 6174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014207317073170733,
+      "loss": 2.706,
+      "step": 6175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014206378986866792,
+      "loss": 2.5006,
+      "step": 6176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001420544090056285,
+      "loss": 2.5384,
+      "step": 6177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014204502814258913,
+      "loss": 2.8462,
+      "step": 6178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014203564727954972,
+      "loss": 2.576,
+      "step": 6179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014202626641651034,
+      "loss": 2.7392,
+      "step": 6180
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014201688555347093,
+      "loss": 2.5596,
+      "step": 6181
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014200750469043155,
+      "loss": 2.7238,
+      "step": 6182
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014199812382739214,
+      "loss": 2.5981,
+      "step": 6183
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014198874296435273,
+      "loss": 2.6306,
+      "step": 6184
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014197936210131332,
+      "loss": 2.4119,
+      "step": 6185
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.0001419699812382739,
+      "loss": 2.498,
+      "step": 6186
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014196060037523453,
+      "loss": 2.4915,
+      "step": 6187
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014195121951219512,
+      "loss": 2.4798,
+      "step": 6188
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014194183864915574,
+      "loss": 2.2504,
+      "step": 6189
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014193245778611633,
+      "loss": 2.6521,
+      "step": 6190
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014192307692307695,
+      "loss": 2.6097,
+      "step": 6191
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014191369606003754,
+      "loss": 2.5575,
+      "step": 6192
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00014190431519699813,
+      "loss": 2.7553,
+      "step": 6193
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014189493433395872,
+      "loss": 2.5713,
+      "step": 6194
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014188555347091934,
+      "loss": 2.7365,
+      "step": 6195
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014187617260787993,
+      "loss": 2.5533,
+      "step": 6196
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014186679174484052,
+      "loss": 2.7198,
+      "step": 6197
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014185741088180114,
+      "loss": 2.4126,
+      "step": 6198
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014184803001876173,
+      "loss": 2.808,
+      "step": 6199
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014183864915572235,
+      "loss": 2.6811,
+      "step": 6200
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014182926829268294,
+      "loss": 2.8438,
+      "step": 6201
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014181988742964353,
+      "loss": 2.846,
+      "step": 6202
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014181050656660412,
+      "loss": 2.5389,
+      "step": 6203
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014180112570356474,
+      "loss": 2.4034,
+      "step": 6204
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014179174484052533,
+      "loss": 2.7792,
+      "step": 6205
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014178236397748595,
+      "loss": 2.6639,
+      "step": 6206
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014177298311444654,
+      "loss": 2.6692,
+      "step": 6207
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014176360225140716,
+      "loss": 2.6287,
+      "step": 6208
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014175422138836775,
+      "loss": 2.7736,
+      "step": 6209
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014174484052532834,
+      "loss": 2.6131,
+      "step": 6210
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014173545966228893,
+      "loss": 2.5441,
+      "step": 6211
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014172607879924952,
+      "loss": 2.5205,
+      "step": 6212
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014171669793621014,
+      "loss": 2.3252,
+      "step": 6213
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.00014170731707317073,
+      "loss": 2.849,
+      "step": 6214
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014169793621013135,
+      "loss": 2.6327,
+      "step": 6215
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014168855534709194,
+      "loss": 2.5867,
+      "step": 6216
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014167917448405256,
+      "loss": 2.6333,
+      "step": 6217
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014166979362101315,
+      "loss": 2.6721,
+      "step": 6218
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014166041275797374,
+      "loss": 2.549,
+      "step": 6219
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014165103189493433,
+      "loss": 2.7973,
+      "step": 6220
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014164165103189495,
+      "loss": 2.6407,
+      "step": 6221
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014163227016885554,
+      "loss": 2.3751,
+      "step": 6222
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014162288930581613,
+      "loss": 2.7725,
+      "step": 6223
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014161350844277675,
+      "loss": 2.4766,
+      "step": 6224
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014160412757973734,
+      "loss": 2.4411,
+      "step": 6225
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014159474671669796,
+      "loss": 2.3234,
+      "step": 6226
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014158536585365855,
+      "loss": 2.2609,
+      "step": 6227
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014157598499061914,
+      "loss": 2.3641,
+      "step": 6228
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014156660412757973,
+      "loss": 2.3267,
+      "step": 6229
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014155722326454035,
+      "loss": 2.5697,
+      "step": 6230
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014154784240150094,
+      "loss": 2.5928,
+      "step": 6231
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014153846153846156,
+      "loss": 2.6599,
+      "step": 6232
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014152908067542215,
+      "loss": 2.7514,
+      "step": 6233
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014151969981238276,
+      "loss": 2.6518,
+      "step": 6234
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014151031894934336,
+      "loss": 2.5933,
+      "step": 6235
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00014150093808630395,
+      "loss": 2.606,
+      "step": 6236
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014149155722326454,
+      "loss": 2.6467,
+      "step": 6237
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014148217636022513,
+      "loss": 2.6456,
+      "step": 6238
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014147279549718575,
+      "loss": 3.023,
+      "step": 6239
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014146341463414634,
+      "loss": 2.8114,
+      "step": 6240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014145403377110696,
+      "loss": 2.8942,
+      "step": 6241
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014144465290806755,
+      "loss": 2.4005,
+      "step": 6242
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014143527204502816,
+      "loss": 2.7135,
+      "step": 6243
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014142589118198875,
+      "loss": 2.3404,
+      "step": 6244
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014141651031894935,
+      "loss": 2.6402,
+      "step": 6245
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014140712945590994,
+      "loss": 2.7222,
+      "step": 6246
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014139774859287055,
+      "loss": 2.3636,
+      "step": 6247
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014138836772983115,
+      "loss": 2.5939,
+      "step": 6248
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014137898686679174,
+      "loss": 2.7086,
+      "step": 6249
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014136960600375235,
+      "loss": 2.409,
+      "step": 6250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014136022514071295,
+      "loss": 2.4882,
+      "step": 6251
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014135084427767356,
+      "loss": 2.6163,
+      "step": 6252
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014134146341463415,
+      "loss": 2.6076,
+      "step": 6253
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014133208255159475,
+      "loss": 2.7338,
+      "step": 6254
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014132270168855534,
+      "loss": 2.5258,
+      "step": 6255
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014131332082551595,
+      "loss": 2.5133,
+      "step": 6256
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 0.00014130393996247655,
+      "loss": 2.7429,
+      "step": 6257
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014129455909943716,
+      "loss": 2.6922,
+      "step": 6258
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014128517823639775,
+      "loss": 2.4763,
+      "step": 6259
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014127579737335837,
+      "loss": 2.6154,
+      "step": 6260
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014126641651031896,
+      "loss": 2.656,
+      "step": 6261
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014125703564727955,
+      "loss": 2.4552,
+      "step": 6262
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014124765478424015,
+      "loss": 2.6457,
+      "step": 6263
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014123827392120074,
+      "loss": 2.3511,
+      "step": 6264
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014122889305816135,
+      "loss": 2.8584,
+      "step": 6265
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014121951219512195,
+      "loss": 2.3703,
+      "step": 6266
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014121013133208256,
+      "loss": 2.6368,
+      "step": 6267
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014120075046904315,
+      "loss": 2.6873,
+      "step": 6268
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014119136960600377,
+      "loss": 2.5483,
+      "step": 6269
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014118198874296436,
+      "loss": 2.4534,
+      "step": 6270
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014117260787992495,
+      "loss": 2.4896,
+      "step": 6271
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014116322701688555,
+      "loss": 2.8252,
+      "step": 6272
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014115384615384616,
+      "loss": 2.6351,
+      "step": 6273
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014114446529080675,
+      "loss": 2.3497,
+      "step": 6274
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014113508442776737,
+      "loss": 2.5716,
+      "step": 6275
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014112570356472796,
+      "loss": 2.6937,
+      "step": 6276
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014111632270168855,
+      "loss": 2.6142,
+      "step": 6277
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00014110694183864917,
+      "loss": 2.4519,
+      "step": 6278
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014109756097560976,
+      "loss": 2.6674,
+      "step": 6279
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014108818011257035,
+      "loss": 2.6986,
+      "step": 6280
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014107879924953095,
+      "loss": 2.6095,
+      "step": 6281
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014106941838649156,
+      "loss": 2.6754,
+      "step": 6282
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014106003752345215,
+      "loss": 2.3889,
+      "step": 6283
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014105065666041277,
+      "loss": 2.6947,
+      "step": 6284
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014104127579737336,
+      "loss": 2.4262,
+      "step": 6285
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014103189493433398,
+      "loss": 2.464,
+      "step": 6286
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014102251407129457,
+      "loss": 2.6417,
+      "step": 6287
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001410131332082552,
+      "loss": 2.7411,
+      "step": 6288
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014100375234521575,
+      "loss": 2.5039,
+      "step": 6289
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014099437148217634,
+      "loss": 2.4869,
+      "step": 6290
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014098499061913696,
+      "loss": 2.3392,
+      "step": 6291
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014097560975609755,
+      "loss": 2.3819,
+      "step": 6292
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014096622889305817,
+      "loss": 2.3049,
+      "step": 6293
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014095684803001876,
+      "loss": 2.8315,
+      "step": 6294
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014094746716697938,
+      "loss": 2.4722,
+      "step": 6295
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014093808630393997,
+      "loss": 2.7136,
+      "step": 6296
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001409287054409006,
+      "loss": 2.5711,
+      "step": 6297
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014091932457786115,
+      "loss": 2.6115,
+      "step": 6298
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014090994371482177,
+      "loss": 2.4716,
+      "step": 6299
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.00014090056285178236,
+      "loss": 2.6883,
+      "step": 6300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014089118198874298,
+      "loss": 2.686,
+      "step": 6301
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014088180112570357,
+      "loss": 2.3378,
+      "step": 6302
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014087242026266416,
+      "loss": 3.0822,
+      "step": 6303
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014086303939962478,
+      "loss": 2.8435,
+      "step": 6304
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014085365853658537,
+      "loss": 2.799,
+      "step": 6305
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.000140844277673546,
+      "loss": 2.4183,
+      "step": 6306
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014083489681050655,
+      "loss": 2.7141,
+      "step": 6307
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014082551594746717,
+      "loss": 2.5838,
+      "step": 6308
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014081613508442776,
+      "loss": 2.5344,
+      "step": 6309
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014080675422138838,
+      "loss": 2.2137,
+      "step": 6310
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014079737335834897,
+      "loss": 2.5979,
+      "step": 6311
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001407879924953096,
+      "loss": 2.7067,
+      "step": 6312
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014077861163227018,
+      "loss": 2.3794,
+      "step": 6313
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001407692307692308,
+      "loss": 2.5113,
+      "step": 6314
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.0001407598499061914,
+      "loss": 2.5394,
+      "step": 6315
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014075046904315195,
+      "loss": 2.7078,
+      "step": 6316
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014074108818011257,
+      "loss": 2.6411,
+      "step": 6317
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014073170731707316,
+      "loss": 2.7923,
+      "step": 6318
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014072232645403378,
+      "loss": 3.0156,
+      "step": 6319
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00014071294559099437,
+      "loss": 2.5182,
+      "step": 6320
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.000140703564727955,
+      "loss": 2.6627,
+      "step": 6321
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00014069418386491558,
+      "loss": 2.7082,
+      "step": 6322
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001406848030018762,
+      "loss": 2.4298,
+      "step": 6323
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001406754221388368,
+      "loss": 2.2967,
+      "step": 6324
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00014066604127579738,
+      "loss": 2.6536,
+      "step": 6325
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00014065666041275797,
+      "loss": 2.5114,
+      "step": 6326
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001406472795497186,
+      "loss": 2.5294,
+      "step": 6327
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00014063789868667918,
+      "loss": 2.3742,
+      "step": 6328
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00014062851782363977,
+      "loss": 2.6824,
+      "step": 6329
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001406191369606004,
+      "loss": 2.535,
+      "step": 6330
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00014060975609756098,
+      "loss": 2.7618,
+      "step": 6331
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001406003752345216,
+      "loss": 2.6648,
+      "step": 6332
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001405909943714822,
+      "loss": 2.6251,
+      "step": 6333
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00014058161350844278,
+      "loss": 2.4772,
+      "step": 6334
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00014057223264540337,
+      "loss": 2.6973,
+      "step": 6335
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000140562851782364,
+      "loss": 2.653,
+      "step": 6336
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00014055347091932458,
+      "loss": 2.3384,
+      "step": 6337
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001405440900562852,
+      "loss": 2.3624,
+      "step": 6338
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001405347091932458,
+      "loss": 2.6365,
+      "step": 6339
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001405253283302064,
+      "loss": 2.863,
+      "step": 6340
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.000140515947467167,
+      "loss": 2.6588,
+      "step": 6341
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.0001405065666041276,
+      "loss": 2.4626,
+      "step": 6342
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00014049718574108818,
+      "loss": 2.4242,
+      "step": 6343
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00014048780487804877,
+      "loss": 2.4552,
+      "step": 6344
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001404784240150094,
+      "loss": 2.6746,
+      "step": 6345
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00014046904315196998,
+      "loss": 2.4792,
+      "step": 6346
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001404596622889306,
+      "loss": 2.9992,
+      "step": 6347
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001404502814258912,
+      "loss": 2.5182,
+      "step": 6348
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001404409005628518,
+      "loss": 2.7116,
+      "step": 6349
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001404315196998124,
+      "loss": 2.7315,
+      "step": 6350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000140422138836773,
+      "loss": 2.443,
+      "step": 6351
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00014041275797373358,
+      "loss": 2.5413,
+      "step": 6352
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001404033771106942,
+      "loss": 2.3752,
+      "step": 6353
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001403939962476548,
+      "loss": 2.378,
+      "step": 6354
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00014038461538461538,
+      "loss": 2.6518,
+      "step": 6355
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.000140375234521576,
+      "loss": 2.3774,
+      "step": 6356
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001403658536585366,
+      "loss": 2.6101,
+      "step": 6357
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001403564727954972,
+      "loss": 2.5694,
+      "step": 6358
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001403470919324578,
+      "loss": 2.7927,
+      "step": 6359
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001403377110694184,
+      "loss": 2.6077,
+      "step": 6360
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.00014032833020637898,
+      "loss": 2.5816,
+      "step": 6361
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001403189493433396,
+      "loss": 2.2901,
+      "step": 6362
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001403095684803002,
+      "loss": 2.7155,
+      "step": 6363
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 0.0001403001876172608,
+      "loss": 2.6723,
+      "step": 6364
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001402908067542214,
+      "loss": 2.3464,
+      "step": 6365
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00014028142589118202,
+      "loss": 2.4609,
+      "step": 6366
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001402720450281426,
+      "loss": 2.7934,
+      "step": 6367
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001402626641651032,
+      "loss": 2.3524,
+      "step": 6368
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001402532833020638,
+      "loss": 2.5498,
+      "step": 6369
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00014024390243902438,
+      "loss": 2.7284,
+      "step": 6370
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000140234521575985,
+      "loss": 2.4702,
+      "step": 6371
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001402251407129456,
+      "loss": 2.7764,
+      "step": 6372
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001402157598499062,
+      "loss": 2.501,
+      "step": 6373
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001402063789868668,
+      "loss": 2.4141,
+      "step": 6374
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00014019699812382742,
+      "loss": 2.8244,
+      "step": 6375
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000140187617260788,
+      "loss": 2.4616,
+      "step": 6376
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001401782363977486,
+      "loss": 2.654,
+      "step": 6377
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001401688555347092,
+      "loss": 2.535,
+      "step": 6378
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001401594746716698,
+      "loss": 2.5045,
+      "step": 6379
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001401500938086304,
+      "loss": 2.5282,
+      "step": 6380
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.000140140712945591,
+      "loss": 2.4829,
+      "step": 6381
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001401313320825516,
+      "loss": 2.4904,
+      "step": 6382
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001401219512195122,
+      "loss": 2.8859,
+      "step": 6383
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00014011257035647282,
+      "loss": 2.9031,
+      "step": 6384
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.0001401031894934334,
+      "loss": 2.8359,
+      "step": 6385
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000140093808630394,
+      "loss": 2.5137,
+      "step": 6386
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001400844277673546,
+      "loss": 2.7034,
+      "step": 6387
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001400750469043152,
+      "loss": 2.6645,
+      "step": 6388
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001400656660412758,
+      "loss": 2.5138,
+      "step": 6389
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00014005628517823641,
+      "loss": 2.6484,
+      "step": 6390
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000140046904315197,
+      "loss": 2.5722,
+      "step": 6391
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00014003752345215762,
+      "loss": 2.5851,
+      "step": 6392
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00014002814258911821,
+      "loss": 2.3055,
+      "step": 6393
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001400187617260788,
+      "loss": 2.6269,
+      "step": 6394
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001400093808630394,
+      "loss": 2.4047,
+      "step": 6395
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00014,
+      "loss": 2.6272,
+      "step": 6396
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001399906191369606,
+      "loss": 2.6302,
+      "step": 6397
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001399812382739212,
+      "loss": 2.5945,
+      "step": 6398
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00013997185741088181,
+      "loss": 2.3986,
+      "step": 6399
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001399624765478424,
+      "loss": 2.4308,
+      "step": 6400
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00013995309568480302,
+      "loss": 2.5508,
+      "step": 6401
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00013994371482176361,
+      "loss": 2.6088,
+      "step": 6402
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001399343339587242,
+      "loss": 2.594,
+      "step": 6403
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0001399249530956848,
+      "loss": 2.4455,
+      "step": 6404
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00013991557223264541,
+      "loss": 2.7646,
+      "step": 6405
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.000139906191369606,
+      "loss": 2.3595,
+      "step": 6406
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001398968105065666,
+      "loss": 2.6903,
+      "step": 6407
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013988742964352721,
+      "loss": 2.2858,
+      "step": 6408
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001398780487804878,
+      "loss": 2.5841,
+      "step": 6409
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013986866791744842,
+      "loss": 2.4841,
+      "step": 6410
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013985928705440901,
+      "loss": 2.3124,
+      "step": 6411
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001398499061913696,
+      "loss": 2.2428,
+      "step": 6412
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001398405253283302,
+      "loss": 2.7809,
+      "step": 6413
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013983114446529081,
+      "loss": 2.4807,
+      "step": 6414
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001398217636022514,
+      "loss": 2.5073,
+      "step": 6415
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013981238273921202,
+      "loss": 2.5869,
+      "step": 6416
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013980300187617261,
+      "loss": 2.6681,
+      "step": 6417
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013979362101313323,
+      "loss": 2.6331,
+      "step": 6418
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013978424015009382,
+      "loss": 2.5252,
+      "step": 6419
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013977485928705441,
+      "loss": 2.5987,
+      "step": 6420
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.000139765478424015,
+      "loss": 2.478,
+      "step": 6421
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001397560975609756,
+      "loss": 2.4844,
+      "step": 6422
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013974671669793621,
+      "loss": 2.56,
+      "step": 6423
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.0001397373358348968,
+      "loss": 2.2012,
+      "step": 6424
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013972795497185742,
+      "loss": 2.7693,
+      "step": 6425
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013971857410881801,
+      "loss": 2.844,
+      "step": 6426
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00013970919324577863,
+      "loss": 2.7316,
+      "step": 6427
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013969981238273922,
+      "loss": 2.508,
+      "step": 6428
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013969043151969981,
+      "loss": 2.4117,
+      "step": 6429
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001396810506566604,
+      "loss": 2.6541,
+      "step": 6430
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013967166979362102,
+      "loss": 2.7057,
+      "step": 6431
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013966228893058161,
+      "loss": 2.6263,
+      "step": 6432
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001396529080675422,
+      "loss": 2.6034,
+      "step": 6433
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013964352720450282,
+      "loss": 2.3467,
+      "step": 6434
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013963414634146341,
+      "loss": 2.4581,
+      "step": 6435
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013962476547842403,
+      "loss": 2.3917,
+      "step": 6436
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013961538461538462,
+      "loss": 2.7264,
+      "step": 6437
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013960600375234521,
+      "loss": 2.6304,
+      "step": 6438
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001395966228893058,
+      "loss": 2.5243,
+      "step": 6439
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013958724202626642,
+      "loss": 2.3857,
+      "step": 6440
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013957786116322701,
+      "loss": 2.7224,
+      "step": 6441
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013956848030018763,
+      "loss": 2.6802,
+      "step": 6442
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013955909943714822,
+      "loss": 2.6714,
+      "step": 6443
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013954971857410884,
+      "loss": 2.518,
+      "step": 6444
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013954033771106943,
+      "loss": 2.5047,
+      "step": 6445
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013953095684803002,
+      "loss": 2.3372,
+      "step": 6446
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013952157598499061,
+      "loss": 2.5255,
+      "step": 6447
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.0001395121951219512,
+      "loss": 2.4112,
+      "step": 6448
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 0.00013950281425891182,
+      "loss": 2.4026,
+      "step": 6449
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013949343339587241,
+      "loss": 2.6067,
+      "step": 6450
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013948405253283303,
+      "loss": 2.3271,
+      "step": 6451
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013947467166979362,
+      "loss": 2.5859,
+      "step": 6452
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013946529080675424,
+      "loss": 2.6217,
+      "step": 6453
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013945590994371483,
+      "loss": 2.5854,
+      "step": 6454
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013944652908067545,
+      "loss": 2.3217,
+      "step": 6455
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.000139437148217636,
+      "loss": 2.4936,
+      "step": 6456
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013942776735459663,
+      "loss": 2.7354,
+      "step": 6457
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013941838649155722,
+      "loss": 2.4964,
+      "step": 6458
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013940900562851784,
+      "loss": 2.4552,
+      "step": 6459
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013939962476547843,
+      "loss": 2.6459,
+      "step": 6460
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013939024390243902,
+      "loss": 2.4547,
+      "step": 6461
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013938086303939964,
+      "loss": 2.424,
+      "step": 6462
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013937148217636023,
+      "loss": 2.6427,
+      "step": 6463
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013936210131332085,
+      "loss": 2.4043,
+      "step": 6464
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.0001393527204502814,
+      "loss": 2.3706,
+      "step": 6465
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013934333958724203,
+      "loss": 2.4891,
+      "step": 6466
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013933395872420262,
+      "loss": 2.4163,
+      "step": 6467
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013932457786116324,
+      "loss": 2.566,
+      "step": 6468
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013931519699812383,
+      "loss": 2.5489,
+      "step": 6469
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00013930581613508445,
+      "loss": 2.3724,
+      "step": 6470
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013929643527204504,
+      "loss": 2.6188,
+      "step": 6471
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013928705440900566,
+      "loss": 2.4022,
+      "step": 6472
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013927767354596625,
+      "loss": 2.1815,
+      "step": 6473
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001392682926829268,
+      "loss": 2.4024,
+      "step": 6474
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013925891181988743,
+      "loss": 2.4677,
+      "step": 6475
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013924953095684802,
+      "loss": 2.5681,
+      "step": 6476
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013924015009380864,
+      "loss": 2.5115,
+      "step": 6477
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013923076923076923,
+      "loss": 2.9312,
+      "step": 6478
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013922138836772985,
+      "loss": 2.349,
+      "step": 6479
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013921200750469044,
+      "loss": 2.5979,
+      "step": 6480
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013920262664165106,
+      "loss": 2.5521,
+      "step": 6481
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013919324577861165,
+      "loss": 2.494,
+      "step": 6482
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013918386491557224,
+      "loss": 2.4431,
+      "step": 6483
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013917448405253283,
+      "loss": 2.3073,
+      "step": 6484
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013916510318949345,
+      "loss": 2.3507,
+      "step": 6485
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013915572232645404,
+      "loss": 2.2591,
+      "step": 6486
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013914634146341463,
+      "loss": 2.1336,
+      "step": 6487
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013913696060037525,
+      "loss": 2.6186,
+      "step": 6488
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013912757973733584,
+      "loss": 2.5518,
+      "step": 6489
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013911819887429646,
+      "loss": 2.5956,
+      "step": 6490
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.00013910881801125705,
+      "loss": 2.538,
+      "step": 6491
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013909943714821764,
+      "loss": 2.1058,
+      "step": 6492
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013909005628517823,
+      "loss": 2.5858,
+      "step": 6493
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013908067542213885,
+      "loss": 2.5004,
+      "step": 6494
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013907129455909944,
+      "loss": 2.491,
+      "step": 6495
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013906191369606006,
+      "loss": 2.8167,
+      "step": 6496
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013905253283302065,
+      "loss": 2.4086,
+      "step": 6497
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013904315196998127,
+      "loss": 2.4316,
+      "step": 6498
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013903377110694186,
+      "loss": 2.5249,
+      "step": 6499
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013902439024390245,
+      "loss": 2.7419,
+      "step": 6500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013901500938086304,
+      "loss": 2.2133,
+      "step": 6501
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013900562851782363,
+      "loss": 2.4469,
+      "step": 6502
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013899624765478425,
+      "loss": 2.754,
+      "step": 6503
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013898686679174484,
+      "loss": 2.4195,
+      "step": 6504
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013897748592870546,
+      "loss": 2.352,
+      "step": 6505
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013896810506566605,
+      "loss": 2.844,
+      "step": 6506
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013895872420262667,
+      "loss": 3.0284,
+      "step": 6507
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013894934333958726,
+      "loss": 2.5308,
+      "step": 6508
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013893996247654785,
+      "loss": 2.7108,
+      "step": 6509
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013893058161350844,
+      "loss": 2.6491,
+      "step": 6510
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013892120075046906,
+      "loss": 2.2851,
+      "step": 6511
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013891181988742965,
+      "loss": 2.4275,
+      "step": 6512
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.00013890243902439024,
+      "loss": 2.6197,
+      "step": 6513
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013889305816135086,
+      "loss": 2.403,
+      "step": 6514
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013888367729831145,
+      "loss": 2.689,
+      "step": 6515
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013887429643527207,
+      "loss": 2.5604,
+      "step": 6516
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013886491557223266,
+      "loss": 2.7616,
+      "step": 6517
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013885553470919325,
+      "loss": 2.7057,
+      "step": 6518
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013884615384615384,
+      "loss": 2.3697,
+      "step": 6519
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013883677298311446,
+      "loss": 2.307,
+      "step": 6520
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013882739212007505,
+      "loss": 2.6708,
+      "step": 6521
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013881801125703567,
+      "loss": 2.4208,
+      "step": 6522
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013880863039399626,
+      "loss": 2.68,
+      "step": 6523
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013879924953095688,
+      "loss": 2.7989,
+      "step": 6524
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013878986866791747,
+      "loss": 2.6131,
+      "step": 6525
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013878048780487806,
+      "loss": 2.8557,
+      "step": 6526
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013877110694183865,
+      "loss": 2.2217,
+      "step": 6527
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013876172607879924,
+      "loss": 2.5677,
+      "step": 6528
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013875234521575986,
+      "loss": 2.5996,
+      "step": 6529
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013874296435272045,
+      "loss": 2.5153,
+      "step": 6530
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013873358348968107,
+      "loss": 2.2348,
+      "step": 6531
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013872420262664166,
+      "loss": 2.4585,
+      "step": 6532
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013871482176360228,
+      "loss": 2.3759,
+      "step": 6533
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00013870544090056287,
+      "loss": 2.615,
+      "step": 6534
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013869606003752346,
+      "loss": 2.8648,
+      "step": 6535
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013868667917448405,
+      "loss": 2.44,
+      "step": 6536
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013867729831144467,
+      "loss": 2.6449,
+      "step": 6537
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013866791744840526,
+      "loss": 2.5748,
+      "step": 6538
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013865853658536585,
+      "loss": 2.4213,
+      "step": 6539
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013864915572232647,
+      "loss": 2.3782,
+      "step": 6540
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013863977485928706,
+      "loss": 2.6397,
+      "step": 6541
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013863039399624767,
+      "loss": 2.4724,
+      "step": 6542
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013862101313320827,
+      "loss": 2.7069,
+      "step": 6543
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013861163227016886,
+      "loss": 2.2634,
+      "step": 6544
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013860225140712945,
+      "loss": 2.6777,
+      "step": 6545
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013859287054409007,
+      "loss": 2.853,
+      "step": 6546
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013858348968105066,
+      "loss": 2.4716,
+      "step": 6547
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013857410881801127,
+      "loss": 2.2328,
+      "step": 6548
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013856472795497187,
+      "loss": 2.4966,
+      "step": 6549
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013855534709193248,
+      "loss": 2.3196,
+      "step": 6550
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013854596622889307,
+      "loss": 2.4324,
+      "step": 6551
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013853658536585367,
+      "loss": 2.5173,
+      "step": 6552
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013852720450281426,
+      "loss": 2.1729,
+      "step": 6553
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013851782363977485,
+      "loss": 2.6911,
+      "step": 6554
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 0.00013850844277673547,
+      "loss": 2.7894,
+      "step": 6555
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013849906191369606,
+      "loss": 2.2538,
+      "step": 6556
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013848968105065667,
+      "loss": 2.5564,
+      "step": 6557
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013848030018761727,
+      "loss": 2.4664,
+      "step": 6558
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013847091932457788,
+      "loss": 2.4213,
+      "step": 6559
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 2.4114,
+      "step": 6560
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013845215759849907,
+      "loss": 2.5601,
+      "step": 6561
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013844277673545966,
+      "loss": 2.6586,
+      "step": 6562
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013843339587242027,
+      "loss": 2.4109,
+      "step": 6563
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013842401500938087,
+      "loss": 2.4158,
+      "step": 6564
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013841463414634146,
+      "loss": 2.7639,
+      "step": 6565
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013840525328330207,
+      "loss": 2.5053,
+      "step": 6566
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013839587242026267,
+      "loss": 2.2412,
+      "step": 6567
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013838649155722328,
+      "loss": 2.5049,
+      "step": 6568
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013837711069418387,
+      "loss": 2.4116,
+      "step": 6569
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013836772983114447,
+      "loss": 2.3457,
+      "step": 6570
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013835834896810506,
+      "loss": 2.3665,
+      "step": 6571
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013834896810506567,
+      "loss": 2.4078,
+      "step": 6572
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013833958724202627,
+      "loss": 2.8561,
+      "step": 6573
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013833020637898688,
+      "loss": 2.5323,
+      "step": 6574
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013832082551594747,
+      "loss": 2.3526,
+      "step": 6575
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0001383114446529081,
+      "loss": 2.4597,
+      "step": 6576
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00013830206378986868,
+      "loss": 2.5318,
+      "step": 6577
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013829268292682927,
+      "loss": 2.7487,
+      "step": 6578
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013828330206378986,
+      "loss": 2.5331,
+      "step": 6579
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013827392120075046,
+      "loss": 2.4292,
+      "step": 6580
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013826454033771107,
+      "loss": 2.3767,
+      "step": 6581
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013825515947467166,
+      "loss": 2.4681,
+      "step": 6582
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013824577861163228,
+      "loss": 2.6433,
+      "step": 6583
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013823639774859287,
+      "loss": 2.4378,
+      "step": 6584
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0001382270168855535,
+      "loss": 2.4561,
+      "step": 6585
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013821763602251408,
+      "loss": 2.7983,
+      "step": 6586
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013820825515947467,
+      "loss": 2.5691,
+      "step": 6587
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013819887429643526,
+      "loss": 2.2574,
+      "step": 6588
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013818949343339588,
+      "loss": 2.4323,
+      "step": 6589
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013818011257035647,
+      "loss": 2.7373,
+      "step": 6590
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013817073170731706,
+      "loss": 2.5051,
+      "step": 6591
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013816135084427768,
+      "loss": 2.4136,
+      "step": 6592
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013815196998123827,
+      "loss": 2.7367,
+      "step": 6593
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.0001381425891181989,
+      "loss": 2.596,
+      "step": 6594
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013813320825515948,
+      "loss": 2.2977,
+      "step": 6595
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013812382739212007,
+      "loss": 2.6154,
+      "step": 6596
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013811444652908066,
+      "loss": 2.4262,
+      "step": 6597
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00013810506566604128,
+      "loss": 2.6871,
+      "step": 6598
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013809568480300187,
+      "loss": 2.467,
+      "step": 6599
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0001380863039399625,
+      "loss": 2.4081,
+      "step": 6600
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013807692307692308,
+      "loss": 2.4485,
+      "step": 6601
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0001380675422138837,
+      "loss": 2.3933,
+      "step": 6602
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0001380581613508443,
+      "loss": 2.6111,
+      "step": 6603
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013804878048780488,
+      "loss": 2.8681,
+      "step": 6604
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013803939962476547,
+      "loss": 2.4301,
+      "step": 6605
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013803001876172606,
+      "loss": 2.3767,
+      "step": 6606
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013802063789868668,
+      "loss": 2.3711,
+      "step": 6607
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013801125703564727,
+      "loss": 2.4749,
+      "step": 6608
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0001380018761726079,
+      "loss": 2.5046,
+      "step": 6609
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013799249530956848,
+      "loss": 2.2658,
+      "step": 6610
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0001379831144465291,
+      "loss": 2.5278,
+      "step": 6611
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0001379737335834897,
+      "loss": 2.4259,
+      "step": 6612
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0001379643527204503,
+      "loss": 2.3943,
+      "step": 6613
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013795497185741087,
+      "loss": 2.6,
+      "step": 6614
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0001379455909943715,
+      "loss": 2.5922,
+      "step": 6615
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013793621013133208,
+      "loss": 2.7054,
+      "step": 6616
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013792682926829267,
+      "loss": 2.6309,
+      "step": 6617
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.0001379174484052533,
+      "loss": 2.2616,
+      "step": 6618
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00013790806754221388,
+      "loss": 2.451,
+      "step": 6619
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001378986866791745,
+      "loss": 2.6466,
+      "step": 6620
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001378893058161351,
+      "loss": 2.5598,
+      "step": 6621
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001378799249530957,
+      "loss": 2.6345,
+      "step": 6622
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00013787054409005627,
+      "loss": 2.6378,
+      "step": 6623
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001378611632270169,
+      "loss": 2.232,
+      "step": 6624
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00013785178236397748,
+      "loss": 2.287,
+      "step": 6625
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001378424015009381,
+      "loss": 2.4508,
+      "step": 6626
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001378330206378987,
+      "loss": 2.5009,
+      "step": 6627
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001378236397748593,
+      "loss": 2.6045,
+      "step": 6628
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001378142589118199,
+      "loss": 2.5665,
+      "step": 6629
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001378048780487805,
+      "loss": 2.5557,
+      "step": 6630
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001377954971857411,
+      "loss": 2.5742,
+      "step": 6631
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00013778611632270167,
+      "loss": 2.5367,
+      "step": 6632
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001377767354596623,
+      "loss": 2.5054,
+      "step": 6633
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00013776735459662288,
+      "loss": 2.7544,
+      "step": 6634
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001377579737335835,
+      "loss": 2.4387,
+      "step": 6635
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001377485928705441,
+      "loss": 2.6236,
+      "step": 6636
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001377392120075047,
+      "loss": 2.4806,
+      "step": 6637
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001377298311444653,
+      "loss": 2.7226,
+      "step": 6638
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.00013772045028142592,
+      "loss": 2.4968,
+      "step": 6639
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001377110694183865,
+      "loss": 2.4129,
+      "step": 6640
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 0.0001377016885553471,
+      "loss": 2.7956,
+      "step": 6641
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001376923076923077,
+      "loss": 2.297,
+      "step": 6642
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00013768292682926828,
+      "loss": 2.4237,
+      "step": 6643
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001376735459662289,
+      "loss": 2.8043,
+      "step": 6644
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001376641651031895,
+      "loss": 2.464,
+      "step": 6645
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001376547842401501,
+      "loss": 2.466,
+      "step": 6646
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001376454033771107,
+      "loss": 2.4651,
+      "step": 6647
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00013763602251407132,
+      "loss": 2.5422,
+      "step": 6648
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001376266416510319,
+      "loss": 2.5311,
+      "step": 6649
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001376172607879925,
+      "loss": 2.5696,
+      "step": 6650
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001376078799249531,
+      "loss": 2.583,
+      "step": 6651
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001375984990619137,
+      "loss": 2.4848,
+      "step": 6652
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001375891181988743,
+      "loss": 2.4764,
+      "step": 6653
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00013757973733583492,
+      "loss": 2.6475,
+      "step": 6654
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001375703564727955,
+      "loss": 2.3857,
+      "step": 6655
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001375609756097561,
+      "loss": 2.1801,
+      "step": 6656
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00013755159474671672,
+      "loss": 2.8109,
+      "step": 6657
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001375422138836773,
+      "loss": 2.4981,
+      "step": 6658
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001375328330206379,
+      "loss": 2.7454,
+      "step": 6659
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001375234521575985,
+      "loss": 2.7911,
+      "step": 6660
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001375140712945591,
+      "loss": 2.5437,
+      "step": 6661
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.0001375046904315197,
+      "loss": 2.6334,
+      "step": 6662
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013749530956848032,
+      "loss": 2.8945,
+      "step": 6663
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001374859287054409,
+      "loss": 2.4042,
+      "step": 6664
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013747654784240153,
+      "loss": 2.406,
+      "step": 6665
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013746716697936212,
+      "loss": 2.4481,
+      "step": 6666
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001374577861163227,
+      "loss": 2.4062,
+      "step": 6667
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001374484052532833,
+      "loss": 2.6334,
+      "step": 6668
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013743902439024392,
+      "loss": 2.7356,
+      "step": 6669
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001374296435272045,
+      "loss": 2.3278,
+      "step": 6670
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001374202626641651,
+      "loss": 2.5061,
+      "step": 6671
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013741088180112572,
+      "loss": 2.5003,
+      "step": 6672
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001374015009380863,
+      "loss": 2.6733,
+      "step": 6673
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013739212007504693,
+      "loss": 2.4429,
+      "step": 6674
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013738273921200752,
+      "loss": 2.4823,
+      "step": 6675
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001373733583489681,
+      "loss": 2.7155,
+      "step": 6676
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001373639774859287,
+      "loss": 2.2816,
+      "step": 6677
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013735459662288932,
+      "loss": 2.6475,
+      "step": 6678
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001373452157598499,
+      "loss": 2.3634,
+      "step": 6679
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013733583489681053,
+      "loss": 2.6482,
+      "step": 6680
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013732645403377112,
+      "loss": 2.779,
+      "step": 6681
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013731707317073173,
+      "loss": 2.5237,
+      "step": 6682
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.00013730769230769233,
+      "loss": 2.4537,
+      "step": 6683
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013729831144465292,
+      "loss": 2.5257,
+      "step": 6684
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0001372889305816135,
+      "loss": 2.566,
+      "step": 6685
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0001372795497185741,
+      "loss": 2.3691,
+      "step": 6686
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013727016885553472,
+      "loss": 2.7755,
+      "step": 6687
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0001372607879924953,
+      "loss": 2.6378,
+      "step": 6688
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013725140712945593,
+      "loss": 2.7891,
+      "step": 6689
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013724202626641652,
+      "loss": 2.3336,
+      "step": 6690
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013723264540337713,
+      "loss": 2.522,
+      "step": 6691
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013722326454033773,
+      "loss": 2.586,
+      "step": 6692
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013721388367729832,
+      "loss": 2.477,
+      "step": 6693
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0001372045028142589,
+      "loss": 2.756,
+      "step": 6694
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013719512195121953,
+      "loss": 2.3947,
+      "step": 6695
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013718574108818012,
+      "loss": 2.3102,
+      "step": 6696
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0001371763602251407,
+      "loss": 2.7275,
+      "step": 6697
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013716697936210133,
+      "loss": 2.3834,
+      "step": 6698
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013715759849906192,
+      "loss": 2.5869,
+      "step": 6699
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013714821763602253,
+      "loss": 2.5156,
+      "step": 6700
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013713883677298313,
+      "loss": 2.4342,
+      "step": 6701
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013712945590994372,
+      "loss": 2.6478,
+      "step": 6702
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.0001371200750469043,
+      "loss": 2.3438,
+      "step": 6703
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013711069418386493,
+      "loss": 2.5684,
+      "step": 6704
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00013710131332082552,
+      "loss": 2.5252,
+      "step": 6705
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013709193245778613,
+      "loss": 2.813,
+      "step": 6706
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013708255159474673,
+      "loss": 2.4076,
+      "step": 6707
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013707317073170734,
+      "loss": 2.6911,
+      "step": 6708
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013706378986866793,
+      "loss": 2.5511,
+      "step": 6709
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013705440900562853,
+      "loss": 2.5538,
+      "step": 6710
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013704502814258912,
+      "loss": 2.4288,
+      "step": 6711
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.0001370356472795497,
+      "loss": 2.5511,
+      "step": 6712
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013702626641651033,
+      "loss": 2.255,
+      "step": 6713
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013701688555347092,
+      "loss": 2.3315,
+      "step": 6714
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013700750469043153,
+      "loss": 2.4607,
+      "step": 6715
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013699812382739213,
+      "loss": 2.4876,
+      "step": 6716
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013698874296435274,
+      "loss": 2.731,
+      "step": 6717
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013697936210131333,
+      "loss": 2.3907,
+      "step": 6718
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013696998123827393,
+      "loss": 2.4275,
+      "step": 6719
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013696060037523452,
+      "loss": 2.5716,
+      "step": 6720
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013695121951219513,
+      "loss": 2.7676,
+      "step": 6721
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013694183864915573,
+      "loss": 2.1109,
+      "step": 6722
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013693245778611632,
+      "loss": 2.5213,
+      "step": 6723
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013692307692307693,
+      "loss": 2.6164,
+      "step": 6724
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013691369606003752,
+      "loss": 2.2606,
+      "step": 6725
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00013690431519699814,
+      "loss": 2.5495,
+      "step": 6726
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013689493433395873,
+      "loss": 2.3478,
+      "step": 6727
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013688555347091932,
+      "loss": 2.3893,
+      "step": 6728
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013687617260787992,
+      "loss": 2.8162,
+      "step": 6729
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013686679174484053,
+      "loss": 2.6803,
+      "step": 6730
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013685741088180112,
+      "loss": 2.7273,
+      "step": 6731
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013684803001876174,
+      "loss": 2.5584,
+      "step": 6732
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013683864915572233,
+      "loss": 2.4814,
+      "step": 6733
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013682926829268295,
+      "loss": 2.6983,
+      "step": 6734
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013681988742964354,
+      "loss": 2.5626,
+      "step": 6735
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013681050656660413,
+      "loss": 2.4375,
+      "step": 6736
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013680112570356472,
+      "loss": 2.4808,
+      "step": 6737
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013679174484052532,
+      "loss": 2.7004,
+      "step": 6738
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013678236397748593,
+      "loss": 2.4362,
+      "step": 6739
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013677298311444652,
+      "loss": 2.4282,
+      "step": 6740
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013676360225140714,
+      "loss": 2.5202,
+      "step": 6741
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013675422138836773,
+      "loss": 2.4903,
+      "step": 6742
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013674484052532835,
+      "loss": 2.3885,
+      "step": 6743
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013673545966228894,
+      "loss": 2.3621,
+      "step": 6744
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013672607879924953,
+      "loss": 2.5647,
+      "step": 6745
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013671669793621012,
+      "loss": 2.4954,
+      "step": 6746
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 0.00013670731707317074,
+      "loss": 2.534,
+      "step": 6747
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013669793621013133,
+      "loss": 2.5594,
+      "step": 6748
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013668855534709192,
+      "loss": 2.4563,
+      "step": 6749
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013667917448405254,
+      "loss": 2.4981,
+      "step": 6750
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013666979362101313,
+      "loss": 2.6297,
+      "step": 6751
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013666041275797375,
+      "loss": 2.3181,
+      "step": 6752
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013665103189493434,
+      "loss": 2.452,
+      "step": 6753
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013664165103189493,
+      "loss": 2.4783,
+      "step": 6754
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013663227016885552,
+      "loss": 2.3676,
+      "step": 6755
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013662288930581614,
+      "loss": 2.6375,
+      "step": 6756
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013661350844277673,
+      "loss": 2.5999,
+      "step": 6757
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013660412757973735,
+      "loss": 2.561,
+      "step": 6758
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013659474671669794,
+      "loss": 2.6068,
+      "step": 6759
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013658536585365856,
+      "loss": 2.8983,
+      "step": 6760
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013657598499061915,
+      "loss": 2.4858,
+      "step": 6761
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013656660412757974,
+      "loss": 2.7819,
+      "step": 6762
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013655722326454033,
+      "loss": 2.7874,
+      "step": 6763
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013654784240150092,
+      "loss": 2.8097,
+      "step": 6764
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013653846153846154,
+      "loss": 2.4224,
+      "step": 6765
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013652908067542213,
+      "loss": 2.6589,
+      "step": 6766
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013651969981238275,
+      "loss": 2.4888,
+      "step": 6767
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013651031894934334,
+      "loss": 2.5692,
+      "step": 6768
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.00013650093808630396,
+      "loss": 2.4961,
+      "step": 6769
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013649155722326455,
+      "loss": 2.3456,
+      "step": 6770
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013648217636022517,
+      "loss": 2.7111,
+      "step": 6771
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013647279549718573,
+      "loss": 2.4307,
+      "step": 6772
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013646341463414635,
+      "loss": 2.3301,
+      "step": 6773
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013645403377110694,
+      "loss": 2.3991,
+      "step": 6774
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013644465290806753,
+      "loss": 2.6158,
+      "step": 6775
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013643527204502815,
+      "loss": 2.7105,
+      "step": 6776
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013642589118198874,
+      "loss": 2.4997,
+      "step": 6777
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013641651031894936,
+      "loss": 2.7135,
+      "step": 6778
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013640712945590995,
+      "loss": 2.4022,
+      "step": 6779
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013639774859287057,
+      "loss": 2.3595,
+      "step": 6780
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013638836772983113,
+      "loss": 2.3688,
+      "step": 6781
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013637898686679175,
+      "loss": 2.4961,
+      "step": 6782
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013636960600375234,
+      "loss": 2.2394,
+      "step": 6783
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013636022514071296,
+      "loss": 2.8853,
+      "step": 6784
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013635084427767355,
+      "loss": 2.4827,
+      "step": 6785
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013634146341463417,
+      "loss": 2.7445,
+      "step": 6786
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013633208255159476,
+      "loss": 2.2774,
+      "step": 6787
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013632270168855535,
+      "loss": 2.521,
+      "step": 6788
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013631332082551597,
+      "loss": 2.596,
+      "step": 6789
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00013630393996247653,
+      "loss": 2.5183,
+      "step": 6790
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013629455909943715,
+      "loss": 2.3691,
+      "step": 6791
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013628517823639774,
+      "loss": 2.5166,
+      "step": 6792
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013627579737335836,
+      "loss": 2.3177,
+      "step": 6793
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013626641651031895,
+      "loss": 2.4903,
+      "step": 6794
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013625703564727957,
+      "loss": 2.631,
+      "step": 6795
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013624765478424016,
+      "loss": 2.5888,
+      "step": 6796
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013623827392120078,
+      "loss": 2.3773,
+      "step": 6797
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013622889305816137,
+      "loss": 2.5533,
+      "step": 6798
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013621951219512196,
+      "loss": 2.8023,
+      "step": 6799
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013621013133208255,
+      "loss": 2.3581,
+      "step": 6800
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013620075046904314,
+      "loss": 2.2911,
+      "step": 6801
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013619136960600376,
+      "loss": 2.8141,
+      "step": 6802
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013618198874296435,
+      "loss": 2.6189,
+      "step": 6803
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013617260787992497,
+      "loss": 2.5261,
+      "step": 6804
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013616322701688556,
+      "loss": 2.3808,
+      "step": 6805
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013615384615384618,
+      "loss": 2.5387,
+      "step": 6806
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013614446529080677,
+      "loss": 2.5545,
+      "step": 6807
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013613508442776736,
+      "loss": 2.6867,
+      "step": 6808
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013612570356472795,
+      "loss": 2.594,
+      "step": 6809
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013611632270168857,
+      "loss": 2.5978,
+      "step": 6810
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00013610694183864916,
+      "loss": 2.6158,
+      "step": 6811
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013609756097560978,
+      "loss": 2.7429,
+      "step": 6812
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013608818011257037,
+      "loss": 2.4183,
+      "step": 6813
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013607879924953096,
+      "loss": 2.5864,
+      "step": 6814
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013606941838649158,
+      "loss": 2.4648,
+      "step": 6815
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013606003752345217,
+      "loss": 2.5851,
+      "step": 6816
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013605065666041276,
+      "loss": 2.4705,
+      "step": 6817
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013604127579737335,
+      "loss": 2.4608,
+      "step": 6818
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013603189493433397,
+      "loss": 2.3756,
+      "step": 6819
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013602251407129456,
+      "loss": 2.3933,
+      "step": 6820
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013601313320825518,
+      "loss": 2.6555,
+      "step": 6821
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013600375234521577,
+      "loss": 2.4078,
+      "step": 6822
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013599437148217639,
+      "loss": 2.3943,
+      "step": 6823
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013598499061913698,
+      "loss": 2.3608,
+      "step": 6824
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013597560975609757,
+      "loss": 2.6307,
+      "step": 6825
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013596622889305816,
+      "loss": 2.8152,
+      "step": 6826
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013595684803001875,
+      "loss": 2.6904,
+      "step": 6827
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013594746716697937,
+      "loss": 2.3515,
+      "step": 6828
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013593808630393996,
+      "loss": 2.6183,
+      "step": 6829
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013592870544090058,
+      "loss": 2.3685,
+      "step": 6830
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013591932457786117,
+      "loss": 2.6682,
+      "step": 6831
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013590994371482179,
+      "loss": 2.3771,
+      "step": 6832
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 0.00013590056285178238,
+      "loss": 2.4237,
+      "step": 6833
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013589118198874297,
+      "loss": 2.4542,
+      "step": 6834
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013588180112570356,
+      "loss": 2.2364,
+      "step": 6835
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013587242026266418,
+      "loss": 2.4962,
+      "step": 6836
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013586303939962477,
+      "loss": 2.6686,
+      "step": 6837
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013585365853658539,
+      "loss": 2.7291,
+      "step": 6838
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013584427767354598,
+      "loss": 2.8796,
+      "step": 6839
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013583489681050657,
+      "loss": 2.6005,
+      "step": 6840
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013582551594746719,
+      "loss": 2.5963,
+      "step": 6841
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013581613508442778,
+      "loss": 2.5014,
+      "step": 6842
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013580675422138837,
+      "loss": 2.4676,
+      "step": 6843
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013579737335834896,
+      "loss": 2.5502,
+      "step": 6844
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013578799249530958,
+      "loss": 2.3737,
+      "step": 6845
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013577861163227017,
+      "loss": 2.5406,
+      "step": 6846
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013576923076923079,
+      "loss": 2.5973,
+      "step": 6847
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013575984990619138,
+      "loss": 2.4928,
+      "step": 6848
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.000135750469043152,
+      "loss": 2.5805,
+      "step": 6849
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013574108818011259,
+      "loss": 2.5388,
+      "step": 6850
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013573170731707318,
+      "loss": 2.3687,
+      "step": 6851
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013572232645403377,
+      "loss": 2.3002,
+      "step": 6852
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013571294559099436,
+      "loss": 2.4543,
+      "step": 6853
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00013570356472795498,
+      "loss": 2.1962,
+      "step": 6854
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013569418386491557,
+      "loss": 2.544,
+      "step": 6855
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013568480300187619,
+      "loss": 2.4125,
+      "step": 6856
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013567542213883678,
+      "loss": 2.6594,
+      "step": 6857
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0001356660412757974,
+      "loss": 2.4732,
+      "step": 6858
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013565666041275799,
+      "loss": 2.5904,
+      "step": 6859
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013564727954971858,
+      "loss": 2.2956,
+      "step": 6860
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013563789868667917,
+      "loss": 2.4829,
+      "step": 6861
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013562851782363979,
+      "loss": 2.3944,
+      "step": 6862
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013561913696060038,
+      "loss": 2.5527,
+      "step": 6863
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.000135609756097561,
+      "loss": 2.6015,
+      "step": 6864
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013560037523452159,
+      "loss": 2.6471,
+      "step": 6865
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013559099437148218,
+      "loss": 2.4036,
+      "step": 6866
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0001355816135084428,
+      "loss": 2.2349,
+      "step": 6867
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013557223264540339,
+      "loss": 2.3572,
+      "step": 6868
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013556285178236398,
+      "loss": 2.4175,
+      "step": 6869
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013555347091932457,
+      "loss": 2.6986,
+      "step": 6870
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013554409005628518,
+      "loss": 2.3682,
+      "step": 6871
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013553470919324578,
+      "loss": 2.4983,
+      "step": 6872
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0001355253283302064,
+      "loss": 2.4408,
+      "step": 6873
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.00013551594746716698,
+      "loss": 2.7235,
+      "step": 6874
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0001355065666041276,
+      "loss": 2.4508,
+      "step": 6875
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0001354971857410882,
+      "loss": 2.448,
+      "step": 6876
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013548780487804878,
+      "loss": 2.4735,
+      "step": 6877
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013547842401500938,
+      "loss": 2.5744,
+      "step": 6878
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013546904315197,
+      "loss": 2.5311,
+      "step": 6879
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013545966228893058,
+      "loss": 2.4604,
+      "step": 6880
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013545028142589118,
+      "loss": 2.8048,
+      "step": 6881
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0001354409005628518,
+      "loss": 2.6484,
+      "step": 6882
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013543151969981238,
+      "loss": 2.7426,
+      "step": 6883
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.000135422138836773,
+      "loss": 2.729,
+      "step": 6884
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0001354127579737336,
+      "loss": 2.2172,
+      "step": 6885
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013540337711069418,
+      "loss": 2.5957,
+      "step": 6886
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013539399624765478,
+      "loss": 2.4648,
+      "step": 6887
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0001353846153846154,
+      "loss": 2.5834,
+      "step": 6888
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013537523452157598,
+      "loss": 2.4373,
+      "step": 6889
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0001353658536585366,
+      "loss": 2.6057,
+      "step": 6890
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0001353564727954972,
+      "loss": 2.4047,
+      "step": 6891
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0001353470919324578,
+      "loss": 2.2497,
+      "step": 6892
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0001353377110694184,
+      "loss": 2.5441,
+      "step": 6893
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.000135328330206379,
+      "loss": 2.4593,
+      "step": 6894
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013531894934333958,
+      "loss": 2.4393,
+      "step": 6895
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00013530956848030018,
+      "loss": 2.5964,
+      "step": 6896
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.0001353001876172608,
+      "loss": 2.7698,
+      "step": 6897
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00013529080675422138,
+      "loss": 2.7468,
+      "step": 6898
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000135281425891182,
+      "loss": 2.3845,
+      "step": 6899
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001352720450281426,
+      "loss": 2.3826,
+      "step": 6900
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001352626641651032,
+      "loss": 2.5948,
+      "step": 6901
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001352532833020638,
+      "loss": 2.3052,
+      "step": 6902
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001352439024390244,
+      "loss": 2.6995,
+      "step": 6903
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00013523452157598498,
+      "loss": 2.5616,
+      "step": 6904
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001352251407129456,
+      "loss": 2.6457,
+      "step": 6905
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001352157598499062,
+      "loss": 2.3454,
+      "step": 6906
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00013520637898686678,
+      "loss": 2.5225,
+      "step": 6907
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001351969981238274,
+      "loss": 2.394,
+      "step": 6908
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000135187617260788,
+      "loss": 2.4781,
+      "step": 6909
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001351782363977486,
+      "loss": 2.2124,
+      "step": 6910
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001351688555347092,
+      "loss": 2.5289,
+      "step": 6911
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001351594746716698,
+      "loss": 2.4502,
+      "step": 6912
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00013515009380863038,
+      "loss": 2.5207,
+      "step": 6913
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.000135140712945591,
+      "loss": 2.4225,
+      "step": 6914
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001351313320825516,
+      "loss": 2.5725,
+      "step": 6915
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001351219512195122,
+      "loss": 2.0918,
+      "step": 6916
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.0001351125703564728,
+      "loss": 2.452,
+      "step": 6917
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00013510318949343342,
+      "loss": 2.6421,
+      "step": 6918
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.000135093808630394,
+      "loss": 2.4956,
+      "step": 6919
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001350844277673546,
+      "loss": 2.8419,
+      "step": 6920
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001350750469043152,
+      "loss": 2.3779,
+      "step": 6921
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00013506566604127578,
+      "loss": 2.4823,
+      "step": 6922
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001350562851782364,
+      "loss": 2.5072,
+      "step": 6923
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.000135046904315197,
+      "loss": 2.7268,
+      "step": 6924
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001350375234521576,
+      "loss": 2.6786,
+      "step": 6925
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001350281425891182,
+      "loss": 2.4934,
+      "step": 6926
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00013501876172607882,
+      "loss": 2.407,
+      "step": 6927
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001350093808630394,
+      "loss": 2.6049,
+      "step": 6928
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00013500000000000003,
+      "loss": 2.5852,
+      "step": 6929
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001349906191369606,
+      "loss": 2.6518,
+      "step": 6930
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001349812382739212,
+      "loss": 2.4487,
+      "step": 6931
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001349718574108818,
+      "loss": 2.3098,
+      "step": 6932
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001349624765478424,
+      "loss": 2.4153,
+      "step": 6933
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.000134953095684803,
+      "loss": 2.5166,
+      "step": 6934
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001349437148217636,
+      "loss": 2.5037,
+      "step": 6935
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00013493433395872422,
+      "loss": 2.328,
+      "step": 6936
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.0001349249530956848,
+      "loss": 2.4616,
+      "step": 6937
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.00013491557223264543,
+      "loss": 2.5252,
+      "step": 6938
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 0.000134906191369606,
+      "loss": 2.3104,
+      "step": 6939
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001348968105065666,
+      "loss": 2.5494,
+      "step": 6940
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001348874296435272,
+      "loss": 2.2493,
+      "step": 6941
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013487804878048782,
+      "loss": 2.4144,
+      "step": 6942
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001348686679174484,
+      "loss": 2.4234,
+      "step": 6943
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013485928705440903,
+      "loss": 2.4175,
+      "step": 6944
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013484990619136962,
+      "loss": 2.5553,
+      "step": 6945
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001348405253283302,
+      "loss": 2.6398,
+      "step": 6946
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013483114446529083,
+      "loss": 2.6108,
+      "step": 6947
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001348217636022514,
+      "loss": 2.5057,
+      "step": 6948
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.000134812382739212,
+      "loss": 2.3215,
+      "step": 6949
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001348030018761726,
+      "loss": 2.6185,
+      "step": 6950
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013479362101313322,
+      "loss": 2.5266,
+      "step": 6951
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001347842401500938,
+      "loss": 2.5338,
+      "step": 6952
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013477485928705443,
+      "loss": 2.648,
+      "step": 6953
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013476547842401502,
+      "loss": 2.4813,
+      "step": 6954
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013475609756097564,
+      "loss": 2.3445,
+      "step": 6955
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013474671669793623,
+      "loss": 2.5981,
+      "step": 6956
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013473733583489682,
+      "loss": 2.6569,
+      "step": 6957
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001347279549718574,
+      "loss": 2.3214,
+      "step": 6958
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.000134718574108818,
+      "loss": 2.513,
+      "step": 6959
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.00013470919324577862,
+      "loss": 2.4134,
+      "step": 6960
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0001346998123827392,
+      "loss": 2.4338,
+      "step": 6961
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013469043151969983,
+      "loss": 2.5988,
+      "step": 6962
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013468105065666042,
+      "loss": 2.4486,
+      "step": 6963
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013467166979362104,
+      "loss": 2.7283,
+      "step": 6964
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013466228893058163,
+      "loss": 2.6538,
+      "step": 6965
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013465290806754222,
+      "loss": 2.3759,
+      "step": 6966
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0001346435272045028,
+      "loss": 2.5113,
+      "step": 6967
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013463414634146343,
+      "loss": 2.2737,
+      "step": 6968
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013462476547842402,
+      "loss": 2.6965,
+      "step": 6969
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013461538461538464,
+      "loss": 2.5736,
+      "step": 6970
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013460600375234523,
+      "loss": 2.8914,
+      "step": 6971
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013459662288930582,
+      "loss": 2.7776,
+      "step": 6972
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013458724202626644,
+      "loss": 2.6722,
+      "step": 6973
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013457786116322703,
+      "loss": 2.4023,
+      "step": 6974
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013456848030018762,
+      "loss": 2.5798,
+      "step": 6975
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.0001345590994371482,
+      "loss": 2.5961,
+      "step": 6976
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013454971857410883,
+      "loss": 2.6868,
+      "step": 6977
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013454033771106942,
+      "loss": 2.4472,
+      "step": 6978
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013453095684803004,
+      "loss": 2.5858,
+      "step": 6979
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013452157598499063,
+      "loss": 2.5875,
+      "step": 6980
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013451219512195125,
+      "loss": 2.4692,
+      "step": 6981
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00013450281425891184,
+      "loss": 2.3292,
+      "step": 6982
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013449343339587243,
+      "loss": 2.5974,
+      "step": 6983
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013448405253283302,
+      "loss": 2.5615,
+      "step": 6984
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.0001344746716697936,
+      "loss": 2.3239,
+      "step": 6985
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013446529080675423,
+      "loss": 2.5638,
+      "step": 6986
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013445590994371482,
+      "loss": 2.6292,
+      "step": 6987
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013444652908067544,
+      "loss": 2.3428,
+      "step": 6988
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013443714821763603,
+      "loss": 2.342,
+      "step": 6989
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013442776735459665,
+      "loss": 2.7529,
+      "step": 6990
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013441838649155724,
+      "loss": 2.6179,
+      "step": 6991
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013440900562851783,
+      "loss": 2.7394,
+      "step": 6992
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013439962476547842,
+      "loss": 2.3883,
+      "step": 6993
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013439024390243904,
+      "loss": 2.5711,
+      "step": 6994
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013438086303939963,
+      "loss": 2.4149,
+      "step": 6995
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013437148217636025,
+      "loss": 2.4532,
+      "step": 6996
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013436210131332084,
+      "loss": 2.4158,
+      "step": 6997
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013435272045028143,
+      "loss": 2.5676,
+      "step": 6998
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013434333958724205,
+      "loss": 2.6183,
+      "step": 6999
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013433395872420264,
+      "loss": 2.5711,
+      "step": 7000
+    },
+    {
+      "epoch": 3.28,
+      "eval_loss": 2.611212968826294,
+      "eval_runtime": 67.6646,
+      "eval_samples_per_second": 28.006,
+      "eval_steps_per_second": 3.503,
+      "step": 7000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013432457786116323,
+      "loss": 2.368,
+      "step": 7001
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013431519699812382,
+      "loss": 2.613,
+      "step": 7002
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00013430581613508444,
+      "loss": 2.4846,
+      "step": 7003
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013429643527204503,
+      "loss": 2.7202,
+      "step": 7004
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013428705440900565,
+      "loss": 2.5813,
+      "step": 7005
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013427767354596624,
+      "loss": 2.5082,
+      "step": 7006
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013426829268292685,
+      "loss": 2.4354,
+      "step": 7007
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013425891181988745,
+      "loss": 2.6202,
+      "step": 7008
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013424953095684804,
+      "loss": 2.5162,
+      "step": 7009
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013424015009380863,
+      "loss": 2.4231,
+      "step": 7010
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013423076923076922,
+      "loss": 2.6997,
+      "step": 7011
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013422138836772984,
+      "loss": 2.2257,
+      "step": 7012
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013421200750469043,
+      "loss": 2.3751,
+      "step": 7013
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013420262664165105,
+      "loss": 2.447,
+      "step": 7014
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013419324577861164,
+      "loss": 2.6247,
+      "step": 7015
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013418386491557225,
+      "loss": 2.3378,
+      "step": 7016
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013417448405253284,
+      "loss": 2.5296,
+      "step": 7017
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013416510318949344,
+      "loss": 2.61,
+      "step": 7018
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013415572232645403,
+      "loss": 2.3858,
+      "step": 7019
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013414634146341464,
+      "loss": 2.6556,
+      "step": 7020
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013413696060037524,
+      "loss": 2.6449,
+      "step": 7021
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013412757973733585,
+      "loss": 2.4297,
+      "step": 7022
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013411819887429644,
+      "loss": 2.3561,
+      "step": 7023
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00013410881801125704,
+      "loss": 2.425,
+      "step": 7024
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013409943714821765,
+      "loss": 2.589,
+      "step": 7025
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013409005628517824,
+      "loss": 2.247,
+      "step": 7026
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013408067542213884,
+      "loss": 2.539,
+      "step": 7027
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013407129455909943,
+      "loss": 2.2437,
+      "step": 7028
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013406191369606004,
+      "loss": 2.5796,
+      "step": 7029
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013405253283302064,
+      "loss": 2.7747,
+      "step": 7030
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013404315196998125,
+      "loss": 2.6646,
+      "step": 7031
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013403377110694184,
+      "loss": 2.5644,
+      "step": 7032
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013402439024390246,
+      "loss": 2.4813,
+      "step": 7033
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013401500938086305,
+      "loss": 2.4884,
+      "step": 7034
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013400562851782364,
+      "loss": 2.4834,
+      "step": 7035
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013399624765478424,
+      "loss": 2.4256,
+      "step": 7036
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013398686679174483,
+      "loss": 2.5765,
+      "step": 7037
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013397748592870544,
+      "loss": 2.6229,
+      "step": 7038
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013396810506566604,
+      "loss": 2.408,
+      "step": 7039
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013395872420262665,
+      "loss": 2.1766,
+      "step": 7040
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013394934333958724,
+      "loss": 2.5754,
+      "step": 7041
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013393996247654786,
+      "loss": 2.4457,
+      "step": 7042
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013393058161350845,
+      "loss": 2.5479,
+      "step": 7043
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013392120075046904,
+      "loss": 2.4598,
+      "step": 7044
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013391181988742964,
+      "loss": 2.6089,
+      "step": 7045
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 0.00013390243902439025,
+      "loss": 2.4943,
+      "step": 7046
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013389305816135084,
+      "loss": 2.6263,
+      "step": 7047
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013388367729831146,
+      "loss": 2.5521,
+      "step": 7048
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013387429643527205,
+      "loss": 2.4785,
+      "step": 7049
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013386491557223264,
+      "loss": 2.1805,
+      "step": 7050
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013385553470919326,
+      "loss": 2.5284,
+      "step": 7051
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013384615384615385,
+      "loss": 2.4854,
+      "step": 7052
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013383677298311444,
+      "loss": 2.5762,
+      "step": 7053
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013382739212007504,
+      "loss": 2.542,
+      "step": 7054
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013381801125703565,
+      "loss": 2.3884,
+      "step": 7055
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013380863039399624,
+      "loss": 2.6222,
+      "step": 7056
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013379924953095686,
+      "loss": 2.4451,
+      "step": 7057
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013378986866791745,
+      "loss": 2.5238,
+      "step": 7058
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013378048780487807,
+      "loss": 2.2645,
+      "step": 7059
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013377110694183866,
+      "loss": 2.6337,
+      "step": 7060
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013376172607879925,
+      "loss": 2.5126,
+      "step": 7061
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013375234521575984,
+      "loss": 2.3821,
+      "step": 7062
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013374296435272043,
+      "loss": 2.3517,
+      "step": 7063
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013373358348968105,
+      "loss": 2.4251,
+      "step": 7064
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013372420262664164,
+      "loss": 2.6333,
+      "step": 7065
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013371482176360226,
+      "loss": 2.719,
+      "step": 7066
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.00013370544090056285,
+      "loss": 2.4987,
+      "step": 7067
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013369606003752347,
+      "loss": 2.5,
+      "step": 7068
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013368667917448406,
+      "loss": 2.5586,
+      "step": 7069
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013367729831144465,
+      "loss": 2.5114,
+      "step": 7070
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013366791744840524,
+      "loss": 2.5376,
+      "step": 7071
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013365853658536586,
+      "loss": 2.5731,
+      "step": 7072
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013364915572232645,
+      "loss": 2.7543,
+      "step": 7073
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013363977485928707,
+      "loss": 2.6675,
+      "step": 7074
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013363039399624766,
+      "loss": 2.5519,
+      "step": 7075
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013362101313320828,
+      "loss": 2.4672,
+      "step": 7076
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013361163227016887,
+      "loss": 2.594,
+      "step": 7077
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013360225140712946,
+      "loss": 2.5758,
+      "step": 7078
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013359287054409005,
+      "loss": 2.2866,
+      "step": 7079
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013358348968105064,
+      "loss": 2.5034,
+      "step": 7080
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013357410881801126,
+      "loss": 2.467,
+      "step": 7081
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013356472795497185,
+      "loss": 2.5554,
+      "step": 7082
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013355534709193247,
+      "loss": 2.6409,
+      "step": 7083
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013354596622889306,
+      "loss": 2.6785,
+      "step": 7084
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013353658536585368,
+      "loss": 2.3791,
+      "step": 7085
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013352720450281427,
+      "loss": 2.6386,
+      "step": 7086
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013351782363977486,
+      "loss": 2.8022,
+      "step": 7087
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00013350844277673545,
+      "loss": 2.6883,
+      "step": 7088
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013349906191369607,
+      "loss": 2.824,
+      "step": 7089
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013348968105065666,
+      "loss": 2.4258,
+      "step": 7090
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013348030018761725,
+      "loss": 2.416,
+      "step": 7091
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013347091932457787,
+      "loss": 2.6407,
+      "step": 7092
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013346153846153846,
+      "loss": 2.5154,
+      "step": 7093
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013345215759849908,
+      "loss": 2.7829,
+      "step": 7094
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013344277673545967,
+      "loss": 2.3445,
+      "step": 7095
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0001334333958724203,
+      "loss": 2.593,
+      "step": 7096
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013342401500938085,
+      "loss": 2.2557,
+      "step": 7097
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013341463414634147,
+      "loss": 2.6095,
+      "step": 7098
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013340525328330206,
+      "loss": 2.2292,
+      "step": 7099
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013339587242026268,
+      "loss": 2.6606,
+      "step": 7100
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013338649155722327,
+      "loss": 2.3897,
+      "step": 7101
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0001333771106941839,
+      "loss": 2.4009,
+      "step": 7102
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013336772983114448,
+      "loss": 2.3329,
+      "step": 7103
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013335834896810507,
+      "loss": 2.524,
+      "step": 7104
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.0001333489681050657,
+      "loss": 2.7393,
+      "step": 7105
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013333958724202625,
+      "loss": 2.7042,
+      "step": 7106
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013333020637898687,
+      "loss": 2.2011,
+      "step": 7107
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013332082551594746,
+      "loss": 2.377,
+      "step": 7108
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013331144465290808,
+      "loss": 2.5091,
+      "step": 7109
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00013330206378986867,
+      "loss": 2.7091,
+      "step": 7110
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0001332926829268293,
+      "loss": 2.7134,
+      "step": 7111
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013328330206378988,
+      "loss": 2.4981,
+      "step": 7112
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0001332739212007505,
+      "loss": 2.2127,
+      "step": 7113
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0001332645403377111,
+      "loss": 2.345,
+      "step": 7114
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013325515947467168,
+      "loss": 2.4704,
+      "step": 7115
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013324577861163227,
+      "loss": 2.8571,
+      "step": 7116
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013323639774859286,
+      "loss": 2.663,
+      "step": 7117
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013322701688555348,
+      "loss": 2.4238,
+      "step": 7118
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013321763602251407,
+      "loss": 2.5373,
+      "step": 7119
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0001332082551594747,
+      "loss": 2.4081,
+      "step": 7120
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013319887429643528,
+      "loss": 2.6243,
+      "step": 7121
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0001331894934333959,
+      "loss": 2.7853,
+      "step": 7122
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0001331801125703565,
+      "loss": 2.3593,
+      "step": 7123
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013317073170731708,
+      "loss": 2.7438,
+      "step": 7124
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013316135084427767,
+      "loss": 2.5292,
+      "step": 7125
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0001331519699812383,
+      "loss": 2.7118,
+      "step": 7126
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013314258911819888,
+      "loss": 2.2982,
+      "step": 7127
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0001331332082551595,
+      "loss": 2.4583,
+      "step": 7128
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0001331238273921201,
+      "loss": 2.5152,
+      "step": 7129
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.00013311444652908068,
+      "loss": 2.7412,
+      "step": 7130
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 0.0001331050656660413,
+      "loss": 2.5381,
+      "step": 7131
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001330956848030019,
+      "loss": 2.2906,
+      "step": 7132
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00013308630393996248,
+      "loss": 2.3317,
+      "step": 7133
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00013307692307692307,
+      "loss": 2.3868,
+      "step": 7134
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001330675422138837,
+      "loss": 2.5633,
+      "step": 7135
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00013305816135084428,
+      "loss": 2.488,
+      "step": 7136
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001330487804878049,
+      "loss": 2.3388,
+      "step": 7137
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001330393996247655,
+      "loss": 2.4864,
+      "step": 7138
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001330300187617261,
+      "loss": 2.4022,
+      "step": 7139
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001330206378986867,
+      "loss": 2.2839,
+      "step": 7140
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001330112570356473,
+      "loss": 2.7058,
+      "step": 7141
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00013300187617260788,
+      "loss": 2.642,
+      "step": 7142
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00013299249530956847,
+      "loss": 2.4962,
+      "step": 7143
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001329831144465291,
+      "loss": 2.0345,
+      "step": 7144
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00013297373358348968,
+      "loss": 2.9104,
+      "step": 7145
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001329643527204503,
+      "loss": 2.7604,
+      "step": 7146
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001329549718574109,
+      "loss": 2.5772,
+      "step": 7147
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001329455909943715,
+      "loss": 2.4791,
+      "step": 7148
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001329362101313321,
+      "loss": 2.4596,
+      "step": 7149
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001329268292682927,
+      "loss": 2.2303,
+      "step": 7150
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.00013291744840525328,
+      "loss": 2.5122,
+      "step": 7151
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001329080675422139,
+      "loss": 2.7836,
+      "step": 7152
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001328986866791745,
+      "loss": 2.5414,
+      "step": 7153
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001328893058161351,
+      "loss": 2.4148,
+      "step": 7154
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001328799249530957,
+      "loss": 2.3731,
+      "step": 7155
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001328705440900563,
+      "loss": 2.376,
+      "step": 7156
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001328611632270169,
+      "loss": 2.3675,
+      "step": 7157
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001328517823639775,
+      "loss": 2.7996,
+      "step": 7158
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001328424015009381,
+      "loss": 2.5899,
+      "step": 7159
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00013283302063789868,
+      "loss": 2.5416,
+      "step": 7160
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001328236397748593,
+      "loss": 2.3764,
+      "step": 7161
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001328142589118199,
+      "loss": 2.3931,
+      "step": 7162
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001328048780487805,
+      "loss": 2.6771,
+      "step": 7163
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001327954971857411,
+      "loss": 2.6151,
+      "step": 7164
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00013278611632270171,
+      "loss": 2.6513,
+      "step": 7165
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001327767354596623,
+      "loss": 2.3686,
+      "step": 7166
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001327673545966229,
+      "loss": 2.4864,
+      "step": 7167
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001327579737335835,
+      "loss": 2.6126,
+      "step": 7168
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00013274859287054408,
+      "loss": 2.7183,
+      "step": 7169
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001327392120075047,
+      "loss": 2.4009,
+      "step": 7170
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001327298311444653,
+      "loss": 2.7636,
+      "step": 7171
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001327204502814259,
+      "loss": 2.5197,
+      "step": 7172
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.0001327110694183865,
+      "loss": 2.3377,
+      "step": 7173
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00013270168855534711,
+      "loss": 2.6788,
+      "step": 7174
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001326923076923077,
+      "loss": 2.396,
+      "step": 7175
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001326829268292683,
+      "loss": 2.4707,
+      "step": 7176
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001326735459662289,
+      "loss": 2.392,
+      "step": 7177
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001326641651031895,
+      "loss": 2.7939,
+      "step": 7178
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001326547842401501,
+      "loss": 2.3563,
+      "step": 7179
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00013264540337711071,
+      "loss": 2.5449,
+      "step": 7180
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001326360225140713,
+      "loss": 2.7525,
+      "step": 7181
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001326266416510319,
+      "loss": 2.4762,
+      "step": 7182
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001326172607879925,
+      "loss": 2.6315,
+      "step": 7183
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001326078799249531,
+      "loss": 2.5296,
+      "step": 7184
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001325984990619137,
+      "loss": 2.583,
+      "step": 7185
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00013258911819887429,
+      "loss": 2.4285,
+      "step": 7186
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001325797373358349,
+      "loss": 2.3708,
+      "step": 7187
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001325703564727955,
+      "loss": 2.2242,
+      "step": 7188
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001325609756097561,
+      "loss": 2.3025,
+      "step": 7189
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001325515947467167,
+      "loss": 2.5227,
+      "step": 7190
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00013254221388367732,
+      "loss": 2.3877,
+      "step": 7191
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001325328330206379,
+      "loss": 2.5661,
+      "step": 7192
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001325234521575985,
+      "loss": 2.6057,
+      "step": 7193
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.0001325140712945591,
+      "loss": 2.417,
+      "step": 7194
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00013250469043151969,
+      "loss": 2.1595,
+      "step": 7195
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001324953095684803,
+      "loss": 2.38,
+      "step": 7196
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001324859287054409,
+      "loss": 2.4009,
+      "step": 7197
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001324765478424015,
+      "loss": 2.5071,
+      "step": 7198
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001324671669793621,
+      "loss": 2.4709,
+      "step": 7199
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00013245778611632272,
+      "loss": 2.4998,
+      "step": 7200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001324484052532833,
+      "loss": 2.3752,
+      "step": 7201
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001324390243902439,
+      "loss": 2.544,
+      "step": 7202
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001324296435272045,
+      "loss": 2.3071,
+      "step": 7203
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001324202626641651,
+      "loss": 2.285,
+      "step": 7204
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001324108818011257,
+      "loss": 2.6946,
+      "step": 7205
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00013240150093808632,
+      "loss": 2.4797,
+      "step": 7206
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001323921200750469,
+      "loss": 2.6299,
+      "step": 7207
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001323827392120075,
+      "loss": 2.2382,
+      "step": 7208
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00013237335834896812,
+      "loss": 2.8249,
+      "step": 7209
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001323639774859287,
+      "loss": 2.3678,
+      "step": 7210
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001323545966228893,
+      "loss": 2.35,
+      "step": 7211
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001323452157598499,
+      "loss": 2.7164,
+      "step": 7212
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001323358348968105,
+      "loss": 2.4586,
+      "step": 7213
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001323264540337711,
+      "loss": 2.2273,
+      "step": 7214
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00013231707317073172,
+      "loss": 2.442,
+      "step": 7215
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.0001323076923076923,
+      "loss": 2.4739,
+      "step": 7216
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00013229831144465293,
+      "loss": 2.2446,
+      "step": 7217
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00013228893058161352,
+      "loss": 2.4861,
+      "step": 7218
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001322795497185741,
+      "loss": 2.5237,
+      "step": 7219
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001322701688555347,
+      "loss": 2.4503,
+      "step": 7220
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001322607879924953,
+      "loss": 2.6993,
+      "step": 7221
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001322514071294559,
+      "loss": 2.6011,
+      "step": 7222
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001322420262664165,
+      "loss": 2.5584,
+      "step": 7223
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00013223264540337712,
+      "loss": 2.4356,
+      "step": 7224
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001322232645403377,
+      "loss": 2.5062,
+      "step": 7225
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00013221388367729833,
+      "loss": 2.4539,
+      "step": 7226
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00013220450281425892,
+      "loss": 2.3179,
+      "step": 7227
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001321951219512195,
+      "loss": 2.3776,
+      "step": 7228
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001321857410881801,
+      "loss": 2.4055,
+      "step": 7229
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00013217636022514072,
+      "loss": 2.4975,
+      "step": 7230
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001321669793621013,
+      "loss": 2.5382,
+      "step": 7231
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00013215759849906193,
+      "loss": 2.688,
+      "step": 7232
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00013214821763602252,
+      "loss": 2.7453,
+      "step": 7233
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001321388367729831,
+      "loss": 2.3626,
+      "step": 7234
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00013212945590994373,
+      "loss": 2.5992,
+      "step": 7235
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.00013212007504690432,
+      "loss": 2.2554,
+      "step": 7236
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001321106941838649,
+      "loss": 2.3281,
+      "step": 7237
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 0.0001321013133208255,
+      "loss": 2.4529,
+      "step": 7238
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013209193245778612,
+      "loss": 2.4898,
+      "step": 7239
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0001320825515947467,
+      "loss": 2.4785,
+      "step": 7240
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013207317073170733,
+      "loss": 2.302,
+      "step": 7241
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013206378986866792,
+      "loss": 2.3991,
+      "step": 7242
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013205440900562854,
+      "loss": 2.6219,
+      "step": 7243
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013204502814258913,
+      "loss": 2.2258,
+      "step": 7244
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013203564727954972,
+      "loss": 2.2749,
+      "step": 7245
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0001320262664165103,
+      "loss": 2.4651,
+      "step": 7246
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0001320168855534709,
+      "loss": 2.3199,
+      "step": 7247
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013200750469043152,
+      "loss": 2.6161,
+      "step": 7248
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0001319981238273921,
+      "loss": 2.4059,
+      "step": 7249
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013198874296435273,
+      "loss": 2.575,
+      "step": 7250
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013197936210131332,
+      "loss": 2.2579,
+      "step": 7251
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013196998123827394,
+      "loss": 2.4649,
+      "step": 7252
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013196060037523453,
+      "loss": 2.6997,
+      "step": 7253
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013195121951219512,
+      "loss": 2.6941,
+      "step": 7254
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0001319418386491557,
+      "loss": 2.5377,
+      "step": 7255
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013193245778611633,
+      "loss": 2.3559,
+      "step": 7256
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013192307692307692,
+      "loss": 2.3188,
+      "step": 7257
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013191369606003754,
+      "loss": 2.6519,
+      "step": 7258
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.00013190431519699813,
+      "loss": 2.5336,
+      "step": 7259
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013189493433395872,
+      "loss": 2.6785,
+      "step": 7260
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013188555347091934,
+      "loss": 2.5287,
+      "step": 7261
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013187617260787993,
+      "loss": 2.5277,
+      "step": 7262
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013186679174484055,
+      "loss": 1.9619,
+      "step": 7263
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0001318574108818011,
+      "loss": 2.4315,
+      "step": 7264
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013184803001876173,
+      "loss": 2.1529,
+      "step": 7265
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013183864915572232,
+      "loss": 2.1949,
+      "step": 7266
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013182926829268294,
+      "loss": 2.4104,
+      "step": 7267
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013181988742964353,
+      "loss": 2.4759,
+      "step": 7268
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013181050656660415,
+      "loss": 2.7436,
+      "step": 7269
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013180112570356474,
+      "loss": 2.721,
+      "step": 7270
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013179174484052536,
+      "loss": 2.2828,
+      "step": 7271
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013178236397748595,
+      "loss": 2.3782,
+      "step": 7272
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.0001317729831144465,
+      "loss": 2.2776,
+      "step": 7273
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013176360225140713,
+      "loss": 2.6231,
+      "step": 7274
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013175422138836772,
+      "loss": 2.5403,
+      "step": 7275
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013174484052532834,
+      "loss": 2.2997,
+      "step": 7276
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013173545966228893,
+      "loss": 2.4221,
+      "step": 7277
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013172607879924955,
+      "loss": 2.5861,
+      "step": 7278
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013171669793621014,
+      "loss": 2.4443,
+      "step": 7279
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00013170731707317076,
+      "loss": 2.4141,
+      "step": 7280
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013169793621013135,
+      "loss": 2.3514,
+      "step": 7281
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013168855534709194,
+      "loss": 2.5267,
+      "step": 7282
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013167917448405253,
+      "loss": 2.4254,
+      "step": 7283
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013166979362101315,
+      "loss": 2.5598,
+      "step": 7284
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013166041275797374,
+      "loss": 2.4326,
+      "step": 7285
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013165103189493436,
+      "loss": 2.1133,
+      "step": 7286
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013164165103189495,
+      "loss": 2.4327,
+      "step": 7287
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013163227016885554,
+      "loss": 2.5318,
+      "step": 7288
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013162288930581616,
+      "loss": 2.4873,
+      "step": 7289
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013161350844277675,
+      "loss": 2.2775,
+      "step": 7290
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013160412757973734,
+      "loss": 2.2435,
+      "step": 7291
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013159474671669793,
+      "loss": 2.3857,
+      "step": 7292
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013158536585365855,
+      "loss": 2.6333,
+      "step": 7293
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013157598499061914,
+      "loss": 2.5383,
+      "step": 7294
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013156660412757976,
+      "loss": 2.4151,
+      "step": 7295
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013155722326454035,
+      "loss": 2.5835,
+      "step": 7296
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013154784240150097,
+      "loss": 2.6077,
+      "step": 7297
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013153846153846156,
+      "loss": 2.5436,
+      "step": 7298
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013152908067542215,
+      "loss": 2.2472,
+      "step": 7299
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013151969981238274,
+      "loss": 2.4087,
+      "step": 7300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013151031894934333,
+      "loss": 2.4298,
+      "step": 7301
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00013150093808630395,
+      "loss": 2.5579,
+      "step": 7302
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013149155722326454,
+      "loss": 2.385,
+      "step": 7303
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013148217636022516,
+      "loss": 2.4249,
+      "step": 7304
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013147279549718575,
+      "loss": 2.345,
+      "step": 7305
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013146341463414636,
+      "loss": 2.3382,
+      "step": 7306
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013145403377110696,
+      "loss": 2.7616,
+      "step": 7307
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013144465290806755,
+      "loss": 2.6127,
+      "step": 7308
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013143527204502814,
+      "loss": 2.6162,
+      "step": 7309
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013142589118198876,
+      "loss": 2.6282,
+      "step": 7310
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013141651031894935,
+      "loss": 2.2519,
+      "step": 7311
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013140712945590996,
+      "loss": 2.577,
+      "step": 7312
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013139774859287056,
+      "loss": 2.5962,
+      "step": 7313
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013138836772983115,
+      "loss": 2.5032,
+      "step": 7314
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013137898686679176,
+      "loss": 2.5353,
+      "step": 7315
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013136960600375236,
+      "loss": 2.5802,
+      "step": 7316
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013136022514071295,
+      "loss": 2.6031,
+      "step": 7317
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013135084427767354,
+      "loss": 2.8734,
+      "step": 7318
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013134146341463416,
+      "loss": 2.6428,
+      "step": 7319
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013133208255159475,
+      "loss": 2.3066,
+      "step": 7320
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013132270168855536,
+      "loss": 2.7238,
+      "step": 7321
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013131332082551596,
+      "loss": 2.4128,
+      "step": 7322
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 0.00013130393996247657,
+      "loss": 2.4085,
+      "step": 7323
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013129455909943716,
+      "loss": 2.6358,
+      "step": 7324
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013128517823639776,
+      "loss": 2.2009,
+      "step": 7325
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013127579737335835,
+      "loss": 2.5234,
+      "step": 7326
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013126641651031894,
+      "loss": 2.3811,
+      "step": 7327
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013125703564727956,
+      "loss": 2.4119,
+      "step": 7328
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013124765478424015,
+      "loss": 2.5364,
+      "step": 7329
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013123827392120076,
+      "loss": 2.3184,
+      "step": 7330
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013122889305816136,
+      "loss": 2.3354,
+      "step": 7331
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013121951219512197,
+      "loss": 2.4939,
+      "step": 7332
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013121013133208256,
+      "loss": 2.5407,
+      "step": 7333
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013120075046904316,
+      "loss": 2.6102,
+      "step": 7334
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013119136960600375,
+      "loss": 2.2399,
+      "step": 7335
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013118198874296436,
+      "loss": 2.3703,
+      "step": 7336
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013117260787992496,
+      "loss": 2.3863,
+      "step": 7337
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013116322701688557,
+      "loss": 2.6048,
+      "step": 7338
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013115384615384616,
+      "loss": 2.4135,
+      "step": 7339
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013114446529080676,
+      "loss": 2.561,
+      "step": 7340
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013113508442776737,
+      "loss": 2.3733,
+      "step": 7341
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013112570356472796,
+      "loss": 2.4162,
+      "step": 7342
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013111632270168856,
+      "loss": 2.2586,
+      "step": 7343
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.00013110694183864915,
+      "loss": 2.6705,
+      "step": 7344
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013109756097560976,
+      "loss": 2.7855,
+      "step": 7345
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013108818011257036,
+      "loss": 2.3109,
+      "step": 7346
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013107879924953097,
+      "loss": 2.5322,
+      "step": 7347
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013106941838649156,
+      "loss": 2.5113,
+      "step": 7348
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013106003752345218,
+      "loss": 2.4512,
+      "step": 7349
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013105065666041277,
+      "loss": 2.338,
+      "step": 7350
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013104127579737336,
+      "loss": 2.553,
+      "step": 7351
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013103189493433395,
+      "loss": 2.4642,
+      "step": 7352
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013102251407129455,
+      "loss": 2.608,
+      "step": 7353
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013101313320825516,
+      "loss": 2.707,
+      "step": 7354
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013100375234521575,
+      "loss": 2.7768,
+      "step": 7355
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013099437148217637,
+      "loss": 2.5241,
+      "step": 7356
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013098499061913696,
+      "loss": 2.8155,
+      "step": 7357
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013097560975609758,
+      "loss": 2.2613,
+      "step": 7358
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013096622889305817,
+      "loss": 2.4228,
+      "step": 7359
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013095684803001876,
+      "loss": 2.4788,
+      "step": 7360
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013094746716697935,
+      "loss": 2.4986,
+      "step": 7361
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013093808630393997,
+      "loss": 2.4448,
+      "step": 7362
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013092870544090056,
+      "loss": 2.5897,
+      "step": 7363
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013091932457786118,
+      "loss": 2.5086,
+      "step": 7364
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013090994371482177,
+      "loss": 2.5396,
+      "step": 7365
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00013090056285178236,
+      "loss": 2.533,
+      "step": 7366
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013089118198874298,
+      "loss": 2.5715,
+      "step": 7367
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013088180112570357,
+      "loss": 2.6796,
+      "step": 7368
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013087242026266416,
+      "loss": 2.5486,
+      "step": 7369
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013086303939962475,
+      "loss": 2.4822,
+      "step": 7370
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013085365853658537,
+      "loss": 2.2719,
+      "step": 7371
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013084427767354596,
+      "loss": 2.332,
+      "step": 7372
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013083489681050658,
+      "loss": 2.5794,
+      "step": 7373
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013082551594746717,
+      "loss": 2.6395,
+      "step": 7374
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0001308161350844278,
+      "loss": 2.6797,
+      "step": 7375
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013080675422138838,
+      "loss": 2.4925,
+      "step": 7376
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013079737335834897,
+      "loss": 2.5022,
+      "step": 7377
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013078799249530956,
+      "loss": 2.4943,
+      "step": 7378
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013077861163227015,
+      "loss": 2.432,
+      "step": 7379
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013076923076923077,
+      "loss": 2.7096,
+      "step": 7380
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013075984990619136,
+      "loss": 2.2865,
+      "step": 7381
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013075046904315198,
+      "loss": 2.532,
+      "step": 7382
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013074108818011257,
+      "loss": 2.7948,
+      "step": 7383
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.0001307317073170732,
+      "loss": 2.8023,
+      "step": 7384
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013072232645403378,
+      "loss": 2.4354,
+      "step": 7385
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013071294559099437,
+      "loss": 2.1143,
+      "step": 7386
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00013070356472795496,
+      "loss": 2.5877,
+      "step": 7387
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013069418386491558,
+      "loss": 2.4471,
+      "step": 7388
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013068480300187617,
+      "loss": 2.3953,
+      "step": 7389
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0001306754221388368,
+      "loss": 2.6136,
+      "step": 7390
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013066604127579738,
+      "loss": 2.3153,
+      "step": 7391
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013065666041275797,
+      "loss": 2.4311,
+      "step": 7392
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0001306472795497186,
+      "loss": 2.3716,
+      "step": 7393
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013063789868667918,
+      "loss": 2.3817,
+      "step": 7394
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013062851782363977,
+      "loss": 2.5463,
+      "step": 7395
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013061913696060036,
+      "loss": 2.4211,
+      "step": 7396
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013060975609756098,
+      "loss": 2.6245,
+      "step": 7397
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013060037523452157,
+      "loss": 2.4776,
+      "step": 7398
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0001305909943714822,
+      "loss": 2.7166,
+      "step": 7399
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013058161350844278,
+      "loss": 2.3828,
+      "step": 7400
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0001305722326454034,
+      "loss": 2.5136,
+      "step": 7401
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.000130562851782364,
+      "loss": 2.486,
+      "step": 7402
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013055347091932458,
+      "loss": 2.6031,
+      "step": 7403
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013054409005628517,
+      "loss": 2.613,
+      "step": 7404
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013053470919324576,
+      "loss": 2.5881,
+      "step": 7405
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013052532833020638,
+      "loss": 2.5297,
+      "step": 7406
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00013051594746716697,
+      "loss": 2.4863,
+      "step": 7407
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.0001305065666041276,
+      "loss": 2.4341,
+      "step": 7408
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00013049718574108818,
+      "loss": 2.7172,
+      "step": 7409
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0001304878048780488,
+      "loss": 2.5695,
+      "step": 7410
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0001304784240150094,
+      "loss": 2.5471,
+      "step": 7411
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00013046904315196998,
+      "loss": 2.6999,
+      "step": 7412
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00013045966228893057,
+      "loss": 2.4641,
+      "step": 7413
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0001304502814258912,
+      "loss": 2.4512,
+      "step": 7414
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00013044090056285178,
+      "loss": 2.4955,
+      "step": 7415
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0001304315196998124,
+      "loss": 2.3705,
+      "step": 7416
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.000130422138836773,
+      "loss": 2.7412,
+      "step": 7417
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00013041275797373358,
+      "loss": 2.7511,
+      "step": 7418
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0001304033771106942,
+      "loss": 2.5662,
+      "step": 7419
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0001303939962476548,
+      "loss": 2.56,
+      "step": 7420
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00013038461538461538,
+      "loss": 2.4312,
+      "step": 7421
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00013037523452157597,
+      "loss": 2.5072,
+      "step": 7422
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0001303658536585366,
+      "loss": 2.4824,
+      "step": 7423
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00013035647279549718,
+      "loss": 2.5336,
+      "step": 7424
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0001303470919324578,
+      "loss": 2.3865,
+      "step": 7425
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0001303377110694184,
+      "loss": 2.811,
+      "step": 7426
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.000130328330206379,
+      "loss": 2.4345,
+      "step": 7427
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.0001303189493433396,
+      "loss": 2.63,
+      "step": 7428
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00013030956848030022,
+      "loss": 2.2859,
+      "step": 7429
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 0.00013030018761726078,
+      "loss": 2.4551,
+      "step": 7430
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00013029080675422137,
+      "loss": 2.4311,
+      "step": 7431
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000130281425891182,
+      "loss": 2.4373,
+      "step": 7432
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00013027204502814258,
+      "loss": 2.5943,
+      "step": 7433
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001302626641651032,
+      "loss": 2.6245,
+      "step": 7434
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001302532833020638,
+      "loss": 2.3441,
+      "step": 7435
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001302439024390244,
+      "loss": 2.6865,
+      "step": 7436
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000130234521575985,
+      "loss": 2.5992,
+      "step": 7437
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00013022514071294562,
+      "loss": 2.3936,
+      "step": 7438
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001302157598499062,
+      "loss": 2.151,
+      "step": 7439
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001302063789868668,
+      "loss": 2.8747,
+      "step": 7440
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001301969981238274,
+      "loss": 2.2854,
+      "step": 7441
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.000130187617260788,
+      "loss": 2.3889,
+      "step": 7442
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001301782363977486,
+      "loss": 2.5399,
+      "step": 7443
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001301688555347092,
+      "loss": 2.4119,
+      "step": 7444
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001301594746716698,
+      "loss": 2.8567,
+      "step": 7445
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001301500938086304,
+      "loss": 2.5123,
+      "step": 7446
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.00013014071294559102,
+      "loss": 2.7354,
+      "step": 7447
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001301313320825516,
+      "loss": 2.6152,
+      "step": 7448
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001301219512195122,
+      "loss": 2.5222,
+      "step": 7449
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001301125703564728,
+      "loss": 2.3019,
+      "step": 7450
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001301031894934334,
+      "loss": 2.6847,
+      "step": 7451
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000130093808630394,
+      "loss": 2.4666,
+      "step": 7452
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00013008442776735462,
+      "loss": 2.6536,
+      "step": 7453
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0001300750469043152,
+      "loss": 2.5018,
+      "step": 7454
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00013006566604127582,
+      "loss": 2.7377,
+      "step": 7455
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00013005628517823642,
+      "loss": 2.8275,
+      "step": 7456
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000130046904315197,
+      "loss": 2.2505,
+      "step": 7457
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0001300375234521576,
+      "loss": 2.3774,
+      "step": 7458
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0001300281425891182,
+      "loss": 2.4949,
+      "step": 7459
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0001300187617260788,
+      "loss": 2.3321,
+      "step": 7460
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0001300093808630394,
+      "loss": 2.3287,
+      "step": 7461
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 2.3233,
+      "step": 7462
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0001299906191369606,
+      "loss": 2.181,
+      "step": 7463
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00012998123827392122,
+      "loss": 2.2549,
+      "step": 7464
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00012997185741088182,
+      "loss": 2.4757,
+      "step": 7465
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0001299624765478424,
+      "loss": 2.7017,
+      "step": 7466
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000129953095684803,
+      "loss": 2.3688,
+      "step": 7467
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00012994371482176362,
+      "loss": 2.3531,
+      "step": 7468
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0001299343339587242,
+      "loss": 2.5474,
+      "step": 7469
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.0001299249530956848,
+      "loss": 2.4436,
+      "step": 7470
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00012991557223264542,
+      "loss": 2.326,
+      "step": 7471
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.000129906191369606,
+      "loss": 2.4822,
+      "step": 7472
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012989681050656662,
+      "loss": 2.1245,
+      "step": 7473
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012988742964352722,
+      "loss": 2.5344,
+      "step": 7474
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0001298780487804878,
+      "loss": 2.495,
+      "step": 7475
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0001298686679174484,
+      "loss": 2.3895,
+      "step": 7476
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012985928705440902,
+      "loss": 2.1156,
+      "step": 7477
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0001298499061913696,
+      "loss": 2.1152,
+      "step": 7478
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012984052532833022,
+      "loss": 2.4216,
+      "step": 7479
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012983114446529082,
+      "loss": 2.6538,
+      "step": 7480
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012982176360225143,
+      "loss": 2.8261,
+      "step": 7481
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012981238273921202,
+      "loss": 2.3846,
+      "step": 7482
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012980300187617262,
+      "loss": 2.4427,
+      "step": 7483
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0001297936210131332,
+      "loss": 2.5723,
+      "step": 7484
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0001297842401500938,
+      "loss": 2.3407,
+      "step": 7485
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012977485928705442,
+      "loss": 2.8666,
+      "step": 7486
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.000129765478424015,
+      "loss": 2.5482,
+      "step": 7487
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012975609756097562,
+      "loss": 2.7381,
+      "step": 7488
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012974671669793622,
+      "loss": 2.6656,
+      "step": 7489
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012973733583489683,
+      "loss": 2.7125,
+      "step": 7490
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012972795497185742,
+      "loss": 2.7706,
+      "step": 7491
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.00012971857410881802,
+      "loss": 2.3516,
+      "step": 7492
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0001297091932457786,
+      "loss": 2.7702,
+      "step": 7493
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012969981238273922,
+      "loss": 2.315,
+      "step": 7494
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012969043151969981,
+      "loss": 2.3554,
+      "step": 7495
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012968105065666043,
+      "loss": 2.713,
+      "step": 7496
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012967166979362102,
+      "loss": 2.4149,
+      "step": 7497
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012966228893058161,
+      "loss": 2.6899,
+      "step": 7498
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012965290806754223,
+      "loss": 2.828,
+      "step": 7499
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012964352720450282,
+      "loss": 2.5223,
+      "step": 7500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012963414634146341,
+      "loss": 2.4487,
+      "step": 7501
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.000129624765478424,
+      "loss": 2.7549,
+      "step": 7502
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012961538461538462,
+      "loss": 2.6333,
+      "step": 7503
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012960600375234521,
+      "loss": 2.5967,
+      "step": 7504
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012959662288930583,
+      "loss": 2.6165,
+      "step": 7505
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012958724202626642,
+      "loss": 2.5264,
+      "step": 7506
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012957786116322704,
+      "loss": 2.2283,
+      "step": 7507
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012956848030018763,
+      "loss": 2.426,
+      "step": 7508
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012955909943714822,
+      "loss": 2.6827,
+      "step": 7509
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012954971857410881,
+      "loss": 2.3483,
+      "step": 7510
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.0001295403377110694,
+      "loss": 2.428,
+      "step": 7511
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012953095684803002,
+      "loss": 2.7684,
+      "step": 7512
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012952157598499061,
+      "loss": 2.2681,
+      "step": 7513
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012951219512195123,
+      "loss": 2.7438,
+      "step": 7514
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 0.00012950281425891182,
+      "loss": 2.497,
+      "step": 7515
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012949343339587244,
+      "loss": 2.3218,
+      "step": 7516
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012948405253283303,
+      "loss": 2.6041,
+      "step": 7517
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012947467166979362,
+      "loss": 2.8284,
+      "step": 7518
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012946529080675421,
+      "loss": 2.2429,
+      "step": 7519
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012945590994371483,
+      "loss": 2.8116,
+      "step": 7520
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012944652908067542,
+      "loss": 2.6137,
+      "step": 7521
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012943714821763604,
+      "loss": 2.7669,
+      "step": 7522
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012942776735459663,
+      "loss": 2.7793,
+      "step": 7523
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012941838649155722,
+      "loss": 2.5581,
+      "step": 7524
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012940900562851784,
+      "loss": 2.3594,
+      "step": 7525
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012939962476547843,
+      "loss": 2.5012,
+      "step": 7526
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012939024390243902,
+      "loss": 2.37,
+      "step": 7527
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012938086303939961,
+      "loss": 2.441,
+      "step": 7528
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012937148217636023,
+      "loss": 2.701,
+      "step": 7529
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012936210131332082,
+      "loss": 2.6708,
+      "step": 7530
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012935272045028144,
+      "loss": 2.5216,
+      "step": 7531
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012934333958724203,
+      "loss": 2.5238,
+      "step": 7532
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012933395872420265,
+      "loss": 2.5095,
+      "step": 7533
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012932457786116324,
+      "loss": 2.5296,
+      "step": 7534
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012931519699812383,
+      "loss": 2.5982,
+      "step": 7535
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.00012930581613508442,
+      "loss": 2.2924,
+      "step": 7536
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012929643527204501,
+      "loss": 2.5189,
+      "step": 7537
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012928705440900563,
+      "loss": 2.3632,
+      "step": 7538
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012927767354596622,
+      "loss": 2.3267,
+      "step": 7539
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012926829268292684,
+      "loss": 2.4282,
+      "step": 7540
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012925891181988743,
+      "loss": 2.432,
+      "step": 7541
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012924953095684805,
+      "loss": 2.7905,
+      "step": 7542
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012924015009380864,
+      "loss": 2.5472,
+      "step": 7543
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012923076923076923,
+      "loss": 2.4729,
+      "step": 7544
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012922138836772982,
+      "loss": 2.6601,
+      "step": 7545
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012921200750469044,
+      "loss": 2.2975,
+      "step": 7546
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012920262664165103,
+      "loss": 2.4096,
+      "step": 7547
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012919324577861165,
+      "loss": 2.4212,
+      "step": 7548
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012918386491557224,
+      "loss": 2.4945,
+      "step": 7549
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012917448405253283,
+      "loss": 2.5667,
+      "step": 7550
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012916510318949345,
+      "loss": 2.6565,
+      "step": 7551
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012915572232645404,
+      "loss": 2.7325,
+      "step": 7552
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012914634146341463,
+      "loss": 2.889,
+      "step": 7553
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012913696060037522,
+      "loss": 2.4468,
+      "step": 7554
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012912757973733584,
+      "loss": 2.6651,
+      "step": 7555
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012911819887429643,
+      "loss": 2.1973,
+      "step": 7556
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00012910881801125705,
+      "loss": 2.3457,
+      "step": 7557
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012909943714821764,
+      "loss": 2.4472,
+      "step": 7558
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012909005628517826,
+      "loss": 2.4691,
+      "step": 7559
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012908067542213885,
+      "loss": 2.3161,
+      "step": 7560
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012907129455909944,
+      "loss": 2.4831,
+      "step": 7561
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012906191369606003,
+      "loss": 2.0994,
+      "step": 7562
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012905253283302062,
+      "loss": 2.4038,
+      "step": 7563
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012904315196998124,
+      "loss": 2.7602,
+      "step": 7564
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012903377110694183,
+      "loss": 2.6964,
+      "step": 7565
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012902439024390245,
+      "loss": 2.5726,
+      "step": 7566
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012901500938086304,
+      "loss": 2.4898,
+      "step": 7567
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012900562851782366,
+      "loss": 2.2751,
+      "step": 7568
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012899624765478425,
+      "loss": 2.4977,
+      "step": 7569
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012898686679174484,
+      "loss": 2.6131,
+      "step": 7570
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012897748592870543,
+      "loss": 2.4517,
+      "step": 7571
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012896810506566605,
+      "loss": 2.2859,
+      "step": 7572
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012895872420262664,
+      "loss": 2.6493,
+      "step": 7573
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012894934333958726,
+      "loss": 2.5257,
+      "step": 7574
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012893996247654785,
+      "loss": 2.5004,
+      "step": 7575
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012893058161350844,
+      "loss": 2.4455,
+      "step": 7576
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012892120075046906,
+      "loss": 2.3999,
+      "step": 7577
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012891181988742965,
+      "loss": 2.3943,
+      "step": 7578
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00012890243902439024,
+      "loss": 2.2519,
+      "step": 7579
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012889305816135083,
+      "loss": 2.5681,
+      "step": 7580
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012888367729831145,
+      "loss": 2.2371,
+      "step": 7581
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012887429643527204,
+      "loss": 2.5321,
+      "step": 7582
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012886491557223266,
+      "loss": 2.6023,
+      "step": 7583
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012885553470919325,
+      "loss": 2.439,
+      "step": 7584
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012884615384615387,
+      "loss": 2.4191,
+      "step": 7585
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012883677298311446,
+      "loss": 2.3106,
+      "step": 7586
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012882739212007508,
+      "loss": 2.6036,
+      "step": 7587
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012881801125703564,
+      "loss": 2.5827,
+      "step": 7588
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012880863039399623,
+      "loss": 2.4117,
+      "step": 7589
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012879924953095685,
+      "loss": 2.6052,
+      "step": 7590
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012878986866791744,
+      "loss": 2.393,
+      "step": 7591
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012878048780487806,
+      "loss": 2.3373,
+      "step": 7592
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012877110694183865,
+      "loss": 2.4582,
+      "step": 7593
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012876172607879927,
+      "loss": 2.4558,
+      "step": 7594
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012875234521575986,
+      "loss": 2.5558,
+      "step": 7595
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012874296435272048,
+      "loss": 2.2839,
+      "step": 7596
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012873358348968104,
+      "loss": 2.101,
+      "step": 7597
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012872420262664166,
+      "loss": 2.6251,
+      "step": 7598
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012871482176360225,
+      "loss": 2.602,
+      "step": 7599
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00012870544090056287,
+      "loss": 2.7056,
+      "step": 7600
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012869606003752346,
+      "loss": 2.3639,
+      "step": 7601
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012868667917448405,
+      "loss": 2.6189,
+      "step": 7602
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012867729831144467,
+      "loss": 2.3492,
+      "step": 7603
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012866791744840526,
+      "loss": 2.5302,
+      "step": 7604
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012865853658536588,
+      "loss": 2.6436,
+      "step": 7605
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012864915572232647,
+      "loss": 2.6884,
+      "step": 7606
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012863977485928706,
+      "loss": 2.2031,
+      "step": 7607
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012863039399624765,
+      "loss": 2.5128,
+      "step": 7608
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012862101313320827,
+      "loss": 2.4935,
+      "step": 7609
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012861163227016886,
+      "loss": 2.7191,
+      "step": 7610
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012860225140712948,
+      "loss": 2.3727,
+      "step": 7611
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012859287054409007,
+      "loss": 2.3158,
+      "step": 7612
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012858348968105068,
+      "loss": 2.3859,
+      "step": 7613
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012857410881801128,
+      "loss": 2.285,
+      "step": 7614
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012856472795497187,
+      "loss": 2.2624,
+      "step": 7615
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012855534709193246,
+      "loss": 2.317,
+      "step": 7616
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012854596622889305,
+      "loss": 2.6448,
+      "step": 7617
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012853658536585367,
+      "loss": 2.3903,
+      "step": 7618
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012852720450281426,
+      "loss": 2.6603,
+      "step": 7619
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012851782363977488,
+      "loss": 2.5284,
+      "step": 7620
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 0.00012850844277673547,
+      "loss": 2.5933,
+      "step": 7621
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012849906191369608,
+      "loss": 2.6293,
+      "step": 7622
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012848968105065668,
+      "loss": 2.5604,
+      "step": 7623
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012848030018761727,
+      "loss": 2.7935,
+      "step": 7624
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012847091932457786,
+      "loss": 2.4527,
+      "step": 7625
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012846153846153848,
+      "loss": 2.5698,
+      "step": 7626
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012845215759849907,
+      "loss": 2.3973,
+      "step": 7627
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012844277673545966,
+      "loss": 2.6072,
+      "step": 7628
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012843339587242028,
+      "loss": 2.4453,
+      "step": 7629
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012842401500938087,
+      "loss": 2.4596,
+      "step": 7630
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012841463414634148,
+      "loss": 2.3637,
+      "step": 7631
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012840525328330208,
+      "loss": 2.4998,
+      "step": 7632
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012839587242026267,
+      "loss": 2.6757,
+      "step": 7633
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012838649155722326,
+      "loss": 2.5482,
+      "step": 7634
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012837711069418388,
+      "loss": 2.5489,
+      "step": 7635
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012836772983114447,
+      "loss": 2.8765,
+      "step": 7636
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012835834896810508,
+      "loss": 2.5257,
+      "step": 7637
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012834896810506568,
+      "loss": 2.5323,
+      "step": 7638
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0001283395872420263,
+      "loss": 2.473,
+      "step": 7639
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012833020637898688,
+      "loss": 2.4981,
+      "step": 7640
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012832082551594747,
+      "loss": 2.4771,
+      "step": 7641
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012831144465290807,
+      "loss": 2.8968,
+      "step": 7642
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.00012830206378986866,
+      "loss": 2.5725,
+      "step": 7643
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012829268292682927,
+      "loss": 2.6145,
+      "step": 7644
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012828330206378987,
+      "loss": 2.5778,
+      "step": 7645
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012827392120075048,
+      "loss": 2.5114,
+      "step": 7646
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012826454033771107,
+      "loss": 2.273,
+      "step": 7647
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0001282551594746717,
+      "loss": 2.6315,
+      "step": 7648
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012824577861163228,
+      "loss": 2.5312,
+      "step": 7649
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012823639774859287,
+      "loss": 2.4542,
+      "step": 7650
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012822701688555347,
+      "loss": 2.4536,
+      "step": 7651
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012821763602251408,
+      "loss": 2.4596,
+      "step": 7652
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012820825515947467,
+      "loss": 2.3862,
+      "step": 7653
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012819887429643527,
+      "loss": 2.7119,
+      "step": 7654
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012818949343339588,
+      "loss": 2.3821,
+      "step": 7655
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012818011257035647,
+      "loss": 2.4835,
+      "step": 7656
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0001281707317073171,
+      "loss": 2.6319,
+      "step": 7657
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012816135084427768,
+      "loss": 2.4767,
+      "step": 7658
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012815196998123827,
+      "loss": 2.6964,
+      "step": 7659
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012814258911819887,
+      "loss": 2.4031,
+      "step": 7660
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012813320825515948,
+      "loss": 2.5307,
+      "step": 7661
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012812382739212007,
+      "loss": 2.3773,
+      "step": 7662
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.0001281144465290807,
+      "loss": 2.5019,
+      "step": 7663
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00012810506566604128,
+      "loss": 2.6134,
+      "step": 7664
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0001280956848030019,
+      "loss": 2.5136,
+      "step": 7665
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0001280863039399625,
+      "loss": 2.4215,
+      "step": 7666
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012807692307692308,
+      "loss": 2.4087,
+      "step": 7667
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012806754221388367,
+      "loss": 2.6925,
+      "step": 7668
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012805816135084427,
+      "loss": 2.371,
+      "step": 7669
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012804878048780488,
+      "loss": 2.6257,
+      "step": 7670
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012803939962476547,
+      "loss": 2.4938,
+      "step": 7671
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0001280300187617261,
+      "loss": 2.3625,
+      "step": 7672
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012802063789868668,
+      "loss": 2.1509,
+      "step": 7673
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0001280112570356473,
+      "loss": 2.5128,
+      "step": 7674
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0001280018761726079,
+      "loss": 2.443,
+      "step": 7675
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012799249530956848,
+      "loss": 2.6064,
+      "step": 7676
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012798311444652907,
+      "loss": 2.4614,
+      "step": 7677
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0001279737335834897,
+      "loss": 2.322,
+      "step": 7678
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012796435272045028,
+      "loss": 2.4278,
+      "step": 7679
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012795497185741087,
+      "loss": 2.5071,
+      "step": 7680
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0001279455909943715,
+      "loss": 2.4756,
+      "step": 7681
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012793621013133208,
+      "loss": 2.3906,
+      "step": 7682
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0001279268292682927,
+      "loss": 2.5678,
+      "step": 7683
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.0001279174484052533,
+      "loss": 2.5455,
+      "step": 7684
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00012790806754221388,
+      "loss": 2.696,
+      "step": 7685
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00012789868667917447,
+      "loss": 2.6394,
+      "step": 7686
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001278893058161351,
+      "loss": 2.6134,
+      "step": 7687
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00012787992495309568,
+      "loss": 2.694,
+      "step": 7688
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001278705440900563,
+      "loss": 2.7616,
+      "step": 7689
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001278611632270169,
+      "loss": 2.5648,
+      "step": 7690
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001278517823639775,
+      "loss": 2.2299,
+      "step": 7691
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001278424015009381,
+      "loss": 2.5804,
+      "step": 7692
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001278330206378987,
+      "loss": 2.7471,
+      "step": 7693
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00012782363977485928,
+      "loss": 2.2991,
+      "step": 7694
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00012781425891181987,
+      "loss": 2.5672,
+      "step": 7695
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001278048780487805,
+      "loss": 2.5869,
+      "step": 7696
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00012779549718574108,
+      "loss": 2.2393,
+      "step": 7697
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001277861163227017,
+      "loss": 2.5793,
+      "step": 7698
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001277767354596623,
+      "loss": 2.3219,
+      "step": 7699
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001277673545966229,
+      "loss": 2.6797,
+      "step": 7700
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001277579737335835,
+      "loss": 2.4601,
+      "step": 7701
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001277485928705441,
+      "loss": 2.4216,
+      "step": 7702
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.00012773921200750468,
+      "loss": 2.6192,
+      "step": 7703
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001277298311444653,
+      "loss": 2.1794,
+      "step": 7704
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001277204502814259,
+      "loss": 2.5551,
+      "step": 7705
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001277110694183865,
+      "loss": 2.6289,
+      "step": 7706
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 0.0001277016885553471,
+      "loss": 2.4465,
+      "step": 7707
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001276923076923077,
+      "loss": 2.4852,
+      "step": 7708
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001276829268292683,
+      "loss": 2.6144,
+      "step": 7709
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001276735459662289,
+      "loss": 2.5495,
+      "step": 7710
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001276641651031895,
+      "loss": 2.2776,
+      "step": 7711
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00012765478424015008,
+      "loss": 2.5004,
+      "step": 7712
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001276454033771107,
+      "loss": 2.4607,
+      "step": 7713
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001276360225140713,
+      "loss": 2.4676,
+      "step": 7714
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001276266416510319,
+      "loss": 2.5729,
+      "step": 7715
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001276172607879925,
+      "loss": 2.6596,
+      "step": 7716
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00012760787992495312,
+      "loss": 2.4819,
+      "step": 7717
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001275984990619137,
+      "loss": 2.2294,
+      "step": 7718
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001275891181988743,
+      "loss": 2.601,
+      "step": 7719
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001275797373358349,
+      "loss": 2.2139,
+      "step": 7720
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00012757035647279548,
+      "loss": 2.643,
+      "step": 7721
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001275609756097561,
+      "loss": 2.7049,
+      "step": 7722
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001275515947467167,
+      "loss": 2.3787,
+      "step": 7723
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001275422138836773,
+      "loss": 2.6235,
+      "step": 7724
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001275328330206379,
+      "loss": 2.6063,
+      "step": 7725
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.00012752345215759852,
+      "loss": 2.3241,
+      "step": 7726
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001275140712945591,
+      "loss": 2.643,
+      "step": 7727
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001275046904315197,
+      "loss": 2.3856,
+      "step": 7728
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001274953095684803,
+      "loss": 2.2018,
+      "step": 7729
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001274859287054409,
+      "loss": 2.3597,
+      "step": 7730
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001274765478424015,
+      "loss": 2.5415,
+      "step": 7731
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00012746716697936212,
+      "loss": 2.1788,
+      "step": 7732
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001274577861163227,
+      "loss": 2.6481,
+      "step": 7733
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001274484052532833,
+      "loss": 2.5146,
+      "step": 7734
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00012743902439024392,
+      "loss": 2.6056,
+      "step": 7735
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001274296435272045,
+      "loss": 2.7224,
+      "step": 7736
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001274202626641651,
+      "loss": 2.4937,
+      "step": 7737
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001274108818011257,
+      "loss": 2.1891,
+      "step": 7738
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001274015009380863,
+      "loss": 2.4111,
+      "step": 7739
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001273921200750469,
+      "loss": 2.6564,
+      "step": 7740
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00012738273921200752,
+      "loss": 2.5034,
+      "step": 7741
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001273733583489681,
+      "loss": 2.4417,
+      "step": 7742
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00012736397748592873,
+      "loss": 2.6369,
+      "step": 7743
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00012735459662288932,
+      "loss": 2.6034,
+      "step": 7744
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00012734521575984994,
+      "loss": 2.7439,
+      "step": 7745
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001273358348968105,
+      "loss": 2.4858,
+      "step": 7746
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001273264540337711,
+      "loss": 2.4763,
+      "step": 7747
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001273170731707317,
+      "loss": 2.6734,
+      "step": 7748
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.0001273076923076923,
+      "loss": 2.4683,
+      "step": 7749
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012729831144465292,
+      "loss": 2.4577,
+      "step": 7750
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0001272889305816135,
+      "loss": 2.5023,
+      "step": 7751
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012727954971857413,
+      "loss": 2.7404,
+      "step": 7752
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012727016885553472,
+      "loss": 2.4992,
+      "step": 7753
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012726078799249534,
+      "loss": 2.3667,
+      "step": 7754
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0001272514071294559,
+      "loss": 2.4716,
+      "step": 7755
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012724202626641652,
+      "loss": 2.5343,
+      "step": 7756
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0001272326454033771,
+      "loss": 2.6662,
+      "step": 7757
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012722326454033773,
+      "loss": 2.7364,
+      "step": 7758
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012721388367729832,
+      "loss": 2.7785,
+      "step": 7759
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0001272045028142589,
+      "loss": 2.4603,
+      "step": 7760
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012719512195121953,
+      "loss": 2.5449,
+      "step": 7761
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012718574108818012,
+      "loss": 2.6195,
+      "step": 7762
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012717636022514074,
+      "loss": 2.3985,
+      "step": 7763
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0001271669793621013,
+      "loss": 2.5428,
+      "step": 7764
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012715759849906192,
+      "loss": 2.4255,
+      "step": 7765
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.0001271482176360225,
+      "loss": 2.7086,
+      "step": 7766
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012713883677298313,
+      "loss": 2.7669,
+      "step": 7767
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012712945590994372,
+      "loss": 2.5557,
+      "step": 7768
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012712007504690434,
+      "loss": 2.4917,
+      "step": 7769
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012711069418386493,
+      "loss": 2.5617,
+      "step": 7770
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00012710131332082554,
+      "loss": 2.284,
+      "step": 7771
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012709193245778614,
+      "loss": 2.3558,
+      "step": 7772
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0001270825515947467,
+      "loss": 2.7508,
+      "step": 7773
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012707317073170732,
+      "loss": 2.4941,
+      "step": 7774
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.0001270637898686679,
+      "loss": 2.5026,
+      "step": 7775
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012705440900562853,
+      "loss": 2.6004,
+      "step": 7776
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012704502814258912,
+      "loss": 2.2735,
+      "step": 7777
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012703564727954974,
+      "loss": 2.6293,
+      "step": 7778
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012702626641651033,
+      "loss": 2.6203,
+      "step": 7779
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012701688555347094,
+      "loss": 2.4665,
+      "step": 7780
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012700750469043154,
+      "loss": 2.5387,
+      "step": 7781
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012699812382739213,
+      "loss": 2.3734,
+      "step": 7782
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012698874296435272,
+      "loss": 2.6467,
+      "step": 7783
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012697936210131334,
+      "loss": 2.5621,
+      "step": 7784
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012696998123827393,
+      "loss": 2.4126,
+      "step": 7785
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012696060037523452,
+      "loss": 2.5115,
+      "step": 7786
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012695121951219513,
+      "loss": 2.5808,
+      "step": 7787
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012694183864915573,
+      "loss": 2.5581,
+      "step": 7788
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012693245778611634,
+      "loss": 2.6819,
+      "step": 7789
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012692307692307693,
+      "loss": 2.2324,
+      "step": 7790
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012691369606003753,
+      "loss": 2.4795,
+      "step": 7791
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00012690431519699812,
+      "loss": 2.5978,
+      "step": 7792
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012689493433395873,
+      "loss": 2.3716,
+      "step": 7793
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012688555347091933,
+      "loss": 2.6723,
+      "step": 7794
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012687617260787994,
+      "loss": 2.5099,
+      "step": 7795
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012686679174484053,
+      "loss": 2.3184,
+      "step": 7796
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012685741088180115,
+      "loss": 2.4723,
+      "step": 7797
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012684803001876174,
+      "loss": 2.4077,
+      "step": 7798
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012683864915572233,
+      "loss": 2.5549,
+      "step": 7799
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012682926829268293,
+      "loss": 2.4236,
+      "step": 7800
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012681988742964352,
+      "loss": 2.3808,
+      "step": 7801
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012681050656660413,
+      "loss": 2.617,
+      "step": 7802
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012680112570356473,
+      "loss": 2.1576,
+      "step": 7803
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012679174484052534,
+      "loss": 2.765,
+      "step": 7804
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012678236397748593,
+      "loss": 2.6999,
+      "step": 7805
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012677298311444655,
+      "loss": 2.3733,
+      "step": 7806
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012676360225140714,
+      "loss": 2.4704,
+      "step": 7807
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012675422138836773,
+      "loss": 2.6868,
+      "step": 7808
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012674484052532833,
+      "loss": 2.5501,
+      "step": 7809
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012673545966228894,
+      "loss": 2.444,
+      "step": 7810
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012672607879924953,
+      "loss": 2.4848,
+      "step": 7811
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012671669793621013,
+      "loss": 2.4867,
+      "step": 7812
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 0.00012670731707317074,
+      "loss": 2.5042,
+      "step": 7813
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012669793621013133,
+      "loss": 2.4368,
+      "step": 7814
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012668855534709195,
+      "loss": 2.4273,
+      "step": 7815
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012667917448405254,
+      "loss": 2.7371,
+      "step": 7816
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012666979362101313,
+      "loss": 2.6052,
+      "step": 7817
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012666041275797373,
+      "loss": 2.4164,
+      "step": 7818
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012665103189493434,
+      "loss": 2.4016,
+      "step": 7819
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012664165103189493,
+      "loss": 2.5138,
+      "step": 7820
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012663227016885555,
+      "loss": 2.5209,
+      "step": 7821
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012662288930581614,
+      "loss": 2.5558,
+      "step": 7822
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012661350844277676,
+      "loss": 2.4332,
+      "step": 7823
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012660412757973735,
+      "loss": 2.367,
+      "step": 7824
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012659474671669794,
+      "loss": 2.7912,
+      "step": 7825
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012658536585365853,
+      "loss": 2.6477,
+      "step": 7826
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012657598499061913,
+      "loss": 2.6535,
+      "step": 7827
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012656660412757974,
+      "loss": 3.1523,
+      "step": 7828
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012655722326454033,
+      "loss": 2.8209,
+      "step": 7829
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012654784240150095,
+      "loss": 2.6184,
+      "step": 7830
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012653846153846154,
+      "loss": 2.798,
+      "step": 7831
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012652908067542216,
+      "loss": 2.5031,
+      "step": 7832
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012651969981238275,
+      "loss": 2.2351,
+      "step": 7833
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012651031894934334,
+      "loss": 2.3842,
+      "step": 7834
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.00012650093808630393,
+      "loss": 2.781,
+      "step": 7835
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012649155722326455,
+      "loss": 2.6925,
+      "step": 7836
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012648217636022514,
+      "loss": 2.4781,
+      "step": 7837
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012647279549718573,
+      "loss": 2.0533,
+      "step": 7838
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012646341463414635,
+      "loss": 2.4876,
+      "step": 7839
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012645403377110694,
+      "loss": 2.2061,
+      "step": 7840
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012644465290806756,
+      "loss": 2.4074,
+      "step": 7841
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012643527204502815,
+      "loss": 2.5078,
+      "step": 7842
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012642589118198874,
+      "loss": 2.6844,
+      "step": 7843
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012641651031894933,
+      "loss": 2.4054,
+      "step": 7844
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012640712945590995,
+      "loss": 2.4118,
+      "step": 7845
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012639774859287054,
+      "loss": 2.2974,
+      "step": 7846
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012638836772983116,
+      "loss": 2.3749,
+      "step": 7847
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012637898686679175,
+      "loss": 2.5664,
+      "step": 7848
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012636960600375237,
+      "loss": 2.3041,
+      "step": 7849
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012636022514071296,
+      "loss": 2.6149,
+      "step": 7850
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012635084427767355,
+      "loss": 2.4171,
+      "step": 7851
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012634146341463414,
+      "loss": 2.6966,
+      "step": 7852
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012633208255159473,
+      "loss": 2.329,
+      "step": 7853
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012632270168855535,
+      "loss": 2.3164,
+      "step": 7854
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012631332082551594,
+      "loss": 2.5279,
+      "step": 7855
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00012630393996247656,
+      "loss": 2.3958,
+      "step": 7856
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012629455909943715,
+      "loss": 2.3688,
+      "step": 7857
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012628517823639777,
+      "loss": 2.3018,
+      "step": 7858
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012627579737335836,
+      "loss": 2.2717,
+      "step": 7859
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012626641651031895,
+      "loss": 2.588,
+      "step": 7860
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012625703564727954,
+      "loss": 2.7086,
+      "step": 7861
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012624765478424016,
+      "loss": 2.1913,
+      "step": 7862
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012623827392120075,
+      "loss": 2.5327,
+      "step": 7863
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012622889305816134,
+      "loss": 2.7927,
+      "step": 7864
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012621951219512196,
+      "loss": 2.8957,
+      "step": 7865
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012621013133208255,
+      "loss": 2.4133,
+      "step": 7866
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012620075046904317,
+      "loss": 2.7109,
+      "step": 7867
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012619136960600376,
+      "loss": 2.5043,
+      "step": 7868
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012618198874296435,
+      "loss": 2.7374,
+      "step": 7869
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012617260787992494,
+      "loss": 2.4116,
+      "step": 7870
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012616322701688556,
+      "loss": 2.5687,
+      "step": 7871
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012615384615384615,
+      "loss": 2.4095,
+      "step": 7872
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012614446529080677,
+      "loss": 2.5739,
+      "step": 7873
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012613508442776736,
+      "loss": 2.4256,
+      "step": 7874
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012612570356472798,
+      "loss": 2.4962,
+      "step": 7875
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012611632270168857,
+      "loss": 2.628,
+      "step": 7876
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00012610694183864916,
+      "loss": 2.5342,
+      "step": 7877
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012609756097560975,
+      "loss": 2.5526,
+      "step": 7878
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012608818011257034,
+      "loss": 2.5515,
+      "step": 7879
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012607879924953096,
+      "loss": 2.631,
+      "step": 7880
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012606941838649155,
+      "loss": 2.3859,
+      "step": 7881
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012606003752345217,
+      "loss": 2.5075,
+      "step": 7882
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012605065666041276,
+      "loss": 2.3192,
+      "step": 7883
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012604127579737338,
+      "loss": 2.4928,
+      "step": 7884
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012603189493433397,
+      "loss": 2.7768,
+      "step": 7885
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012602251407129456,
+      "loss": 2.696,
+      "step": 7886
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012601313320825515,
+      "loss": 2.6774,
+      "step": 7887
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012600375234521577,
+      "loss": 2.7887,
+      "step": 7888
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012599437148217636,
+      "loss": 2.4842,
+      "step": 7889
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012598499061913695,
+      "loss": 2.1728,
+      "step": 7890
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012597560975609757,
+      "loss": 2.4395,
+      "step": 7891
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012596622889305816,
+      "loss": 2.4791,
+      "step": 7892
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012595684803001878,
+      "loss": 2.4642,
+      "step": 7893
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012594746716697937,
+      "loss": 2.6104,
+      "step": 7894
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012593808630393996,
+      "loss": 2.7404,
+      "step": 7895
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012592870544090055,
+      "loss": 2.4268,
+      "step": 7896
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012591932457786117,
+      "loss": 2.5997,
+      "step": 7897
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012590994371482176,
+      "loss": 2.1536,
+      "step": 7898
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 0.00012590056285178238,
+      "loss": 2.7447,
+      "step": 7899
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012589118198874297,
+      "loss": 2.6973,
+      "step": 7900
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0001258818011257036,
+      "loss": 2.5721,
+      "step": 7901
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012587242026266418,
+      "loss": 2.3775,
+      "step": 7902
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0001258630393996248,
+      "loss": 2.2949,
+      "step": 7903
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012585365853658536,
+      "loss": 2.4069,
+      "step": 7904
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012584427767354595,
+      "loss": 2.393,
+      "step": 7905
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012583489681050657,
+      "loss": 2.7123,
+      "step": 7906
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012582551594746716,
+      "loss": 2.4163,
+      "step": 7907
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012581613508442778,
+      "loss": 2.3228,
+      "step": 7908
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012580675422138837,
+      "loss": 2.4182,
+      "step": 7909
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012579737335834899,
+      "loss": 2.3332,
+      "step": 7910
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012578799249530958,
+      "loss": 2.8288,
+      "step": 7911
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0001257786116322702,
+      "loss": 2.6083,
+      "step": 7912
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012576923076923076,
+      "loss": 2.6648,
+      "step": 7913
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012575984990619138,
+      "loss": 2.4962,
+      "step": 7914
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012575046904315197,
+      "loss": 2.546,
+      "step": 7915
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012574108818011259,
+      "loss": 2.6043,
+      "step": 7916
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012573170731707318,
+      "loss": 2.4599,
+      "step": 7917
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012572232645403377,
+      "loss": 2.4483,
+      "step": 7918
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012571294559099439,
+      "loss": 2.1927,
+      "step": 7919
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.00012570356472795498,
+      "loss": 2.8088,
+      "step": 7920
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0001256941838649156,
+      "loss": 2.4427,
+      "step": 7921
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012568480300187616,
+      "loss": 2.6559,
+      "step": 7922
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012567542213883678,
+      "loss": 2.5967,
+      "step": 7923
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012566604127579737,
+      "loss": 2.29,
+      "step": 7924
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012565666041275799,
+      "loss": 2.4849,
+      "step": 7925
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012564727954971858,
+      "loss": 2.2487,
+      "step": 7926
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0001256378986866792,
+      "loss": 2.4631,
+      "step": 7927
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012562851782363979,
+      "loss": 2.5453,
+      "step": 7928
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0001256191369606004,
+      "loss": 2.1628,
+      "step": 7929
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.000125609756097561,
+      "loss": 2.6226,
+      "step": 7930
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012560037523452156,
+      "loss": 2.8858,
+      "step": 7931
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012559099437148218,
+      "loss": 2.5256,
+      "step": 7932
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012558161350844277,
+      "loss": 2.5506,
+      "step": 7933
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012557223264540339,
+      "loss": 2.4618,
+      "step": 7934
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012556285178236398,
+      "loss": 2.5234,
+      "step": 7935
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0001255534709193246,
+      "loss": 2.2613,
+      "step": 7936
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012554409005628519,
+      "loss": 2.4678,
+      "step": 7937
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0001255347091932458,
+      "loss": 2.667,
+      "step": 7938
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.0001255253283302064,
+      "loss": 2.4813,
+      "step": 7939
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012551594746716699,
+      "loss": 2.4959,
+      "step": 7940
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00012550656660412758,
+      "loss": 2.5259,
+      "step": 7941
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0001254971857410882,
+      "loss": 2.5414,
+      "step": 7942
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00012548780487804879,
+      "loss": 2.6567,
+      "step": 7943
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00012547842401500938,
+      "loss": 2.5514,
+      "step": 7944
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00012546904315197,
+      "loss": 2.4826,
+      "step": 7945
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00012545966228893059,
+      "loss": 2.3704,
+      "step": 7946
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0001254502814258912,
+      "loss": 2.4762,
+      "step": 7947
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0001254409005628518,
+      "loss": 2.5986,
+      "step": 7948
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00012543151969981239,
+      "loss": 2.3702,
+      "step": 7949
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00012542213883677298,
+      "loss": 2.4663,
+      "step": 7950
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0001254127579737336,
+      "loss": 2.4796,
+      "step": 7951
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00012540337711069419,
+      "loss": 2.7394,
+      "step": 7952
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0001253939962476548,
+      "loss": 2.5717,
+      "step": 7953
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0001253846153846154,
+      "loss": 2.3552,
+      "step": 7954
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.000125375234521576,
+      "loss": 2.4703,
+      "step": 7955
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0001253658536585366,
+      "loss": 2.6138,
+      "step": 7956
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0001253564727954972,
+      "loss": 2.5771,
+      "step": 7957
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00012534709193245779,
+      "loss": 2.5671,
+      "step": 7958
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00012533771106941838,
+      "loss": 2.347,
+      "step": 7959
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.000125328330206379,
+      "loss": 2.663,
+      "step": 7960
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00012531894934333959,
+      "loss": 2.5751,
+      "step": 7961
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0001253095684803002,
+      "loss": 2.5811,
+      "step": 7962
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.0001253001876172608,
+      "loss": 2.2643,
+      "step": 7963
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001252908067542214,
+      "loss": 2.3708,
+      "step": 7964
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.000125281425891182,
+      "loss": 2.5937,
+      "step": 7965
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001252720450281426,
+      "loss": 2.4706,
+      "step": 7966
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00012526266416510319,
+      "loss": 2.5686,
+      "step": 7967
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001252532833020638,
+      "loss": 2.9176,
+      "step": 7968
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001252439024390244,
+      "loss": 2.491,
+      "step": 7969
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00012523452157598499,
+      "loss": 2.32,
+      "step": 7970
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001252251407129456,
+      "loss": 2.5501,
+      "step": 7971
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001252157598499062,
+      "loss": 2.735,
+      "step": 7972
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001252063789868668,
+      "loss": 2.3552,
+      "step": 7973
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001251969981238274,
+      "loss": 2.5591,
+      "step": 7974
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.000125187617260788,
+      "loss": 2.4031,
+      "step": 7975
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00012517823639774858,
+      "loss": 2.513,
+      "step": 7976
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001251688555347092,
+      "loss": 2.4021,
+      "step": 7977
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001251594746716698,
+      "loss": 2.1497,
+      "step": 7978
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001251500938086304,
+      "loss": 2.464,
+      "step": 7979
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.000125140712945591,
+      "loss": 2.3945,
+      "step": 7980
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00012513133208255162,
+      "loss": 2.3465,
+      "step": 7981
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001251219512195122,
+      "loss": 2.4201,
+      "step": 7982
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001251125703564728,
+      "loss": 2.4946,
+      "step": 7983
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.0001251031894934334,
+      "loss": 2.5527,
+      "step": 7984
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00012509380863039398,
+      "loss": 2.6482,
+      "step": 7985
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001250844277673546,
+      "loss": 2.5285,
+      "step": 7986
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001250750469043152,
+      "loss": 2.2569,
+      "step": 7987
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001250656660412758,
+      "loss": 2.7889,
+      "step": 7988
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001250562851782364,
+      "loss": 2.307,
+      "step": 7989
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00012504690431519702,
+      "loss": 2.5441,
+      "step": 7990
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001250375234521576,
+      "loss": 2.907,
+      "step": 7991
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001250281425891182,
+      "loss": 2.3156,
+      "step": 7992
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001250187617260788,
+      "loss": 2.4308,
+      "step": 7993
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001250093808630394,
+      "loss": 2.3228,
+      "step": 7994
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.000125,
+      "loss": 2.3377,
+      "step": 7995
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001249906191369606,
+      "loss": 2.5237,
+      "step": 7996
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001249812382739212,
+      "loss": 2.573,
+      "step": 7997
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001249718574108818,
+      "loss": 2.3316,
+      "step": 7998
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00012496247654784242,
+      "loss": 2.323,
+      "step": 7999
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.000124953095684803,
+      "loss": 2.2693,
+      "step": 8000
+    },
+    {
+      "epoch": 3.75,
+      "eval_loss": 2.606384038925171,
+      "eval_runtime": 67.6984,
+      "eval_samples_per_second": 27.992,
+      "eval_steps_per_second": 3.501,
+      "step": 8000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001249437148217636,
+      "loss": 2.536,
+      "step": 8001
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001249343339587242,
+      "loss": 2.4118,
+      "step": 8002
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001249249530956848,
+      "loss": 2.6276,
+      "step": 8003
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.0001249155722326454,
+      "loss": 2.6681,
+      "step": 8004
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 0.00012490619136960602,
+      "loss": 2.7068,
+      "step": 8005
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0001248968105065666,
+      "loss": 2.7921,
+      "step": 8006
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00012488742964352723,
+      "loss": 2.6041,
+      "step": 8007
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00012487804878048782,
+      "loss": 2.4601,
+      "step": 8008
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0001248686679174484,
+      "loss": 2.4463,
+      "step": 8009
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.000124859287054409,
+      "loss": 2.3046,
+      "step": 8010
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0001248499061913696,
+      "loss": 2.5629,
+      "step": 8011
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0001248405253283302,
+      "loss": 2.5214,
+      "step": 8012
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0001248311444652908,
+      "loss": 2.5403,
+      "step": 8013
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00012482176360225142,
+      "loss": 2.5179,
+      "step": 8014
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.000124812382739212,
+      "loss": 2.4495,
+      "step": 8015
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00012480300187617263,
+      "loss": 2.439,
+      "step": 8016
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00012479362101313322,
+      "loss": 2.6331,
+      "step": 8017
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0001247842401500938,
+      "loss": 2.5547,
+      "step": 8018
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0001247748592870544,
+      "loss": 2.318,
+      "step": 8019
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00012476547842401502,
+      "loss": 2.5278,
+      "step": 8020
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0001247560975609756,
+      "loss": 2.4697,
+      "step": 8021
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0001247467166979362,
+      "loss": 2.703,
+      "step": 8022
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00012473733583489682,
+      "loss": 2.3417,
+      "step": 8023
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.0001247279549718574,
+      "loss": 2.4527,
+      "step": 8024
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00012471857410881803,
+      "loss": 2.3451,
+      "step": 8025
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00012470919324577862,
+      "loss": 2.4863,
+      "step": 8026
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0001246998123827392,
+      "loss": 2.7478,
+      "step": 8027
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0001246904315196998,
+      "loss": 2.544,
+      "step": 8028
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012468105065666042,
+      "loss": 2.3281,
+      "step": 8029
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.000124671669793621,
+      "loss": 2.77,
+      "step": 8030
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012466228893058163,
+      "loss": 2.487,
+      "step": 8031
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012465290806754222,
+      "loss": 2.3991,
+      "step": 8032
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012464352720450284,
+      "loss": 2.6982,
+      "step": 8033
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012463414634146343,
+      "loss": 2.4892,
+      "step": 8034
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012462476547842402,
+      "loss": 2.6184,
+      "step": 8035
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0001246153846153846,
+      "loss": 2.3565,
+      "step": 8036
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0001246060037523452,
+      "loss": 2.4531,
+      "step": 8037
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012459662288930582,
+      "loss": 2.5905,
+      "step": 8038
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0001245872420262664,
+      "loss": 2.4796,
+      "step": 8039
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012457786116322703,
+      "loss": 2.2736,
+      "step": 8040
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012456848030018762,
+      "loss": 2.2372,
+      "step": 8041
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012455909943714824,
+      "loss": 2.4399,
+      "step": 8042
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012454971857410883,
+      "loss": 2.2548,
+      "step": 8043
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012454033771106942,
+      "loss": 2.294,
+      "step": 8044
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012453095684803,
+      "loss": 2.4197,
+      "step": 8045
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012452157598499063,
+      "loss": 2.6304,
+      "step": 8046
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00012451219512195122,
+      "loss": 2.1936,
+      "step": 8047
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.0001245028142589118,
+      "loss": 2.3367,
+      "step": 8048
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012449343339587243,
+      "loss": 2.5038,
+      "step": 8049
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012448405253283302,
+      "loss": 2.2476,
+      "step": 8050
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012447467166979364,
+      "loss": 2.5827,
+      "step": 8051
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012446529080675423,
+      "loss": 2.3706,
+      "step": 8052
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012445590994371482,
+      "loss": 2.2084,
+      "step": 8053
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0001244465290806754,
+      "loss": 2.6935,
+      "step": 8054
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012443714821763603,
+      "loss": 2.5221,
+      "step": 8055
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012442776735459662,
+      "loss": 2.4109,
+      "step": 8056
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012441838649155724,
+      "loss": 2.4281,
+      "step": 8057
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012440900562851783,
+      "loss": 2.3678,
+      "step": 8058
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012439962476547845,
+      "loss": 2.4057,
+      "step": 8059
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012439024390243904,
+      "loss": 2.5054,
+      "step": 8060
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012438086303939963,
+      "loss": 2.3882,
+      "step": 8061
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012437148217636022,
+      "loss": 2.2671,
+      "step": 8062
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.0001243621013133208,
+      "loss": 2.2914,
+      "step": 8063
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012435272045028143,
+      "loss": 2.6534,
+      "step": 8064
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012434333958724202,
+      "loss": 2.6532,
+      "step": 8065
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012433395872420264,
+      "loss": 2.6012,
+      "step": 8066
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012432457786116323,
+      "loss": 2.5888,
+      "step": 8067
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012431519699812385,
+      "loss": 2.4121,
+      "step": 8068
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00012430581613508444,
+      "loss": 2.5067,
+      "step": 8069
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012429643527204506,
+      "loss": 2.2939,
+      "step": 8070
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012428705440900562,
+      "loss": 2.488,
+      "step": 8071
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012427767354596624,
+      "loss": 2.6546,
+      "step": 8072
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012426829268292683,
+      "loss": 2.3013,
+      "step": 8073
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012425891181988742,
+      "loss": 2.6647,
+      "step": 8074
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012424953095684804,
+      "loss": 2.5897,
+      "step": 8075
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012424015009380863,
+      "loss": 2.7949,
+      "step": 8076
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012423076923076925,
+      "loss": 2.9481,
+      "step": 8077
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012422138836772984,
+      "loss": 2.595,
+      "step": 8078
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012421200750469045,
+      "loss": 2.311,
+      "step": 8079
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012420262664165102,
+      "loss": 2.5864,
+      "step": 8080
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012419324577861164,
+      "loss": 2.1713,
+      "step": 8081
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012418386491557223,
+      "loss": 2.4581,
+      "step": 8082
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012417448405253285,
+      "loss": 2.3621,
+      "step": 8083
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012416510318949344,
+      "loss": 2.4613,
+      "step": 8084
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012415572232645405,
+      "loss": 2.3841,
+      "step": 8085
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012414634146341465,
+      "loss": 2.6057,
+      "step": 8086
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012413696060037524,
+      "loss": 2.4502,
+      "step": 8087
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012412757973733585,
+      "loss": 2.6124,
+      "step": 8088
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012411819887429642,
+      "loss": 2.2158,
+      "step": 8089
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.00012410881801125704,
+      "loss": 2.3719,
+      "step": 8090
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012409943714821763,
+      "loss": 2.4722,
+      "step": 8091
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012409005628517825,
+      "loss": 2.7793,
+      "step": 8092
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012408067542213884,
+      "loss": 2.5053,
+      "step": 8093
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012407129455909945,
+      "loss": 2.524,
+      "step": 8094
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012406191369606005,
+      "loss": 2.3994,
+      "step": 8095
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012405253283302066,
+      "loss": 2.4599,
+      "step": 8096
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012404315196998125,
+      "loss": 2.5339,
+      "step": 8097
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012403377110694185,
+      "loss": 2.4314,
+      "step": 8098
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012402439024390244,
+      "loss": 2.5806,
+      "step": 8099
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012401500938086305,
+      "loss": 2.3951,
+      "step": 8100
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012400562851782365,
+      "loss": 2.7739,
+      "step": 8101
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012399624765478424,
+      "loss": 2.4818,
+      "step": 8102
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012398686679174485,
+      "loss": 2.3231,
+      "step": 8103
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012397748592870545,
+      "loss": 2.8478,
+      "step": 8104
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012396810506566606,
+      "loss": 2.3461,
+      "step": 8105
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012395872420262665,
+      "loss": 2.1407,
+      "step": 8106
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012394934333958725,
+      "loss": 2.5009,
+      "step": 8107
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012393996247654784,
+      "loss": 2.4756,
+      "step": 8108
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012393058161350845,
+      "loss": 2.6034,
+      "step": 8109
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012392120075046905,
+      "loss": 2.4339,
+      "step": 8110
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012391181988742966,
+      "loss": 2.2761,
+      "step": 8111
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 0.00012390243902439025,
+      "loss": 2.7119,
+      "step": 8112
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012389305816135087,
+      "loss": 2.5215,
+      "step": 8113
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012388367729831146,
+      "loss": 2.6768,
+      "step": 8114
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012387429643527205,
+      "loss": 2.4111,
+      "step": 8115
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012386491557223265,
+      "loss": 2.2743,
+      "step": 8116
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012385553470919324,
+      "loss": 2.6701,
+      "step": 8117
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012384615384615385,
+      "loss": 2.7495,
+      "step": 8118
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012383677298311445,
+      "loss": 2.4439,
+      "step": 8119
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012382739212007506,
+      "loss": 2.3672,
+      "step": 8120
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012381801125703565,
+      "loss": 2.7665,
+      "step": 8121
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012380863039399627,
+      "loss": 2.7688,
+      "step": 8122
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012379924953095686,
+      "loss": 2.5201,
+      "step": 8123
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012378986866791745,
+      "loss": 2.6537,
+      "step": 8124
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012378048780487804,
+      "loss": 2.456,
+      "step": 8125
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012377110694183866,
+      "loss": 2.5284,
+      "step": 8126
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012376172607879925,
+      "loss": 2.5974,
+      "step": 8127
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012375234521575984,
+      "loss": 2.6266,
+      "step": 8128
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012374296435272046,
+      "loss": 2.6836,
+      "step": 8129
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012373358348968105,
+      "loss": 2.6393,
+      "step": 8130
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012372420262664167,
+      "loss": 2.4308,
+      "step": 8131
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012371482176360226,
+      "loss": 2.3734,
+      "step": 8132
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00012370544090056285,
+      "loss": 2.7027,
+      "step": 8133
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012369606003752344,
+      "loss": 2.572,
+      "step": 8134
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012368667917448406,
+      "loss": 2.9219,
+      "step": 8135
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012367729831144465,
+      "loss": 2.4993,
+      "step": 8136
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012366791744840527,
+      "loss": 2.3892,
+      "step": 8137
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012365853658536586,
+      "loss": 2.5566,
+      "step": 8138
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012364915572232648,
+      "loss": 2.3252,
+      "step": 8139
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012363977485928707,
+      "loss": 2.6984,
+      "step": 8140
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012363039399624766,
+      "loss": 2.3991,
+      "step": 8141
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012362101313320825,
+      "loss": 2.4486,
+      "step": 8142
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012361163227016884,
+      "loss": 2.5362,
+      "step": 8143
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012360225140712946,
+      "loss": 2.6018,
+      "step": 8144
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012359287054409005,
+      "loss": 2.6552,
+      "step": 8145
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012358348968105067,
+      "loss": 2.4626,
+      "step": 8146
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012357410881801126,
+      "loss": 2.4151,
+      "step": 8147
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012356472795497188,
+      "loss": 2.7974,
+      "step": 8148
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012355534709193247,
+      "loss": 2.1929,
+      "step": 8149
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012354596622889306,
+      "loss": 2.4623,
+      "step": 8150
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012353658536585365,
+      "loss": 2.5084,
+      "step": 8151
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012352720450281427,
+      "loss": 2.4467,
+      "step": 8152
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012351782363977486,
+      "loss": 2.2679,
+      "step": 8153
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00012350844277673545,
+      "loss": 2.5512,
+      "step": 8154
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012349906191369607,
+      "loss": 2.4043,
+      "step": 8155
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012348968105065666,
+      "loss": 2.4377,
+      "step": 8156
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012348030018761728,
+      "loss": 2.507,
+      "step": 8157
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012347091932457787,
+      "loss": 2.3532,
+      "step": 8158
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012346153846153846,
+      "loss": 2.2163,
+      "step": 8159
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012345215759849905,
+      "loss": 2.4055,
+      "step": 8160
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012344277673545967,
+      "loss": 2.5074,
+      "step": 8161
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012343339587242026,
+      "loss": 2.5935,
+      "step": 8162
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012342401500938088,
+      "loss": 2.2677,
+      "step": 8163
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012341463414634147,
+      "loss": 2.3768,
+      "step": 8164
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0001234052532833021,
+      "loss": 2.8203,
+      "step": 8165
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012339587242026268,
+      "loss": 2.5339,
+      "step": 8166
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012338649155722327,
+      "loss": 2.3907,
+      "step": 8167
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012337711069418386,
+      "loss": 2.4357,
+      "step": 8168
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012336772983114445,
+      "loss": 2.6253,
+      "step": 8169
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012335834896810507,
+      "loss": 2.2542,
+      "step": 8170
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012334896810506566,
+      "loss": 2.5142,
+      "step": 8171
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012333958724202628,
+      "loss": 2.5017,
+      "step": 8172
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012333020637898687,
+      "loss": 2.4989,
+      "step": 8173
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0001233208255159475,
+      "loss": 2.1877,
+      "step": 8174
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012331144465290808,
+      "loss": 3.0546,
+      "step": 8175
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.00012330206378986867,
+      "loss": 2.7519,
+      "step": 8176
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012329268292682926,
+      "loss": 2.4382,
+      "step": 8177
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012328330206378988,
+      "loss": 2.6018,
+      "step": 8178
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012327392120075047,
+      "loss": 2.5527,
+      "step": 8179
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012326454033771106,
+      "loss": 2.3124,
+      "step": 8180
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012325515947467168,
+      "loss": 2.4078,
+      "step": 8181
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012324577861163227,
+      "loss": 2.2548,
+      "step": 8182
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0001232363977485929,
+      "loss": 2.4501,
+      "step": 8183
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012322701688555348,
+      "loss": 2.5161,
+      "step": 8184
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012321763602251407,
+      "loss": 2.3893,
+      "step": 8185
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012320825515947466,
+      "loss": 2.5831,
+      "step": 8186
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012319887429643528,
+      "loss": 2.4843,
+      "step": 8187
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012318949343339587,
+      "loss": 2.4784,
+      "step": 8188
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0001231801125703565,
+      "loss": 2.4119,
+      "step": 8189
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012317073170731708,
+      "loss": 2.6307,
+      "step": 8190
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0001231613508442777,
+      "loss": 2.7119,
+      "step": 8191
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.0001231519699812383,
+      "loss": 2.7281,
+      "step": 8192
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012314258911819888,
+      "loss": 2.6104,
+      "step": 8193
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012313320825515947,
+      "loss": 2.6504,
+      "step": 8194
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012312382739212006,
+      "loss": 2.6141,
+      "step": 8195
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012311444652908068,
+      "loss": 2.4598,
+      "step": 8196
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 0.00012310506566604127,
+      "loss": 2.4018,
+      "step": 8197
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001230956848030019,
+      "loss": 2.2631,
+      "step": 8198
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00012308630393996248,
+      "loss": 2.8972,
+      "step": 8199
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001230769230769231,
+      "loss": 2.5595,
+      "step": 8200
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001230675422138837,
+      "loss": 2.3255,
+      "step": 8201
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00012305816135084428,
+      "loss": 2.3957,
+      "step": 8202
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00012304878048780487,
+      "loss": 2.485,
+      "step": 8203
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001230393996247655,
+      "loss": 2.7545,
+      "step": 8204
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00012303001876172608,
+      "loss": 2.5327,
+      "step": 8205
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00012302063789868667,
+      "loss": 2.43,
+      "step": 8206
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001230112570356473,
+      "loss": 2.3806,
+      "step": 8207
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00012300187617260788,
+      "loss": 2.4316,
+      "step": 8208
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001229924953095685,
+      "loss": 2.2748,
+      "step": 8209
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001229831144465291,
+      "loss": 2.6164,
+      "step": 8210
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00012297373358348968,
+      "loss": 2.499,
+      "step": 8211
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00012296435272045027,
+      "loss": 2.6632,
+      "step": 8212
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001229549718574109,
+      "loss": 2.4181,
+      "step": 8213
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00012294559099437148,
+      "loss": 2.6959,
+      "step": 8214
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001229362101313321,
+      "loss": 2.6605,
+      "step": 8215
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001229268292682927,
+      "loss": 2.5371,
+      "step": 8216
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001229174484052533,
+      "loss": 2.4859,
+      "step": 8217
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.0001229080675422139,
+      "loss": 2.595,
+      "step": 8218
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001228986866791745,
+      "loss": 2.2927,
+      "step": 8219
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00012288930581613508,
+      "loss": 2.299,
+      "step": 8220
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00012287992495309567,
+      "loss": 2.3172,
+      "step": 8221
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001228705440900563,
+      "loss": 2.6008,
+      "step": 8222
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00012286116322701688,
+      "loss": 2.1889,
+      "step": 8223
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001228517823639775,
+      "loss": 2.5935,
+      "step": 8224
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001228424015009381,
+      "loss": 2.5736,
+      "step": 8225
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001228330206378987,
+      "loss": 2.4796,
+      "step": 8226
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001228236397748593,
+      "loss": 2.4158,
+      "step": 8227
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00012281425891181991,
+      "loss": 2.4371,
+      "step": 8228
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00012280487804878048,
+      "loss": 2.5701,
+      "step": 8229
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001227954971857411,
+      "loss": 2.373,
+      "step": 8230
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001227861163227017,
+      "loss": 2.3101,
+      "step": 8231
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00012277673545966228,
+      "loss": 2.4016,
+      "step": 8232
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001227673545966229,
+      "loss": 2.3397,
+      "step": 8233
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001227579737335835,
+      "loss": 2.5197,
+      "step": 8234
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001227485928705441,
+      "loss": 2.511,
+      "step": 8235
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001227392120075047,
+      "loss": 2.3947,
+      "step": 8236
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00012272983114446531,
+      "loss": 2.2676,
+      "step": 8237
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00012272045028142588,
+      "loss": 2.6398,
+      "step": 8238
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001227110694183865,
+      "loss": 2.5181,
+      "step": 8239
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.0001227016885553471,
+      "loss": 2.0208,
+      "step": 8240
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001226923076923077,
+      "loss": 2.9445,
+      "step": 8241
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001226829268292683,
+      "loss": 2.5828,
+      "step": 8242
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00012267354596622891,
+      "loss": 2.6344,
+      "step": 8243
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001226641651031895,
+      "loss": 2.3126,
+      "step": 8244
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001226547842401501,
+      "loss": 2.5394,
+      "step": 8245
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00012264540337711071,
+      "loss": 2.31,
+      "step": 8246
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00012263602251407128,
+      "loss": 2.5878,
+      "step": 8247
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001226266416510319,
+      "loss": 2.2532,
+      "step": 8248
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001226172607879925,
+      "loss": 2.6972,
+      "step": 8249
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001226078799249531,
+      "loss": 2.6706,
+      "step": 8250
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001225984990619137,
+      "loss": 2.5249,
+      "step": 8251
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00012258911819887431,
+      "loss": 2.521,
+      "step": 8252
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001225797373358349,
+      "loss": 2.3485,
+      "step": 8253
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00012257035647279552,
+      "loss": 2.5945,
+      "step": 8254
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00012256097560975611,
+      "loss": 2.2715,
+      "step": 8255
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001225515947467167,
+      "loss": 2.4656,
+      "step": 8256
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001225422138836773,
+      "loss": 2.6105,
+      "step": 8257
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001225328330206379,
+      "loss": 2.5069,
+      "step": 8258
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001225234521575985,
+      "loss": 2.2788,
+      "step": 8259
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001225140712945591,
+      "loss": 2.4232,
+      "step": 8260
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.00012250469043151971,
+      "loss": 2.4667,
+      "step": 8261
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001224953095684803,
+      "loss": 2.1758,
+      "step": 8262
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012248592870544092,
+      "loss": 2.7713,
+      "step": 8263
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012247654784240151,
+      "loss": 2.7114,
+      "step": 8264
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001224671669793621,
+      "loss": 2.4944,
+      "step": 8265
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001224577861163227,
+      "loss": 2.6723,
+      "step": 8266
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012244840525328331,
+      "loss": 2.7137,
+      "step": 8267
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001224390243902439,
+      "loss": 2.5591,
+      "step": 8268
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012242964352720452,
+      "loss": 2.5915,
+      "step": 8269
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012242026266416511,
+      "loss": 2.5809,
+      "step": 8270
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001224108818011257,
+      "loss": 2.8094,
+      "step": 8271
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012240150093808632,
+      "loss": 2.4729,
+      "step": 8272
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012239212007504691,
+      "loss": 2.7729,
+      "step": 8273
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001223827392120075,
+      "loss": 2.5641,
+      "step": 8274
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001223733583489681,
+      "loss": 2.4283,
+      "step": 8275
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012236397748592871,
+      "loss": 2.5441,
+      "step": 8276
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.0001223545966228893,
+      "loss": 2.2898,
+      "step": 8277
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012234521575984992,
+      "loss": 2.5805,
+      "step": 8278
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012233583489681051,
+      "loss": 2.4322,
+      "step": 8279
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012232645403377113,
+      "loss": 2.6424,
+      "step": 8280
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012231707317073172,
+      "loss": 2.4967,
+      "step": 8281
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00012230769230769231,
+      "loss": 2.3142,
+      "step": 8282
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001222983114446529,
+      "loss": 2.7523,
+      "step": 8283
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001222889305816135,
+      "loss": 2.6075,
+      "step": 8284
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012227954971857411,
+      "loss": 2.5886,
+      "step": 8285
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001222701688555347,
+      "loss": 2.3069,
+      "step": 8286
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012226078799249532,
+      "loss": 2.388,
+      "step": 8287
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001222514071294559,
+      "loss": 2.4969,
+      "step": 8288
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012224202626641653,
+      "loss": 2.4707,
+      "step": 8289
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012223264540337712,
+      "loss": 2.6136,
+      "step": 8290
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001222232645403377,
+      "loss": 2.2945,
+      "step": 8291
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001222138836772983,
+      "loss": 2.2372,
+      "step": 8292
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012220450281425892,
+      "loss": 2.2198,
+      "step": 8293
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001221951219512195,
+      "loss": 2.4905,
+      "step": 8294
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012218574108818013,
+      "loss": 2.4281,
+      "step": 8295
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012217636022514072,
+      "loss": 2.4659,
+      "step": 8296
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001221669793621013,
+      "loss": 2.5331,
+      "step": 8297
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012215759849906193,
+      "loss": 2.5683,
+      "step": 8298
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012214821763602252,
+      "loss": 2.4731,
+      "step": 8299
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001221388367729831,
+      "loss": 2.627,
+      "step": 8300
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001221294559099437,
+      "loss": 2.3048,
+      "step": 8301
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012212007504690432,
+      "loss": 2.6204,
+      "step": 8302
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.0001221106941838649,
+      "loss": 2.3053,
+      "step": 8303
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 0.00012210131332082553,
+      "loss": 2.5956,
+      "step": 8304
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012209193245778612,
+      "loss": 2.3496,
+      "step": 8305
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012208255159474674,
+      "loss": 2.2094,
+      "step": 8306
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012207317073170733,
+      "loss": 2.4694,
+      "step": 8307
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012206378986866794,
+      "loss": 2.5761,
+      "step": 8308
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012205440900562853,
+      "loss": 2.3763,
+      "step": 8309
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012204502814258913,
+      "loss": 2.6251,
+      "step": 8310
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012203564727954972,
+      "loss": 2.4036,
+      "step": 8311
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012202626641651031,
+      "loss": 2.3356,
+      "step": 8312
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012201688555347093,
+      "loss": 2.4985,
+      "step": 8313
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012200750469043152,
+      "loss": 2.2256,
+      "step": 8314
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012199812382739213,
+      "loss": 2.7866,
+      "step": 8315
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012198874296435272,
+      "loss": 2.6929,
+      "step": 8316
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012197936210131334,
+      "loss": 2.2978,
+      "step": 8317
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012196998123827393,
+      "loss": 2.3429,
+      "step": 8318
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012196060037523453,
+      "loss": 2.4565,
+      "step": 8319
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012195121951219512,
+      "loss": 2.6697,
+      "step": 8320
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012194183864915574,
+      "loss": 2.414,
+      "step": 8321
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012193245778611633,
+      "loss": 2.6715,
+      "step": 8322
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012192307692307694,
+      "loss": 2.3786,
+      "step": 8323
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012191369606003753,
+      "loss": 2.7655,
+      "step": 8324
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00012190431519699812,
+      "loss": 2.2968,
+      "step": 8325
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012189493433395874,
+      "loss": 2.5053,
+      "step": 8326
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012188555347091933,
+      "loss": 2.6012,
+      "step": 8327
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012187617260787993,
+      "loss": 2.4609,
+      "step": 8328
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012186679174484052,
+      "loss": 2.2466,
+      "step": 8329
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012185741088180114,
+      "loss": 2.7238,
+      "step": 8330
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012184803001876173,
+      "loss": 2.2101,
+      "step": 8331
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012183864915572234,
+      "loss": 2.7362,
+      "step": 8332
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012182926829268293,
+      "loss": 2.656,
+      "step": 8333
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012181988742964354,
+      "loss": 2.3036,
+      "step": 8334
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012181050656660414,
+      "loss": 2.4166,
+      "step": 8335
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012180112570356474,
+      "loss": 2.7801,
+      "step": 8336
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012179174484052533,
+      "loss": 2.3045,
+      "step": 8337
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012178236397748592,
+      "loss": 2.7084,
+      "step": 8338
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012177298311444654,
+      "loss": 2.3161,
+      "step": 8339
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012176360225140713,
+      "loss": 2.2842,
+      "step": 8340
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012175422138836773,
+      "loss": 2.6253,
+      "step": 8341
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012174484052532833,
+      "loss": 2.2817,
+      "step": 8342
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012173545966228894,
+      "loss": 2.4289,
+      "step": 8343
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012172607879924953,
+      "loss": 2.488,
+      "step": 8344
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012171669793621014,
+      "loss": 2.5049,
+      "step": 8345
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.00012170731707317073,
+      "loss": 2.3974,
+      "step": 8346
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012169793621013135,
+      "loss": 2.451,
+      "step": 8347
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012168855534709194,
+      "loss": 2.4501,
+      "step": 8348
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012167917448405254,
+      "loss": 2.3275,
+      "step": 8349
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012166979362101313,
+      "loss": 2.4998,
+      "step": 8350
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012166041275797373,
+      "loss": 2.328,
+      "step": 8351
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012165103189493434,
+      "loss": 2.5383,
+      "step": 8352
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012164165103189493,
+      "loss": 2.4585,
+      "step": 8353
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012163227016885554,
+      "loss": 2.4471,
+      "step": 8354
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012162288930581613,
+      "loss": 2.4495,
+      "step": 8355
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012161350844277675,
+      "loss": 2.5344,
+      "step": 8356
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012160412757973734,
+      "loss": 2.378,
+      "step": 8357
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012159474671669794,
+      "loss": 2.4225,
+      "step": 8358
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012158536585365853,
+      "loss": 2.522,
+      "step": 8359
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012157598499061915,
+      "loss": 2.4087,
+      "step": 8360
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012156660412757974,
+      "loss": 2.403,
+      "step": 8361
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012155722326454035,
+      "loss": 2.5018,
+      "step": 8362
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012154784240150094,
+      "loss": 2.3667,
+      "step": 8363
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012153846153846153,
+      "loss": 2.6671,
+      "step": 8364
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012152908067542215,
+      "loss": 2.795,
+      "step": 8365
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012151969981238274,
+      "loss": 2.521,
+      "step": 8366
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012151031894934334,
+      "loss": 2.5353,
+      "step": 8367
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00012150093808630393,
+      "loss": 2.6295,
+      "step": 8368
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012149155722326455,
+      "loss": 2.2987,
+      "step": 8369
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012148217636022514,
+      "loss": 2.61,
+      "step": 8370
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012147279549718575,
+      "loss": 2.299,
+      "step": 8371
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012146341463414634,
+      "loss": 2.4683,
+      "step": 8372
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012145403377110696,
+      "loss": 2.501,
+      "step": 8373
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012144465290806755,
+      "loss": 2.6732,
+      "step": 8374
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012143527204502815,
+      "loss": 2.7671,
+      "step": 8375
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012142589118198874,
+      "loss": 2.6155,
+      "step": 8376
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012141651031894933,
+      "loss": 2.6456,
+      "step": 8377
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012140712945590995,
+      "loss": 2.2759,
+      "step": 8378
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012139774859287054,
+      "loss": 2.5554,
+      "step": 8379
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012138836772983115,
+      "loss": 2.3798,
+      "step": 8380
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012137898686679174,
+      "loss": 2.3874,
+      "step": 8381
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012136960600375236,
+      "loss": 2.3463,
+      "step": 8382
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012136022514071295,
+      "loss": 2.2722,
+      "step": 8383
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012135084427767355,
+      "loss": 2.276,
+      "step": 8384
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012134146341463414,
+      "loss": 2.515,
+      "step": 8385
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012133208255159476,
+      "loss": 2.6932,
+      "step": 8386
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012132270168855535,
+      "loss": 2.2948,
+      "step": 8387
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012131332082551597,
+      "loss": 2.64,
+      "step": 8388
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 0.00012130393996247655,
+      "loss": 2.7455,
+      "step": 8389
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012129455909943714,
+      "loss": 2.5962,
+      "step": 8390
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012128517823639776,
+      "loss": 2.5862,
+      "step": 8391
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012127579737335835,
+      "loss": 2.8022,
+      "step": 8392
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012126641651031897,
+      "loss": 2.4435,
+      "step": 8393
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012125703564727954,
+      "loss": 2.6729,
+      "step": 8394
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012124765478424016,
+      "loss": 2.5467,
+      "step": 8395
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012123827392120075,
+      "loss": 2.493,
+      "step": 8396
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012122889305816137,
+      "loss": 2.5523,
+      "step": 8397
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012121951219512195,
+      "loss": 2.7147,
+      "step": 8398
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012121013133208257,
+      "loss": 2.6223,
+      "step": 8399
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012120075046904316,
+      "loss": 2.345,
+      "step": 8400
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012119136960600377,
+      "loss": 2.3695,
+      "step": 8401
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012118198874296437,
+      "loss": 2.4611,
+      "step": 8402
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012117260787992494,
+      "loss": 2.4904,
+      "step": 8403
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012116322701688556,
+      "loss": 2.5827,
+      "step": 8404
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012115384615384615,
+      "loss": 2.4167,
+      "step": 8405
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012114446529080677,
+      "loss": 2.4535,
+      "step": 8406
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012113508442776735,
+      "loss": 2.5196,
+      "step": 8407
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012112570356472797,
+      "loss": 2.5015,
+      "step": 8408
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012111632270168856,
+      "loss": 2.3689,
+      "step": 8409
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00012110694183864917,
+      "loss": 2.3497,
+      "step": 8410
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012109756097560976,
+      "loss": 2.4138,
+      "step": 8411
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012108818011257037,
+      "loss": 2.7201,
+      "step": 8412
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012107879924953096,
+      "loss": 2.447,
+      "step": 8413
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012106941838649158,
+      "loss": 2.5237,
+      "step": 8414
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012106003752345217,
+      "loss": 2.6601,
+      "step": 8415
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012105065666041275,
+      "loss": 2.6522,
+      "step": 8416
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012104127579737336,
+      "loss": 2.4891,
+      "step": 8417
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012103189493433396,
+      "loss": 2.5192,
+      "step": 8418
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012102251407129457,
+      "loss": 2.1796,
+      "step": 8419
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012101313320825516,
+      "loss": 2.5242,
+      "step": 8420
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012100375234521577,
+      "loss": 2.3626,
+      "step": 8421
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012099437148217636,
+      "loss": 2.6269,
+      "step": 8422
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012098499061913698,
+      "loss": 2.4821,
+      "step": 8423
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012097560975609757,
+      "loss": 2.2898,
+      "step": 8424
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012096622889305817,
+      "loss": 2.5295,
+      "step": 8425
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012095684803001876,
+      "loss": 2.5319,
+      "step": 8426
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012094746716697938,
+      "loss": 2.4168,
+      "step": 8427
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012093808630393997,
+      "loss": 2.326,
+      "step": 8428
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012092870544090056,
+      "loss": 2.3017,
+      "step": 8429
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012091932457786117,
+      "loss": 2.3217,
+      "step": 8430
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012090994371482176,
+      "loss": 2.4139,
+      "step": 8431
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00012090056285178238,
+      "loss": 2.3623,
+      "step": 8432
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012089118198874297,
+      "loss": 2.4498,
+      "step": 8433
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012088180112570357,
+      "loss": 2.4311,
+      "step": 8434
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012087242026266416,
+      "loss": 2.2521,
+      "step": 8435
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012086303939962478,
+      "loss": 2.6145,
+      "step": 8436
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012085365853658537,
+      "loss": 2.3495,
+      "step": 8437
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012084427767354598,
+      "loss": 2.5589,
+      "step": 8438
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012083489681050657,
+      "loss": 2.4397,
+      "step": 8439
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012082551594746719,
+      "loss": 2.2938,
+      "step": 8440
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012081613508442778,
+      "loss": 2.3041,
+      "step": 8441
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012080675422138837,
+      "loss": 2.501,
+      "step": 8442
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012079737335834897,
+      "loss": 2.5278,
+      "step": 8443
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012078799249530956,
+      "loss": 2.6134,
+      "step": 8444
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012077861163227018,
+      "loss": 2.3818,
+      "step": 8445
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012076923076923077,
+      "loss": 2.5184,
+      "step": 8446
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012075984990619138,
+      "loss": 2.5337,
+      "step": 8447
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012075046904315197,
+      "loss": 2.6177,
+      "step": 8448
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012074108818011259,
+      "loss": 2.6709,
+      "step": 8449
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012073170731707318,
+      "loss": 2.8051,
+      "step": 8450
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012072232645403378,
+      "loss": 2.5099,
+      "step": 8451
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012071294559099437,
+      "loss": 2.2961,
+      "step": 8452
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.00012070356472795499,
+      "loss": 2.6608,
+      "step": 8453
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012069418386491558,
+      "loss": 2.6463,
+      "step": 8454
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012068480300187617,
+      "loss": 2.3035,
+      "step": 8455
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012067542213883678,
+      "loss": 2.4842,
+      "step": 8456
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012066604127579737,
+      "loss": 2.4094,
+      "step": 8457
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012065666041275799,
+      "loss": 2.5734,
+      "step": 8458
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012064727954971858,
+      "loss": 2.7579,
+      "step": 8459
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012063789868667918,
+      "loss": 2.2904,
+      "step": 8460
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012062851782363977,
+      "loss": 2.3513,
+      "step": 8461
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012061913696060039,
+      "loss": 2.4514,
+      "step": 8462
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012060975609756098,
+      "loss": 2.2743,
+      "step": 8463
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012060037523452159,
+      "loss": 2.2729,
+      "step": 8464
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012059099437148218,
+      "loss": 2.3715,
+      "step": 8465
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.0001205816135084428,
+      "loss": 2.5482,
+      "step": 8466
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012057223264540339,
+      "loss": 2.2531,
+      "step": 8467
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012056285178236398,
+      "loss": 2.7674,
+      "step": 8468
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012055347091932458,
+      "loss": 2.4236,
+      "step": 8469
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012054409005628517,
+      "loss": 2.4627,
+      "step": 8470
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012053470919324579,
+      "loss": 2.5534,
+      "step": 8471
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012052532833020638,
+      "loss": 2.6444,
+      "step": 8472
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012051594746716699,
+      "loss": 2.5501,
+      "step": 8473
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00012050656660412758,
+      "loss": 2.6418,
+      "step": 8474
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001204971857410882,
+      "loss": 2.4039,
+      "step": 8475
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012048780487804879,
+      "loss": 2.6758,
+      "step": 8476
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012047842401500939,
+      "loss": 3.009,
+      "step": 8477
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012046904315196998,
+      "loss": 2.3951,
+      "step": 8478
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001204596622889306,
+      "loss": 2.36,
+      "step": 8479
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012045028142589119,
+      "loss": 2.5511,
+      "step": 8480
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012044090056285178,
+      "loss": 2.4289,
+      "step": 8481
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012043151969981239,
+      "loss": 2.1869,
+      "step": 8482
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012042213883677298,
+      "loss": 2.4141,
+      "step": 8483
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001204127579737336,
+      "loss": 2.4436,
+      "step": 8484
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012040337711069419,
+      "loss": 2.5453,
+      "step": 8485
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012039399624765479,
+      "loss": 2.6053,
+      "step": 8486
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012038461538461538,
+      "loss": 2.3842,
+      "step": 8487
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.000120375234521576,
+      "loss": 2.4829,
+      "step": 8488
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012036585365853659,
+      "loss": 2.5166,
+      "step": 8489
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001203564727954972,
+      "loss": 2.2815,
+      "step": 8490
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012034709193245779,
+      "loss": 2.4352,
+      "step": 8491
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.0001203377110694184,
+      "loss": 2.1018,
+      "step": 8492
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.000120328330206379,
+      "loss": 2.4422,
+      "step": 8493
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012031894934333959,
+      "loss": 2.4313,
+      "step": 8494
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012030956848030019,
+      "loss": 2.3296,
+      "step": 8495
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 0.00012030018761726078,
+      "loss": 2.4416,
+      "step": 8496
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001202908067542214,
+      "loss": 2.6777,
+      "step": 8497
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00012028142589118199,
+      "loss": 2.426,
+      "step": 8498
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001202720450281426,
+      "loss": 2.8188,
+      "step": 8499
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00012026266416510319,
+      "loss": 2.6059,
+      "step": 8500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001202532833020638,
+      "loss": 2.3653,
+      "step": 8501
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001202439024390244,
+      "loss": 2.4922,
+      "step": 8502
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.000120234521575985,
+      "loss": 2.6114,
+      "step": 8503
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00012022514071294559,
+      "loss": 2.7095,
+      "step": 8504
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00012021575984990621,
+      "loss": 2.7177,
+      "step": 8505
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001202063789868668,
+      "loss": 2.1852,
+      "step": 8506
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00012019699812382739,
+      "loss": 2.5082,
+      "step": 8507
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.000120187617260788,
+      "loss": 2.4706,
+      "step": 8508
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00012017823639774859,
+      "loss": 2.6564,
+      "step": 8509
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001201688555347092,
+      "loss": 2.6184,
+      "step": 8510
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001201594746716698,
+      "loss": 2.3968,
+      "step": 8511
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001201500938086304,
+      "loss": 2.4444,
+      "step": 8512
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00012014071294559099,
+      "loss": 2.7377,
+      "step": 8513
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00012013133208255161,
+      "loss": 2.7133,
+      "step": 8514
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001201219512195122,
+      "loss": 2.4289,
+      "step": 8515
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001201125703564728,
+      "loss": 2.6073,
+      "step": 8516
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.0001201031894934334,
+      "loss": 2.6275,
+      "step": 8517
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00012009380863039401,
+      "loss": 2.4748,
+      "step": 8518
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001200844277673546,
+      "loss": 2.2931,
+      "step": 8519
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00012007504690431521,
+      "loss": 2.546,
+      "step": 8520
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001200656660412758,
+      "loss": 2.4156,
+      "step": 8521
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00012005628517823639,
+      "loss": 2.808,
+      "step": 8522
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00012004690431519701,
+      "loss": 2.1425,
+      "step": 8523
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001200375234521576,
+      "loss": 2.4939,
+      "step": 8524
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001200281425891182,
+      "loss": 2.3387,
+      "step": 8525
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001200187617260788,
+      "loss": 2.8025,
+      "step": 8526
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00012000938086303941,
+      "loss": 2.4015,
+      "step": 8527
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00012,
+      "loss": 2.2868,
+      "step": 8528
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00011999061913696061,
+      "loss": 2.3299,
+      "step": 8529
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001199812382739212,
+      "loss": 2.5092,
+      "step": 8530
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00011997185741088182,
+      "loss": 2.4383,
+      "step": 8531
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00011996247654784241,
+      "loss": 2.5143,
+      "step": 8532
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00011995309568480301,
+      "loss": 2.6233,
+      "step": 8533
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001199437148217636,
+      "loss": 2.6189,
+      "step": 8534
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001199343339587242,
+      "loss": 2.4217,
+      "step": 8535
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00011992495309568481,
+      "loss": 2.3872,
+      "step": 8536
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0001199155722326454,
+      "loss": 2.4339,
+      "step": 8537
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00011990619136960601,
+      "loss": 2.3704,
+      "step": 8538
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001198968105065666,
+      "loss": 2.6205,
+      "step": 8539
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011988742964352722,
+      "loss": 2.4097,
+      "step": 8540
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011987804878048781,
+      "loss": 2.4816,
+      "step": 8541
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011986866791744841,
+      "loss": 2.3095,
+      "step": 8542
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.000119859287054409,
+      "loss": 2.4981,
+      "step": 8543
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011984990619136962,
+      "loss": 2.5982,
+      "step": 8544
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011984052532833021,
+      "loss": 2.3167,
+      "step": 8545
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011983114446529083,
+      "loss": 2.2766,
+      "step": 8546
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011982176360225141,
+      "loss": 2.2924,
+      "step": 8547
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.000119812382739212,
+      "loss": 2.4857,
+      "step": 8548
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011980300187617262,
+      "loss": 2.3947,
+      "step": 8549
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011979362101313321,
+      "loss": 2.3751,
+      "step": 8550
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011978424015009381,
+      "loss": 2.4275,
+      "step": 8551
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001197748592870544,
+      "loss": 2.4244,
+      "step": 8552
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011976547842401502,
+      "loss": 2.4725,
+      "step": 8553
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011975609756097561,
+      "loss": 2.5162,
+      "step": 8554
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011974671669793623,
+      "loss": 2.2266,
+      "step": 8555
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011973733583489681,
+      "loss": 2.2035,
+      "step": 8556
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011972795497185742,
+      "loss": 2.4952,
+      "step": 8557
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011971857410881802,
+      "loss": 2.3903,
+      "step": 8558
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.00011970919324577863,
+      "loss": 2.4023,
+      "step": 8559
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011969981238273922,
+      "loss": 2.4809,
+      "step": 8560
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001196904315196998,
+      "loss": 2.5675,
+      "step": 8561
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011968105065666042,
+      "loss": 2.4945,
+      "step": 8562
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011967166979362101,
+      "loss": 2.2212,
+      "step": 8563
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011966228893058163,
+      "loss": 2.4549,
+      "step": 8564
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001196529080675422,
+      "loss": 2.6884,
+      "step": 8565
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011964352720450282,
+      "loss": 2.6219,
+      "step": 8566
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011963414634146342,
+      "loss": 2.3592,
+      "step": 8567
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011962476547842403,
+      "loss": 2.3625,
+      "step": 8568
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011961538461538462,
+      "loss": 2.5423,
+      "step": 8569
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011960600375234523,
+      "loss": 2.6475,
+      "step": 8570
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011959662288930582,
+      "loss": 2.6463,
+      "step": 8571
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011958724202626644,
+      "loss": 2.3009,
+      "step": 8572
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011957786116322703,
+      "loss": 2.2397,
+      "step": 8573
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.0001195684803001876,
+      "loss": 2.2821,
+      "step": 8574
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011955909943714822,
+      "loss": 2.5069,
+      "step": 8575
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011954971857410882,
+      "loss": 2.402,
+      "step": 8576
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011954033771106943,
+      "loss": 2.332,
+      "step": 8577
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011953095684803002,
+      "loss": 2.3845,
+      "step": 8578
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011952157598499063,
+      "loss": 2.4261,
+      "step": 8579
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011951219512195122,
+      "loss": 2.3904,
+      "step": 8580
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 0.00011950281425891184,
+      "loss": 2.2966,
+      "step": 8581
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011949343339587243,
+      "loss": 2.3111,
+      "step": 8582
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011948405253283303,
+      "loss": 2.0656,
+      "step": 8583
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011947467166979362,
+      "loss": 2.4328,
+      "step": 8584
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011946529080675424,
+      "loss": 2.5056,
+      "step": 8585
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011945590994371483,
+      "loss": 2.3799,
+      "step": 8586
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011944652908067542,
+      "loss": 2.4642,
+      "step": 8587
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011943714821763603,
+      "loss": 2.5046,
+      "step": 8588
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011942776735459662,
+      "loss": 2.1797,
+      "step": 8589
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011941838649155724,
+      "loss": 2.4606,
+      "step": 8590
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011940900562851783,
+      "loss": 2.5262,
+      "step": 8591
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011939962476547843,
+      "loss": 2.4694,
+      "step": 8592
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011939024390243902,
+      "loss": 2.7564,
+      "step": 8593
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011938086303939964,
+      "loss": 2.4486,
+      "step": 8594
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011937148217636023,
+      "loss": 2.554,
+      "step": 8595
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011936210131332084,
+      "loss": 2.6049,
+      "step": 8596
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011935272045028143,
+      "loss": 2.6652,
+      "step": 8597
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011934333958724205,
+      "loss": 2.8836,
+      "step": 8598
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011933395872420264,
+      "loss": 2.1868,
+      "step": 8599
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011932457786116323,
+      "loss": 2.3698,
+      "step": 8600
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011931519699812383,
+      "loss": 2.3573,
+      "step": 8601
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00011930581613508442,
+      "loss": 2.3974,
+      "step": 8602
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011929643527204504,
+      "loss": 2.4432,
+      "step": 8603
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011928705440900563,
+      "loss": 2.7684,
+      "step": 8604
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011927767354596624,
+      "loss": 2.2236,
+      "step": 8605
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011926829268292683,
+      "loss": 2.5458,
+      "step": 8606
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011925891181988745,
+      "loss": 2.5158,
+      "step": 8607
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011924953095684804,
+      "loss": 2.4457,
+      "step": 8608
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011924015009380864,
+      "loss": 2.4374,
+      "step": 8609
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011923076923076923,
+      "loss": 2.4423,
+      "step": 8610
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011922138836772985,
+      "loss": 2.3272,
+      "step": 8611
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011921200750469044,
+      "loss": 2.5695,
+      "step": 8612
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011920262664165103,
+      "loss": 2.334,
+      "step": 8613
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011919324577861164,
+      "loss": 2.5701,
+      "step": 8614
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011918386491557223,
+      "loss": 2.3892,
+      "step": 8615
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011917448405253285,
+      "loss": 2.399,
+      "step": 8616
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011916510318949344,
+      "loss": 2.3981,
+      "step": 8617
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011915572232645404,
+      "loss": 2.5571,
+      "step": 8618
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011914634146341463,
+      "loss": 2.4687,
+      "step": 8619
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011913696060037525,
+      "loss": 2.2064,
+      "step": 8620
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011912757973733584,
+      "loss": 2.431,
+      "step": 8621
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011911819887429645,
+      "loss": 2.6254,
+      "step": 8622
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00011910881801125704,
+      "loss": 2.1794,
+      "step": 8623
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011909943714821766,
+      "loss": 2.3922,
+      "step": 8624
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011909005628517825,
+      "loss": 2.3176,
+      "step": 8625
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011908067542213884,
+      "loss": 2.8023,
+      "step": 8626
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011907129455909944,
+      "loss": 2.4858,
+      "step": 8627
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011906191369606003,
+      "loss": 2.5659,
+      "step": 8628
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011905253283302065,
+      "loss": 2.65,
+      "step": 8629
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011904315196998124,
+      "loss": 2.2069,
+      "step": 8630
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011903377110694185,
+      "loss": 2.6705,
+      "step": 8631
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011902439024390244,
+      "loss": 2.4188,
+      "step": 8632
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011901500938086305,
+      "loss": 2.4339,
+      "step": 8633
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011900562851782365,
+      "loss": 2.4039,
+      "step": 8634
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011899624765478425,
+      "loss": 2.387,
+      "step": 8635
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011898686679174484,
+      "loss": 2.4668,
+      "step": 8636
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011897748592870546,
+      "loss": 2.5449,
+      "step": 8637
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011896810506566605,
+      "loss": 2.2707,
+      "step": 8638
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011895872420262664,
+      "loss": 2.5618,
+      "step": 8639
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011894934333958725,
+      "loss": 2.4537,
+      "step": 8640
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011893996247654784,
+      "loss": 2.3039,
+      "step": 8641
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011893058161350845,
+      "loss": 2.6779,
+      "step": 8642
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011892120075046905,
+      "loss": 2.4437,
+      "step": 8643
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011891181988742965,
+      "loss": 2.2418,
+      "step": 8644
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00011890243902439024,
+      "loss": 2.4114,
+      "step": 8645
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011889305816135086,
+      "loss": 2.6306,
+      "step": 8646
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011888367729831145,
+      "loss": 2.1053,
+      "step": 8647
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011887429643527205,
+      "loss": 2.5379,
+      "step": 8648
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011886491557223265,
+      "loss": 2.4497,
+      "step": 8649
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011885553470919326,
+      "loss": 2.1334,
+      "step": 8650
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011884615384615385,
+      "loss": 2.6415,
+      "step": 8651
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011883677298311445,
+      "loss": 2.6262,
+      "step": 8652
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011882739212007505,
+      "loss": 2.6661,
+      "step": 8653
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011881801125703564,
+      "loss": 2.2866,
+      "step": 8654
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011880863039399626,
+      "loss": 2.4961,
+      "step": 8655
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011879924953095685,
+      "loss": 2.3263,
+      "step": 8656
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011878986866791745,
+      "loss": 2.4867,
+      "step": 8657
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011878048780487805,
+      "loss": 2.3242,
+      "step": 8658
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011877110694183866,
+      "loss": 2.492,
+      "step": 8659
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011876172607879925,
+      "loss": 2.4054,
+      "step": 8660
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011875234521575986,
+      "loss": 2.5398,
+      "step": 8661
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011874296435272045,
+      "loss": 2.4369,
+      "step": 8662
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011873358348968107,
+      "loss": 2.491,
+      "step": 8663
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011872420262664166,
+      "loss": 2.2721,
+      "step": 8664
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011871482176360225,
+      "loss": 2.3278,
+      "step": 8665
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.00011870544090056285,
+      "loss": 2.2705,
+      "step": 8666
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011869606003752345,
+      "loss": 2.8129,
+      "step": 8667
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011868667917448406,
+      "loss": 2.225,
+      "step": 8668
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011867729831144465,
+      "loss": 2.2202,
+      "step": 8669
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011866791744840526,
+      "loss": 2.5605,
+      "step": 8670
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011865853658536585,
+      "loss": 2.522,
+      "step": 8671
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011864915572232647,
+      "loss": 2.4334,
+      "step": 8672
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011863977485928706,
+      "loss": 2.6089,
+      "step": 8673
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011863039399624766,
+      "loss": 2.4855,
+      "step": 8674
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011862101313320825,
+      "loss": 2.3995,
+      "step": 8675
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011861163227016887,
+      "loss": 2.5637,
+      "step": 8676
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011860225140712946,
+      "loss": 2.3879,
+      "step": 8677
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011859287054409005,
+      "loss": 2.4612,
+      "step": 8678
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011858348968105066,
+      "loss": 2.2898,
+      "step": 8679
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011857410881801125,
+      "loss": 2.6386,
+      "step": 8680
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011856472795497187,
+      "loss": 2.3277,
+      "step": 8681
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011855534709193246,
+      "loss": 2.491,
+      "step": 8682
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011854596622889306,
+      "loss": 2.4832,
+      "step": 8683
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011853658536585365,
+      "loss": 2.6296,
+      "step": 8684
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011852720450281427,
+      "loss": 2.3682,
+      "step": 8685
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011851782363977486,
+      "loss": 2.259,
+      "step": 8686
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 0.00011850844277673547,
+      "loss": 2.6307,
+      "step": 8687
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011849906191369606,
+      "loss": 2.3559,
+      "step": 8688
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011848968105065668,
+      "loss": 2.3428,
+      "step": 8689
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011848030018761727,
+      "loss": 2.2431,
+      "step": 8690
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011847091932457786,
+      "loss": 2.6528,
+      "step": 8691
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011846153846153846,
+      "loss": 2.3037,
+      "step": 8692
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011845215759849905,
+      "loss": 2.2768,
+      "step": 8693
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011844277673545967,
+      "loss": 2.3054,
+      "step": 8694
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011843339587242026,
+      "loss": 2.3743,
+      "step": 8695
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011842401500938087,
+      "loss": 2.4269,
+      "step": 8696
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011841463414634146,
+      "loss": 2.6854,
+      "step": 8697
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011840525328330208,
+      "loss": 2.3976,
+      "step": 8698
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011839587242026267,
+      "loss": 2.5445,
+      "step": 8699
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011838649155722327,
+      "loss": 2.57,
+      "step": 8700
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011837711069418386,
+      "loss": 2.3708,
+      "step": 8701
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011836772983114448,
+      "loss": 2.624,
+      "step": 8702
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011835834896810507,
+      "loss": 2.6611,
+      "step": 8703
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011834896810506566,
+      "loss": 2.7908,
+      "step": 8704
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011833958724202627,
+      "loss": 2.4972,
+      "step": 8705
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011833020637898686,
+      "loss": 2.7612,
+      "step": 8706
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011832082551594748,
+      "loss": 2.596,
+      "step": 8707
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011831144465290807,
+      "loss": 2.258,
+      "step": 8708
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00011830206378986867,
+      "loss": 2.5283,
+      "step": 8709
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011829268292682926,
+      "loss": 2.4229,
+      "step": 8710
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011828330206378988,
+      "loss": 2.3253,
+      "step": 8711
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011827392120075047,
+      "loss": 2.5252,
+      "step": 8712
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011826454033771109,
+      "loss": 2.3386,
+      "step": 8713
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011825515947467167,
+      "loss": 2.494,
+      "step": 8714
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011824577861163228,
+      "loss": 2.4884,
+      "step": 8715
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011823639774859288,
+      "loss": 2.4738,
+      "step": 8716
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001182270168855535,
+      "loss": 2.2057,
+      "step": 8717
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011821763602251407,
+      "loss": 2.3853,
+      "step": 8718
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011820825515947466,
+      "loss": 2.3371,
+      "step": 8719
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011819887429643528,
+      "loss": 2.156,
+      "step": 8720
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011818949343339587,
+      "loss": 2.3494,
+      "step": 8721
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011818011257035649,
+      "loss": 2.3913,
+      "step": 8722
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011817073170731707,
+      "loss": 2.8654,
+      "step": 8723
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011816135084427768,
+      "loss": 2.4425,
+      "step": 8724
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011815196998123828,
+      "loss": 2.4193,
+      "step": 8725
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001181425891181989,
+      "loss": 2.5376,
+      "step": 8726
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011813320825515947,
+      "loss": 2.4639,
+      "step": 8727
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011812382739212009,
+      "loss": 2.5162,
+      "step": 8728
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00011811444652908068,
+      "loss": 2.5628,
+      "step": 8729
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.0001181050656660413,
+      "loss": 2.382,
+      "step": 8730
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011809568480300189,
+      "loss": 2.7085,
+      "step": 8731
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011808630393996247,
+      "loss": 2.3682,
+      "step": 8732
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011807692307692308,
+      "loss": 2.2913,
+      "step": 8733
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011806754221388368,
+      "loss": 2.5912,
+      "step": 8734
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001180581613508443,
+      "loss": 2.6371,
+      "step": 8735
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011804878048780488,
+      "loss": 2.6499,
+      "step": 8736
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011803939962476549,
+      "loss": 2.4096,
+      "step": 8737
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011803001876172608,
+      "loss": 2.023,
+      "step": 8738
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001180206378986867,
+      "loss": 2.6562,
+      "step": 8739
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011801125703564729,
+      "loss": 2.3832,
+      "step": 8740
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011800187617260789,
+      "loss": 2.4765,
+      "step": 8741
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011799249530956848,
+      "loss": 2.5324,
+      "step": 8742
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001179831144465291,
+      "loss": 2.3207,
+      "step": 8743
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011797373358348969,
+      "loss": 2.349,
+      "step": 8744
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011796435272045028,
+      "loss": 2.4876,
+      "step": 8745
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011795497185741089,
+      "loss": 2.4542,
+      "step": 8746
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011794559099437148,
+      "loss": 2.4548,
+      "step": 8747
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001179362101313321,
+      "loss": 2.5097,
+      "step": 8748
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011792682926829269,
+      "loss": 2.4862,
+      "step": 8749
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011791744840525329,
+      "loss": 2.2838,
+      "step": 8750
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.00011790806754221388,
+      "loss": 2.4756,
+      "step": 8751
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001178986866791745,
+      "loss": 2.4873,
+      "step": 8752
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011788930581613509,
+      "loss": 2.5646,
+      "step": 8753
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001178799249530957,
+      "loss": 2.1939,
+      "step": 8754
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011787054409005629,
+      "loss": 2.6211,
+      "step": 8755
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001178611632270169,
+      "loss": 2.5833,
+      "step": 8756
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001178517823639775,
+      "loss": 2.3564,
+      "step": 8757
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011784240150093809,
+      "loss": 2.5671,
+      "step": 8758
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011783302063789869,
+      "loss": 2.195,
+      "step": 8759
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011782363977485928,
+      "loss": 2.1473,
+      "step": 8760
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001178142589118199,
+      "loss": 2.1815,
+      "step": 8761
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011780487804878049,
+      "loss": 2.5928,
+      "step": 8762
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001177954971857411,
+      "loss": 2.2985,
+      "step": 8763
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011778611632270169,
+      "loss": 2.3716,
+      "step": 8764
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001177767354596623,
+      "loss": 2.4817,
+      "step": 8765
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001177673545966229,
+      "loss": 2.6376,
+      "step": 8766
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001177579737335835,
+      "loss": 2.3413,
+      "step": 8767
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011774859287054409,
+      "loss": 2.4117,
+      "step": 8768
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011773921200750471,
+      "loss": 2.4161,
+      "step": 8769
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001177298311444653,
+      "loss": 2.4246,
+      "step": 8770
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011772045028142589,
+      "loss": 2.524,
+      "step": 8771
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.0001177110694183865,
+      "loss": 2.7781,
+      "step": 8772
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 0.00011770168855534709,
+      "loss": 2.4964,
+      "step": 8773
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001176923076923077,
+      "loss": 2.3847,
+      "step": 8774
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001176829268292683,
+      "loss": 2.3734,
+      "step": 8775
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001176735459662289,
+      "loss": 2.5787,
+      "step": 8776
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00011766416510318949,
+      "loss": 2.6797,
+      "step": 8777
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00011765478424015011,
+      "loss": 2.4367,
+      "step": 8778
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001176454033771107,
+      "loss": 2.388,
+      "step": 8779
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001176360225140713,
+      "loss": 2.2085,
+      "step": 8780
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001176266416510319,
+      "loss": 2.268,
+      "step": 8781
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00011761726078799251,
+      "loss": 2.6046,
+      "step": 8782
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001176078799249531,
+      "loss": 2.29,
+      "step": 8783
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001175984990619137,
+      "loss": 2.7242,
+      "step": 8784
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001175891181988743,
+      "loss": 2.7644,
+      "step": 8785
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00011757973733583489,
+      "loss": 2.4603,
+      "step": 8786
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00011757035647279551,
+      "loss": 2.4071,
+      "step": 8787
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001175609756097561,
+      "loss": 2.2155,
+      "step": 8788
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001175515947467167,
+      "loss": 2.3508,
+      "step": 8789
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001175422138836773,
+      "loss": 2.5398,
+      "step": 8790
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00011753283302063791,
+      "loss": 2.4039,
+      "step": 8791
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001175234521575985,
+      "loss": 2.2661,
+      "step": 8792
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00011751407129455911,
+      "loss": 2.3602,
+      "step": 8793
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.0001175046904315197,
+      "loss": 2.6067,
+      "step": 8794
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011749530956848032,
+      "loss": 2.3884,
+      "step": 8795
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011748592870544091,
+      "loss": 2.8268,
+      "step": 8796
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001174765478424015,
+      "loss": 2.4126,
+      "step": 8797
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001174671669793621,
+      "loss": 2.6003,
+      "step": 8798
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001174577861163227,
+      "loss": 2.5453,
+      "step": 8799
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011744840525328331,
+      "loss": 2.4373,
+      "step": 8800
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001174390243902439,
+      "loss": 2.3911,
+      "step": 8801
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011742964352720451,
+      "loss": 2.4797,
+      "step": 8802
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001174202626641651,
+      "loss": 2.2878,
+      "step": 8803
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011741088180112572,
+      "loss": 2.5319,
+      "step": 8804
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011740150093808631,
+      "loss": 2.2512,
+      "step": 8805
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011739212007504691,
+      "loss": 2.2735,
+      "step": 8806
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001173827392120075,
+      "loss": 2.5368,
+      "step": 8807
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011737335834896812,
+      "loss": 2.2856,
+      "step": 8808
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011736397748592871,
+      "loss": 2.2549,
+      "step": 8809
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001173545966228893,
+      "loss": 2.5114,
+      "step": 8810
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011734521575984991,
+      "loss": 2.4311,
+      "step": 8811
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.0001173358348968105,
+      "loss": 2.4209,
+      "step": 8812
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011732645403377112,
+      "loss": 2.2898,
+      "step": 8813
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011731707317073171,
+      "loss": 2.5487,
+      "step": 8814
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00011730769230769231,
+      "loss": 2.2075,
+      "step": 8815
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001172983114446529,
+      "loss": 2.3418,
+      "step": 8816
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011728893058161352,
+      "loss": 2.2364,
+      "step": 8817
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011727954971857411,
+      "loss": 2.4539,
+      "step": 8818
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011727016885553472,
+      "loss": 2.3297,
+      "step": 8819
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011726078799249531,
+      "loss": 2.4306,
+      "step": 8820
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011725140712945593,
+      "loss": 2.3286,
+      "step": 8821
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011724202626641652,
+      "loss": 2.4701,
+      "step": 8822
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011723264540337711,
+      "loss": 2.6393,
+      "step": 8823
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011722326454033771,
+      "loss": 2.7676,
+      "step": 8824
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.0001172138836772983,
+      "loss": 2.3157,
+      "step": 8825
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011720450281425892,
+      "loss": 2.0578,
+      "step": 8826
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011719512195121951,
+      "loss": 2.4907,
+      "step": 8827
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011718574108818012,
+      "loss": 2.3472,
+      "step": 8828
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011717636022514071,
+      "loss": 2.2623,
+      "step": 8829
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011716697936210133,
+      "loss": 2.4782,
+      "step": 8830
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011715759849906192,
+      "loss": 2.5772,
+      "step": 8831
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011714821763602252,
+      "loss": 2.4777,
+      "step": 8832
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011713883677298311,
+      "loss": 2.4896,
+      "step": 8833
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011712945590994373,
+      "loss": 2.4411,
+      "step": 8834
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011712007504690432,
+      "loss": 2.2501,
+      "step": 8835
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011711069418386491,
+      "loss": 2.6414,
+      "step": 8836
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00011710131332082552,
+      "loss": 2.4635,
+      "step": 8837
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011709193245778611,
+      "loss": 2.2285,
+      "step": 8838
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011708255159474673,
+      "loss": 2.3477,
+      "step": 8839
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011707317073170732,
+      "loss": 2.3201,
+      "step": 8840
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011706378986866792,
+      "loss": 2.7304,
+      "step": 8841
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011705440900562851,
+      "loss": 2.4736,
+      "step": 8842
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011704502814258913,
+      "loss": 2.4537,
+      "step": 8843
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011703564727954972,
+      "loss": 2.4471,
+      "step": 8844
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011702626641651033,
+      "loss": 2.4464,
+      "step": 8845
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011701688555347092,
+      "loss": 2.0799,
+      "step": 8846
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011700750469043154,
+      "loss": 2.5093,
+      "step": 8847
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011699812382739213,
+      "loss": 2.6859,
+      "step": 8848
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011698874296435272,
+      "loss": 2.5741,
+      "step": 8849
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011697936210131332,
+      "loss": 2.6673,
+      "step": 8850
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011696998123827391,
+      "loss": 2.428,
+      "step": 8851
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011696060037523453,
+      "loss": 2.7467,
+      "step": 8852
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011695121951219512,
+      "loss": 2.2424,
+      "step": 8853
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011694183864915573,
+      "loss": 2.3582,
+      "step": 8854
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011693245778611632,
+      "loss": 2.3853,
+      "step": 8855
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011692307692307694,
+      "loss": 2.4069,
+      "step": 8856
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011691369606003753,
+      "loss": 2.4565,
+      "step": 8857
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.00011690431519699813,
+      "loss": 2.5526,
+      "step": 8858
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011689493433395872,
+      "loss": 2.4888,
+      "step": 8859
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011688555347091934,
+      "loss": 2.1633,
+      "step": 8860
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011687617260787993,
+      "loss": 2.3622,
+      "step": 8861
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011686679174484052,
+      "loss": 2.3053,
+      "step": 8862
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011685741088180113,
+      "loss": 2.2636,
+      "step": 8863
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011684803001876172,
+      "loss": 2.2359,
+      "step": 8864
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011683864915572234,
+      "loss": 2.3495,
+      "step": 8865
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011682926829268293,
+      "loss": 2.598,
+      "step": 8866
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011681988742964353,
+      "loss": 2.2502,
+      "step": 8867
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011681050656660412,
+      "loss": 2.8057,
+      "step": 8868
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011680112570356474,
+      "loss": 2.5991,
+      "step": 8869
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011679174484052533,
+      "loss": 2.5579,
+      "step": 8870
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011678236397748595,
+      "loss": 2.5408,
+      "step": 8871
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011677298311444653,
+      "loss": 2.542,
+      "step": 8872
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011676360225140714,
+      "loss": 2.3759,
+      "step": 8873
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011675422138836774,
+      "loss": 2.5534,
+      "step": 8874
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011674484052532833,
+      "loss": 2.2718,
+      "step": 8875
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011673545966228893,
+      "loss": 2.2986,
+      "step": 8876
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011672607879924952,
+      "loss": 2.5872,
+      "step": 8877
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011671669793621014,
+      "loss": 2.4521,
+      "step": 8878
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 0.00011670731707317073,
+      "loss": 2.3031,
+      "step": 8879
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011669793621013135,
+      "loss": 2.6617,
+      "step": 8880
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011668855534709193,
+      "loss": 2.5855,
+      "step": 8881
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011667917448405254,
+      "loss": 2.3844,
+      "step": 8882
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011666979362101314,
+      "loss": 2.4932,
+      "step": 8883
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011666041275797375,
+      "loss": 2.2541,
+      "step": 8884
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011665103189493433,
+      "loss": 2.2659,
+      "step": 8885
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011664165103189495,
+      "loss": 2.5077,
+      "step": 8886
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011663227016885554,
+      "loss": 2.3877,
+      "step": 8887
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011662288930581613,
+      "loss": 2.4203,
+      "step": 8888
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011661350844277675,
+      "loss": 2.3385,
+      "step": 8889
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011660412757973733,
+      "loss": 2.3748,
+      "step": 8890
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011659474671669794,
+      "loss": 2.4565,
+      "step": 8891
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011658536585365853,
+      "loss": 2.3984,
+      "step": 8892
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011657598499061915,
+      "loss": 2.3329,
+      "step": 8893
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011656660412757973,
+      "loss": 2.5947,
+      "step": 8894
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011655722326454035,
+      "loss": 2.4929,
+      "step": 8895
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011654784240150094,
+      "loss": 2.4164,
+      "step": 8896
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011653846153846156,
+      "loss": 2.3921,
+      "step": 8897
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011652908067542215,
+      "loss": 2.1664,
+      "step": 8898
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011651969981238275,
+      "loss": 2.3554,
+      "step": 8899
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011651031894934334,
+      "loss": 2.4369,
+      "step": 8900
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00011650093808630393,
+      "loss": 2.2723,
+      "step": 8901
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011649155722326455,
+      "loss": 2.3551,
+      "step": 8902
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011648217636022514,
+      "loss": 2.4256,
+      "step": 8903
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011647279549718575,
+      "loss": 2.5003,
+      "step": 8904
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011646341463414634,
+      "loss": 2.6496,
+      "step": 8905
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011645403377110696,
+      "loss": 2.3322,
+      "step": 8906
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011644465290806755,
+      "loss": 2.1458,
+      "step": 8907
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011643527204502815,
+      "loss": 2.5293,
+      "step": 8908
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011642589118198874,
+      "loss": 2.5498,
+      "step": 8909
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011641651031894936,
+      "loss": 2.3783,
+      "step": 8910
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011640712945590995,
+      "loss": 2.4532,
+      "step": 8911
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011639774859287056,
+      "loss": 2.5193,
+      "step": 8912
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011638836772983115,
+      "loss": 2.6229,
+      "step": 8913
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011637898686679174,
+      "loss": 2.5572,
+      "step": 8914
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011636960600375236,
+      "loss": 2.5071,
+      "step": 8915
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011636022514071295,
+      "loss": 2.3633,
+      "step": 8916
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011635084427767355,
+      "loss": 2.5476,
+      "step": 8917
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011634146341463414,
+      "loss": 2.4244,
+      "step": 8918
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011633208255159476,
+      "loss": 2.3265,
+      "step": 8919
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011632270168855535,
+      "loss": 2.3338,
+      "step": 8920
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011631332082551596,
+      "loss": 2.5135,
+      "step": 8921
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00011630393996247655,
+      "loss": 2.4296,
+      "step": 8922
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011629455909943717,
+      "loss": 2.5732,
+      "step": 8923
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011628517823639776,
+      "loss": 2.4848,
+      "step": 8924
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011627579737335836,
+      "loss": 2.2871,
+      "step": 8925
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011626641651031895,
+      "loss": 2.2711,
+      "step": 8926
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011625703564727957,
+      "loss": 2.3813,
+      "step": 8927
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011624765478424016,
+      "loss": 2.3916,
+      "step": 8928
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011623827392120075,
+      "loss": 2.2738,
+      "step": 8929
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011622889305816136,
+      "loss": 2.5542,
+      "step": 8930
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011621951219512195,
+      "loss": 2.6961,
+      "step": 8931
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011621013133208257,
+      "loss": 2.5914,
+      "step": 8932
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011620075046904316,
+      "loss": 2.4436,
+      "step": 8933
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011619136960600376,
+      "loss": 2.3474,
+      "step": 8934
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011618198874296435,
+      "loss": 2.4731,
+      "step": 8935
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011617260787992497,
+      "loss": 2.3464,
+      "step": 8936
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011616322701688556,
+      "loss": 2.5086,
+      "step": 8937
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011615384615384617,
+      "loss": 2.3922,
+      "step": 8938
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011614446529080676,
+      "loss": 2.5,
+      "step": 8939
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011613508442776737,
+      "loss": 2.4783,
+      "step": 8940
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011612570356472797,
+      "loss": 2.3428,
+      "step": 8941
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011611632270168856,
+      "loss": 2.3743,
+      "step": 8942
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.00011610694183864916,
+      "loss": 2.3614,
+      "step": 8943
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011609756097560975,
+      "loss": 2.2623,
+      "step": 8944
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011608818011257037,
+      "loss": 2.6701,
+      "step": 8945
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011607879924953096,
+      "loss": 2.3509,
+      "step": 8946
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011606941838649157,
+      "loss": 2.5193,
+      "step": 8947
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011606003752345216,
+      "loss": 2.5042,
+      "step": 8948
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011605065666041277,
+      "loss": 2.4002,
+      "step": 8949
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011604127579737337,
+      "loss": 2.6145,
+      "step": 8950
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011603189493433397,
+      "loss": 2.3661,
+      "step": 8951
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011602251407129456,
+      "loss": 2.725,
+      "step": 8952
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011601313320825518,
+      "loss": 2.2954,
+      "step": 8953
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011600375234521577,
+      "loss": 2.2007,
+      "step": 8954
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011599437148217636,
+      "loss": 2.4723,
+      "step": 8955
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011598499061913697,
+      "loss": 2.4197,
+      "step": 8956
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011597560975609756,
+      "loss": 2.294,
+      "step": 8957
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011596622889305817,
+      "loss": 2.4105,
+      "step": 8958
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011595684803001877,
+      "loss": 2.5099,
+      "step": 8959
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011594746716697937,
+      "loss": 2.4319,
+      "step": 8960
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011593808630393996,
+      "loss": 2.2626,
+      "step": 8961
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011592870544090058,
+      "loss": 2.488,
+      "step": 8962
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011591932457786117,
+      "loss": 2.2126,
+      "step": 8963
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011590994371482177,
+      "loss": 2.4405,
+      "step": 8964
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 0.00011590056285178236,
+      "loss": 2.2782,
+      "step": 8965
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011589118198874298,
+      "loss": 2.3234,
+      "step": 8966
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011588180112570357,
+      "loss": 2.5431,
+      "step": 8967
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011587242026266416,
+      "loss": 2.6542,
+      "step": 8968
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011586303939962477,
+      "loss": 2.3928,
+      "step": 8969
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011585365853658536,
+      "loss": 2.3589,
+      "step": 8970
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011584427767354598,
+      "loss": 2.2463,
+      "step": 8971
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011583489681050657,
+      "loss": 2.7384,
+      "step": 8972
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011582551594746717,
+      "loss": 2.4114,
+      "step": 8973
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011581613508442776,
+      "loss": 2.4178,
+      "step": 8974
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011580675422138838,
+      "loss": 2.3746,
+      "step": 8975
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011579737335834897,
+      "loss": 2.2171,
+      "step": 8976
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011578799249530958,
+      "loss": 2.6807,
+      "step": 8977
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011577861163227017,
+      "loss": 2.7678,
+      "step": 8978
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011576923076923079,
+      "loss": 2.3645,
+      "step": 8979
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011575984990619138,
+      "loss": 2.2364,
+      "step": 8980
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011575046904315197,
+      "loss": 2.1894,
+      "step": 8981
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011574108818011257,
+      "loss": 2.5085,
+      "step": 8982
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011573170731707316,
+      "loss": 2.3519,
+      "step": 8983
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011572232645403378,
+      "loss": 2.3911,
+      "step": 8984
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011571294559099437,
+      "loss": 2.4717,
+      "step": 8985
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00011570356472795498,
+      "loss": 2.4359,
+      "step": 8986
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011569418386491557,
+      "loss": 2.3703,
+      "step": 8987
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011568480300187619,
+      "loss": 2.6206,
+      "step": 8988
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011567542213883678,
+      "loss": 2.4394,
+      "step": 8989
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011566604127579738,
+      "loss": 2.5234,
+      "step": 8990
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011565666041275797,
+      "loss": 2.5031,
+      "step": 8991
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011564727954971859,
+      "loss": 2.4767,
+      "step": 8992
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011563789868667918,
+      "loss": 2.3038,
+      "step": 8993
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011562851782363977,
+      "loss": 2.484,
+      "step": 8994
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011561913696060038,
+      "loss": 2.1056,
+      "step": 8995
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011560975609756097,
+      "loss": 2.3149,
+      "step": 8996
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011560037523452159,
+      "loss": 2.5632,
+      "step": 8997
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011559099437148218,
+      "loss": 2.5445,
+      "step": 8998
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011558161350844278,
+      "loss": 2.2159,
+      "step": 8999
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011557223264540337,
+      "loss": 2.4139,
+      "step": 9000
+    },
+    {
+      "epoch": 4.22,
+      "eval_loss": 2.601161241531372,
+      "eval_runtime": 67.8572,
+      "eval_samples_per_second": 27.926,
+      "eval_steps_per_second": 3.493,
+      "step": 9000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011556285178236399,
+      "loss": 2.7441,
+      "step": 9001
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011555347091932458,
+      "loss": 2.2337,
+      "step": 9002
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011554409005628519,
+      "loss": 2.5304,
+      "step": 9003
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011553470919324578,
+      "loss": 2.3725,
+      "step": 9004
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.0001155253283302064,
+      "loss": 2.5707,
+      "step": 9005
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011551594746716699,
+      "loss": 2.384,
+      "step": 9006
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00011550656660412758,
+      "loss": 2.2776,
+      "step": 9007
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011549718574108818,
+      "loss": 2.3543,
+      "step": 9008
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011548780487804877,
+      "loss": 2.4472,
+      "step": 9009
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011547842401500939,
+      "loss": 2.4962,
+      "step": 9010
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011546904315196998,
+      "loss": 2.4391,
+      "step": 9011
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011545966228893059,
+      "loss": 2.3759,
+      "step": 9012
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011545028142589118,
+      "loss": 2.5518,
+      "step": 9013
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001154409005628518,
+      "loss": 2.4083,
+      "step": 9014
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011543151969981239,
+      "loss": 2.5479,
+      "step": 9015
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011542213883677299,
+      "loss": 2.232,
+      "step": 9016
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011541275797373358,
+      "loss": 2.2728,
+      "step": 9017
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001154033771106942,
+      "loss": 2.2429,
+      "step": 9018
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011539399624765479,
+      "loss": 2.4405,
+      "step": 9019
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 2.376,
+      "step": 9020
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011537523452157599,
+      "loss": 2.3847,
+      "step": 9021
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011536585365853658,
+      "loss": 2.6416,
+      "step": 9022
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001153564727954972,
+      "loss": 2.5733,
+      "step": 9023
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011534709193245779,
+      "loss": 2.5342,
+      "step": 9024
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011533771106941839,
+      "loss": 2.1969,
+      "step": 9025
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011532833020637898,
+      "loss": 2.4565,
+      "step": 9026
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001153189493433396,
+      "loss": 2.3587,
+      "step": 9027
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011530956848030019,
+      "loss": 2.62,
+      "step": 9028
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.00011530018761726081,
+      "loss": 2.5986,
+      "step": 9029
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011529080675422139,
+      "loss": 2.3724,
+      "step": 9030
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.000115281425891182,
+      "loss": 2.5454,
+      "step": 9031
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001152720450281426,
+      "loss": 2.5763,
+      "step": 9032
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011526266416510319,
+      "loss": 2.6235,
+      "step": 9033
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011525328330206379,
+      "loss": 2.3993,
+      "step": 9034
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011524390243902438,
+      "loss": 2.5083,
+      "step": 9035
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.000115234521575985,
+      "loss": 2.7352,
+      "step": 9036
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011522514071294559,
+      "loss": 2.1005,
+      "step": 9037
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011521575984990621,
+      "loss": 2.1758,
+      "step": 9038
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011520637898686679,
+      "loss": 2.3457,
+      "step": 9039
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001151969981238274,
+      "loss": 2.4177,
+      "step": 9040
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.000115187617260788,
+      "loss": 2.4171,
+      "step": 9041
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011517823639774861,
+      "loss": 2.37,
+      "step": 9042
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011516885553470919,
+      "loss": 2.4551,
+      "step": 9043
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011515947467166981,
+      "loss": 2.3563,
+      "step": 9044
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001151500938086304,
+      "loss": 2.4186,
+      "step": 9045
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011514071294559099,
+      "loss": 2.4851,
+      "step": 9046
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011513133208255161,
+      "loss": 2.0927,
+      "step": 9047
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00011512195121951219,
+      "loss": 2.7221,
+      "step": 9048
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001151125703564728,
+      "loss": 2.5781,
+      "step": 9049
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.0001151031894934334,
+      "loss": 2.4815,
+      "step": 9050
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011509380863039401,
+      "loss": 2.5253,
+      "step": 9051
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011508442776735459,
+      "loss": 2.3879,
+      "step": 9052
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011507504690431521,
+      "loss": 2.6126,
+      "step": 9053
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001150656660412758,
+      "loss": 2.4576,
+      "step": 9054
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011505628517823642,
+      "loss": 2.4693,
+      "step": 9055
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011504690431519701,
+      "loss": 2.457,
+      "step": 9056
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011503752345215761,
+      "loss": 2.3421,
+      "step": 9057
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001150281425891182,
+      "loss": 2.6476,
+      "step": 9058
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001150187617260788,
+      "loss": 2.2157,
+      "step": 9059
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011500938086303941,
+      "loss": 2.3361,
+      "step": 9060
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011499999999999999,
+      "loss": 2.5131,
+      "step": 9061
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011499061913696061,
+      "loss": 2.5203,
+      "step": 9062
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001149812382739212,
+      "loss": 2.499,
+      "step": 9063
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011497185741088182,
+      "loss": 2.1332,
+      "step": 9064
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011496247654784241,
+      "loss": 2.5054,
+      "step": 9065
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011495309568480301,
+      "loss": 2.3634,
+      "step": 9066
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.0001149437148217636,
+      "loss": 2.5561,
+      "step": 9067
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011493433395872422,
+      "loss": 2.2278,
+      "step": 9068
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011492495309568481,
+      "loss": 2.5961,
+      "step": 9069
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011491557223264542,
+      "loss": 2.256,
+      "step": 9070
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 0.00011490619136960601,
+      "loss": 2.4566,
+      "step": 9071
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001148968105065666,
+      "loss": 2.2618,
+      "step": 9072
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011488742964352722,
+      "loss": 2.4294,
+      "step": 9073
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011487804878048781,
+      "loss": 2.1675,
+      "step": 9074
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011486866791744841,
+      "loss": 2.2998,
+      "step": 9075
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.000114859287054409,
+      "loss": 2.5295,
+      "step": 9076
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011484990619136962,
+      "loss": 2.524,
+      "step": 9077
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011484052532833021,
+      "loss": 2.2283,
+      "step": 9078
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011483114446529082,
+      "loss": 2.5071,
+      "step": 9079
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011482176360225141,
+      "loss": 2.4387,
+      "step": 9080
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011481238273921203,
+      "loss": 2.1745,
+      "step": 9081
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011480300187617262,
+      "loss": 2.3043,
+      "step": 9082
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011479362101313322,
+      "loss": 2.875,
+      "step": 9083
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011478424015009381,
+      "loss": 2.335,
+      "step": 9084
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.0001147748592870544,
+      "loss": 2.6644,
+      "step": 9085
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011476547842401502,
+      "loss": 2.3569,
+      "step": 9086
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011475609756097561,
+      "loss": 2.5002,
+      "step": 9087
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011474671669793622,
+      "loss": 2.5345,
+      "step": 9088
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011473733583489681,
+      "loss": 2.5743,
+      "step": 9089
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011472795497185743,
+      "loss": 2.5032,
+      "step": 9090
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011471857410881802,
+      "loss": 2.5265,
+      "step": 9091
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00011470919324577862,
+      "loss": 2.3644,
+      "step": 9092
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011469981238273921,
+      "loss": 2.3298,
+      "step": 9093
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011469043151969983,
+      "loss": 2.4106,
+      "step": 9094
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011468105065666042,
+      "loss": 2.435,
+      "step": 9095
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011467166979362103,
+      "loss": 2.473,
+      "step": 9096
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011466228893058162,
+      "loss": 2.2035,
+      "step": 9097
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011465290806754221,
+      "loss": 2.15,
+      "step": 9098
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011464352720450283,
+      "loss": 2.4029,
+      "step": 9099
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011463414634146342,
+      "loss": 2.307,
+      "step": 9100
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011462476547842402,
+      "loss": 2.2177,
+      "step": 9101
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011461538461538461,
+      "loss": 2.3455,
+      "step": 9102
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011460600375234523,
+      "loss": 2.8191,
+      "step": 9103
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011459662288930582,
+      "loss": 2.5072,
+      "step": 9104
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011458724202626643,
+      "loss": 2.3278,
+      "step": 9105
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011457786116322702,
+      "loss": 2.4782,
+      "step": 9106
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011456848030018763,
+      "loss": 2.6495,
+      "step": 9107
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011455909943714823,
+      "loss": 2.4427,
+      "step": 9108
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011454971857410883,
+      "loss": 2.3165,
+      "step": 9109
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011454033771106942,
+      "loss": 2.2539,
+      "step": 9110
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011453095684803001,
+      "loss": 2.5591,
+      "step": 9111
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011452157598499063,
+      "loss": 2.2686,
+      "step": 9112
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011451219512195122,
+      "loss": 2.7227,
+      "step": 9113
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.00011450281425891182,
+      "loss": 2.2268,
+      "step": 9114
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011449343339587242,
+      "loss": 2.6491,
+      "step": 9115
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011448405253283303,
+      "loss": 2.4507,
+      "step": 9116
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011447467166979362,
+      "loss": 2.3659,
+      "step": 9117
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011446529080675423,
+      "loss": 2.3296,
+      "step": 9118
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011445590994371482,
+      "loss": 2.3498,
+      "step": 9119
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011444652908067544,
+      "loss": 2.49,
+      "step": 9120
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011443714821763603,
+      "loss": 2.5263,
+      "step": 9121
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011442776735459663,
+      "loss": 2.3659,
+      "step": 9122
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011441838649155722,
+      "loss": 2.5971,
+      "step": 9123
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011440900562851784,
+      "loss": 2.3504,
+      "step": 9124
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011439962476547843,
+      "loss": 2.4075,
+      "step": 9125
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011439024390243902,
+      "loss": 2.5306,
+      "step": 9126
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011438086303939963,
+      "loss": 2.3259,
+      "step": 9127
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011437148217636022,
+      "loss": 2.5385,
+      "step": 9128
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011436210131332084,
+      "loss": 2.4915,
+      "step": 9129
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011435272045028143,
+      "loss": 2.1975,
+      "step": 9130
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011434333958724203,
+      "loss": 2.5825,
+      "step": 9131
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011433395872420262,
+      "loss": 2.4565,
+      "step": 9132
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011432457786116324,
+      "loss": 2.4759,
+      "step": 9133
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011431519699812383,
+      "loss": 2.306,
+      "step": 9134
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00011430581613508444,
+      "loss": 2.6276,
+      "step": 9135
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011429643527204503,
+      "loss": 2.4158,
+      "step": 9136
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011428705440900565,
+      "loss": 2.6122,
+      "step": 9137
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011427767354596624,
+      "loss": 2.5053,
+      "step": 9138
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011426829268292683,
+      "loss": 2.4866,
+      "step": 9139
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011425891181988743,
+      "loss": 2.344,
+      "step": 9140
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011424953095684802,
+      "loss": 2.158,
+      "step": 9141
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011424015009380864,
+      "loss": 2.4135,
+      "step": 9142
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011423076923076923,
+      "loss": 2.4571,
+      "step": 9143
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011422138836772984,
+      "loss": 2.4195,
+      "step": 9144
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011421200750469043,
+      "loss": 2.4505,
+      "step": 9145
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011420262664165105,
+      "loss": 2.553,
+      "step": 9146
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011419324577861164,
+      "loss": 2.3263,
+      "step": 9147
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011418386491557224,
+      "loss": 2.366,
+      "step": 9148
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011417448405253283,
+      "loss": 2.6621,
+      "step": 9149
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011416510318949345,
+      "loss": 2.3928,
+      "step": 9150
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011415572232645404,
+      "loss": 2.2195,
+      "step": 9151
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011414634146341463,
+      "loss": 2.4165,
+      "step": 9152
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011413696060037524,
+      "loss": 2.3263,
+      "step": 9153
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011412757973733583,
+      "loss": 2.2625,
+      "step": 9154
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011411819887429645,
+      "loss": 2.6318,
+      "step": 9155
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00011410881801125704,
+      "loss": 2.4251,
+      "step": 9156
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011409943714821764,
+      "loss": 2.5591,
+      "step": 9157
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011409005628517823,
+      "loss": 2.8843,
+      "step": 9158
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011408067542213885,
+      "loss": 2.5283,
+      "step": 9159
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011407129455909944,
+      "loss": 2.5208,
+      "step": 9160
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011406191369606005,
+      "loss": 2.5775,
+      "step": 9161
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011405253283302064,
+      "loss": 2.4369,
+      "step": 9162
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011404315196998126,
+      "loss": 2.4757,
+      "step": 9163
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011403377110694185,
+      "loss": 2.2976,
+      "step": 9164
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011402439024390244,
+      "loss": 2.2442,
+      "step": 9165
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011401500938086304,
+      "loss": 2.2747,
+      "step": 9166
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011400562851782363,
+      "loss": 2.6607,
+      "step": 9167
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011399624765478425,
+      "loss": 2.2601,
+      "step": 9168
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011398686679174484,
+      "loss": 2.4397,
+      "step": 9169
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011397748592870545,
+      "loss": 2.5926,
+      "step": 9170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011396810506566604,
+      "loss": 2.5975,
+      "step": 9171
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011395872420262666,
+      "loss": 2.4447,
+      "step": 9172
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011394934333958725,
+      "loss": 2.5018,
+      "step": 9173
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011393996247654785,
+      "loss": 2.6324,
+      "step": 9174
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011393058161350844,
+      "loss": 2.3234,
+      "step": 9175
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011392120075046906,
+      "loss": 2.3895,
+      "step": 9176
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011391181988742965,
+      "loss": 2.4546,
+      "step": 9177
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 0.00011390243902439024,
+      "loss": 2.2939,
+      "step": 9178
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011389305816135085,
+      "loss": 2.338,
+      "step": 9179
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011388367729831144,
+      "loss": 2.6907,
+      "step": 9180
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011387429643527206,
+      "loss": 2.6379,
+      "step": 9181
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011386491557223265,
+      "loss": 2.3986,
+      "step": 9182
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011385553470919325,
+      "loss": 2.2071,
+      "step": 9183
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011384615384615384,
+      "loss": 2.3378,
+      "step": 9184
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011383677298311446,
+      "loss": 2.2852,
+      "step": 9185
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011382739212007505,
+      "loss": 2.4971,
+      "step": 9186
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011381801125703565,
+      "loss": 2.389,
+      "step": 9187
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011380863039399625,
+      "loss": 2.4636,
+      "step": 9188
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011379924953095686,
+      "loss": 2.3061,
+      "step": 9189
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011378986866791745,
+      "loss": 2.3634,
+      "step": 9190
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011378048780487805,
+      "loss": 2.5462,
+      "step": 9191
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011377110694183865,
+      "loss": 2.3762,
+      "step": 9192
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011376172607879924,
+      "loss": 2.6003,
+      "step": 9193
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011375234521575986,
+      "loss": 2.4822,
+      "step": 9194
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011374296435272045,
+      "loss": 2.5455,
+      "step": 9195
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011373358348968107,
+      "loss": 2.2188,
+      "step": 9196
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011372420262664165,
+      "loss": 2.3621,
+      "step": 9197
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011371482176360226,
+      "loss": 2.605,
+      "step": 9198
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00011370544090056285,
+      "loss": 2.4951,
+      "step": 9199
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011369606003752347,
+      "loss": 2.477,
+      "step": 9200
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011368667917448405,
+      "loss": 2.6097,
+      "step": 9201
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011367729831144467,
+      "loss": 2.3367,
+      "step": 9202
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011366791744840526,
+      "loss": 2.4009,
+      "step": 9203
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011365853658536585,
+      "loss": 2.2119,
+      "step": 9204
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011364915572232647,
+      "loss": 2.4985,
+      "step": 9205
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011363977485928705,
+      "loss": 2.6237,
+      "step": 9206
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011363039399624766,
+      "loss": 2.2127,
+      "step": 9207
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011362101313320825,
+      "loss": 2.6953,
+      "step": 9208
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011361163227016887,
+      "loss": 2.7177,
+      "step": 9209
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011360225140712945,
+      "loss": 2.4321,
+      "step": 9210
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011359287054409007,
+      "loss": 2.376,
+      "step": 9211
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011358348968105066,
+      "loss": 2.2388,
+      "step": 9212
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011357410881801128,
+      "loss": 2.2862,
+      "step": 9213
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011356472795497187,
+      "loss": 2.4034,
+      "step": 9214
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011355534709193247,
+      "loss": 2.5184,
+      "step": 9215
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011354596622889306,
+      "loss": 2.4136,
+      "step": 9216
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011353658536585365,
+      "loss": 2.4207,
+      "step": 9217
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011352720450281427,
+      "loss": 2.3691,
+      "step": 9218
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011351782363977485,
+      "loss": 2.2943,
+      "step": 9219
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.00011350844277673547,
+      "loss": 2.4,
+      "step": 9220
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011349906191369606,
+      "loss": 2.2483,
+      "step": 9221
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011348968105065668,
+      "loss": 2.3578,
+      "step": 9222
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011348030018761727,
+      "loss": 2.2542,
+      "step": 9223
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011347091932457787,
+      "loss": 2.4657,
+      "step": 9224
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011346153846153846,
+      "loss": 2.289,
+      "step": 9225
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011345215759849908,
+      "loss": 2.3861,
+      "step": 9226
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011344277673545967,
+      "loss": 2.5327,
+      "step": 9227
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011343339587242028,
+      "loss": 2.3223,
+      "step": 9228
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011342401500938087,
+      "loss": 2.3369,
+      "step": 9229
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011341463414634146,
+      "loss": 2.3633,
+      "step": 9230
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011340525328330208,
+      "loss": 2.3337,
+      "step": 9231
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011339587242026267,
+      "loss": 2.341,
+      "step": 9232
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011338649155722327,
+      "loss": 2.4389,
+      "step": 9233
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011337711069418386,
+      "loss": 2.3055,
+      "step": 9234
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011336772983114448,
+      "loss": 2.4686,
+      "step": 9235
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011335834896810507,
+      "loss": 2.6857,
+      "step": 9236
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011334896810506568,
+      "loss": 2.5667,
+      "step": 9237
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011333958724202627,
+      "loss": 2.3049,
+      "step": 9238
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011333020637898689,
+      "loss": 2.4776,
+      "step": 9239
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011332082551594748,
+      "loss": 2.4459,
+      "step": 9240
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011331144465290808,
+      "loss": 2.691,
+      "step": 9241
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00011330206378986867,
+      "loss": 2.3675,
+      "step": 9242
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011329268292682926,
+      "loss": 2.4396,
+      "step": 9243
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011328330206378988,
+      "loss": 2.383,
+      "step": 9244
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011327392120075047,
+      "loss": 2.4762,
+      "step": 9245
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011326454033771108,
+      "loss": 2.4226,
+      "step": 9246
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011325515947467167,
+      "loss": 2.633,
+      "step": 9247
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011324577861163229,
+      "loss": 2.1849,
+      "step": 9248
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011323639774859288,
+      "loss": 2.4296,
+      "step": 9249
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011322701688555348,
+      "loss": 2.5896,
+      "step": 9250
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011321763602251407,
+      "loss": 2.3695,
+      "step": 9251
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011320825515947469,
+      "loss": 2.2254,
+      "step": 9252
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011319887429643528,
+      "loss": 2.3953,
+      "step": 9253
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011318949343339589,
+      "loss": 2.7324,
+      "step": 9254
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011318011257035648,
+      "loss": 2.5106,
+      "step": 9255
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011317073170731707,
+      "loss": 2.4651,
+      "step": 9256
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011316135084427768,
+      "loss": 2.4623,
+      "step": 9257
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011315196998123828,
+      "loss": 2.4295,
+      "step": 9258
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011314258911819888,
+      "loss": 2.3608,
+      "step": 9259
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011313320825515947,
+      "loss": 2.4531,
+      "step": 9260
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011312382739212009,
+      "loss": 2.3395,
+      "step": 9261
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011311444652908068,
+      "loss": 2.1914,
+      "step": 9262
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 0.00011310506566604128,
+      "loss": 2.3789,
+      "step": 9263
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011309568480300188,
+      "loss": 2.4215,
+      "step": 9264
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001130863039399625,
+      "loss": 2.4451,
+      "step": 9265
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011307692307692308,
+      "loss": 2.179,
+      "step": 9266
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011306754221388369,
+      "loss": 2.4931,
+      "step": 9267
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011305816135084428,
+      "loss": 2.3483,
+      "step": 9268
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011304878048780487,
+      "loss": 2.3432,
+      "step": 9269
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011303939962476549,
+      "loss": 2.3611,
+      "step": 9270
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011303001876172608,
+      "loss": 2.4105,
+      "step": 9271
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011302063789868668,
+      "loss": 2.4537,
+      "step": 9272
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011301125703564728,
+      "loss": 2.5061,
+      "step": 9273
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001130018761726079,
+      "loss": 2.4451,
+      "step": 9274
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011299249530956848,
+      "loss": 2.4129,
+      "step": 9275
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011298311444652909,
+      "loss": 2.3706,
+      "step": 9276
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011297373358348968,
+      "loss": 2.6769,
+      "step": 9277
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001129643527204503,
+      "loss": 2.37,
+      "step": 9278
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011295497185741089,
+      "loss": 2.2257,
+      "step": 9279
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001129455909943715,
+      "loss": 2.5256,
+      "step": 9280
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011293621013133208,
+      "loss": 2.2665,
+      "step": 9281
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011292682926829268,
+      "loss": 2.3412,
+      "step": 9282
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.0001129174484052533,
+      "loss": 2.4082,
+      "step": 9283
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00011290806754221388,
+      "loss": 2.2843,
+      "step": 9284
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011289868667917449,
+      "loss": 2.3484,
+      "step": 9285
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011288930581613508,
+      "loss": 2.4058,
+      "step": 9286
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001128799249530957,
+      "loss": 2.5461,
+      "step": 9287
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011287054409005629,
+      "loss": 2.4808,
+      "step": 9288
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001128611632270169,
+      "loss": 2.2704,
+      "step": 9289
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011285178236397748,
+      "loss": 2.5769,
+      "step": 9290
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001128424015009381,
+      "loss": 2.508,
+      "step": 9291
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011283302063789869,
+      "loss": 2.6077,
+      "step": 9292
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001128236397748593,
+      "loss": 2.5693,
+      "step": 9293
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011281425891181989,
+      "loss": 2.302,
+      "step": 9294
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011280487804878048,
+      "loss": 2.4479,
+      "step": 9295
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001127954971857411,
+      "loss": 2.5225,
+      "step": 9296
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011278611632270169,
+      "loss": 2.3592,
+      "step": 9297
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011277673545966229,
+      "loss": 2.4818,
+      "step": 9298
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011276735459662288,
+      "loss": 2.3265,
+      "step": 9299
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001127579737335835,
+      "loss": 2.2912,
+      "step": 9300
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011274859287054409,
+      "loss": 2.6894,
+      "step": 9301
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001127392120075047,
+      "loss": 2.4964,
+      "step": 9302
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.00011272983114446529,
+      "loss": 2.6662,
+      "step": 9303
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001127204502814259,
+      "loss": 2.5095,
+      "step": 9304
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001127110694183865,
+      "loss": 2.5646,
+      "step": 9305
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001127016885553471,
+      "loss": 2.2973,
+      "step": 9306
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00011269230769230769,
+      "loss": 2.4361,
+      "step": 9307
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00011268292682926828,
+      "loss": 2.3749,
+      "step": 9308
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001126735459662289,
+      "loss": 2.3001,
+      "step": 9309
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00011266416510318949,
+      "loss": 2.3576,
+      "step": 9310
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001126547842401501,
+      "loss": 2.4383,
+      "step": 9311
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00011264540337711069,
+      "loss": 2.5445,
+      "step": 9312
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001126360225140713,
+      "loss": 2.4912,
+      "step": 9313
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001126266416510319,
+      "loss": 2.4809,
+      "step": 9314
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001126172607879925,
+      "loss": 2.3764,
+      "step": 9315
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00011260787992495309,
+      "loss": 2.4347,
+      "step": 9316
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00011259849906191371,
+      "loss": 2.2942,
+      "step": 9317
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001125891181988743,
+      "loss": 2.5443,
+      "step": 9318
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001125797373358349,
+      "loss": 2.6704,
+      "step": 9319
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001125703564727955,
+      "loss": 2.3021,
+      "step": 9320
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00011256097560975609,
+      "loss": 2.4322,
+      "step": 9321
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001125515947467167,
+      "loss": 2.3536,
+      "step": 9322
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001125422138836773,
+      "loss": 2.4981,
+      "step": 9323
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001125328330206379,
+      "loss": 2.432,
+      "step": 9324
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00011252345215759849,
+      "loss": 2.1503,
+      "step": 9325
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00011251407129455911,
+      "loss": 2.702,
+      "step": 9326
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.0001125046904315197,
+      "loss": 2.4301,
+      "step": 9327
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001124953095684803,
+      "loss": 2.4016,
+      "step": 9328
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001124859287054409,
+      "loss": 2.7178,
+      "step": 9329
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011247654784240151,
+      "loss": 2.2372,
+      "step": 9330
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001124671669793621,
+      "loss": 2.472,
+      "step": 9331
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011245778611632271,
+      "loss": 2.4221,
+      "step": 9332
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001124484052532833,
+      "loss": 2.4194,
+      "step": 9333
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011243902439024392,
+      "loss": 2.4406,
+      "step": 9334
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011242964352720451,
+      "loss": 2.2233,
+      "step": 9335
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001124202626641651,
+      "loss": 2.3853,
+      "step": 9336
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001124108818011257,
+      "loss": 2.395,
+      "step": 9337
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001124015009380863,
+      "loss": 2.4688,
+      "step": 9338
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011239212007504691,
+      "loss": 2.424,
+      "step": 9339
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001123827392120075,
+      "loss": 2.1845,
+      "step": 9340
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011237335834896811,
+      "loss": 2.3511,
+      "step": 9341
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001123639774859287,
+      "loss": 2.7675,
+      "step": 9342
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011235459662288932,
+      "loss": 2.449,
+      "step": 9343
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011234521575984991,
+      "loss": 2.489,
+      "step": 9344
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011233583489681051,
+      "loss": 2.6571,
+      "step": 9345
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.0001123264540337711,
+      "loss": 2.2702,
+      "step": 9346
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011231707317073172,
+      "loss": 2.2806,
+      "step": 9347
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00011230769230769231,
+      "loss": 2.419,
+      "step": 9348
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001122983114446529,
+      "loss": 2.3644,
+      "step": 9349
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011228893058161351,
+      "loss": 2.3178,
+      "step": 9350
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001122795497185741,
+      "loss": 2.2937,
+      "step": 9351
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011227016885553472,
+      "loss": 2.6021,
+      "step": 9352
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011226078799249531,
+      "loss": 2.4114,
+      "step": 9353
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011225140712945591,
+      "loss": 2.4465,
+      "step": 9354
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001122420262664165,
+      "loss": 2.5843,
+      "step": 9355
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011223264540337712,
+      "loss": 2.4385,
+      "step": 9356
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011222326454033771,
+      "loss": 2.6393,
+      "step": 9357
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011221388367729833,
+      "loss": 2.3348,
+      "step": 9358
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011220450281425891,
+      "loss": 2.5349,
+      "step": 9359
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011219512195121953,
+      "loss": 2.5777,
+      "step": 9360
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011218574108818012,
+      "loss": 2.5492,
+      "step": 9361
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011217636022514071,
+      "loss": 2.3984,
+      "step": 9362
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011216697936210131,
+      "loss": 2.4366,
+      "step": 9363
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.0001121575984990619,
+      "loss": 2.6986,
+      "step": 9364
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011214821763602252,
+      "loss": 2.3886,
+      "step": 9365
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011213883677298311,
+      "loss": 2.4468,
+      "step": 9366
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011212945590994373,
+      "loss": 2.4136,
+      "step": 9367
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011212007504690431,
+      "loss": 2.3017,
+      "step": 9368
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011211069418386493,
+      "loss": 2.698,
+      "step": 9369
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 0.00011210131332082552,
+      "loss": 2.3053,
+      "step": 9370
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011209193245778614,
+      "loss": 2.6911,
+      "step": 9371
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011208255159474673,
+      "loss": 2.6518,
+      "step": 9372
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011207317073170733,
+      "loss": 2.5149,
+      "step": 9373
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011206378986866792,
+      "loss": 2.424,
+      "step": 9374
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011205440900562851,
+      "loss": 2.5561,
+      "step": 9375
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011204502814258913,
+      "loss": 2.5725,
+      "step": 9376
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011203564727954971,
+      "loss": 2.2844,
+      "step": 9377
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011202626641651033,
+      "loss": 2.5155,
+      "step": 9378
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011201688555347092,
+      "loss": 2.6286,
+      "step": 9379
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011200750469043154,
+      "loss": 2.2495,
+      "step": 9380
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011199812382739213,
+      "loss": 2.2685,
+      "step": 9381
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011198874296435273,
+      "loss": 2.4139,
+      "step": 9382
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011197936210131332,
+      "loss": 2.4598,
+      "step": 9383
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011196998123827394,
+      "loss": 2.389,
+      "step": 9384
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011196060037523453,
+      "loss": 2.3883,
+      "step": 9385
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011195121951219514,
+      "loss": 2.539,
+      "step": 9386
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011194183864915573,
+      "loss": 2.6778,
+      "step": 9387
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011193245778611632,
+      "loss": 2.4687,
+      "step": 9388
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011192307692307694,
+      "loss": 2.7575,
+      "step": 9389
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011191369606003753,
+      "loss": 2.287,
+      "step": 9390
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00011190431519699813,
+      "loss": 2.4892,
+      "step": 9391
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011189493433395872,
+      "loss": 2.5067,
+      "step": 9392
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011188555347091934,
+      "loss": 2.4117,
+      "step": 9393
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011187617260787993,
+      "loss": 2.4168,
+      "step": 9394
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011186679174484054,
+      "loss": 2.8307,
+      "step": 9395
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011185741088180113,
+      "loss": 2.4879,
+      "step": 9396
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011184803001876175,
+      "loss": 2.2854,
+      "step": 9397
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011183864915572234,
+      "loss": 2.5188,
+      "step": 9398
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011182926829268294,
+      "loss": 2.4025,
+      "step": 9399
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011181988742964353,
+      "loss": 2.2726,
+      "step": 9400
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011181050656660412,
+      "loss": 2.5882,
+      "step": 9401
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011180112570356474,
+      "loss": 2.6153,
+      "step": 9402
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011179174484052533,
+      "loss": 2.1541,
+      "step": 9403
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011178236397748594,
+      "loss": 2.5873,
+      "step": 9404
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011177298311444653,
+      "loss": 2.5203,
+      "step": 9405
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011176360225140714,
+      "loss": 2.3404,
+      "step": 9406
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011175422138836774,
+      "loss": 2.4722,
+      "step": 9407
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011174484052532834,
+      "loss": 2.5117,
+      "step": 9408
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011173545966228893,
+      "loss": 2.6931,
+      "step": 9409
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011172607879924955,
+      "loss": 2.4453,
+      "step": 9410
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011171669793621014,
+      "loss": 2.4744,
+      "step": 9411
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.00011170731707317074,
+      "loss": 2.283,
+      "step": 9412
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011169793621013134,
+      "loss": 2.6775,
+      "step": 9413
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011168855534709193,
+      "loss": 2.8207,
+      "step": 9414
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011167917448405254,
+      "loss": 2.4259,
+      "step": 9415
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011166979362101314,
+      "loss": 2.9835,
+      "step": 9416
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011166041275797374,
+      "loss": 2.3866,
+      "step": 9417
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011165103189493433,
+      "loss": 2.3833,
+      "step": 9418
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011164165103189495,
+      "loss": 2.2369,
+      "step": 9419
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011163227016885554,
+      "loss": 2.4742,
+      "step": 9420
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011162288930581614,
+      "loss": 2.4855,
+      "step": 9421
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011161350844277674,
+      "loss": 2.7523,
+      "step": 9422
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011160412757973735,
+      "loss": 2.4586,
+      "step": 9423
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011159474671669794,
+      "loss": 2.3986,
+      "step": 9424
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011158536585365855,
+      "loss": 2.2162,
+      "step": 9425
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011157598499061914,
+      "loss": 2.5072,
+      "step": 9426
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011156660412757973,
+      "loss": 2.6993,
+      "step": 9427
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011155722326454035,
+      "loss": 2.6166,
+      "step": 9428
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011154784240150094,
+      "loss": 2.376,
+      "step": 9429
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011153846153846154,
+      "loss": 2.502,
+      "step": 9430
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011152908067542214,
+      "loss": 2.2774,
+      "step": 9431
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011151969981238275,
+      "loss": 2.3834,
+      "step": 9432
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011151031894934334,
+      "loss": 2.5517,
+      "step": 9433
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00011150093808630395,
+      "loss": 2.517,
+      "step": 9434
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011149155722326454,
+      "loss": 2.7171,
+      "step": 9435
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011148217636022516,
+      "loss": 2.5245,
+      "step": 9436
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011147279549718575,
+      "loss": 2.5386,
+      "step": 9437
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011146341463414635,
+      "loss": 2.3171,
+      "step": 9438
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011145403377110694,
+      "loss": 2.3337,
+      "step": 9439
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011144465290806754,
+      "loss": 2.2628,
+      "step": 9440
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011143527204502815,
+      "loss": 2.368,
+      "step": 9441
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011142589118198874,
+      "loss": 2.6292,
+      "step": 9442
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011141651031894935,
+      "loss": 2.1623,
+      "step": 9443
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011140712945590994,
+      "loss": 2.3058,
+      "step": 9444
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011139774859287056,
+      "loss": 2.5086,
+      "step": 9445
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011138836772983115,
+      "loss": 2.2884,
+      "step": 9446
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011137898686679175,
+      "loss": 2.4771,
+      "step": 9447
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011136960600375234,
+      "loss": 2.4846,
+      "step": 9448
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011136022514071296,
+      "loss": 2.6453,
+      "step": 9449
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011135084427767355,
+      "loss": 2.6198,
+      "step": 9450
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011134146341463416,
+      "loss": 2.3468,
+      "step": 9451
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011133208255159475,
+      "loss": 2.4747,
+      "step": 9452
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011132270168855534,
+      "loss": 2.5112,
+      "step": 9453
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011131332082551596,
+      "loss": 2.2933,
+      "step": 9454
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 0.00011130393996247655,
+      "loss": 2.4907,
+      "step": 9455
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011129455909943715,
+      "loss": 2.3797,
+      "step": 9456
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011128517823639774,
+      "loss": 2.5185,
+      "step": 9457
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011127579737335836,
+      "loss": 2.5159,
+      "step": 9458
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011126641651031895,
+      "loss": 2.4995,
+      "step": 9459
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011125703564727956,
+      "loss": 2.3642,
+      "step": 9460
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011124765478424015,
+      "loss": 2.4164,
+      "step": 9461
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011123827392120077,
+      "loss": 2.2545,
+      "step": 9462
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011122889305816136,
+      "loss": 2.4202,
+      "step": 9463
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011121951219512196,
+      "loss": 2.4644,
+      "step": 9464
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011121013133208255,
+      "loss": 2.3175,
+      "step": 9465
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011120075046904314,
+      "loss": 2.564,
+      "step": 9466
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011119136960600376,
+      "loss": 2.3136,
+      "step": 9467
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011118198874296435,
+      "loss": 2.2301,
+      "step": 9468
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011117260787992496,
+      "loss": 2.2135,
+      "step": 9469
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011116322701688555,
+      "loss": 2.6914,
+      "step": 9470
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011115384615384617,
+      "loss": 2.5068,
+      "step": 9471
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011114446529080676,
+      "loss": 2.3935,
+      "step": 9472
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011113508442776736,
+      "loss": 2.2824,
+      "step": 9473
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011112570356472795,
+      "loss": 2.2831,
+      "step": 9474
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011111632270168857,
+      "loss": 2.6104,
+      "step": 9475
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00011110694183864916,
+      "loss": 2.4379,
+      "step": 9476
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011109756097560977,
+      "loss": 2.3585,
+      "step": 9477
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011108818011257036,
+      "loss": 2.6795,
+      "step": 9478
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011107879924953095,
+      "loss": 2.8627,
+      "step": 9479
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011106941838649157,
+      "loss": 2.2674,
+      "step": 9480
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011106003752345216,
+      "loss": 2.4796,
+      "step": 9481
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011105065666041276,
+      "loss": 2.3654,
+      "step": 9482
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011104127579737335,
+      "loss": 2.5628,
+      "step": 9483
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011103189493433397,
+      "loss": 2.6075,
+      "step": 9484
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011102251407129456,
+      "loss": 2.361,
+      "step": 9485
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011101313320825517,
+      "loss": 2.6014,
+      "step": 9486
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011100375234521576,
+      "loss": 2.2301,
+      "step": 9487
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011099437148217637,
+      "loss": 2.4201,
+      "step": 9488
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011098499061913697,
+      "loss": 2.4496,
+      "step": 9489
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011097560975609757,
+      "loss": 2.4152,
+      "step": 9490
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011096622889305816,
+      "loss": 2.2449,
+      "step": 9491
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011095684803001875,
+      "loss": 2.3493,
+      "step": 9492
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011094746716697937,
+      "loss": 2.3304,
+      "step": 9493
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011093808630393996,
+      "loss": 2.2935,
+      "step": 9494
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011092870544090057,
+      "loss": 2.3252,
+      "step": 9495
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011091932457786116,
+      "loss": 2.5734,
+      "step": 9496
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011090994371482177,
+      "loss": 2.1839,
+      "step": 9497
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.00011090056285178237,
+      "loss": 2.2845,
+      "step": 9498
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011089118198874297,
+      "loss": 2.3014,
+      "step": 9499
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011088180112570356,
+      "loss": 2.5971,
+      "step": 9500
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011087242026266418,
+      "loss": 2.4042,
+      "step": 9501
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011086303939962477,
+      "loss": 2.2901,
+      "step": 9502
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011085365853658537,
+      "loss": 2.2532,
+      "step": 9503
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011084427767354597,
+      "loss": 2.3493,
+      "step": 9504
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011083489681050656,
+      "loss": 2.5343,
+      "step": 9505
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011082551594746717,
+      "loss": 2.4466,
+      "step": 9506
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011081613508442777,
+      "loss": 2.4852,
+      "step": 9507
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011080675422138837,
+      "loss": 2.2229,
+      "step": 9508
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011079737335834896,
+      "loss": 2.5172,
+      "step": 9509
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011078799249530958,
+      "loss": 2.2699,
+      "step": 9510
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011077861163227017,
+      "loss": 2.4986,
+      "step": 9511
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011076923076923077,
+      "loss": 2.1946,
+      "step": 9512
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011075984990619137,
+      "loss": 2.2353,
+      "step": 9513
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011075046904315198,
+      "loss": 2.2048,
+      "step": 9514
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011074108818011257,
+      "loss": 2.1958,
+      "step": 9515
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011073170731707319,
+      "loss": 2.4111,
+      "step": 9516
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011072232645403377,
+      "loss": 2.2389,
+      "step": 9517
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011071294559099436,
+      "loss": 2.3795,
+      "step": 9518
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.00011070356472795498,
+      "loss": 2.2433,
+      "step": 9519
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011069418386491557,
+      "loss": 2.3058,
+      "step": 9520
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011068480300187617,
+      "loss": 2.3975,
+      "step": 9521
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011067542213883676,
+      "loss": 2.3015,
+      "step": 9522
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011066604127579738,
+      "loss": 2.6572,
+      "step": 9523
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011065666041275797,
+      "loss": 2.4977,
+      "step": 9524
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011064727954971859,
+      "loss": 2.4353,
+      "step": 9525
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011063789868667917,
+      "loss": 2.555,
+      "step": 9526
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011062851782363979,
+      "loss": 2.5683,
+      "step": 9527
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011061913696060038,
+      "loss": 2.5735,
+      "step": 9528
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.000110609756097561,
+      "loss": 2.7012,
+      "step": 9529
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011060037523452157,
+      "loss": 2.4845,
+      "step": 9530
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011059099437148216,
+      "loss": 2.4273,
+      "step": 9531
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011058161350844278,
+      "loss": 2.3007,
+      "step": 9532
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011057223264540337,
+      "loss": 2.2082,
+      "step": 9533
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011056285178236399,
+      "loss": 2.507,
+      "step": 9534
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011055347091932457,
+      "loss": 2.393,
+      "step": 9535
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011054409005628519,
+      "loss": 2.631,
+      "step": 9536
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011053470919324578,
+      "loss": 2.2576,
+      "step": 9537
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.0001105253283302064,
+      "loss": 2.2534,
+      "step": 9538
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011051594746716699,
+      "loss": 2.3589,
+      "step": 9539
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00011050656660412759,
+      "loss": 2.4275,
+      "step": 9540
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011049718574108818,
+      "loss": 2.5507,
+      "step": 9541
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001104878048780488,
+      "loss": 2.657,
+      "step": 9542
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011047842401500939,
+      "loss": 2.5521,
+      "step": 9543
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011046904315197,
+      "loss": 2.4566,
+      "step": 9544
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011045966228893059,
+      "loss": 2.4503,
+      "step": 9545
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011045028142589118,
+      "loss": 2.4949,
+      "step": 9546
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001104409005628518,
+      "loss": 2.3197,
+      "step": 9547
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011043151969981239,
+      "loss": 2.4278,
+      "step": 9548
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011042213883677299,
+      "loss": 2.1514,
+      "step": 9549
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011041275797373358,
+      "loss": 2.5106,
+      "step": 9550
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001104033771106942,
+      "loss": 2.5115,
+      "step": 9551
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011039399624765479,
+      "loss": 2.6112,
+      "step": 9552
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001103846153846154,
+      "loss": 2.3851,
+      "step": 9553
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011037523452157599,
+      "loss": 2.4247,
+      "step": 9554
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001103658536585366,
+      "loss": 2.5466,
+      "step": 9555
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001103564727954972,
+      "loss": 2.4133,
+      "step": 9556
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001103470919324578,
+      "loss": 2.4939,
+      "step": 9557
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011033771106941839,
+      "loss": 2.6802,
+      "step": 9558
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011032833020637898,
+      "loss": 2.5016,
+      "step": 9559
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001103189493433396,
+      "loss": 2.313,
+      "step": 9560
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.00011030956848030019,
+      "loss": 2.3999,
+      "step": 9561
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 0.0001103001876172608,
+      "loss": 2.1531,
+      "step": 9562
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00011029080675422139,
+      "loss": 2.5182,
+      "step": 9563
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.000110281425891182,
+      "loss": 2.6013,
+      "step": 9564
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001102720450281426,
+      "loss": 2.2712,
+      "step": 9565
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001102626641651032,
+      "loss": 2.5246,
+      "step": 9566
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00011025328330206379,
+      "loss": 2.4379,
+      "step": 9567
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00011024390243902441,
+      "loss": 2.605,
+      "step": 9568
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.000110234521575985,
+      "loss": 2.6591,
+      "step": 9569
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001102251407129456,
+      "loss": 2.5486,
+      "step": 9570
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001102157598499062,
+      "loss": 2.51,
+      "step": 9571
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00011020637898686679,
+      "loss": 2.4681,
+      "step": 9572
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001101969981238274,
+      "loss": 2.3758,
+      "step": 9573
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.000110187617260788,
+      "loss": 2.5794,
+      "step": 9574
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001101782363977486,
+      "loss": 2.3318,
+      "step": 9575
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00011016885553470919,
+      "loss": 2.3037,
+      "step": 9576
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00011015947467166981,
+      "loss": 2.5505,
+      "step": 9577
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001101500938086304,
+      "loss": 2.3377,
+      "step": 9578
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.000110140712945591,
+      "loss": 2.3485,
+      "step": 9579
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001101313320825516,
+      "loss": 2.6021,
+      "step": 9580
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00011012195121951221,
+      "loss": 2.4147,
+      "step": 9581
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.0001101125703564728,
+      "loss": 2.4523,
+      "step": 9582
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00011010318949343341,
+      "loss": 2.4469,
+      "step": 9583
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.000110093808630394,
+      "loss": 2.2551,
+      "step": 9584
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00011008442776735459,
+      "loss": 2.3487,
+      "step": 9585
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00011007504690431521,
+      "loss": 2.5481,
+      "step": 9586
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001100656660412758,
+      "loss": 2.3798,
+      "step": 9587
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001100562851782364,
+      "loss": 2.4871,
+      "step": 9588
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.000110046904315197,
+      "loss": 2.5572,
+      "step": 9589
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00011003752345215761,
+      "loss": 2.6194,
+      "step": 9590
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001100281425891182,
+      "loss": 2.288,
+      "step": 9591
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00011001876172607881,
+      "loss": 2.2299,
+      "step": 9592
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001100093808630394,
+      "loss": 2.5392,
+      "step": 9593
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 2.5251,
+      "step": 9594
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00010999061913696061,
+      "loss": 2.3918,
+      "step": 9595
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00010998123827392121,
+      "loss": 2.5593,
+      "step": 9596
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001099718574108818,
+      "loss": 2.2866,
+      "step": 9597
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001099624765478424,
+      "loss": 2.4904,
+      "step": 9598
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00010995309568480301,
+      "loss": 2.4291,
+      "step": 9599
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001099437148217636,
+      "loss": 2.381,
+      "step": 9600
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00010993433395872421,
+      "loss": 2.6444,
+      "step": 9601
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.0001099249530956848,
+      "loss": 2.3989,
+      "step": 9602
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00010991557223264542,
+      "loss": 2.4465,
+      "step": 9603
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00010990619136960601,
+      "loss": 2.5971,
+      "step": 9604
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010989681050656661,
+      "loss": 2.5057,
+      "step": 9605
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0001098874296435272,
+      "loss": 2.2751,
+      "step": 9606
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010987804878048782,
+      "loss": 2.5127,
+      "step": 9607
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010986866791744841,
+      "loss": 2.4028,
+      "step": 9608
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010985928705440902,
+      "loss": 2.6204,
+      "step": 9609
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010984990619136961,
+      "loss": 2.4489,
+      "step": 9610
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0001098405253283302,
+      "loss": 2.3147,
+      "step": 9611
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010983114446529082,
+      "loss": 2.4144,
+      "step": 9612
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010982176360225141,
+      "loss": 2.4528,
+      "step": 9613
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010981238273921201,
+      "loss": 2.4298,
+      "step": 9614
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.0001098030018761726,
+      "loss": 2.6166,
+      "step": 9615
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010979362101313322,
+      "loss": 2.1376,
+      "step": 9616
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010978424015009381,
+      "loss": 2.312,
+      "step": 9617
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010977485928705442,
+      "loss": 2.4193,
+      "step": 9618
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010976547842401501,
+      "loss": 2.6225,
+      "step": 9619
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010975609756097563,
+      "loss": 2.559,
+      "step": 9620
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010974671669793622,
+      "loss": 2.2344,
+      "step": 9621
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010973733583489682,
+      "loss": 2.5447,
+      "step": 9622
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010972795497185741,
+      "loss": 2.4744,
+      "step": 9623
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.000109718574108818,
+      "loss": 2.4878,
+      "step": 9624
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00010970919324577862,
+      "loss": 2.5194,
+      "step": 9625
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010969981238273921,
+      "loss": 2.6624,
+      "step": 9626
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010969043151969982,
+      "loss": 2.5212,
+      "step": 9627
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010968105065666041,
+      "loss": 2.4555,
+      "step": 9628
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010967166979362103,
+      "loss": 2.5049,
+      "step": 9629
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010966228893058162,
+      "loss": 2.5553,
+      "step": 9630
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010965290806754222,
+      "loss": 2.5909,
+      "step": 9631
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010964352720450281,
+      "loss": 2.2962,
+      "step": 9632
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010963414634146343,
+      "loss": 2.5912,
+      "step": 9633
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010962476547842402,
+      "loss": 2.6163,
+      "step": 9634
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010961538461538463,
+      "loss": 2.7781,
+      "step": 9635
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010960600375234522,
+      "loss": 2.503,
+      "step": 9636
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010959662288930581,
+      "loss": 2.4026,
+      "step": 9637
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010958724202626643,
+      "loss": 2.1945,
+      "step": 9638
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010957786116322702,
+      "loss": 2.43,
+      "step": 9639
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010956848030018762,
+      "loss": 2.4207,
+      "step": 9640
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010955909943714821,
+      "loss": 2.209,
+      "step": 9641
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010954971857410883,
+      "loss": 2.4508,
+      "step": 9642
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010954033771106942,
+      "loss": 2.3728,
+      "step": 9643
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010953095684803003,
+      "loss": 2.5115,
+      "step": 9644
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010952157598499062,
+      "loss": 2.4091,
+      "step": 9645
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010951219512195123,
+      "loss": 2.5407,
+      "step": 9646
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 0.00010950281425891183,
+      "loss": 2.309,
+      "step": 9647
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010949343339587243,
+      "loss": 2.3052,
+      "step": 9648
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010948405253283302,
+      "loss": 2.8406,
+      "step": 9649
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010947467166979361,
+      "loss": 2.5287,
+      "step": 9650
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010946529080675423,
+      "loss": 2.7274,
+      "step": 9651
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010945590994371482,
+      "loss": 2.5359,
+      "step": 9652
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010944652908067543,
+      "loss": 2.4642,
+      "step": 9653
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010943714821763602,
+      "loss": 2.5966,
+      "step": 9654
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010942776735459663,
+      "loss": 2.341,
+      "step": 9655
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010941838649155723,
+      "loss": 2.5005,
+      "step": 9656
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010940900562851783,
+      "loss": 2.4197,
+      "step": 9657
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010939962476547842,
+      "loss": 2.271,
+      "step": 9658
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010939024390243904,
+      "loss": 2.4495,
+      "step": 9659
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010938086303939963,
+      "loss": 2.1892,
+      "step": 9660
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010937148217636023,
+      "loss": 2.4102,
+      "step": 9661
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010936210131332082,
+      "loss": 2.377,
+      "step": 9662
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010935272045028142,
+      "loss": 2.693,
+      "step": 9663
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010934333958724203,
+      "loss": 2.4121,
+      "step": 9664
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010933395872420262,
+      "loss": 2.4271,
+      "step": 9665
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010932457786116323,
+      "loss": 2.7349,
+      "step": 9666
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010931519699812382,
+      "loss": 2.7282,
+      "step": 9667
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00010930581613508444,
+      "loss": 2.3655,
+      "step": 9668
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010929643527204503,
+      "loss": 2.308,
+      "step": 9669
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010928705440900563,
+      "loss": 2.3523,
+      "step": 9670
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010927767354596622,
+      "loss": 2.7572,
+      "step": 9671
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010926829268292684,
+      "loss": 2.3926,
+      "step": 9672
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010925891181988743,
+      "loss": 2.6194,
+      "step": 9673
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010924953095684805,
+      "loss": 2.4758,
+      "step": 9674
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010924015009380863,
+      "loss": 2.5886,
+      "step": 9675
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010923076923076922,
+      "loss": 2.1763,
+      "step": 9676
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010922138836772984,
+      "loss": 2.5201,
+      "step": 9677
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010921200750469043,
+      "loss": 2.4501,
+      "step": 9678
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010920262664165103,
+      "loss": 2.2757,
+      "step": 9679
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010919324577861162,
+      "loss": 2.6837,
+      "step": 9680
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010918386491557224,
+      "loss": 2.2908,
+      "step": 9681
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010917448405253283,
+      "loss": 2.6681,
+      "step": 9682
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010916510318949345,
+      "loss": 2.4252,
+      "step": 9683
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010915572232645403,
+      "loss": 2.6045,
+      "step": 9684
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010914634146341465,
+      "loss": 2.3483,
+      "step": 9685
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010913696060037524,
+      "loss": 2.3416,
+      "step": 9686
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010912757973733586,
+      "loss": 2.698,
+      "step": 9687
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010911819887429643,
+      "loss": 2.3129,
+      "step": 9688
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00010910881801125702,
+      "loss": 2.382,
+      "step": 9689
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010909943714821764,
+      "loss": 2.4557,
+      "step": 9690
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010909005628517823,
+      "loss": 2.3714,
+      "step": 9691
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010908067542213885,
+      "loss": 2.5103,
+      "step": 9692
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010907129455909943,
+      "loss": 2.2345,
+      "step": 9693
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010906191369606005,
+      "loss": 2.3925,
+      "step": 9694
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010905253283302064,
+      "loss": 2.5156,
+      "step": 9695
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010904315196998126,
+      "loss": 2.2524,
+      "step": 9696
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010903377110694183,
+      "loss": 2.3857,
+      "step": 9697
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010902439024390245,
+      "loss": 2.3552,
+      "step": 9698
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010901500938086304,
+      "loss": 2.3648,
+      "step": 9699
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010900562851782366,
+      "loss": 2.3977,
+      "step": 9700
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010899624765478425,
+      "loss": 2.4644,
+      "step": 9701
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010898686679174483,
+      "loss": 2.4908,
+      "step": 9702
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010897748592870545,
+      "loss": 2.275,
+      "step": 9703
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010896810506566604,
+      "loss": 2.4153,
+      "step": 9704
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010895872420262666,
+      "loss": 2.6083,
+      "step": 9705
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010894934333958725,
+      "loss": 2.5066,
+      "step": 9706
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010893996247654785,
+      "loss": 2.44,
+      "step": 9707
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010893058161350844,
+      "loss": 2.5776,
+      "step": 9708
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010892120075046906,
+      "loss": 2.516,
+      "step": 9709
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010891181988742965,
+      "loss": 2.4971,
+      "step": 9710
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.00010890243902439026,
+      "loss": 2.586,
+      "step": 9711
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010889305816135085,
+      "loss": 2.7053,
+      "step": 9712
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010888367729831146,
+      "loss": 2.0598,
+      "step": 9713
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010887429643527206,
+      "loss": 2.1752,
+      "step": 9714
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010886491557223265,
+      "loss": 2.4988,
+      "step": 9715
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010885553470919325,
+      "loss": 2.3748,
+      "step": 9716
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010884615384615384,
+      "loss": 2.4185,
+      "step": 9717
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010883677298311446,
+      "loss": 2.8851,
+      "step": 9718
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010882739212007505,
+      "loss": 2.3899,
+      "step": 9719
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010881801125703566,
+      "loss": 2.4988,
+      "step": 9720
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010880863039399625,
+      "loss": 2.2756,
+      "step": 9721
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010879924953095686,
+      "loss": 2.5584,
+      "step": 9722
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010878986866791746,
+      "loss": 2.492,
+      "step": 9723
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010878048780487806,
+      "loss": 2.3544,
+      "step": 9724
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010877110694183865,
+      "loss": 2.6304,
+      "step": 9725
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010876172607879927,
+      "loss": 2.7345,
+      "step": 9726
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010875234521575986,
+      "loss": 2.1783,
+      "step": 9727
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010874296435272045,
+      "loss": 2.1391,
+      "step": 9728
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010873358348968106,
+      "loss": 2.1662,
+      "step": 9729
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010872420262664165,
+      "loss": 2.4958,
+      "step": 9730
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010871482176360226,
+      "loss": 2.5255,
+      "step": 9731
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00010870544090056286,
+      "loss": 2.3847,
+      "step": 9732
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010869606003752346,
+      "loss": 2.4541,
+      "step": 9733
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010868667917448405,
+      "loss": 2.29,
+      "step": 9734
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010867729831144467,
+      "loss": 2.4085,
+      "step": 9735
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010866791744840526,
+      "loss": 2.3345,
+      "step": 9736
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010865853658536586,
+      "loss": 2.4962,
+      "step": 9737
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010864915572232645,
+      "loss": 2.3274,
+      "step": 9738
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010863977485928707,
+      "loss": 2.5693,
+      "step": 9739
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010863039399624766,
+      "loss": 2.5876,
+      "step": 9740
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010862101313320827,
+      "loss": 2.4556,
+      "step": 9741
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010861163227016886,
+      "loss": 2.3315,
+      "step": 9742
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010860225140712945,
+      "loss": 2.5981,
+      "step": 9743
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010859287054409007,
+      "loss": 2.5351,
+      "step": 9744
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010858348968105066,
+      "loss": 2.2181,
+      "step": 9745
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010857410881801126,
+      "loss": 2.609,
+      "step": 9746
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010856472795497185,
+      "loss": 2.4217,
+      "step": 9747
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010855534709193247,
+      "loss": 2.4954,
+      "step": 9748
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010854596622889306,
+      "loss": 2.2738,
+      "step": 9749
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010853658536585367,
+      "loss": 2.6165,
+      "step": 9750
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010852720450281426,
+      "loss": 2.6483,
+      "step": 9751
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010851782363977488,
+      "loss": 2.6977,
+      "step": 9752
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 0.00010850844277673547,
+      "loss": 2.3707,
+      "step": 9753
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010849906191369607,
+      "loss": 2.6836,
+      "step": 9754
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010848968105065666,
+      "loss": 2.5147,
+      "step": 9755
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010848030018761725,
+      "loss": 2.5819,
+      "step": 9756
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010847091932457787,
+      "loss": 2.1753,
+      "step": 9757
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010846153846153846,
+      "loss": 2.3559,
+      "step": 9758
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010845215759849907,
+      "loss": 2.316,
+      "step": 9759
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010844277673545966,
+      "loss": 2.4574,
+      "step": 9760
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010843339587242028,
+      "loss": 2.5204,
+      "step": 9761
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010842401500938087,
+      "loss": 2.4093,
+      "step": 9762
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010841463414634147,
+      "loss": 2.4662,
+      "step": 9763
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010840525328330206,
+      "loss": 2.4989,
+      "step": 9764
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010839587242026268,
+      "loss": 2.3985,
+      "step": 9765
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010838649155722327,
+      "loss": 2.5292,
+      "step": 9766
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010837711069418388,
+      "loss": 2.6072,
+      "step": 9767
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010836772983114447,
+      "loss": 2.693,
+      "step": 9768
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010835834896810506,
+      "loss": 2.5109,
+      "step": 9769
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010834896810506568,
+      "loss": 2.2668,
+      "step": 9770
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010833958724202627,
+      "loss": 2.5157,
+      "step": 9771
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010833020637898687,
+      "loss": 2.6777,
+      "step": 9772
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010832082551594746,
+      "loss": 2.3346,
+      "step": 9773
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010831144465290808,
+      "loss": 2.5586,
+      "step": 9774
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.00010830206378986867,
+      "loss": 2.5155,
+      "step": 9775
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010829268292682928,
+      "loss": 2.5377,
+      "step": 9776
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010828330206378987,
+      "loss": 2.3401,
+      "step": 9777
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010827392120075049,
+      "loss": 2.3935,
+      "step": 9778
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010826454033771108,
+      "loss": 2.4989,
+      "step": 9779
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010825515947467168,
+      "loss": 2.3195,
+      "step": 9780
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010824577861163227,
+      "loss": 2.2512,
+      "step": 9781
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010823639774859286,
+      "loss": 2.3395,
+      "step": 9782
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010822701688555348,
+      "loss": 2.4072,
+      "step": 9783
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010821763602251407,
+      "loss": 2.5995,
+      "step": 9784
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010820825515947468,
+      "loss": 2.7583,
+      "step": 9785
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010819887429643527,
+      "loss": 2.5274,
+      "step": 9786
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010818949343339589,
+      "loss": 2.3886,
+      "step": 9787
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010818011257035648,
+      "loss": 2.4346,
+      "step": 9788
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010817073170731708,
+      "loss": 2.3111,
+      "step": 9789
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010816135084427767,
+      "loss": 2.4705,
+      "step": 9790
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010815196998123829,
+      "loss": 2.1996,
+      "step": 9791
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010814258911819888,
+      "loss": 2.2788,
+      "step": 9792
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010813320825515949,
+      "loss": 2.472,
+      "step": 9793
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010812382739212008,
+      "loss": 2.4199,
+      "step": 9794
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010811444652908067,
+      "loss": 2.5225,
+      "step": 9795
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00010810506566604129,
+      "loss": 2.4727,
+      "step": 9796
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010809568480300188,
+      "loss": 2.4718,
+      "step": 9797
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010808630393996248,
+      "loss": 2.1586,
+      "step": 9798
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010807692307692307,
+      "loss": 2.5545,
+      "step": 9799
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010806754221388369,
+      "loss": 2.6093,
+      "step": 9800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010805816135084428,
+      "loss": 2.3362,
+      "step": 9801
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010804878048780489,
+      "loss": 2.1396,
+      "step": 9802
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010803939962476548,
+      "loss": 2.1371,
+      "step": 9803
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0001080300187617261,
+      "loss": 2.2736,
+      "step": 9804
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010802063789868669,
+      "loss": 2.4619,
+      "step": 9805
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010801125703564729,
+      "loss": 2.876,
+      "step": 9806
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010800187617260788,
+      "loss": 2.6454,
+      "step": 9807
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010799249530956847,
+      "loss": 2.3504,
+      "step": 9808
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010798311444652909,
+      "loss": 2.4142,
+      "step": 9809
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010797373358348968,
+      "loss": 2.2995,
+      "step": 9810
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010796435272045028,
+      "loss": 2.3381,
+      "step": 9811
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010795497185741088,
+      "loss": 2.1807,
+      "step": 9812
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0001079455909943715,
+      "loss": 2.3237,
+      "step": 9813
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010793621013133208,
+      "loss": 2.8631,
+      "step": 9814
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010792682926829269,
+      "loss": 2.4157,
+      "step": 9815
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00010791744840525328,
+      "loss": 2.6075,
+      "step": 9816
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.0001079080675422139,
+      "loss": 2.6786,
+      "step": 9817
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010789868667917449,
+      "loss": 2.549,
+      "step": 9818
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001078893058161351,
+      "loss": 2.336,
+      "step": 9819
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010787992495309568,
+      "loss": 2.3505,
+      "step": 9820
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010787054409005628,
+      "loss": 2.5441,
+      "step": 9821
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001078611632270169,
+      "loss": 2.4352,
+      "step": 9822
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010785178236397748,
+      "loss": 2.3947,
+      "step": 9823
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010784240150093809,
+      "loss": 2.5884,
+      "step": 9824
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010783302063789868,
+      "loss": 2.52,
+      "step": 9825
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001078236397748593,
+      "loss": 2.3592,
+      "step": 9826
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010781425891181989,
+      "loss": 2.5197,
+      "step": 9827
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001078048780487805,
+      "loss": 2.6564,
+      "step": 9828
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010779549718574108,
+      "loss": 2.4199,
+      "step": 9829
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001077861163227017,
+      "loss": 2.4571,
+      "step": 9830
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001077767354596623,
+      "loss": 2.4475,
+      "step": 9831
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010776735459662291,
+      "loss": 2.5994,
+      "step": 9832
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010775797373358349,
+      "loss": 2.3289,
+      "step": 9833
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010774859287054408,
+      "loss": 2.364,
+      "step": 9834
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001077392120075047,
+      "loss": 2.756,
+      "step": 9835
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010772983114446529,
+      "loss": 2.3589,
+      "step": 9836
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001077204502814259,
+      "loss": 2.2331,
+      "step": 9837
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.00010771106941838648,
+      "loss": 2.5318,
+      "step": 9838
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 0.0001077016885553471,
+      "loss": 2.4984,
+      "step": 9839
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001076923076923077,
+      "loss": 2.4986,
+      "step": 9840
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010768292682926831,
+      "loss": 2.5207,
+      "step": 9841
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010767354596622889,
+      "loss": 2.253,
+      "step": 9842
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010766416510318951,
+      "loss": 2.5666,
+      "step": 9843
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001076547842401501,
+      "loss": 2.0924,
+      "step": 9844
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010764540337711072,
+      "loss": 2.376,
+      "step": 9845
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010763602251407129,
+      "loss": 2.4325,
+      "step": 9846
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010762664165103188,
+      "loss": 2.223,
+      "step": 9847
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001076172607879925,
+      "loss": 2.6239,
+      "step": 9848
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010760787992495309,
+      "loss": 2.3624,
+      "step": 9849
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010759849906191371,
+      "loss": 2.5547,
+      "step": 9850
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010758911819887429,
+      "loss": 2.2441,
+      "step": 9851
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001075797373358349,
+      "loss": 2.5267,
+      "step": 9852
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001075703564727955,
+      "loss": 2.7944,
+      "step": 9853
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010756097560975612,
+      "loss": 2.3797,
+      "step": 9854
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010755159474671669,
+      "loss": 2.3713,
+      "step": 9855
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010754221388367731,
+      "loss": 2.486,
+      "step": 9856
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.0001075328330206379,
+      "loss": 2.4025,
+      "step": 9857
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010752345215759852,
+      "loss": 2.3619,
+      "step": 9858
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010751407129455911,
+      "loss": 2.252,
+      "step": 9859
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00010750469043151969,
+      "loss": 2.3264,
+      "step": 9860
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001074953095684803,
+      "loss": 2.62,
+      "step": 9861
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001074859287054409,
+      "loss": 2.492,
+      "step": 9862
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010747654784240152,
+      "loss": 2.6921,
+      "step": 9863
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010746716697936209,
+      "loss": 2.7119,
+      "step": 9864
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010745778611632271,
+      "loss": 2.5389,
+      "step": 9865
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001074484052532833,
+      "loss": 2.4596,
+      "step": 9866
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010743902439024392,
+      "loss": 2.4193,
+      "step": 9867
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010742964352720451,
+      "loss": 2.7152,
+      "step": 9868
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010742026266416512,
+      "loss": 2.2493,
+      "step": 9869
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001074108818011257,
+      "loss": 2.5755,
+      "step": 9870
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010740150093808632,
+      "loss": 2.6677,
+      "step": 9871
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010739212007504692,
+      "loss": 2.4487,
+      "step": 9872
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010738273921200749,
+      "loss": 2.6214,
+      "step": 9873
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010737335834896811,
+      "loss": 2.2642,
+      "step": 9874
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001073639774859287,
+      "loss": 2.3883,
+      "step": 9875
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010735459662288932,
+      "loss": 2.5354,
+      "step": 9876
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010734521575984991,
+      "loss": 2.2553,
+      "step": 9877
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010733583489681052,
+      "loss": 2.3067,
+      "step": 9878
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001073264540337711,
+      "loss": 2.4035,
+      "step": 9879
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010731707317073172,
+      "loss": 2.679,
+      "step": 9880
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.00010730769230769231,
+      "loss": 2.5192,
+      "step": 9881
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010729831144465292,
+      "loss": 2.2561,
+      "step": 9882
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010728893058161351,
+      "loss": 2.448,
+      "step": 9883
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010727954971857413,
+      "loss": 2.3664,
+      "step": 9884
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010727016885553472,
+      "loss": 2.3928,
+      "step": 9885
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010726078799249531,
+      "loss": 2.5355,
+      "step": 9886
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010725140712945591,
+      "loss": 2.6956,
+      "step": 9887
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.0001072420262664165,
+      "loss": 2.5795,
+      "step": 9888
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010723264540337712,
+      "loss": 2.4436,
+      "step": 9889
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010722326454033771,
+      "loss": 2.1239,
+      "step": 9890
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010721388367729832,
+      "loss": 2.6205,
+      "step": 9891
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010720450281425891,
+      "loss": 2.4856,
+      "step": 9892
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010719512195121953,
+      "loss": 2.6449,
+      "step": 9893
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010718574108818012,
+      "loss": 2.3688,
+      "step": 9894
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010717636022514072,
+      "loss": 2.4874,
+      "step": 9895
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010716697936210131,
+      "loss": 2.5166,
+      "step": 9896
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010715759849906193,
+      "loss": 2.301,
+      "step": 9897
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010714821763602252,
+      "loss": 2.6314,
+      "step": 9898
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010713883677298311,
+      "loss": 2.6006,
+      "step": 9899
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010712945590994372,
+      "loss": 2.5097,
+      "step": 9900
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010712007504690431,
+      "loss": 2.2509,
+      "step": 9901
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010711069418386493,
+      "loss": 2.3199,
+      "step": 9902
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00010710131332082552,
+      "loss": 2.9008,
+      "step": 9903
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010709193245778612,
+      "loss": 2.6122,
+      "step": 9904
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010708255159474671,
+      "loss": 2.3136,
+      "step": 9905
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010707317073170733,
+      "loss": 2.4415,
+      "step": 9906
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010706378986866792,
+      "loss": 2.6071,
+      "step": 9907
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010705440900562853,
+      "loss": 2.4494,
+      "step": 9908
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010704502814258912,
+      "loss": 2.61,
+      "step": 9909
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010703564727954974,
+      "loss": 2.2048,
+      "step": 9910
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010702626641651033,
+      "loss": 2.4422,
+      "step": 9911
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010701688555347092,
+      "loss": 2.3039,
+      "step": 9912
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010700750469043152,
+      "loss": 2.6107,
+      "step": 9913
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010699812382739211,
+      "loss": 2.3303,
+      "step": 9914
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010698874296435273,
+      "loss": 2.4741,
+      "step": 9915
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010697936210131332,
+      "loss": 2.3551,
+      "step": 9916
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010696998123827393,
+      "loss": 2.6423,
+      "step": 9917
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010696060037523452,
+      "loss": 2.225,
+      "step": 9918
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010695121951219514,
+      "loss": 2.4152,
+      "step": 9919
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010694183864915573,
+      "loss": 2.5562,
+      "step": 9920
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010693245778611633,
+      "loss": 2.29,
+      "step": 9921
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010692307692307692,
+      "loss": 2.5315,
+      "step": 9922
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010691369606003754,
+      "loss": 2.5813,
+      "step": 9923
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00010690431519699813,
+      "loss": 2.4245,
+      "step": 9924
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010689493433395872,
+      "loss": 2.4371,
+      "step": 9925
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010688555347091933,
+      "loss": 2.3793,
+      "step": 9926
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010687617260787992,
+      "loss": 2.4479,
+      "step": 9927
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010686679174484054,
+      "loss": 2.5965,
+      "step": 9928
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010685741088180113,
+      "loss": 2.4332,
+      "step": 9929
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010684803001876173,
+      "loss": 2.3975,
+      "step": 9930
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010683864915572232,
+      "loss": 2.1806,
+      "step": 9931
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010682926829268294,
+      "loss": 2.4142,
+      "step": 9932
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010681988742964353,
+      "loss": 2.2365,
+      "step": 9933
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010681050656660414,
+      "loss": 2.3327,
+      "step": 9934
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010680112570356473,
+      "loss": 2.4863,
+      "step": 9935
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010679174484052535,
+      "loss": 2.4778,
+      "step": 9936
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010678236397748594,
+      "loss": 2.4881,
+      "step": 9937
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010677298311444653,
+      "loss": 2.5259,
+      "step": 9938
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010676360225140713,
+      "loss": 2.5259,
+      "step": 9939
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010675422138836772,
+      "loss": 2.2445,
+      "step": 9940
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010674484052532834,
+      "loss": 2.4778,
+      "step": 9941
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010673545966228893,
+      "loss": 2.4695,
+      "step": 9942
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010672607879924954,
+      "loss": 2.4536,
+      "step": 9943
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010671669793621013,
+      "loss": 2.7155,
+      "step": 9944
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 0.00010670731707317075,
+      "loss": 2.5496,
+      "step": 9945
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010669793621013134,
+      "loss": 2.5495,
+      "step": 9946
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010668855534709194,
+      "loss": 2.4548,
+      "step": 9947
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010667917448405253,
+      "loss": 2.3879,
+      "step": 9948
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010666979362101315,
+      "loss": 2.6529,
+      "step": 9949
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010666041275797374,
+      "loss": 2.6875,
+      "step": 9950
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010665103189493435,
+      "loss": 2.5643,
+      "step": 9951
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010664165103189494,
+      "loss": 2.3517,
+      "step": 9952
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010663227016885553,
+      "loss": 2.6465,
+      "step": 9953
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010662288930581614,
+      "loss": 2.3304,
+      "step": 9954
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010661350844277674,
+      "loss": 2.3941,
+      "step": 9955
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010660412757973734,
+      "loss": 2.5677,
+      "step": 9956
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010659474671669793,
+      "loss": 2.5063,
+      "step": 9957
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010658536585365855,
+      "loss": 2.5386,
+      "step": 9958
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010657598499061914,
+      "loss": 2.6583,
+      "step": 9959
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010656660412757974,
+      "loss": 2.1755,
+      "step": 9960
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010655722326454034,
+      "loss": 2.2087,
+      "step": 9961
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010654784240150095,
+      "loss": 2.4634,
+      "step": 9962
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010653846153846154,
+      "loss": 2.533,
+      "step": 9963
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010652908067542215,
+      "loss": 2.2716,
+      "step": 9964
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010651969981238274,
+      "loss": 2.6478,
+      "step": 9965
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010651031894934333,
+      "loss": 2.7063,
+      "step": 9966
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.00010650093808630395,
+      "loss": 2.4734,
+      "step": 9967
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010649155722326454,
+      "loss": 2.7659,
+      "step": 9968
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010648217636022514,
+      "loss": 2.432,
+      "step": 9969
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010647279549718574,
+      "loss": 2.2335,
+      "step": 9970
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010646341463414635,
+      "loss": 2.581,
+      "step": 9971
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010645403377110694,
+      "loss": 2.3119,
+      "step": 9972
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010644465290806755,
+      "loss": 2.5913,
+      "step": 9973
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010643527204502814,
+      "loss": 2.4282,
+      "step": 9974
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010642589118198876,
+      "loss": 2.4464,
+      "step": 9975
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010641651031894935,
+      "loss": 2.4459,
+      "step": 9976
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010640712945590995,
+      "loss": 2.5347,
+      "step": 9977
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010639774859287054,
+      "loss": 2.2771,
+      "step": 9978
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010638836772983114,
+      "loss": 2.5199,
+      "step": 9979
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010637898686679175,
+      "loss": 2.336,
+      "step": 9980
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010636960600375234,
+      "loss": 2.2474,
+      "step": 9981
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010636022514071295,
+      "loss": 2.5542,
+      "step": 9982
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010635084427767354,
+      "loss": 2.2023,
+      "step": 9983
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010634146341463416,
+      "loss": 2.6676,
+      "step": 9984
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010633208255159475,
+      "loss": 2.5257,
+      "step": 9985
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010632270168855535,
+      "loss": 2.7666,
+      "step": 9986
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010631332082551594,
+      "loss": 2.3179,
+      "step": 9987
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00010630393996247656,
+      "loss": 2.512,
+      "step": 9988
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010629455909943715,
+      "loss": 2.7129,
+      "step": 9989
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010628517823639776,
+      "loss": 2.2754,
+      "step": 9990
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010627579737335835,
+      "loss": 2.4403,
+      "step": 9991
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010626641651031894,
+      "loss": 2.4589,
+      "step": 9992
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010625703564727956,
+      "loss": 2.5729,
+      "step": 9993
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010624765478424015,
+      "loss": 2.4647,
+      "step": 9994
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010623827392120075,
+      "loss": 2.5543,
+      "step": 9995
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010622889305816134,
+      "loss": 2.4196,
+      "step": 9996
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010621951219512196,
+      "loss": 2.4611,
+      "step": 9997
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010621013133208255,
+      "loss": 2.3032,
+      "step": 9998
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010620075046904316,
+      "loss": 2.6724,
+      "step": 9999
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010619136960600375,
+      "loss": 2.3365,
+      "step": 10000
+    },
+    {
+      "epoch": 4.69,
+      "eval_loss": 2.5976531505584717,
+      "eval_runtime": 67.8146,
+      "eval_samples_per_second": 27.944,
+      "eval_steps_per_second": 3.495,
+      "step": 10000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010618198874296437,
+      "loss": 2.3393,
+      "step": 10001
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010617260787992496,
+      "loss": 2.4314,
+      "step": 10002
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010616322701688558,
+      "loss": 2.458,
+      "step": 10003
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010615384615384615,
+      "loss": 2.3798,
+      "step": 10004
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010614446529080674,
+      "loss": 2.6023,
+      "step": 10005
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010613508442776736,
+      "loss": 2.3294,
+      "step": 10006
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010612570356472795,
+      "loss": 2.519,
+      "step": 10007
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010611632270168857,
+      "loss": 2.486,
+      "step": 10008
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00010610694183864915,
+      "loss": 2.3873,
+      "step": 10009
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010609756097560977,
+      "loss": 2.5108,
+      "step": 10010
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010608818011257036,
+      "loss": 2.7744,
+      "step": 10011
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010607879924953098,
+      "loss": 2.5124,
+      "step": 10012
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010606941838649155,
+      "loss": 2.7196,
+      "step": 10013
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010606003752345217,
+      "loss": 2.2371,
+      "step": 10014
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010605065666041276,
+      "loss": 2.4848,
+      "step": 10015
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010604127579737338,
+      "loss": 2.3432,
+      "step": 10016
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010603189493433397,
+      "loss": 2.326,
+      "step": 10017
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010602251407129455,
+      "loss": 2.419,
+      "step": 10018
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010601313320825517,
+      "loss": 2.4068,
+      "step": 10019
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010600375234521576,
+      "loss": 2.3348,
+      "step": 10020
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010599437148217638,
+      "loss": 2.2003,
+      "step": 10021
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010598499061913695,
+      "loss": 2.6197,
+      "step": 10022
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010597560975609757,
+      "loss": 2.3618,
+      "step": 10023
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010596622889305816,
+      "loss": 2.1593,
+      "step": 10024
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010595684803001878,
+      "loss": 2.2584,
+      "step": 10025
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010594746716697937,
+      "loss": 2.5277,
+      "step": 10026
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010593808630393997,
+      "loss": 2.3101,
+      "step": 10027
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010592870544090057,
+      "loss": 2.2674,
+      "step": 10028
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010591932457786118,
+      "loss": 2.2967,
+      "step": 10029
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010590994371482177,
+      "loss": 2.4471,
+      "step": 10030
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 0.00010590056285178235,
+      "loss": 2.7188,
+      "step": 10031
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010589118198874297,
+      "loss": 2.3921,
+      "step": 10032
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010588180112570356,
+      "loss": 2.3056,
+      "step": 10033
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010587242026266418,
+      "loss": 2.4032,
+      "step": 10034
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010586303939962477,
+      "loss": 2.535,
+      "step": 10035
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010585365853658537,
+      "loss": 2.2226,
+      "step": 10036
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010584427767354597,
+      "loss": 2.348,
+      "step": 10037
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010583489681050658,
+      "loss": 2.0534,
+      "step": 10038
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010582551594746717,
+      "loss": 2.4417,
+      "step": 10039
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010581613508442778,
+      "loss": 2.3848,
+      "step": 10040
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010580675422138837,
+      "loss": 2.5778,
+      "step": 10041
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010579737335834899,
+      "loss": 2.5079,
+      "step": 10042
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010578799249530958,
+      "loss": 2.5053,
+      "step": 10043
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010577861163227017,
+      "loss": 2.371,
+      "step": 10044
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010576923076923077,
+      "loss": 2.4571,
+      "step": 10045
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010575984990619137,
+      "loss": 2.4686,
+      "step": 10046
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010575046904315198,
+      "loss": 2.3967,
+      "step": 10047
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010574108818011257,
+      "loss": 2.3569,
+      "step": 10048
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010573170731707318,
+      "loss": 2.4269,
+      "step": 10049
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010572232645403377,
+      "loss": 2.187,
+      "step": 10050
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010571294559099439,
+      "loss": 2.2818,
+      "step": 10051
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00010570356472795498,
+      "loss": 2.2894,
+      "step": 10052
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010569418386491558,
+      "loss": 2.4231,
+      "step": 10053
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010568480300187617,
+      "loss": 2.4697,
+      "step": 10054
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010567542213883679,
+      "loss": 2.5117,
+      "step": 10055
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010566604127579738,
+      "loss": 2.6339,
+      "step": 10056
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010565666041275797,
+      "loss": 2.3722,
+      "step": 10057
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010564727954971858,
+      "loss": 2.7999,
+      "step": 10058
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010563789868667917,
+      "loss": 2.5995,
+      "step": 10059
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010562851782363979,
+      "loss": 2.4547,
+      "step": 10060
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010561913696060038,
+      "loss": 2.3844,
+      "step": 10061
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010560975609756098,
+      "loss": 2.1793,
+      "step": 10062
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010560037523452157,
+      "loss": 2.4024,
+      "step": 10063
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010559099437148219,
+      "loss": 2.4651,
+      "step": 10064
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010558161350844278,
+      "loss": 2.0464,
+      "step": 10065
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010557223264540339,
+      "loss": 2.2802,
+      "step": 10066
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010556285178236398,
+      "loss": 2.3814,
+      "step": 10067
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0001055534709193246,
+      "loss": 2.3574,
+      "step": 10068
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010554409005628519,
+      "loss": 2.4111,
+      "step": 10069
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010553470919324578,
+      "loss": 2.1987,
+      "step": 10070
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010552532833020638,
+      "loss": 2.4864,
+      "step": 10071
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010551594746716697,
+      "loss": 2.452,
+      "step": 10072
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.00010550656660412759,
+      "loss": 2.4022,
+      "step": 10073
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010549718574108818,
+      "loss": 2.534,
+      "step": 10074
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010548780487804879,
+      "loss": 2.2662,
+      "step": 10075
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010547842401500938,
+      "loss": 2.1431,
+      "step": 10076
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010546904315197,
+      "loss": 2.1298,
+      "step": 10077
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010545966228893059,
+      "loss": 2.3516,
+      "step": 10078
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010545028142589119,
+      "loss": 2.2392,
+      "step": 10079
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010544090056285178,
+      "loss": 2.5391,
+      "step": 10080
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001054315196998124,
+      "loss": 2.385,
+      "step": 10081
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010542213883677299,
+      "loss": 2.3962,
+      "step": 10082
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010541275797373358,
+      "loss": 2.34,
+      "step": 10083
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010540337711069419,
+      "loss": 2.2214,
+      "step": 10084
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010539399624765478,
+      "loss": 2.5822,
+      "step": 10085
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001053846153846154,
+      "loss": 2.3788,
+      "step": 10086
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010537523452157599,
+      "loss": 2.4297,
+      "step": 10087
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010536585365853659,
+      "loss": 2.5433,
+      "step": 10088
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010535647279549718,
+      "loss": 2.4839,
+      "step": 10089
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001053470919324578,
+      "loss": 2.4773,
+      "step": 10090
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010533771106941839,
+      "loss": 2.3341,
+      "step": 10091
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.000105328330206379,
+      "loss": 2.269,
+      "step": 10092
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00010531894934333959,
+      "loss": 2.6442,
+      "step": 10093
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001053095684803002,
+      "loss": 2.3341,
+      "step": 10094
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.0001053001876172608,
+      "loss": 2.214,
+      "step": 10095
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010529080675422139,
+      "loss": 2.4515,
+      "step": 10096
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010528142589118199,
+      "loss": 2.3173,
+      "step": 10097
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010527204502814258,
+      "loss": 2.5147,
+      "step": 10098
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001052626641651032,
+      "loss": 2.5699,
+      "step": 10099
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010525328330206379,
+      "loss": 2.3653,
+      "step": 10100
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001052439024390244,
+      "loss": 2.6759,
+      "step": 10101
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010523452157598499,
+      "loss": 2.3894,
+      "step": 10102
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001052251407129456,
+      "loss": 2.3081,
+      "step": 10103
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001052157598499062,
+      "loss": 2.3287,
+      "step": 10104
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001052063789868668,
+      "loss": 2.3931,
+      "step": 10105
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010519699812382739,
+      "loss": 2.5299,
+      "step": 10106
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010518761726078801,
+      "loss": 2.461,
+      "step": 10107
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001051782363977486,
+      "loss": 2.2262,
+      "step": 10108
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010516885553470919,
+      "loss": 2.2891,
+      "step": 10109
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001051594746716698,
+      "loss": 2.4767,
+      "step": 10110
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010515009380863039,
+      "loss": 2.4391,
+      "step": 10111
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.000105140712945591,
+      "loss": 2.5846,
+      "step": 10112
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001051313320825516,
+      "loss": 2.4517,
+      "step": 10113
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.0001051219512195122,
+      "loss": 2.2308,
+      "step": 10114
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010511257035647279,
+      "loss": 2.5399,
+      "step": 10115
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00010510318949343341,
+      "loss": 2.2911,
+      "step": 10116
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.000105093808630394,
+      "loss": 2.5306,
+      "step": 10117
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001050844277673546,
+      "loss": 2.5135,
+      "step": 10118
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001050750469043152,
+      "loss": 2.3891,
+      "step": 10119
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00010506566604127581,
+      "loss": 2.6277,
+      "step": 10120
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001050562851782364,
+      "loss": 2.5243,
+      "step": 10121
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.000105046904315197,
+      "loss": 2.3648,
+      "step": 10122
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001050375234521576,
+      "loss": 2.6709,
+      "step": 10123
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00010502814258911819,
+      "loss": 2.3466,
+      "step": 10124
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00010501876172607881,
+      "loss": 2.4146,
+      "step": 10125
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001050093808630394,
+      "loss": 2.4161,
+      "step": 10126
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.000105,
+      "loss": 2.6166,
+      "step": 10127
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001049906191369606,
+      "loss": 2.5275,
+      "step": 10128
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00010498123827392121,
+      "loss": 2.203,
+      "step": 10129
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001049718574108818,
+      "loss": 2.2379,
+      "step": 10130
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00010496247654784241,
+      "loss": 2.4467,
+      "step": 10131
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.000104953095684803,
+      "loss": 2.531,
+      "step": 10132
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00010494371482176362,
+      "loss": 2.4544,
+      "step": 10133
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.00010493433395872421,
+      "loss": 2.5158,
+      "step": 10134
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001049249530956848,
+      "loss": 2.2757,
+      "step": 10135
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.0001049155722326454,
+      "loss": 2.5517,
+      "step": 10136
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 0.000104906191369606,
+      "loss": 2.4264,
+      "step": 10137
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010489681050656661,
+      "loss": 2.5506,
+      "step": 10138
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001048874296435272,
+      "loss": 2.4643,
+      "step": 10139
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010487804878048781,
+      "loss": 2.6335,
+      "step": 10140
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001048686679174484,
+      "loss": 2.4349,
+      "step": 10141
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010485928705440902,
+      "loss": 2.3597,
+      "step": 10142
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010484990619136961,
+      "loss": 2.3028,
+      "step": 10143
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010484052532833021,
+      "loss": 2.3727,
+      "step": 10144
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001048311444652908,
+      "loss": 2.2379,
+      "step": 10145
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010482176360225142,
+      "loss": 2.5775,
+      "step": 10146
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010481238273921201,
+      "loss": 2.385,
+      "step": 10147
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010480300187617262,
+      "loss": 2.4533,
+      "step": 10148
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010479362101313321,
+      "loss": 2.3315,
+      "step": 10149
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001047842401500938,
+      "loss": 2.6209,
+      "step": 10150
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010477485928705442,
+      "loss": 2.6249,
+      "step": 10151
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010476547842401501,
+      "loss": 2.3625,
+      "step": 10152
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010475609756097561,
+      "loss": 2.4018,
+      "step": 10153
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001047467166979362,
+      "loss": 2.4587,
+      "step": 10154
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010473733583489682,
+      "loss": 2.5049,
+      "step": 10155
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010472795497185741,
+      "loss": 2.3078,
+      "step": 10156
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010471857410881802,
+      "loss": 2.2089,
+      "step": 10157
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.00010470919324577861,
+      "loss": 2.5953,
+      "step": 10158
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010469981238273923,
+      "loss": 2.4177,
+      "step": 10159
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010469043151969982,
+      "loss": 2.455,
+      "step": 10160
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010468105065666044,
+      "loss": 2.524,
+      "step": 10161
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010467166979362101,
+      "loss": 2.3678,
+      "step": 10162
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.0001046622889305816,
+      "loss": 2.4764,
+      "step": 10163
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010465290806754222,
+      "loss": 2.4851,
+      "step": 10164
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010464352720450281,
+      "loss": 2.3356,
+      "step": 10165
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010463414634146342,
+      "loss": 2.7627,
+      "step": 10166
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010462476547842401,
+      "loss": 2.4761,
+      "step": 10167
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010461538461538463,
+      "loss": 2.5945,
+      "step": 10168
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010460600375234522,
+      "loss": 2.3931,
+      "step": 10169
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010459662288930584,
+      "loss": 2.5103,
+      "step": 10170
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010458724202626641,
+      "loss": 2.4087,
+      "step": 10171
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010457786116322703,
+      "loss": 2.4155,
+      "step": 10172
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010456848030018762,
+      "loss": 2.4681,
+      "step": 10173
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010455909943714824,
+      "loss": 2.5637,
+      "step": 10174
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010454971857410883,
+      "loss": 2.3098,
+      "step": 10175
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010454033771106941,
+      "loss": 2.7447,
+      "step": 10176
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010453095684803003,
+      "loss": 2.5568,
+      "step": 10177
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010452157598499062,
+      "loss": 2.1461,
+      "step": 10178
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010451219512195123,
+      "loss": 2.5142,
+      "step": 10179
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00010450281425891181,
+      "loss": 2.4294,
+      "step": 10180
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010449343339587243,
+      "loss": 2.6304,
+      "step": 10181
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010448405253283302,
+      "loss": 2.2131,
+      "step": 10182
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010447467166979364,
+      "loss": 2.1247,
+      "step": 10183
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010446529080675423,
+      "loss": 2.6811,
+      "step": 10184
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010445590994371483,
+      "loss": 2.4014,
+      "step": 10185
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010444652908067543,
+      "loss": 2.5171,
+      "step": 10186
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010443714821763604,
+      "loss": 2.405,
+      "step": 10187
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010442776735459663,
+      "loss": 2.4214,
+      "step": 10188
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010441838649155721,
+      "loss": 2.472,
+      "step": 10189
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010440900562851783,
+      "loss": 2.2986,
+      "step": 10190
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010439962476547842,
+      "loss": 2.5423,
+      "step": 10191
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010439024390243904,
+      "loss": 2.2966,
+      "step": 10192
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010438086303939963,
+      "loss": 2.6881,
+      "step": 10193
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010437148217636023,
+      "loss": 2.5273,
+      "step": 10194
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010436210131332083,
+      "loss": 2.309,
+      "step": 10195
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010435272045028144,
+      "loss": 2.4695,
+      "step": 10196
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010434333958724203,
+      "loss": 2.421,
+      "step": 10197
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010433395872420264,
+      "loss": 2.4949,
+      "step": 10198
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010432457786116323,
+      "loss": 2.7103,
+      "step": 10199
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010431519699812385,
+      "loss": 2.3158,
+      "step": 10200
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00010430581613508444,
+      "loss": 2.557,
+      "step": 10201
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010429643527204503,
+      "loss": 2.5544,
+      "step": 10202
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010428705440900563,
+      "loss": 2.4302,
+      "step": 10203
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010427767354596623,
+      "loss": 2.6005,
+      "step": 10204
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010426829268292684,
+      "loss": 2.2731,
+      "step": 10205
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010425891181988743,
+      "loss": 2.2609,
+      "step": 10206
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010424953095684804,
+      "loss": 2.4264,
+      "step": 10207
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010424015009380863,
+      "loss": 2.3858,
+      "step": 10208
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010423076923076925,
+      "loss": 2.5292,
+      "step": 10209
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010422138836772984,
+      "loss": 2.692,
+      "step": 10210
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010421200750469044,
+      "loss": 2.5988,
+      "step": 10211
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010420262664165103,
+      "loss": 2.5484,
+      "step": 10212
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010419324577861165,
+      "loss": 2.5617,
+      "step": 10213
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010418386491557224,
+      "loss": 2.3667,
+      "step": 10214
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010417448405253283,
+      "loss": 2.423,
+      "step": 10215
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010416510318949344,
+      "loss": 2.5191,
+      "step": 10216
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010415572232645403,
+      "loss": 2.4796,
+      "step": 10217
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010414634146341465,
+      "loss": 2.28,
+      "step": 10218
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010413696060037524,
+      "loss": 2.4761,
+      "step": 10219
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010412757973733584,
+      "loss": 2.2714,
+      "step": 10220
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010411819887429643,
+      "loss": 2.341,
+      "step": 10221
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00010410881801125705,
+      "loss": 2.4821,
+      "step": 10222
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010409943714821764,
+      "loss": 2.3619,
+      "step": 10223
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010409005628517825,
+      "loss": 2.2687,
+      "step": 10224
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010408067542213884,
+      "loss": 2.2344,
+      "step": 10225
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010407129455909946,
+      "loss": 2.3161,
+      "step": 10226
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010406191369606005,
+      "loss": 2.5212,
+      "step": 10227
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010405253283302064,
+      "loss": 2.3985,
+      "step": 10228
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010404315196998124,
+      "loss": 2.4196,
+      "step": 10229
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010403377110694183,
+      "loss": 2.2572,
+      "step": 10230
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010402439024390245,
+      "loss": 2.6129,
+      "step": 10231
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010401500938086304,
+      "loss": 2.3756,
+      "step": 10232
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010400562851782365,
+      "loss": 2.402,
+      "step": 10233
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010399624765478424,
+      "loss": 2.2187,
+      "step": 10234
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010398686679174486,
+      "loss": 2.2732,
+      "step": 10235
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010397748592870545,
+      "loss": 2.2711,
+      "step": 10236
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010396810506566605,
+      "loss": 2.4806,
+      "step": 10237
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010395872420262664,
+      "loss": 2.4453,
+      "step": 10238
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010394934333958726,
+      "loss": 2.6328,
+      "step": 10239
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010393996247654785,
+      "loss": 2.5221,
+      "step": 10240
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010393058161350844,
+      "loss": 2.3934,
+      "step": 10241
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010392120075046905,
+      "loss": 2.4686,
+      "step": 10242
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010391181988742964,
+      "loss": 2.4855,
+      "step": 10243
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 0.00010390243902439026,
+      "loss": 2.9259,
+      "step": 10244
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010389305816135085,
+      "loss": 2.5607,
+      "step": 10245
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010388367729831145,
+      "loss": 2.4952,
+      "step": 10246
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010387429643527204,
+      "loss": 2.5651,
+      "step": 10247
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010386491557223266,
+      "loss": 2.3662,
+      "step": 10248
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010385553470919325,
+      "loss": 2.2742,
+      "step": 10249
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010384615384615386,
+      "loss": 2.2551,
+      "step": 10250
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010383677298311445,
+      "loss": 2.6069,
+      "step": 10251
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010382739212007506,
+      "loss": 2.3527,
+      "step": 10252
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010381801125703566,
+      "loss": 2.5283,
+      "step": 10253
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010380863039399625,
+      "loss": 2.6916,
+      "step": 10254
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010379924953095685,
+      "loss": 2.5879,
+      "step": 10255
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010378986866791744,
+      "loss": 2.6693,
+      "step": 10256
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010378048780487806,
+      "loss": 2.4747,
+      "step": 10257
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010377110694183865,
+      "loss": 2.5218,
+      "step": 10258
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010376172607879926,
+      "loss": 2.463,
+      "step": 10259
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010375234521575985,
+      "loss": 2.502,
+      "step": 10260
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010374296435272046,
+      "loss": 2.5399,
+      "step": 10261
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010373358348968106,
+      "loss": 2.4182,
+      "step": 10262
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010372420262664166,
+      "loss": 2.287,
+      "step": 10263
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010371482176360225,
+      "loss": 2.2825,
+      "step": 10264
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.00010370544090056287,
+      "loss": 2.4871,
+      "step": 10265
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010369606003752346,
+      "loss": 2.3797,
+      "step": 10266
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010368667917448405,
+      "loss": 2.1791,
+      "step": 10267
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010367729831144466,
+      "loss": 2.3136,
+      "step": 10268
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010366791744840525,
+      "loss": 2.5777,
+      "step": 10269
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010365853658536586,
+      "loss": 2.426,
+      "step": 10270
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010364915572232646,
+      "loss": 2.5379,
+      "step": 10271
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010363977485928706,
+      "loss": 2.4992,
+      "step": 10272
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010363039399624765,
+      "loss": 2.1286,
+      "step": 10273
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010362101313320827,
+      "loss": 2.3169,
+      "step": 10274
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010361163227016886,
+      "loss": 2.53,
+      "step": 10275
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010360225140712946,
+      "loss": 2.3384,
+      "step": 10276
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010359287054409006,
+      "loss": 2.3756,
+      "step": 10277
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010358348968105067,
+      "loss": 2.4875,
+      "step": 10278
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010357410881801126,
+      "loss": 2.4706,
+      "step": 10279
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010356472795497186,
+      "loss": 2.518,
+      "step": 10280
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010355534709193246,
+      "loss": 2.4559,
+      "step": 10281
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010354596622889305,
+      "loss": 2.3056,
+      "step": 10282
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010353658536585367,
+      "loss": 2.2996,
+      "step": 10283
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010352720450281426,
+      "loss": 2.3949,
+      "step": 10284
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010351782363977486,
+      "loss": 2.3123,
+      "step": 10285
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00010350844277673546,
+      "loss": 2.3889,
+      "step": 10286
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010349906191369607,
+      "loss": 2.6097,
+      "step": 10287
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010348968105065666,
+      "loss": 2.7546,
+      "step": 10288
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010348030018761727,
+      "loss": 2.2841,
+      "step": 10289
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010347091932457786,
+      "loss": 2.417,
+      "step": 10290
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010346153846153848,
+      "loss": 2.4281,
+      "step": 10291
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010345215759849907,
+      "loss": 2.4046,
+      "step": 10292
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010344277673545966,
+      "loss": 2.2368,
+      "step": 10293
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010343339587242026,
+      "loss": 2.4152,
+      "step": 10294
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010342401500938085,
+      "loss": 2.19,
+      "step": 10295
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010341463414634147,
+      "loss": 2.474,
+      "step": 10296
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010340525328330206,
+      "loss": 2.4008,
+      "step": 10297
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010339587242026267,
+      "loss": 2.4732,
+      "step": 10298
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010338649155722326,
+      "loss": 2.4398,
+      "step": 10299
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010337711069418388,
+      "loss": 2.445,
+      "step": 10300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010336772983114447,
+      "loss": 2.6778,
+      "step": 10301
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010335834896810507,
+      "loss": 2.2331,
+      "step": 10302
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010334896810506566,
+      "loss": 2.2915,
+      "step": 10303
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010333958724202628,
+      "loss": 2.34,
+      "step": 10304
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010333020637898687,
+      "loss": 2.4915,
+      "step": 10305
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010332082551594746,
+      "loss": 2.4168,
+      "step": 10306
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010331144465290807,
+      "loss": 2.412,
+      "step": 10307
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00010330206378986866,
+      "loss": 2.3291,
+      "step": 10308
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010329268292682928,
+      "loss": 2.4699,
+      "step": 10309
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010328330206378987,
+      "loss": 2.3985,
+      "step": 10310
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010327392120075047,
+      "loss": 2.4996,
+      "step": 10311
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010326454033771106,
+      "loss": 2.5817,
+      "step": 10312
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010325515947467168,
+      "loss": 2.6002,
+      "step": 10313
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010324577861163227,
+      "loss": 2.3967,
+      "step": 10314
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010323639774859288,
+      "loss": 2.631,
+      "step": 10315
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010322701688555347,
+      "loss": 2.5517,
+      "step": 10316
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010321763602251409,
+      "loss": 2.3682,
+      "step": 10317
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010320825515947468,
+      "loss": 2.3344,
+      "step": 10318
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010319887429643527,
+      "loss": 2.6459,
+      "step": 10319
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010318949343339587,
+      "loss": 2.343,
+      "step": 10320
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010318011257035646,
+      "loss": 2.3654,
+      "step": 10321
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010317073170731708,
+      "loss": 2.3621,
+      "step": 10322
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010316135084427767,
+      "loss": 2.468,
+      "step": 10323
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010315196998123828,
+      "loss": 2.3117,
+      "step": 10324
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010314258911819887,
+      "loss": 2.3083,
+      "step": 10325
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010313320825515949,
+      "loss": 2.3278,
+      "step": 10326
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010312382739212008,
+      "loss": 2.22,
+      "step": 10327
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.0001031144465290807,
+      "loss": 2.6516,
+      "step": 10328
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 0.00010310506566604127,
+      "loss": 2.4673,
+      "step": 10329
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010309568480300189,
+      "loss": 2.4612,
+      "step": 10330
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010308630393996248,
+      "loss": 2.2889,
+      "step": 10331
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010307692307692307,
+      "loss": 2.4131,
+      "step": 10332
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010306754221388368,
+      "loss": 2.2839,
+      "step": 10333
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010305816135084427,
+      "loss": 2.4224,
+      "step": 10334
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010304878048780489,
+      "loss": 2.4358,
+      "step": 10335
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010303939962476548,
+      "loss": 2.452,
+      "step": 10336
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001030300187617261,
+      "loss": 2.572,
+      "step": 10337
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010302063789868667,
+      "loss": 2.5678,
+      "step": 10338
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010301125703564729,
+      "loss": 2.4847,
+      "step": 10339
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010300187617260788,
+      "loss": 2.3712,
+      "step": 10340
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001029924953095685,
+      "loss": 2.4415,
+      "step": 10341
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010298311444652909,
+      "loss": 2.4723,
+      "step": 10342
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001029737335834897,
+      "loss": 2.3496,
+      "step": 10343
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010296435272045029,
+      "loss": 2.5792,
+      "step": 10344
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010295497185741088,
+      "loss": 2.6611,
+      "step": 10345
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001029455909943715,
+      "loss": 2.209,
+      "step": 10346
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010293621013133207,
+      "loss": 2.5265,
+      "step": 10347
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010292682926829269,
+      "loss": 2.7815,
+      "step": 10348
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.00010291744840525328,
+      "loss": 2.4769,
+      "step": 10349
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001029080675422139,
+      "loss": 2.6411,
+      "step": 10350
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00010289868667917449,
+      "loss": 2.4034,
+      "step": 10351
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001028893058161351,
+      "loss": 2.6199,
+      "step": 10352
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00010287992495309569,
+      "loss": 2.5307,
+      "step": 10353
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001028705440900563,
+      "loss": 2.5117,
+      "step": 10354
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001028611632270169,
+      "loss": 2.3432,
+      "step": 10355
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001028517823639775,
+      "loss": 2.2773,
+      "step": 10356
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00010284240150093809,
+      "loss": 2.3992,
+      "step": 10357
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00010283302063789871,
+      "loss": 2.3614,
+      "step": 10358
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001028236397748593,
+      "loss": 2.2206,
+      "step": 10359
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00010281425891181989,
+      "loss": 2.5737,
+      "step": 10360
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001028048780487805,
+      "loss": 2.2787,
+      "step": 10361
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00010279549718574108,
+      "loss": 2.8252,
+      "step": 10362
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001027861163227017,
+      "loss": 2.6387,
+      "step": 10363
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001027767354596623,
+      "loss": 2.407,
+      "step": 10364
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001027673545966229,
+      "loss": 2.3804,
+      "step": 10365
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00010275797373358349,
+      "loss": 2.4557,
+      "step": 10366
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00010274859287054411,
+      "loss": 2.371,
+      "step": 10367
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001027392120075047,
+      "loss": 2.1921,
+      "step": 10368
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001027298311444653,
+      "loss": 2.4958,
+      "step": 10369
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001027204502814259,
+      "loss": 2.5364,
+      "step": 10370
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00010271106941838651,
+      "loss": 2.4336,
+      "step": 10371
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.0001027016885553471,
+      "loss": 2.2544,
+      "step": 10372
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001026923076923077,
+      "loss": 2.3761,
+      "step": 10373
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001026829268292683,
+      "loss": 2.5883,
+      "step": 10374
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00010267354596622889,
+      "loss": 2.4775,
+      "step": 10375
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00010266416510318951,
+      "loss": 2.5395,
+      "step": 10376
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001026547842401501,
+      "loss": 2.4193,
+      "step": 10377
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001026454033771107,
+      "loss": 2.5604,
+      "step": 10378
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001026360225140713,
+      "loss": 2.3834,
+      "step": 10379
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00010262664165103191,
+      "loss": 2.6701,
+      "step": 10380
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001026172607879925,
+      "loss": 2.3245,
+      "step": 10381
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00010260787992495311,
+      "loss": 2.5152,
+      "step": 10382
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001025984990619137,
+      "loss": 2.529,
+      "step": 10383
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00010258911819887432,
+      "loss": 2.1496,
+      "step": 10384
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00010257973733583491,
+      "loss": 2.3621,
+      "step": 10385
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001025703564727955,
+      "loss": 2.6254,
+      "step": 10386
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001025609756097561,
+      "loss": 2.5163,
+      "step": 10387
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001025515947467167,
+      "loss": 2.4302,
+      "step": 10388
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00010254221388367731,
+      "loss": 2.3362,
+      "step": 10389
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001025328330206379,
+      "loss": 2.2905,
+      "step": 10390
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00010252345215759851,
+      "loss": 2.2658,
+      "step": 10391
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.0001025140712945591,
+      "loss": 2.3526,
+      "step": 10392
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00010250469043151972,
+      "loss": 2.5434,
+      "step": 10393
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010249530956848031,
+      "loss": 2.4028,
+      "step": 10394
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010248592870544091,
+      "loss": 2.2688,
+      "step": 10395
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001024765478424015,
+      "loss": 2.3552,
+      "step": 10396
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010246716697936212,
+      "loss": 2.2091,
+      "step": 10397
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010245778611632271,
+      "loss": 2.5978,
+      "step": 10398
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001024484052532833,
+      "loss": 2.6147,
+      "step": 10399
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001024390243902439,
+      "loss": 2.4999,
+      "step": 10400
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001024296435272045,
+      "loss": 2.3347,
+      "step": 10401
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010242026266416512,
+      "loss": 2.5861,
+      "step": 10402
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001024108818011257,
+      "loss": 2.4352,
+      "step": 10403
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010240150093808631,
+      "loss": 2.2565,
+      "step": 10404
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001023921200750469,
+      "loss": 2.3763,
+      "step": 10405
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010238273921200752,
+      "loss": 2.2883,
+      "step": 10406
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010237335834896811,
+      "loss": 2.3905,
+      "step": 10407
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010236397748592872,
+      "loss": 2.4104,
+      "step": 10408
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001023545966228893,
+      "loss": 2.5455,
+      "step": 10409
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010234521575984992,
+      "loss": 2.5325,
+      "step": 10410
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010233583489681052,
+      "loss": 2.4278,
+      "step": 10411
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001023264540337711,
+      "loss": 2.404,
+      "step": 10412
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00010231707317073171,
+      "loss": 2.6025,
+      "step": 10413
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.0001023076923076923,
+      "loss": 2.3885,
+      "step": 10414
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010229831144465292,
+      "loss": 2.1227,
+      "step": 10415
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010228893058161351,
+      "loss": 2.4611,
+      "step": 10416
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010227954971857412,
+      "loss": 2.3459,
+      "step": 10417
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0001022701688555347,
+      "loss": 2.4764,
+      "step": 10418
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010226078799249532,
+      "loss": 2.3305,
+      "step": 10419
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010225140712945592,
+      "loss": 2.3195,
+      "step": 10420
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010224202626641652,
+      "loss": 2.5448,
+      "step": 10421
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010223264540337711,
+      "loss": 2.1431,
+      "step": 10422
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010222326454033773,
+      "loss": 2.454,
+      "step": 10423
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010221388367729832,
+      "loss": 2.1492,
+      "step": 10424
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010220450281425891,
+      "loss": 2.436,
+      "step": 10425
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010219512195121952,
+      "loss": 2.7761,
+      "step": 10426
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.0001021857410881801,
+      "loss": 2.5703,
+      "step": 10427
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010217636022514072,
+      "loss": 2.4621,
+      "step": 10428
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010216697936210132,
+      "loss": 2.3979,
+      "step": 10429
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010215759849906192,
+      "loss": 2.5474,
+      "step": 10430
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010214821763602251,
+      "loss": 2.4133,
+      "step": 10431
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010213883677298313,
+      "loss": 2.411,
+      "step": 10432
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010212945590994372,
+      "loss": 2.4212,
+      "step": 10433
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010212007504690432,
+      "loss": 2.4792,
+      "step": 10434
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010211069418386491,
+      "loss": 2.4495,
+      "step": 10435
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 0.00010210131332082553,
+      "loss": 2.5607,
+      "step": 10436
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010209193245778612,
+      "loss": 2.6602,
+      "step": 10437
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010208255159474671,
+      "loss": 2.2654,
+      "step": 10438
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010207317073170732,
+      "loss": 2.5675,
+      "step": 10439
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010206378986866791,
+      "loss": 2.3276,
+      "step": 10440
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010205440900562853,
+      "loss": 2.4344,
+      "step": 10441
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010204502814258912,
+      "loss": 2.7859,
+      "step": 10442
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010203564727954972,
+      "loss": 2.4861,
+      "step": 10443
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010202626641651031,
+      "loss": 2.4181,
+      "step": 10444
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010201688555347093,
+      "loss": 2.2707,
+      "step": 10445
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010200750469043152,
+      "loss": 2.5027,
+      "step": 10446
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010199812382739213,
+      "loss": 2.7107,
+      "step": 10447
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010198874296435272,
+      "loss": 2.4029,
+      "step": 10448
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010197936210131334,
+      "loss": 2.2776,
+      "step": 10449
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010196998123827393,
+      "loss": 2.2453,
+      "step": 10450
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010196060037523452,
+      "loss": 2.2628,
+      "step": 10451
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010195121951219512,
+      "loss": 2.4414,
+      "step": 10452
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010194183864915571,
+      "loss": 2.4574,
+      "step": 10453
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010193245778611633,
+      "loss": 2.4516,
+      "step": 10454
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010192307692307692,
+      "loss": 2.5135,
+      "step": 10455
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010191369606003753,
+      "loss": 2.6785,
+      "step": 10456
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.00010190431519699812,
+      "loss": 2.3595,
+      "step": 10457
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010189493433395874,
+      "loss": 2.634,
+      "step": 10458
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010188555347091933,
+      "loss": 2.2709,
+      "step": 10459
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010187617260787993,
+      "loss": 2.4687,
+      "step": 10460
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010186679174484052,
+      "loss": 2.5371,
+      "step": 10461
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010185741088180114,
+      "loss": 2.5086,
+      "step": 10462
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010184803001876173,
+      "loss": 2.4009,
+      "step": 10463
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010183864915572232,
+      "loss": 2.3537,
+      "step": 10464
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010182926829268293,
+      "loss": 2.7478,
+      "step": 10465
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010181988742964352,
+      "loss": 2.5152,
+      "step": 10466
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010181050656660414,
+      "loss": 2.4973,
+      "step": 10467
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010180112570356473,
+      "loss": 2.4617,
+      "step": 10468
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010179174484052533,
+      "loss": 2.1336,
+      "step": 10469
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010178236397748592,
+      "loss": 2.4994,
+      "step": 10470
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010177298311444654,
+      "loss": 2.2966,
+      "step": 10471
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010176360225140713,
+      "loss": 2.4408,
+      "step": 10472
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010175422138836774,
+      "loss": 2.5377,
+      "step": 10473
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010174484052532833,
+      "loss": 2.3632,
+      "step": 10474
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010173545966228895,
+      "loss": 2.2864,
+      "step": 10475
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010172607879924954,
+      "loss": 2.5977,
+      "step": 10476
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010171669793621013,
+      "loss": 2.2139,
+      "step": 10477
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00010170731707317073,
+      "loss": 2.5061,
+      "step": 10478
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010169793621013132,
+      "loss": 2.5614,
+      "step": 10479
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010168855534709194,
+      "loss": 2.4327,
+      "step": 10480
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010167917448405253,
+      "loss": 2.4811,
+      "step": 10481
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010166979362101314,
+      "loss": 2.3068,
+      "step": 10482
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010166041275797373,
+      "loss": 2.3489,
+      "step": 10483
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010165103189493435,
+      "loss": 2.4035,
+      "step": 10484
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010164165103189494,
+      "loss": 2.7193,
+      "step": 10485
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010163227016885555,
+      "loss": 2.3755,
+      "step": 10486
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010162288930581613,
+      "loss": 2.545,
+      "step": 10487
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010161350844277675,
+      "loss": 2.4183,
+      "step": 10488
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010160412757973734,
+      "loss": 2.2687,
+      "step": 10489
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010159474671669793,
+      "loss": 2.1149,
+      "step": 10490
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010158536585365854,
+      "loss": 2.332,
+      "step": 10491
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010157598499061913,
+      "loss": 2.4877,
+      "step": 10492
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010156660412757975,
+      "loss": 2.3008,
+      "step": 10493
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010155722326454034,
+      "loss": 2.5928,
+      "step": 10494
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010154784240150095,
+      "loss": 2.334,
+      "step": 10495
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010153846153846153,
+      "loss": 2.1736,
+      "step": 10496
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010152908067542215,
+      "loss": 2.0035,
+      "step": 10497
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010151969981238274,
+      "loss": 2.2317,
+      "step": 10498
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010151031894934336,
+      "loss": 2.4778,
+      "step": 10499
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00010150093808630394,
+      "loss": 2.4424,
+      "step": 10500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010149155722326455,
+      "loss": 2.3881,
+      "step": 10501
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010148217636022515,
+      "loss": 2.4141,
+      "step": 10502
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010147279549718574,
+      "loss": 2.7661,
+      "step": 10503
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010146341463414635,
+      "loss": 2.4568,
+      "step": 10504
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010145403377110693,
+      "loss": 2.2986,
+      "step": 10505
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010144465290806755,
+      "loss": 2.8654,
+      "step": 10506
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010143527204502814,
+      "loss": 2.3971,
+      "step": 10507
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010142589118198876,
+      "loss": 2.2951,
+      "step": 10508
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010141651031894934,
+      "loss": 2.7104,
+      "step": 10509
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010140712945590995,
+      "loss": 2.7274,
+      "step": 10510
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010139774859287054,
+      "loss": 2.5243,
+      "step": 10511
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010138836772983116,
+      "loss": 2.3401,
+      "step": 10512
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010137898686679175,
+      "loss": 2.1731,
+      "step": 10513
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010136960600375236,
+      "loss": 2.2014,
+      "step": 10514
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010136022514071295,
+      "loss": 2.613,
+      "step": 10515
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010135084427767354,
+      "loss": 2.4131,
+      "step": 10516
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010134146341463416,
+      "loss": 2.4302,
+      "step": 10517
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010133208255159475,
+      "loss": 2.5759,
+      "step": 10518
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010132270168855535,
+      "loss": 2.7025,
+      "step": 10519
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010131332082551594,
+      "loss": 2.3796,
+      "step": 10520
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 0.00010130393996247656,
+      "loss": 2.3946,
+      "step": 10521
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010129455909943715,
+      "loss": 2.5685,
+      "step": 10522
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010128517823639776,
+      "loss": 2.6668,
+      "step": 10523
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010127579737335835,
+      "loss": 2.6101,
+      "step": 10524
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010126641651031897,
+      "loss": 2.6689,
+      "step": 10525
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010125703564727956,
+      "loss": 2.534,
+      "step": 10526
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010124765478424016,
+      "loss": 2.4287,
+      "step": 10527
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010123827392120075,
+      "loss": 2.5291,
+      "step": 10528
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010122889305816134,
+      "loss": 2.338,
+      "step": 10529
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010121951219512196,
+      "loss": 2.9538,
+      "step": 10530
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010121013133208255,
+      "loss": 2.3855,
+      "step": 10531
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010120075046904316,
+      "loss": 2.4153,
+      "step": 10532
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010119136960600375,
+      "loss": 2.3981,
+      "step": 10533
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010118198874296437,
+      "loss": 2.3958,
+      "step": 10534
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010117260787992496,
+      "loss": 2.4064,
+      "step": 10535
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010116322701688556,
+      "loss": 2.323,
+      "step": 10536
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010115384615384615,
+      "loss": 2.4002,
+      "step": 10537
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010114446529080677,
+      "loss": 2.4369,
+      "step": 10538
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010113508442776736,
+      "loss": 2.324,
+      "step": 10539
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010112570356472797,
+      "loss": 2.544,
+      "step": 10540
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010111632270168856,
+      "loss": 2.3759,
+      "step": 10541
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.00010110694183864915,
+      "loss": 2.4803,
+      "step": 10542
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010109756097560977,
+      "loss": 2.2737,
+      "step": 10543
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010108818011257036,
+      "loss": 2.4652,
+      "step": 10544
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010107879924953096,
+      "loss": 2.4514,
+      "step": 10545
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010106941838649155,
+      "loss": 2.2724,
+      "step": 10546
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010106003752345217,
+      "loss": 2.3421,
+      "step": 10547
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010105065666041276,
+      "loss": 2.5522,
+      "step": 10548
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010104127579737337,
+      "loss": 2.5189,
+      "step": 10549
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010103189493433396,
+      "loss": 2.4105,
+      "step": 10550
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010102251407129458,
+      "loss": 2.5143,
+      "step": 10551
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010101313320825517,
+      "loss": 2.5999,
+      "step": 10552
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010100375234521577,
+      "loss": 2.3135,
+      "step": 10553
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010099437148217636,
+      "loss": 2.3007,
+      "step": 10554
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010098499061913695,
+      "loss": 2.2383,
+      "step": 10555
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010097560975609757,
+      "loss": 2.4127,
+      "step": 10556
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010096622889305816,
+      "loss": 2.4538,
+      "step": 10557
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010095684803001877,
+      "loss": 2.3594,
+      "step": 10558
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010094746716697936,
+      "loss": 2.4156,
+      "step": 10559
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010093808630393998,
+      "loss": 2.5729,
+      "step": 10560
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010092870544090057,
+      "loss": 2.1976,
+      "step": 10561
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010091932457786117,
+      "loss": 2.4753,
+      "step": 10562
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010090994371482176,
+      "loss": 2.506,
+      "step": 10563
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00010090056285178238,
+      "loss": 2.4094,
+      "step": 10564
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010089118198874297,
+      "loss": 2.3645,
+      "step": 10565
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010088180112570358,
+      "loss": 2.3871,
+      "step": 10566
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010087242026266417,
+      "loss": 2.2296,
+      "step": 10567
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010086303939962478,
+      "loss": 2.5311,
+      "step": 10568
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010085365853658538,
+      "loss": 2.4345,
+      "step": 10569
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010084427767354597,
+      "loss": 2.5024,
+      "step": 10570
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010083489681050657,
+      "loss": 2.3607,
+      "step": 10571
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010082551594746716,
+      "loss": 2.577,
+      "step": 10572
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010081613508442778,
+      "loss": 2.506,
+      "step": 10573
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010080675422138837,
+      "loss": 2.4011,
+      "step": 10574
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010079737335834898,
+      "loss": 2.2009,
+      "step": 10575
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010078799249530957,
+      "loss": 2.5732,
+      "step": 10576
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010077861163227018,
+      "loss": 2.6196,
+      "step": 10577
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010076923076923077,
+      "loss": 2.2855,
+      "step": 10578
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010075984990619138,
+      "loss": 2.5684,
+      "step": 10579
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010075046904315197,
+      "loss": 2.4537,
+      "step": 10580
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010074108818011259,
+      "loss": 2.3631,
+      "step": 10581
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010073170731707318,
+      "loss": 2.0967,
+      "step": 10582
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010072232645403377,
+      "loss": 2.5459,
+      "step": 10583
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010071294559099437,
+      "loss": 2.4301,
+      "step": 10584
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00010070356472795497,
+      "loss": 2.4343,
+      "step": 10585
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010069418386491558,
+      "loss": 2.3799,
+      "step": 10586
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010068480300187617,
+      "loss": 2.4324,
+      "step": 10587
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010067542213883678,
+      "loss": 2.3537,
+      "step": 10588
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010066604127579737,
+      "loss": 2.6268,
+      "step": 10589
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010065666041275799,
+      "loss": 2.3947,
+      "step": 10590
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010064727954971858,
+      "loss": 2.3691,
+      "step": 10591
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010063789868667918,
+      "loss": 2.686,
+      "step": 10592
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010062851782363977,
+      "loss": 2.2488,
+      "step": 10593
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010061913696060039,
+      "loss": 2.3492,
+      "step": 10594
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010060975609756098,
+      "loss": 2.4302,
+      "step": 10595
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010060037523452157,
+      "loss": 2.5908,
+      "step": 10596
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010059099437148218,
+      "loss": 2.245,
+      "step": 10597
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010058161350844277,
+      "loss": 2.4218,
+      "step": 10598
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010057223264540339,
+      "loss": 2.1624,
+      "step": 10599
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010056285178236398,
+      "loss": 2.3519,
+      "step": 10600
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010055347091932458,
+      "loss": 2.1349,
+      "step": 10601
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010054409005628517,
+      "loss": 2.5696,
+      "step": 10602
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010053470919324579,
+      "loss": 2.1771,
+      "step": 10603
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010052532833020638,
+      "loss": 2.5286,
+      "step": 10604
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010051594746716699,
+      "loss": 2.1507,
+      "step": 10605
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00010050656660412758,
+      "loss": 2.3194,
+      "step": 10606
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0001004971857410882,
+      "loss": 2.7202,
+      "step": 10607
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010048780487804879,
+      "loss": 2.9384,
+      "step": 10608
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010047842401500938,
+      "loss": 2.5927,
+      "step": 10609
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010046904315196998,
+      "loss": 2.4598,
+      "step": 10610
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010045966228893057,
+      "loss": 2.3703,
+      "step": 10611
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010045028142589119,
+      "loss": 2.4257,
+      "step": 10612
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010044090056285178,
+      "loss": 2.2951,
+      "step": 10613
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010043151969981239,
+      "loss": 2.4098,
+      "step": 10614
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010042213883677298,
+      "loss": 2.754,
+      "step": 10615
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.0001004127579737336,
+      "loss": 2.3223,
+      "step": 10616
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010040337711069419,
+      "loss": 2.3099,
+      "step": 10617
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010039399624765479,
+      "loss": 2.2888,
+      "step": 10618
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010038461538461538,
+      "loss": 2.4308,
+      "step": 10619
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.000100375234521576,
+      "loss": 2.4596,
+      "step": 10620
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010036585365853659,
+      "loss": 2.3366,
+      "step": 10621
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010035647279549718,
+      "loss": 2.599,
+      "step": 10622
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010034709193245779,
+      "loss": 2.4607,
+      "step": 10623
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010033771106941838,
+      "loss": 2.5773,
+      "step": 10624
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.000100328330206379,
+      "loss": 2.5139,
+      "step": 10625
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010031894934333959,
+      "loss": 2.4935,
+      "step": 10626
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010030956848030019,
+      "loss": 2.3601,
+      "step": 10627
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 0.00010030018761726078,
+      "loss": 2.45,
+      "step": 10628
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001002908067542214,
+      "loss": 2.4471,
+      "step": 10629
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010028142589118199,
+      "loss": 2.269,
+      "step": 10630
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001002720450281426,
+      "loss": 2.3745,
+      "step": 10631
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010026266416510319,
+      "loss": 2.5321,
+      "step": 10632
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001002532833020638,
+      "loss": 2.4489,
+      "step": 10633
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001002439024390244,
+      "loss": 2.3582,
+      "step": 10634
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010023452157598499,
+      "loss": 2.4124,
+      "step": 10635
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010022514071294559,
+      "loss": 2.2419,
+      "step": 10636
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010021575984990618,
+      "loss": 2.4345,
+      "step": 10637
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001002063789868668,
+      "loss": 2.2717,
+      "step": 10638
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010019699812382739,
+      "loss": 2.3654,
+      "step": 10639
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.000100187617260788,
+      "loss": 2.3284,
+      "step": 10640
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010017823639774859,
+      "loss": 2.48,
+      "step": 10641
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001001688555347092,
+      "loss": 2.3305,
+      "step": 10642
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001001594746716698,
+      "loss": 2.2005,
+      "step": 10643
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010015009380863041,
+      "loss": 2.4696,
+      "step": 10644
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010014071294559099,
+      "loss": 2.4411,
+      "step": 10645
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010013133208255161,
+      "loss": 2.5123,
+      "step": 10646
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001001219512195122,
+      "loss": 2.2376,
+      "step": 10647
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.00010011257035647279,
+      "loss": 2.3581,
+      "step": 10648
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001001031894934334,
+      "loss": 2.6945,
+      "step": 10649
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00010009380863039399,
+      "loss": 2.3751,
+      "step": 10650
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001000844277673546,
+      "loss": 2.3038,
+      "step": 10651
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001000750469043152,
+      "loss": 2.3614,
+      "step": 10652
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00010006566604127581,
+      "loss": 2.3878,
+      "step": 10653
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00010005628517823639,
+      "loss": 2.3171,
+      "step": 10654
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00010004690431519701,
+      "loss": 2.3404,
+      "step": 10655
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001000375234521576,
+      "loss": 2.4715,
+      "step": 10656
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00010002814258911822,
+      "loss": 2.3766,
+      "step": 10657
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001000187617260788,
+      "loss": 2.5032,
+      "step": 10658
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00010000938086303941,
+      "loss": 2.1974,
+      "step": 10659
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0001,
+      "loss": 2.3144,
+      "step": 10660
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.999061913696061e-05,
+      "loss": 2.3066,
+      "step": 10661
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.998123827392121e-05,
+      "loss": 2.507,
+      "step": 10662
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.99718574108818e-05,
+      "loss": 2.3142,
+      "step": 10663
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.996247654784241e-05,
+      "loss": 2.5497,
+      "step": 10664
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.995309568480301e-05,
+      "loss": 2.1918,
+      "step": 10665
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.994371482176362e-05,
+      "loss": 2.4003,
+      "step": 10666
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.99343339587242e-05,
+      "loss": 2.3328,
+      "step": 10667
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.99249530956848e-05,
+      "loss": 2.2279,
+      "step": 10668
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.99155722326454e-05,
+      "loss": 2.3093,
+      "step": 10669
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.990619136960601e-05,
+      "loss": 2.1286,
+      "step": 10670
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.989681050656661e-05,
+      "loss": 2.4792,
+      "step": 10671
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.98874296435272e-05,
+      "loss": 2.1393,
+      "step": 10672
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.987804878048781e-05,
+      "loss": 2.2799,
+      "step": 10673
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.986866791744841e-05,
+      "loss": 2.0393,
+      "step": 10674
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.985928705440902e-05,
+      "loss": 2.5095,
+      "step": 10675
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.984990619136961e-05,
+      "loss": 2.622,
+      "step": 10676
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.984052532833021e-05,
+      "loss": 2.4496,
+      "step": 10677
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.983114446529082e-05,
+      "loss": 2.2776,
+      "step": 10678
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.982176360225142e-05,
+      "loss": 2.3949,
+      "step": 10679
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.981238273921201e-05,
+      "loss": 2.5121,
+      "step": 10680
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.98030018761726e-05,
+      "loss": 2.4692,
+      "step": 10681
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.979362101313321e-05,
+      "loss": 2.4078,
+      "step": 10682
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.978424015009381e-05,
+      "loss": 2.2168,
+      "step": 10683
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.977485928705442e-05,
+      "loss": 2.6108,
+      "step": 10684
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.976547842401501e-05,
+      "loss": 2.2961,
+      "step": 10685
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.975609756097561e-05,
+      "loss": 2.6211,
+      "step": 10686
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.974671669793622e-05,
+      "loss": 2.3247,
+      "step": 10687
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.973733583489682e-05,
+      "loss": 2.4268,
+      "step": 10688
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.972795497185741e-05,
+      "loss": 2.3291,
+      "step": 10689
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.971857410881802e-05,
+      "loss": 2.5924,
+      "step": 10690
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.970919324577862e-05,
+      "loss": 2.2474,
+      "step": 10691
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.969981238273923e-05,
+      "loss": 2.4746,
+      "step": 10692
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.969043151969982e-05,
+      "loss": 2.4073,
+      "step": 10693
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.968105065666041e-05,
+      "loss": 2.4273,
+      "step": 10694
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.967166979362101e-05,
+      "loss": 2.368,
+      "step": 10695
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.966228893058162e-05,
+      "loss": 2.5355,
+      "step": 10696
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.965290806754222e-05,
+      "loss": 2.4182,
+      "step": 10697
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.964352720450281e-05,
+      "loss": 2.3035,
+      "step": 10698
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.963414634146342e-05,
+      "loss": 2.3906,
+      "step": 10699
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.962476547842402e-05,
+      "loss": 2.2321,
+      "step": 10700
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.961538461538463e-05,
+      "loss": 2.6718,
+      "step": 10701
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.960600375234522e-05,
+      "loss": 2.5245,
+      "step": 10702
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.959662288930582e-05,
+      "loss": 2.5784,
+      "step": 10703
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.958724202626643e-05,
+      "loss": 2.3787,
+      "step": 10704
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.957786116322703e-05,
+      "loss": 2.2877,
+      "step": 10705
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.956848030018762e-05,
+      "loss": 2.5907,
+      "step": 10706
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.955909943714821e-05,
+      "loss": 2.565,
+      "step": 10707
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.954971857410882e-05,
+      "loss": 2.2625,
+      "step": 10708
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.954033771106942e-05,
+      "loss": 2.2727,
+      "step": 10709
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.953095684803003e-05,
+      "loss": 2.1476,
+      "step": 10710
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.952157598499062e-05,
+      "loss": 2.2652,
+      "step": 10711
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.951219512195122e-05,
+      "loss": 2.3331,
+      "step": 10712
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.950281425891183e-05,
+      "loss": 2.5393,
+      "step": 10713
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.949343339587243e-05,
+      "loss": 2.1949,
+      "step": 10714
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.948405253283302e-05,
+      "loss": 2.3939,
+      "step": 10715
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.947467166979363e-05,
+      "loss": 2.4747,
+      "step": 10716
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.946529080675423e-05,
+      "loss": 2.3229,
+      "step": 10717
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.945590994371484e-05,
+      "loss": 2.4453,
+      "step": 10718
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.944652908067543e-05,
+      "loss": 2.3809,
+      "step": 10719
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.943714821763602e-05,
+      "loss": 2.3075,
+      "step": 10720
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.942776735459662e-05,
+      "loss": 2.4807,
+      "step": 10721
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.941838649155723e-05,
+      "loss": 2.3018,
+      "step": 10722
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.940900562851783e-05,
+      "loss": 2.5415,
+      "step": 10723
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.939962476547842e-05,
+      "loss": 2.5278,
+      "step": 10724
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.939024390243903e-05,
+      "loss": 2.3041,
+      "step": 10725
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.938086303939963e-05,
+      "loss": 2.2936,
+      "step": 10726
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.937148217636023e-05,
+      "loss": 2.4516,
+      "step": 10727
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.936210131332083e-05,
+      "loss": 2.4133,
+      "step": 10728
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.935272045028143e-05,
+      "loss": 2.3884,
+      "step": 10729
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.934333958724203e-05,
+      "loss": 2.4804,
+      "step": 10730
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.933395872420264e-05,
+      "loss": 2.3651,
+      "step": 10731
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.932457786116324e-05,
+      "loss": 2.3705,
+      "step": 10732
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.931519699812382e-05,
+      "loss": 2.0484,
+      "step": 10733
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.930581613508443e-05,
+      "loss": 2.3939,
+      "step": 10734
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.929643527204503e-05,
+      "loss": 2.112,
+      "step": 10735
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.928705440900563e-05,
+      "loss": 2.3612,
+      "step": 10736
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.927767354596623e-05,
+      "loss": 2.4481,
+      "step": 10737
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.926829268292683e-05,
+      "loss": 2.4431,
+      "step": 10738
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.925891181988743e-05,
+      "loss": 2.3111,
+      "step": 10739
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.924953095684804e-05,
+      "loss": 2.2418,
+      "step": 10740
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.924015009380864e-05,
+      "loss": 2.5782,
+      "step": 10741
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.923076923076923e-05,
+      "loss": 2.3098,
+      "step": 10742
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.922138836772984e-05,
+      "loss": 2.3805,
+      "step": 10743
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.921200750469044e-05,
+      "loss": 2.6046,
+      "step": 10744
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.920262664165105e-05,
+      "loss": 2.348,
+      "step": 10745
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.919324577861163e-05,
+      "loss": 1.922,
+      "step": 10746
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.918386491557223e-05,
+      "loss": 2.1531,
+      "step": 10747
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.917448405253283e-05,
+      "loss": 2.2801,
+      "step": 10748
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.916510318949344e-05,
+      "loss": 2.4692,
+      "step": 10749
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.915572232645404e-05,
+      "loss": 2.2788,
+      "step": 10750
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.914634146341463e-05,
+      "loss": 2.4335,
+      "step": 10751
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.913696060037524e-05,
+      "loss": 2.3382,
+      "step": 10752
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.912757973733584e-05,
+      "loss": 2.2197,
+      "step": 10753
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.911819887429645e-05,
+      "loss": 2.4994,
+      "step": 10754
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.910881801125704e-05,
+      "loss": 2.3933,
+      "step": 10755
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.909943714821764e-05,
+      "loss": 2.3071,
+      "step": 10756
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.909005628517825e-05,
+      "loss": 2.3016,
+      "step": 10757
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.908067542213885e-05,
+      "loss": 2.3615,
+      "step": 10758
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.907129455909944e-05,
+      "loss": 2.3055,
+      "step": 10759
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.906191369606003e-05,
+      "loss": 2.4042,
+      "step": 10760
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.905253283302064e-05,
+      "loss": 2.6039,
+      "step": 10761
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.904315196998124e-05,
+      "loss": 2.3661,
+      "step": 10762
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.903377110694185e-05,
+      "loss": 2.4738,
+      "step": 10763
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.902439024390244e-05,
+      "loss": 2.3095,
+      "step": 10764
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.901500938086304e-05,
+      "loss": 2.6181,
+      "step": 10765
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.900562851782365e-05,
+      "loss": 2.3526,
+      "step": 10766
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.899624765478425e-05,
+      "loss": 2.5042,
+      "step": 10767
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.898686679174484e-05,
+      "loss": 2.3879,
+      "step": 10768
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.897748592870545e-05,
+      "loss": 2.3133,
+      "step": 10769
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.896810506566605e-05,
+      "loss": 2.3735,
+      "step": 10770
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.895872420262666e-05,
+      "loss": 2.4628,
+      "step": 10771
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.894934333958725e-05,
+      "loss": 2.4224,
+      "step": 10772
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.893996247654784e-05,
+      "loss": 2.5886,
+      "step": 10773
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.893058161350844e-05,
+      "loss": 2.5776,
+      "step": 10774
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.892120075046905e-05,
+      "loss": 2.5305,
+      "step": 10775
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.891181988742965e-05,
+      "loss": 2.4998,
+      "step": 10776
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.890243902439024e-05,
+      "loss": 2.2911,
+      "step": 10777
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.889305816135085e-05,
+      "loss": 2.735,
+      "step": 10778
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.888367729831145e-05,
+      "loss": 2.5474,
+      "step": 10779
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.887429643527206e-05,
+      "loss": 2.2506,
+      "step": 10780
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.886491557223265e-05,
+      "loss": 2.5033,
+      "step": 10781
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.885553470919325e-05,
+      "loss": 2.122,
+      "step": 10782
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.884615384615386e-05,
+      "loss": 2.5697,
+      "step": 10783
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.883677298311446e-05,
+      "loss": 2.2809,
+      "step": 10784
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.882739212007505e-05,
+      "loss": 2.5039,
+      "step": 10785
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.881801125703564e-05,
+      "loss": 2.5731,
+      "step": 10786
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.880863039399625e-05,
+      "loss": 2.198,
+      "step": 10787
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.879924953095685e-05,
+      "loss": 2.214,
+      "step": 10788
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.878986866791746e-05,
+      "loss": 2.3014,
+      "step": 10789
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.878048780487805e-05,
+      "loss": 2.4677,
+      "step": 10790
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.877110694183865e-05,
+      "loss": 2.4149,
+      "step": 10791
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.876172607879926e-05,
+      "loss": 2.2463,
+      "step": 10792
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.875234521575986e-05,
+      "loss": 2.3403,
+      "step": 10793
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.874296435272045e-05,
+      "loss": 2.3263,
+      "step": 10794
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.873358348968106e-05,
+      "loss": 2.2898,
+      "step": 10795
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.872420262664166e-05,
+      "loss": 2.5954,
+      "step": 10796
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.871482176360226e-05,
+      "loss": 2.2819,
+      "step": 10797
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.870544090056286e-05,
+      "loss": 2.3859,
+      "step": 10798
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.869606003752345e-05,
+      "loss": 2.275,
+      "step": 10799
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.868667917448405e-05,
+      "loss": 2.262,
+      "step": 10800
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.867729831144466e-05,
+      "loss": 2.4785,
+      "step": 10801
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.866791744840526e-05,
+      "loss": 2.2416,
+      "step": 10802
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.865853658536585e-05,
+      "loss": 2.5802,
+      "step": 10803
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.864915572232646e-05,
+      "loss": 2.4163,
+      "step": 10804
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.863977485928706e-05,
+      "loss": 2.3192,
+      "step": 10805
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.863039399624766e-05,
+      "loss": 2.3413,
+      "step": 10806
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.862101313320826e-05,
+      "loss": 2.4675,
+      "step": 10807
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.861163227016886e-05,
+      "loss": 2.3184,
+      "step": 10808
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.860225140712946e-05,
+      "loss": 2.5754,
+      "step": 10809
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.859287054409007e-05,
+      "loss": 2.2786,
+      "step": 10810
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.858348968105066e-05,
+      "loss": 2.1754,
+      "step": 10811
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.857410881801125e-05,
+      "loss": 2.2411,
+      "step": 10812
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.856472795497186e-05,
+      "loss": 2.0482,
+      "step": 10813
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.855534709193246e-05,
+      "loss": 2.2884,
+      "step": 10814
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.854596622889306e-05,
+      "loss": 2.1996,
+      "step": 10815
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.853658536585366e-05,
+      "loss": 2.6157,
+      "step": 10816
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.852720450281426e-05,
+      "loss": 2.4106,
+      "step": 10817
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.851782363977486e-05,
+      "loss": 2.3984,
+      "step": 10818
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.850844277673547e-05,
+      "loss": 2.6287,
+      "step": 10819
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.849906191369607e-05,
+      "loss": 2.3047,
+      "step": 10820
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.848968105065666e-05,
+      "loss": 2.2832,
+      "step": 10821
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.848030018761727e-05,
+      "loss": 2.2359,
+      "step": 10822
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.847091932457787e-05,
+      "loss": 2.1713,
+      "step": 10823
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.846153846153848e-05,
+      "loss": 2.1179,
+      "step": 10824
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.845215759849906e-05,
+      "loss": 2.4397,
+      "step": 10825
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.844277673545966e-05,
+      "loss": 2.3645,
+      "step": 10826
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.843339587242026e-05,
+      "loss": 2.2399,
+      "step": 10827
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.842401500938087e-05,
+      "loss": 2.2383,
+      "step": 10828
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.841463414634147e-05,
+      "loss": 2.2434,
+      "step": 10829
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.840525328330206e-05,
+      "loss": 2.4457,
+      "step": 10830
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.839587242026267e-05,
+      "loss": 2.3109,
+      "step": 10831
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.838649155722327e-05,
+      "loss": 2.1439,
+      "step": 10832
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.837711069418388e-05,
+      "loss": 2.2548,
+      "step": 10833
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.836772983114447e-05,
+      "loss": 2.3971,
+      "step": 10834
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.835834896810507e-05,
+      "loss": 2.3441,
+      "step": 10835
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.834896810506568e-05,
+      "loss": 2.4768,
+      "step": 10836
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.833958724202628e-05,
+      "loss": 2.3516,
+      "step": 10837
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.833020637898687e-05,
+      "loss": 2.5387,
+      "step": 10838
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.832082551594746e-05,
+      "loss": 2.3844,
+      "step": 10839
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.831144465290807e-05,
+      "loss": 2.5459,
+      "step": 10840
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.830206378986867e-05,
+      "loss": 2.1172,
+      "step": 10841
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.829268292682928e-05,
+      "loss": 2.4264,
+      "step": 10842
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.828330206378987e-05,
+      "loss": 2.3431,
+      "step": 10843
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.827392120075047e-05,
+      "loss": 2.3074,
+      "step": 10844
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.826454033771108e-05,
+      "loss": 2.457,
+      "step": 10845
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.825515947467168e-05,
+      "loss": 2.3464,
+      "step": 10846
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.824577861163227e-05,
+      "loss": 2.5718,
+      "step": 10847
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.823639774859288e-05,
+      "loss": 2.3185,
+      "step": 10848
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.822701688555348e-05,
+      "loss": 2.468,
+      "step": 10849
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.821763602251409e-05,
+      "loss": 2.2412,
+      "step": 10850
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.820825515947468e-05,
+      "loss": 2.4418,
+      "step": 10851
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.819887429643527e-05,
+      "loss": 2.3606,
+      "step": 10852
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.818949343339587e-05,
+      "loss": 2.2381,
+      "step": 10853
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.818011257035648e-05,
+      "loss": 2.1596,
+      "step": 10854
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.817073170731708e-05,
+      "loss": 2.4244,
+      "step": 10855
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.816135084427767e-05,
+      "loss": 2.2474,
+      "step": 10856
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.815196998123828e-05,
+      "loss": 2.5633,
+      "step": 10857
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.814258911819888e-05,
+      "loss": 2.4721,
+      "step": 10858
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.813320825515949e-05,
+      "loss": 2.2133,
+      "step": 10859
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.812382739212008e-05,
+      "loss": 2.2324,
+      "step": 10860
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.811444652908068e-05,
+      "loss": 2.3519,
+      "step": 10861
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.810506566604129e-05,
+      "loss": 2.1361,
+      "step": 10862
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.809568480300189e-05,
+      "loss": 2.2994,
+      "step": 10863
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.808630393996248e-05,
+      "loss": 2.4603,
+      "step": 10864
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.807692307692307e-05,
+      "loss": 2.405,
+      "step": 10865
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.806754221388368e-05,
+      "loss": 2.3315,
+      "step": 10866
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.805816135084428e-05,
+      "loss": 2.2101,
+      "step": 10867
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.804878048780489e-05,
+      "loss": 2.4651,
+      "step": 10868
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.803939962476548e-05,
+      "loss": 2.4767,
+      "step": 10869
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.803001876172608e-05,
+      "loss": 2.2272,
+      "step": 10870
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.802063789868669e-05,
+      "loss": 2.1936,
+      "step": 10871
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.801125703564729e-05,
+      "loss": 2.4535,
+      "step": 10872
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.800187617260788e-05,
+      "loss": 2.3904,
+      "step": 10873
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.799249530956849e-05,
+      "loss": 2.4771,
+      "step": 10874
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.798311444652909e-05,
+      "loss": 2.4368,
+      "step": 10875
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.79737335834897e-05,
+      "loss": 2.3627,
+      "step": 10876
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.796435272045029e-05,
+      "loss": 2.4828,
+      "step": 10877
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.795497185741088e-05,
+      "loss": 2.5019,
+      "step": 10878
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.794559099437148e-05,
+      "loss": 2.4962,
+      "step": 10879
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.793621013133209e-05,
+      "loss": 2.5054,
+      "step": 10880
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.792682926829269e-05,
+      "loss": 2.5013,
+      "step": 10881
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.791744840525328e-05,
+      "loss": 2.2265,
+      "step": 10882
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.790806754221389e-05,
+      "loss": 2.3377,
+      "step": 10883
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.789868667917449e-05,
+      "loss": 2.6533,
+      "step": 10884
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.78893058161351e-05,
+      "loss": 2.394,
+      "step": 10885
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.787992495309569e-05,
+      "loss": 2.3733,
+      "step": 10886
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.787054409005629e-05,
+      "loss": 2.4283,
+      "step": 10887
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.78611632270169e-05,
+      "loss": 2.2175,
+      "step": 10888
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.78517823639775e-05,
+      "loss": 2.442,
+      "step": 10889
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.784240150093809e-05,
+      "loss": 2.4699,
+      "step": 10890
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.783302063789868e-05,
+      "loss": 2.4081,
+      "step": 10891
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.782363977485929e-05,
+      "loss": 2.3702,
+      "step": 10892
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.781425891181989e-05,
+      "loss": 2.6265,
+      "step": 10893
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.78048780487805e-05,
+      "loss": 2.5213,
+      "step": 10894
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.779549718574109e-05,
+      "loss": 2.3457,
+      "step": 10895
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.778611632270169e-05,
+      "loss": 2.5631,
+      "step": 10896
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.77767354596623e-05,
+      "loss": 2.3699,
+      "step": 10897
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.77673545966229e-05,
+      "loss": 2.644,
+      "step": 10898
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.77579737335835e-05,
+      "loss": 2.2341,
+      "step": 10899
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.77485928705441e-05,
+      "loss": 2.3639,
+      "step": 10900
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.77392120075047e-05,
+      "loss": 2.4843,
+      "step": 10901
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.77298311444653e-05,
+      "loss": 2.7168,
+      "step": 10902
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.77204502814259e-05,
+      "loss": 2.3631,
+      "step": 10903
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.771106941838649e-05,
+      "loss": 2.3664,
+      "step": 10904
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.770168855534709e-05,
+      "loss": 2.508,
+      "step": 10905
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.76923076923077e-05,
+      "loss": 2.4268,
+      "step": 10906
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.76829268292683e-05,
+      "loss": 2.321,
+      "step": 10907
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.76735459662289e-05,
+      "loss": 2.2304,
+      "step": 10908
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.76641651031895e-05,
+      "loss": 2.6101,
+      "step": 10909
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.76547842401501e-05,
+      "loss": 2.1798,
+      "step": 10910
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.76454033771107e-05,
+      "loss": 2.3031,
+      "step": 10911
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.763602251407131e-05,
+      "loss": 2.2399,
+      "step": 10912
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.76266416510319e-05,
+      "loss": 2.2992,
+      "step": 10913
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.76172607879925e-05,
+      "loss": 2.4075,
+      "step": 10914
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.760787992495311e-05,
+      "loss": 2.6468,
+      "step": 10915
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.75984990619137e-05,
+      "loss": 2.4567,
+      "step": 10916
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.75891181988743e-05,
+      "loss": 2.4806,
+      "step": 10917
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.75797373358349e-05,
+      "loss": 2.4002,
+      "step": 10918
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.75703564727955e-05,
+      "loss": 2.3371,
+      "step": 10919
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.75609756097561e-05,
+      "loss": 2.2135,
+      "step": 10920
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.755159474671671e-05,
+      "loss": 2.5065,
+      "step": 10921
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.75422138836773e-05,
+      "loss": 2.2887,
+      "step": 10922
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.75328330206379e-05,
+      "loss": 2.1372,
+      "step": 10923
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.752345215759851e-05,
+      "loss": 2.4457,
+      "step": 10924
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.751407129455911e-05,
+      "loss": 2.2923,
+      "step": 10925
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.75046904315197e-05,
+      "loss": 2.5558,
+      "step": 10926
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.749530956848031e-05,
+      "loss": 2.1423,
+      "step": 10927
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.748592870544091e-05,
+      "loss": 2.4674,
+      "step": 10928
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.747654784240152e-05,
+      "loss": 2.3706,
+      "step": 10929
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.746716697936211e-05,
+      "loss": 2.2026,
+      "step": 10930
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.74577861163227e-05,
+      "loss": 2.3531,
+      "step": 10931
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.74484052532833e-05,
+      "loss": 2.5127,
+      "step": 10932
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.743902439024391e-05,
+      "loss": 2.4551,
+      "step": 10933
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.742964352720451e-05,
+      "loss": 2.3832,
+      "step": 10934
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.74202626641651e-05,
+      "loss": 2.3753,
+      "step": 10935
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.741088180112571e-05,
+      "loss": 2.3559,
+      "step": 10936
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.740150093808631e-05,
+      "loss": 2.2407,
+      "step": 10937
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.739212007504692e-05,
+      "loss": 2.3323,
+      "step": 10938
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.738273921200751e-05,
+      "loss": 2.2129,
+      "step": 10939
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.737335834896811e-05,
+      "loss": 2.3672,
+      "step": 10940
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.736397748592872e-05,
+      "loss": 2.455,
+      "step": 10941
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.735459662288932e-05,
+      "loss": 2.4232,
+      "step": 10942
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.734521575984991e-05,
+      "loss": 2.1886,
+      "step": 10943
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.73358348968105e-05,
+      "loss": 2.4974,
+      "step": 10944
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.732645403377111e-05,
+      "loss": 2.3399,
+      "step": 10945
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.731707317073171e-05,
+      "loss": 2.1655,
+      "step": 10946
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.730769230769232e-05,
+      "loss": 2.4547,
+      "step": 10947
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.729831144465291e-05,
+      "loss": 2.529,
+      "step": 10948
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.728893058161351e-05,
+      "loss": 2.2504,
+      "step": 10949
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.727954971857412e-05,
+      "loss": 2.4453,
+      "step": 10950
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.727016885553472e-05,
+      "loss": 2.194,
+      "step": 10951
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.726078799249531e-05,
+      "loss": 2.4563,
+      "step": 10952
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.725140712945592e-05,
+      "loss": 2.4173,
+      "step": 10953
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.724202626641652e-05,
+      "loss": 2.2294,
+      "step": 10954
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.723264540337712e-05,
+      "loss": 2.6044,
+      "step": 10955
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.722326454033772e-05,
+      "loss": 2.1823,
+      "step": 10956
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.72138836772983e-05,
+      "loss": 2.358,
+      "step": 10957
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.720450281425891e-05,
+      "loss": 2.3557,
+      "step": 10958
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.719512195121952e-05,
+      "loss": 2.1075,
+      "step": 10959
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.718574108818012e-05,
+      "loss": 2.3525,
+      "step": 10960
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.717636022514071e-05,
+      "loss": 2.3957,
+      "step": 10961
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.716697936210132e-05,
+      "loss": 2.2323,
+      "step": 10962
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.715759849906192e-05,
+      "loss": 2.3415,
+      "step": 10963
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.714821763602252e-05,
+      "loss": 2.3913,
+      "step": 10964
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.713883677298312e-05,
+      "loss": 2.3662,
+      "step": 10965
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.712945590994372e-05,
+      "loss": 2.2085,
+      "step": 10966
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.712007504690432e-05,
+      "loss": 2.5946,
+      "step": 10967
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.711069418386493e-05,
+      "loss": 2.1447,
+      "step": 10968
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.710131332082552e-05,
+      "loss": 2.3917,
+      "step": 10969
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.709193245778611e-05,
+      "loss": 2.3288,
+      "step": 10970
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.708255159474672e-05,
+      "loss": 2.3117,
+      "step": 10971
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.707317073170732e-05,
+      "loss": 2.3796,
+      "step": 10972
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.706378986866792e-05,
+      "loss": 2.2701,
+      "step": 10973
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.705440900562852e-05,
+      "loss": 2.1957,
+      "step": 10974
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.704502814258912e-05,
+      "loss": 2.2829,
+      "step": 10975
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.703564727954972e-05,
+      "loss": 2.5432,
+      "step": 10976
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.702626641651033e-05,
+      "loss": 2.3166,
+      "step": 10977
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.701688555347092e-05,
+      "loss": 2.5642,
+      "step": 10978
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.700750469043152e-05,
+      "loss": 2.382,
+      "step": 10979
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.699812382739213e-05,
+      "loss": 2.3243,
+      "step": 10980
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.698874296435273e-05,
+      "loss": 2.4849,
+      "step": 10981
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.697936210131332e-05,
+      "loss": 2.3382,
+      "step": 10982
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.696998123827392e-05,
+      "loss": 2.5291,
+      "step": 10983
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.696060037523452e-05,
+      "loss": 2.4613,
+      "step": 10984
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.695121951219512e-05,
+      "loss": 2.2436,
+      "step": 10985
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.694183864915573e-05,
+      "loss": 2.2971,
+      "step": 10986
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.693245778611633e-05,
+      "loss": 2.33,
+      "step": 10987
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.692307692307692e-05,
+      "loss": 2.4755,
+      "step": 10988
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.691369606003753e-05,
+      "loss": 2.5894,
+      "step": 10989
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.690431519699813e-05,
+      "loss": 2.37,
+      "step": 10990
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.689493433395874e-05,
+      "loss": 2.6568,
+      "step": 10991
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.688555347091933e-05,
+      "loss": 2.2122,
+      "step": 10992
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.687617260787993e-05,
+      "loss": 2.4632,
+      "step": 10993
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.686679174484054e-05,
+      "loss": 2.5511,
+      "step": 10994
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.685741088180113e-05,
+      "loss": 2.3585,
+      "step": 10995
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.684803001876173e-05,
+      "loss": 2.7535,
+      "step": 10996
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.683864915572232e-05,
+      "loss": 2.4273,
+      "step": 10997
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.682926829268293e-05,
+      "loss": 2.4625,
+      "step": 10998
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.681988742964353e-05,
+      "loss": 2.2734,
+      "step": 10999
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.681050656660414e-05,
+      "loss": 2.3035,
+      "step": 11000
+    },
+    {
+      "epoch": 5.16,
+      "eval_loss": 2.6019270420074463,
+      "eval_runtime": 67.8366,
+      "eval_samples_per_second": 27.935,
+      "eval_steps_per_second": 3.494,
+      "step": 11000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.680112570356473e-05,
+      "loss": 2.3937,
+      "step": 11001
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.679174484052533e-05,
+      "loss": 2.4159,
+      "step": 11002
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.678236397748594e-05,
+      "loss": 2.4415,
+      "step": 11003
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.677298311444654e-05,
+      "loss": 2.2838,
+      "step": 11004
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.676360225140713e-05,
+      "loss": 2.2637,
+      "step": 11005
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.675422138836774e-05,
+      "loss": 2.3231,
+      "step": 11006
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.674484052532834e-05,
+      "loss": 2.1573,
+      "step": 11007
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.673545966228893e-05,
+      "loss": 2.3055,
+      "step": 11008
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.672607879924954e-05,
+      "loss": 2.342,
+      "step": 11009
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.671669793621013e-05,
+      "loss": 2.4124,
+      "step": 11010
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.670731707317073e-05,
+      "loss": 2.211,
+      "step": 11011
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.669793621013134e-05,
+      "loss": 2.3883,
+      "step": 11012
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.668855534709194e-05,
+      "loss": 2.6452,
+      "step": 11013
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.667917448405253e-05,
+      "loss": 2.2819,
+      "step": 11014
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.666979362101314e-05,
+      "loss": 2.4024,
+      "step": 11015
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.666041275797374e-05,
+      "loss": 2.4715,
+      "step": 11016
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.665103189493435e-05,
+      "loss": 2.3855,
+      "step": 11017
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.664165103189494e-05,
+      "loss": 2.2913,
+      "step": 11018
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.663227016885554e-05,
+      "loss": 2.263,
+      "step": 11019
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.662288930581615e-05,
+      "loss": 2.4822,
+      "step": 11020
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.661350844277675e-05,
+      "loss": 2.4535,
+      "step": 11021
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.660412757973734e-05,
+      "loss": 2.3582,
+      "step": 11022
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.659474671669793e-05,
+      "loss": 2.3339,
+      "step": 11023
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.658536585365854e-05,
+      "loss": 2.7157,
+      "step": 11024
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.657598499061914e-05,
+      "loss": 2.5391,
+      "step": 11025
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.656660412757975e-05,
+      "loss": 2.4622,
+      "step": 11026
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.655722326454034e-05,
+      "loss": 2.3617,
+      "step": 11027
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.654784240150094e-05,
+      "loss": 2.5347,
+      "step": 11028
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.653846153846155e-05,
+      "loss": 2.2281,
+      "step": 11029
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.652908067542215e-05,
+      "loss": 2.3074,
+      "step": 11030
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.651969981238274e-05,
+      "loss": 2.0992,
+      "step": 11031
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.651031894934335e-05,
+      "loss": 2.357,
+      "step": 11032
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.650093808630395e-05,
+      "loss": 2.5373,
+      "step": 11033
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.649155722326455e-05,
+      "loss": 2.327,
+      "step": 11034
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.648217636022515e-05,
+      "loss": 2.2559,
+      "step": 11035
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.647279549718574e-05,
+      "loss": 2.6795,
+      "step": 11036
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.646341463414634e-05,
+      "loss": 2.3782,
+      "step": 11037
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.645403377110695e-05,
+      "loss": 2.4515,
+      "step": 11038
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.644465290806755e-05,
+      "loss": 2.6927,
+      "step": 11039
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.643527204502814e-05,
+      "loss": 2.3234,
+      "step": 11040
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.642589118198875e-05,
+      "loss": 2.3318,
+      "step": 11041
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.641651031894935e-05,
+      "loss": 2.1987,
+      "step": 11042
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.640712945590995e-05,
+      "loss": 2.4498,
+      "step": 11043
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.639774859287055e-05,
+      "loss": 2.4285,
+      "step": 11044
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.638836772983115e-05,
+      "loss": 2.4837,
+      "step": 11045
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.637898686679175e-05,
+      "loss": 2.5644,
+      "step": 11046
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.636960600375236e-05,
+      "loss": 2.5752,
+      "step": 11047
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.636022514071295e-05,
+      "loss": 2.5305,
+      "step": 11048
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.635084427767354e-05,
+      "loss": 2.4433,
+      "step": 11049
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.634146341463415e-05,
+      "loss": 2.4475,
+      "step": 11050
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.633208255159475e-05,
+      "loss": 2.1249,
+      "step": 11051
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.632270168855535e-05,
+      "loss": 2.214,
+      "step": 11052
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.631332082551595e-05,
+      "loss": 2.4842,
+      "step": 11053
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.630393996247655e-05,
+      "loss": 2.3074,
+      "step": 11054
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.629455909943715e-05,
+      "loss": 2.3095,
+      "step": 11055
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.628517823639776e-05,
+      "loss": 2.4234,
+      "step": 11056
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.627579737335835e-05,
+      "loss": 2.322,
+      "step": 11057
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.626641651031895e-05,
+      "loss": 2.176,
+      "step": 11058
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.625703564727956e-05,
+      "loss": 2.3923,
+      "step": 11059
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.624765478424016e-05,
+      "loss": 2.452,
+      "step": 11060
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.623827392120075e-05,
+      "loss": 2.4744,
+      "step": 11061
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.622889305816134e-05,
+      "loss": 2.4846,
+      "step": 11062
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.621951219512195e-05,
+      "loss": 2.0636,
+      "step": 11063
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.621013133208255e-05,
+      "loss": 2.6018,
+      "step": 11064
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.620075046904316e-05,
+      "loss": 2.2245,
+      "step": 11065
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.619136960600376e-05,
+      "loss": 2.3709,
+      "step": 11066
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.618198874296435e-05,
+      "loss": 2.2803,
+      "step": 11067
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.617260787992496e-05,
+      "loss": 2.4283,
+      "step": 11068
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.616322701688556e-05,
+      "loss": 2.4099,
+      "step": 11069
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.615384615384617e-05,
+      "loss": 2.1983,
+      "step": 11070
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.614446529080676e-05,
+      "loss": 2.4947,
+      "step": 11071
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.613508442776736e-05,
+      "loss": 2.2303,
+      "step": 11072
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.612570356472797e-05,
+      "loss": 2.3039,
+      "step": 11073
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.611632270168856e-05,
+      "loss": 2.5222,
+      "step": 11074
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.610694183864916e-05,
+      "loss": 2.307,
+      "step": 11075
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.609756097560975e-05,
+      "loss": 2.463,
+      "step": 11076
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.608818011257036e-05,
+      "loss": 2.4734,
+      "step": 11077
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.607879924953096e-05,
+      "loss": 2.5341,
+      "step": 11078
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.606941838649157e-05,
+      "loss": 2.4923,
+      "step": 11079
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.606003752345216e-05,
+      "loss": 2.4885,
+      "step": 11080
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.605065666041276e-05,
+      "loss": 2.2884,
+      "step": 11081
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.604127579737337e-05,
+      "loss": 2.2688,
+      "step": 11082
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.603189493433397e-05,
+      "loss": 2.2079,
+      "step": 11083
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.602251407129456e-05,
+      "loss": 2.4454,
+      "step": 11084
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.601313320825517e-05,
+      "loss": 2.5431,
+      "step": 11085
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.600375234521577e-05,
+      "loss": 2.2584,
+      "step": 11086
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.599437148217636e-05,
+      "loss": 2.1451,
+      "step": 11087
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.598499061913697e-05,
+      "loss": 2.4314,
+      "step": 11088
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.597560975609756e-05,
+      "loss": 2.4145,
+      "step": 11089
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.596622889305816e-05,
+      "loss": 2.5485,
+      "step": 11090
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.595684803001877e-05,
+      "loss": 2.39,
+      "step": 11091
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.594746716697937e-05,
+      "loss": 2.3247,
+      "step": 11092
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.593808630393996e-05,
+      "loss": 2.8055,
+      "step": 11093
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.592870544090057e-05,
+      "loss": 2.312,
+      "step": 11094
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.591932457786117e-05,
+      "loss": 2.1093,
+      "step": 11095
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.590994371482178e-05,
+      "loss": 2.4912,
+      "step": 11096
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.590056285178237e-05,
+      "loss": 2.2967,
+      "step": 11097
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.589118198874297e-05,
+      "loss": 2.3198,
+      "step": 11098
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.588180112570358e-05,
+      "loss": 2.6934,
+      "step": 11099
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.587242026266417e-05,
+      "loss": 2.4142,
+      "step": 11100
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.586303939962477e-05,
+      "loss": 2.4531,
+      "step": 11101
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.585365853658536e-05,
+      "loss": 2.3415,
+      "step": 11102
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.584427767354597e-05,
+      "loss": 2.673,
+      "step": 11103
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.583489681050657e-05,
+      "loss": 2.228,
+      "step": 11104
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.582551594746718e-05,
+      "loss": 2.4031,
+      "step": 11105
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.581613508442777e-05,
+      "loss": 2.3973,
+      "step": 11106
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.580675422138837e-05,
+      "loss": 2.4658,
+      "step": 11107
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.579737335834898e-05,
+      "loss": 2.1369,
+      "step": 11108
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.578799249530958e-05,
+      "loss": 2.2882,
+      "step": 11109
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.577861163227017e-05,
+      "loss": 2.4457,
+      "step": 11110
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.576923076923078e-05,
+      "loss": 2.7538,
+      "step": 11111
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.575984990619138e-05,
+      "loss": 2.2999,
+      "step": 11112
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.575046904315197e-05,
+      "loss": 2.2526,
+      "step": 11113
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.574108818011258e-05,
+      "loss": 2.2338,
+      "step": 11114
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.573170731707317e-05,
+      "loss": 2.5287,
+      "step": 11115
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.572232645403377e-05,
+      "loss": 2.1803,
+      "step": 11116
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.571294559099438e-05,
+      "loss": 2.6436,
+      "step": 11117
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.570356472795498e-05,
+      "loss": 2.1853,
+      "step": 11118
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.569418386491557e-05,
+      "loss": 2.5029,
+      "step": 11119
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.568480300187618e-05,
+      "loss": 2.6148,
+      "step": 11120
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.567542213883678e-05,
+      "loss": 2.3291,
+      "step": 11121
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.566604127579738e-05,
+      "loss": 2.1874,
+      "step": 11122
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.565666041275798e-05,
+      "loss": 2.5709,
+      "step": 11123
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.564727954971858e-05,
+      "loss": 2.4677,
+      "step": 11124
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.563789868667918e-05,
+      "loss": 2.2924,
+      "step": 11125
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.562851782363979e-05,
+      "loss": 2.6034,
+      "step": 11126
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.561913696060038e-05,
+      "loss": 2.1392,
+      "step": 11127
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.560975609756097e-05,
+      "loss": 2.4045,
+      "step": 11128
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.560037523452158e-05,
+      "loss": 2.3798,
+      "step": 11129
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.559099437148218e-05,
+      "loss": 2.3284,
+      "step": 11130
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.558161350844278e-05,
+      "loss": 2.4633,
+      "step": 11131
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.557223264540337e-05,
+      "loss": 2.2112,
+      "step": 11132
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.556285178236398e-05,
+      "loss": 2.5977,
+      "step": 11133
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.555347091932458e-05,
+      "loss": 2.5632,
+      "step": 11134
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.554409005628519e-05,
+      "loss": 2.5719,
+      "step": 11135
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.553470919324578e-05,
+      "loss": 2.4843,
+      "step": 11136
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.552532833020638e-05,
+      "loss": 2.6135,
+      "step": 11137
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.551594746716699e-05,
+      "loss": 2.342,
+      "step": 11138
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.550656660412759e-05,
+      "loss": 2.1192,
+      "step": 11139
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.549718574108818e-05,
+      "loss": 2.2833,
+      "step": 11140
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.548780487804877e-05,
+      "loss": 2.4249,
+      "step": 11141
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.547842401500938e-05,
+      "loss": 2.3853,
+      "step": 11142
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.546904315196998e-05,
+      "loss": 2.4654,
+      "step": 11143
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.545966228893059e-05,
+      "loss": 2.3665,
+      "step": 11144
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.545028142589118e-05,
+      "loss": 2.497,
+      "step": 11145
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.544090056285178e-05,
+      "loss": 2.5536,
+      "step": 11146
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.543151969981239e-05,
+      "loss": 2.4426,
+      "step": 11147
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.542213883677299e-05,
+      "loss": 2.5675,
+      "step": 11148
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.54127579737336e-05,
+      "loss": 2.2126,
+      "step": 11149
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.540337711069419e-05,
+      "loss": 2.623,
+      "step": 11150
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.539399624765479e-05,
+      "loss": 2.2927,
+      "step": 11151
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.53846153846154e-05,
+      "loss": 2.389,
+      "step": 11152
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.537523452157599e-05,
+      "loss": 2.2959,
+      "step": 11153
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.536585365853659e-05,
+      "loss": 2.2904,
+      "step": 11154
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.535647279549718e-05,
+      "loss": 2.2568,
+      "step": 11155
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.534709193245779e-05,
+      "loss": 2.2138,
+      "step": 11156
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.533771106941839e-05,
+      "loss": 2.2995,
+      "step": 11157
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.5328330206379e-05,
+      "loss": 2.1975,
+      "step": 11158
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.531894934333959e-05,
+      "loss": 2.3365,
+      "step": 11159
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.530956848030019e-05,
+      "loss": 2.1252,
+      "step": 11160
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.53001876172608e-05,
+      "loss": 2.6012,
+      "step": 11161
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.52908067542214e-05,
+      "loss": 2.5356,
+      "step": 11162
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.528142589118199e-05,
+      "loss": 2.6454,
+      "step": 11163
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.52720450281426e-05,
+      "loss": 2.4404,
+      "step": 11164
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.52626641651032e-05,
+      "loss": 2.144,
+      "step": 11165
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.525328330206379e-05,
+      "loss": 2.3896,
+      "step": 11166
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.52439024390244e-05,
+      "loss": 2.2143,
+      "step": 11167
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.523452157598499e-05,
+      "loss": 2.42,
+      "step": 11168
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.522514071294559e-05,
+      "loss": 2.6628,
+      "step": 11169
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.52157598499062e-05,
+      "loss": 2.4082,
+      "step": 11170
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.52063789868668e-05,
+      "loss": 2.2904,
+      "step": 11171
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.519699812382739e-05,
+      "loss": 2.491,
+      "step": 11172
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.5187617260788e-05,
+      "loss": 2.2446,
+      "step": 11173
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.51782363977486e-05,
+      "loss": 2.4007,
+      "step": 11174
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.51688555347092e-05,
+      "loss": 2.6309,
+      "step": 11175
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.51594746716698e-05,
+      "loss": 2.5098,
+      "step": 11176
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.51500938086304e-05,
+      "loss": 2.3218,
+      "step": 11177
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.5140712945591e-05,
+      "loss": 2.5294,
+      "step": 11178
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.51313320825516e-05,
+      "loss": 2.3656,
+      "step": 11179
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.51219512195122e-05,
+      "loss": 2.5409,
+      "step": 11180
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.511257035647279e-05,
+      "loss": 2.403,
+      "step": 11181
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.51031894934334e-05,
+      "loss": 2.4919,
+      "step": 11182
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.5093808630394e-05,
+      "loss": 2.5239,
+      "step": 11183
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.50844277673546e-05,
+      "loss": 2.5465,
+      "step": 11184
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.50750469043152e-05,
+      "loss": 2.7086,
+      "step": 11185
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.50656660412758e-05,
+      "loss": 2.2668,
+      "step": 11186
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.50562851782364e-05,
+      "loss": 2.6142,
+      "step": 11187
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.504690431519701e-05,
+      "loss": 2.3616,
+      "step": 11188
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.50375234521576e-05,
+      "loss": 2.4161,
+      "step": 11189
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.50281425891182e-05,
+      "loss": 2.49,
+      "step": 11190
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.501876172607881e-05,
+      "loss": 2.3239,
+      "step": 11191
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.50093808630394e-05,
+      "loss": 2.1427,
+      "step": 11192
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.5e-05,
+      "loss": 2.5599,
+      "step": 11193
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.49906191369606e-05,
+      "loss": 2.1706,
+      "step": 11194
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.49812382739212e-05,
+      "loss": 2.268,
+      "step": 11195
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.49718574108818e-05,
+      "loss": 2.213,
+      "step": 11196
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.496247654784241e-05,
+      "loss": 2.4348,
+      "step": 11197
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.4953095684803e-05,
+      "loss": 2.3317,
+      "step": 11198
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.49437148217636e-05,
+      "loss": 2.5302,
+      "step": 11199
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.493433395872421e-05,
+      "loss": 2.0753,
+      "step": 11200
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.492495309568481e-05,
+      "loss": 2.3452,
+      "step": 11201
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.49155722326454e-05,
+      "loss": 2.5662,
+      "step": 11202
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.490619136960601e-05,
+      "loss": 2.5548,
+      "step": 11203
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.489681050656661e-05,
+      "loss": 2.5821,
+      "step": 11204
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.48874296435272e-05,
+      "loss": 2.3953,
+      "step": 11205
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.487804878048781e-05,
+      "loss": 2.5423,
+      "step": 11206
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.48686679174484e-05,
+      "loss": 2.4743,
+      "step": 11207
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.4859287054409e-05,
+      "loss": 2.4243,
+      "step": 11208
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.484990619136961e-05,
+      "loss": 2.4702,
+      "step": 11209
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.484052532833021e-05,
+      "loss": 2.525,
+      "step": 11210
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.48311444652908e-05,
+      "loss": 2.577,
+      "step": 11211
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.482176360225141e-05,
+      "loss": 2.5282,
+      "step": 11212
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.481238273921201e-05,
+      "loss": 2.4033,
+      "step": 11213
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.480300187617262e-05,
+      "loss": 2.3277,
+      "step": 11214
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.479362101313321e-05,
+      "loss": 2.1481,
+      "step": 11215
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.478424015009381e-05,
+      "loss": 2.5314,
+      "step": 11216
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.477485928705442e-05,
+      "loss": 2.6889,
+      "step": 11217
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.476547842401501e-05,
+      "loss": 2.3223,
+      "step": 11218
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.475609756097561e-05,
+      "loss": 2.3367,
+      "step": 11219
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.47467166979362e-05,
+      "loss": 2.1617,
+      "step": 11220
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.473733583489681e-05,
+      "loss": 2.2917,
+      "step": 11221
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.472795497185741e-05,
+      "loss": 2.5695,
+      "step": 11222
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.471857410881802e-05,
+      "loss": 2.3722,
+      "step": 11223
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.470919324577861e-05,
+      "loss": 2.2882,
+      "step": 11224
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.469981238273921e-05,
+      "loss": 2.4922,
+      "step": 11225
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.469043151969982e-05,
+      "loss": 2.0965,
+      "step": 11226
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.468105065666042e-05,
+      "loss": 2.2793,
+      "step": 11227
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.467166979362103e-05,
+      "loss": 2.3391,
+      "step": 11228
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.466228893058162e-05,
+      "loss": 2.1294,
+      "step": 11229
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.465290806754222e-05,
+      "loss": 2.6361,
+      "step": 11230
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.464352720450283e-05,
+      "loss": 2.3562,
+      "step": 11231
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.463414634146342e-05,
+      "loss": 2.3799,
+      "step": 11232
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.462476547842401e-05,
+      "loss": 2.4622,
+      "step": 11233
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.461538461538461e-05,
+      "loss": 2.0764,
+      "step": 11234
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.460600375234522e-05,
+      "loss": 2.3754,
+      "step": 11235
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.459662288930582e-05,
+      "loss": 2.2436,
+      "step": 11236
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.458724202626643e-05,
+      "loss": 2.4163,
+      "step": 11237
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.457786116322702e-05,
+      "loss": 2.4921,
+      "step": 11238
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.456848030018762e-05,
+      "loss": 2.4746,
+      "step": 11239
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.455909943714823e-05,
+      "loss": 2.5063,
+      "step": 11240
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.454971857410883e-05,
+      "loss": 2.3776,
+      "step": 11241
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.454033771106942e-05,
+      "loss": 2.238,
+      "step": 11242
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.453095684803003e-05,
+      "loss": 2.2779,
+      "step": 11243
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.452157598499063e-05,
+      "loss": 2.4828,
+      "step": 11244
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.451219512195122e-05,
+      "loss": 2.0675,
+      "step": 11245
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.450281425891183e-05,
+      "loss": 2.5071,
+      "step": 11246
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.449343339587242e-05,
+      "loss": 2.4525,
+      "step": 11247
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.448405253283302e-05,
+      "loss": 2.7054,
+      "step": 11248
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.447467166979363e-05,
+      "loss": 2.4183,
+      "step": 11249
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.446529080675423e-05,
+      "loss": 2.1297,
+      "step": 11250
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.445590994371482e-05,
+      "loss": 2.1478,
+      "step": 11251
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.444652908067543e-05,
+      "loss": 2.4438,
+      "step": 11252
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.443714821763603e-05,
+      "loss": 2.564,
+      "step": 11253
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.442776735459664e-05,
+      "loss": 2.2213,
+      "step": 11254
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.441838649155723e-05,
+      "loss": 2.422,
+      "step": 11255
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.440900562851783e-05,
+      "loss": 2.4966,
+      "step": 11256
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.439962476547844e-05,
+      "loss": 2.394,
+      "step": 11257
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.439024390243903e-05,
+      "loss": 2.6177,
+      "step": 11258
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.438086303939963e-05,
+      "loss": 2.2497,
+      "step": 11259
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.437148217636022e-05,
+      "loss": 2.2631,
+      "step": 11260
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.436210131332083e-05,
+      "loss": 2.3644,
+      "step": 11261
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.435272045028143e-05,
+      "loss": 2.7236,
+      "step": 11262
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.434333958724204e-05,
+      "loss": 2.5038,
+      "step": 11263
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.433395872420263e-05,
+      "loss": 2.3975,
+      "step": 11264
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.432457786116323e-05,
+      "loss": 2.2761,
+      "step": 11265
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.431519699812384e-05,
+      "loss": 2.2804,
+      "step": 11266
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.430581613508444e-05,
+      "loss": 2.489,
+      "step": 11267
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.429643527204503e-05,
+      "loss": 2.5145,
+      "step": 11268
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.428705440900564e-05,
+      "loss": 2.2659,
+      "step": 11269
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.427767354596624e-05,
+      "loss": 2.5779,
+      "step": 11270
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.426829268292683e-05,
+      "loss": 2.3375,
+      "step": 11271
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.425891181988744e-05,
+      "loss": 2.5358,
+      "step": 11272
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.424953095684803e-05,
+      "loss": 2.2036,
+      "step": 11273
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.424015009380863e-05,
+      "loss": 2.2494,
+      "step": 11274
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.423076923076924e-05,
+      "loss": 2.2422,
+      "step": 11275
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.422138836772984e-05,
+      "loss": 2.2427,
+      "step": 11276
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.421200750469043e-05,
+      "loss": 2.3143,
+      "step": 11277
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.420262664165103e-05,
+      "loss": 2.2926,
+      "step": 11278
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.419324577861164e-05,
+      "loss": 2.09,
+      "step": 11279
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.418386491557224e-05,
+      "loss": 2.1323,
+      "step": 11280
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.417448405253283e-05,
+      "loss": 2.4071,
+      "step": 11281
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.416510318949344e-05,
+      "loss": 2.3091,
+      "step": 11282
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.415572232645404e-05,
+      "loss": 2.6396,
+      "step": 11283
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.414634146341463e-05,
+      "loss": 2.3253,
+      "step": 11284
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.413696060037524e-05,
+      "loss": 2.6179,
+      "step": 11285
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.412757973733583e-05,
+      "loss": 2.6806,
+      "step": 11286
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.411819887429643e-05,
+      "loss": 2.4549,
+      "step": 11287
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.410881801125704e-05,
+      "loss": 2.61,
+      "step": 11288
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.409943714821764e-05,
+      "loss": 2.2125,
+      "step": 11289
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.409005628517823e-05,
+      "loss": 2.4531,
+      "step": 11290
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.408067542213884e-05,
+      "loss": 2.2491,
+      "step": 11291
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.407129455909944e-05,
+      "loss": 2.0852,
+      "step": 11292
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.406191369606005e-05,
+      "loss": 2.2515,
+      "step": 11293
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.405253283302064e-05,
+      "loss": 2.4818,
+      "step": 11294
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.404315196998124e-05,
+      "loss": 2.5097,
+      "step": 11295
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.403377110694185e-05,
+      "loss": 2.4345,
+      "step": 11296
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.402439024390244e-05,
+      "loss": 2.4721,
+      "step": 11297
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.401500938086304e-05,
+      "loss": 2.401,
+      "step": 11298
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.400562851782363e-05,
+      "loss": 2.2555,
+      "step": 11299
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.399624765478424e-05,
+      "loss": 2.395,
+      "step": 11300
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.398686679174484e-05,
+      "loss": 2.1647,
+      "step": 11301
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.397748592870545e-05,
+      "loss": 2.2504,
+      "step": 11302
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.396810506566604e-05,
+      "loss": 2.6293,
+      "step": 11303
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.395872420262664e-05,
+      "loss": 2.5645,
+      "step": 11304
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.394934333958725e-05,
+      "loss": 2.3338,
+      "step": 11305
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.393996247654785e-05,
+      "loss": 2.2057,
+      "step": 11306
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.393058161350846e-05,
+      "loss": 2.4519,
+      "step": 11307
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.392120075046905e-05,
+      "loss": 2.4299,
+      "step": 11308
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.391181988742965e-05,
+      "loss": 2.295,
+      "step": 11309
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.390243902439024e-05,
+      "loss": 2.1632,
+      "step": 11310
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.389305816135085e-05,
+      "loss": 2.162,
+      "step": 11311
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.388367729831144e-05,
+      "loss": 2.1933,
+      "step": 11312
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.387429643527204e-05,
+      "loss": 2.4811,
+      "step": 11313
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.386491557223265e-05,
+      "loss": 2.2446,
+      "step": 11314
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.385553470919325e-05,
+      "loss": 2.3802,
+      "step": 11315
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.384615384615386e-05,
+      "loss": 2.4915,
+      "step": 11316
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.383677298311445e-05,
+      "loss": 2.5819,
+      "step": 11317
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.382739212007505e-05,
+      "loss": 2.3523,
+      "step": 11318
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.381801125703566e-05,
+      "loss": 2.1585,
+      "step": 11319
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.380863039399626e-05,
+      "loss": 2.1721,
+      "step": 11320
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.379924953095685e-05,
+      "loss": 2.293,
+      "step": 11321
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.378986866791746e-05,
+      "loss": 2.251,
+      "step": 11322
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.378048780487805e-05,
+      "loss": 2.5678,
+      "step": 11323
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.377110694183865e-05,
+      "loss": 2.2707,
+      "step": 11324
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.376172607879926e-05,
+      "loss": 2.6118,
+      "step": 11325
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.375234521575985e-05,
+      "loss": 2.4911,
+      "step": 11326
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.374296435272045e-05,
+      "loss": 2.6168,
+      "step": 11327
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.373358348968106e-05,
+      "loss": 2.2845,
+      "step": 11328
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.372420262664166e-05,
+      "loss": 2.4807,
+      "step": 11329
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.371482176360225e-05,
+      "loss": 2.4099,
+      "step": 11330
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.370544090056286e-05,
+      "loss": 2.4012,
+      "step": 11331
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.369606003752346e-05,
+      "loss": 2.2926,
+      "step": 11332
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.368667917448407e-05,
+      "loss": 2.2657,
+      "step": 11333
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.367729831144466e-05,
+      "loss": 2.2749,
+      "step": 11334
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.366791744840526e-05,
+      "loss": 2.4398,
+      "step": 11335
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.365853658536587e-05,
+      "loss": 2.4695,
+      "step": 11336
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.364915572232646e-05,
+      "loss": 2.4338,
+      "step": 11337
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.363977485928706e-05,
+      "loss": 2.424,
+      "step": 11338
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.363039399624765e-05,
+      "loss": 2.2324,
+      "step": 11339
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.362101313320826e-05,
+      "loss": 2.0736,
+      "step": 11340
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.361163227016886e-05,
+      "loss": 2.4102,
+      "step": 11341
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.360225140712947e-05,
+      "loss": 2.6522,
+      "step": 11342
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.359287054409006e-05,
+      "loss": 2.4318,
+      "step": 11343
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.358348968105066e-05,
+      "loss": 2.3384,
+      "step": 11344
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.357410881801127e-05,
+      "loss": 2.2973,
+      "step": 11345
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.356472795497187e-05,
+      "loss": 2.4486,
+      "step": 11346
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.355534709193246e-05,
+      "loss": 2.3638,
+      "step": 11347
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.354596622889307e-05,
+      "loss": 2.6425,
+      "step": 11348
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.353658536585367e-05,
+      "loss": 2.4376,
+      "step": 11349
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.352720450281426e-05,
+      "loss": 2.3302,
+      "step": 11350
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.351782363977486e-05,
+      "loss": 2.296,
+      "step": 11351
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.350844277673546e-05,
+      "loss": 2.215,
+      "step": 11352
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.349906191369606e-05,
+      "loss": 2.4673,
+      "step": 11353
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.348968105065666e-05,
+      "loss": 2.2768,
+      "step": 11354
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.348030018761727e-05,
+      "loss": 2.3583,
+      "step": 11355
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.347091932457786e-05,
+      "loss": 2.4888,
+      "step": 11356
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.346153846153846e-05,
+      "loss": 2.3057,
+      "step": 11357
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.345215759849907e-05,
+      "loss": 2.4746,
+      "step": 11358
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.344277673545967e-05,
+      "loss": 2.6079,
+      "step": 11359
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.343339587242026e-05,
+      "loss": 2.1737,
+      "step": 11360
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.342401500938087e-05,
+      "loss": 2.5033,
+      "step": 11361
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.341463414634147e-05,
+      "loss": 2.5567,
+      "step": 11362
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.340525328330206e-05,
+      "loss": 2.4021,
+      "step": 11363
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.339587242026267e-05,
+      "loss": 2.52,
+      "step": 11364
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.338649155722326e-05,
+      "loss": 2.4691,
+      "step": 11365
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.337711069418386e-05,
+      "loss": 2.3101,
+      "step": 11366
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.336772983114447e-05,
+      "loss": 2.538,
+      "step": 11367
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.335834896810507e-05,
+      "loss": 2.4597,
+      "step": 11368
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.334896810506566e-05,
+      "loss": 2.7116,
+      "step": 11369
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.333958724202627e-05,
+      "loss": 2.6728,
+      "step": 11370
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.333020637898687e-05,
+      "loss": 2.4594,
+      "step": 11371
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.332082551594748e-05,
+      "loss": 2.5982,
+      "step": 11372
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.331144465290807e-05,
+      "loss": 2.1877,
+      "step": 11373
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.330206378986867e-05,
+      "loss": 2.5982,
+      "step": 11374
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.329268292682928e-05,
+      "loss": 2.2573,
+      "step": 11375
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.328330206378987e-05,
+      "loss": 2.3222,
+      "step": 11376
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.327392120075047e-05,
+      "loss": 2.3727,
+      "step": 11377
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.326454033771106e-05,
+      "loss": 2.2453,
+      "step": 11378
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.325515947467167e-05,
+      "loss": 2.2456,
+      "step": 11379
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.324577861163227e-05,
+      "loss": 2.5101,
+      "step": 11380
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.323639774859288e-05,
+      "loss": 2.3218,
+      "step": 11381
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.322701688555347e-05,
+      "loss": 2.4921,
+      "step": 11382
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.321763602251407e-05,
+      "loss": 2.4794,
+      "step": 11383
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.320825515947468e-05,
+      "loss": 2.3674,
+      "step": 11384
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.319887429643528e-05,
+      "loss": 2.4599,
+      "step": 11385
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.318949343339589e-05,
+      "loss": 2.369,
+      "step": 11386
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.318011257035648e-05,
+      "loss": 2.2718,
+      "step": 11387
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.317073170731708e-05,
+      "loss": 2.3387,
+      "step": 11388
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.316135084427767e-05,
+      "loss": 2.5196,
+      "step": 11389
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.315196998123828e-05,
+      "loss": 2.3648,
+      "step": 11390
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.314258911819887e-05,
+      "loss": 2.3464,
+      "step": 11391
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.313320825515947e-05,
+      "loss": 2.6723,
+      "step": 11392
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.312382739212008e-05,
+      "loss": 2.2994,
+      "step": 11393
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.311444652908068e-05,
+      "loss": 2.4371,
+      "step": 11394
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.310506566604129e-05,
+      "loss": 2.3387,
+      "step": 11395
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.309568480300188e-05,
+      "loss": 2.3409,
+      "step": 11396
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.308630393996248e-05,
+      "loss": 2.2765,
+      "step": 11397
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.307692307692309e-05,
+      "loss": 2.3367,
+      "step": 11398
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.306754221388369e-05,
+      "loss": 2.5304,
+      "step": 11399
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.305816135084428e-05,
+      "loss": 2.4109,
+      "step": 11400
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.304878048780489e-05,
+      "loss": 2.1901,
+      "step": 11401
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.303939962476548e-05,
+      "loss": 2.6613,
+      "step": 11402
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.303001876172608e-05,
+      "loss": 2.6985,
+      "step": 11403
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.302063789868669e-05,
+      "loss": 2.5463,
+      "step": 11404
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.301125703564728e-05,
+      "loss": 2.482,
+      "step": 11405
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.300187617260788e-05,
+      "loss": 2.2083,
+      "step": 11406
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.299249530956849e-05,
+      "loss": 2.4724,
+      "step": 11407
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.298311444652909e-05,
+      "loss": 2.2856,
+      "step": 11408
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.297373358348968e-05,
+      "loss": 2.2796,
+      "step": 11409
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.296435272045029e-05,
+      "loss": 2.4004,
+      "step": 11410
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.295497185741089e-05,
+      "loss": 2.6657,
+      "step": 11411
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.29455909943715e-05,
+      "loss": 2.3496,
+      "step": 11412
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.293621013133209e-05,
+      "loss": 2.4373,
+      "step": 11413
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.292682926829269e-05,
+      "loss": 2.2163,
+      "step": 11414
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.291744840525328e-05,
+      "loss": 2.4263,
+      "step": 11415
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.290806754221389e-05,
+      "loss": 2.2743,
+      "step": 11416
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.289868667917449e-05,
+      "loss": 2.5354,
+      "step": 11417
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.288930581613508e-05,
+      "loss": 2.4342,
+      "step": 11418
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.287992495309569e-05,
+      "loss": 2.6076,
+      "step": 11419
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.287054409005629e-05,
+      "loss": 2.5749,
+      "step": 11420
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.28611632270169e-05,
+      "loss": 2.3547,
+      "step": 11421
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.285178236397749e-05,
+      "loss": 2.5317,
+      "step": 11422
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.284240150093809e-05,
+      "loss": 2.4271,
+      "step": 11423
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.28330206378987e-05,
+      "loss": 2.2814,
+      "step": 11424
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.28236397748593e-05,
+      "loss": 2.1756,
+      "step": 11425
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.281425891181989e-05,
+      "loss": 2.5085,
+      "step": 11426
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.28048780487805e-05,
+      "loss": 2.3891,
+      "step": 11427
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.279549718574109e-05,
+      "loss": 2.3756,
+      "step": 11428
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.278611632270169e-05,
+      "loss": 2.1422,
+      "step": 11429
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.27767354596623e-05,
+      "loss": 2.207,
+      "step": 11430
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.276735459662289e-05,
+      "loss": 2.5903,
+      "step": 11431
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.275797373358349e-05,
+      "loss": 2.457,
+      "step": 11432
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.27485928705441e-05,
+      "loss": 2.3787,
+      "step": 11433
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.27392120075047e-05,
+      "loss": 2.3075,
+      "step": 11434
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.272983114446529e-05,
+      "loss": 2.2386,
+      "step": 11435
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.27204502814259e-05,
+      "loss": 2.4714,
+      "step": 11436
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.27110694183865e-05,
+      "loss": 2.4205,
+      "step": 11437
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.27016885553471e-05,
+      "loss": 2.3611,
+      "step": 11438
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.26923076923077e-05,
+      "loss": 2.2542,
+      "step": 11439
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.26829268292683e-05,
+      "loss": 2.3342,
+      "step": 11440
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.26735459662289e-05,
+      "loss": 2.2942,
+      "step": 11441
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.26641651031895e-05,
+      "loss": 2.3005,
+      "step": 11442
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.26547842401501e-05,
+      "loss": 2.1863,
+      "step": 11443
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.264540337711069e-05,
+      "loss": 2.2018,
+      "step": 11444
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.26360225140713e-05,
+      "loss": 2.1741,
+      "step": 11445
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.26266416510319e-05,
+      "loss": 2.4072,
+      "step": 11446
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.26172607879925e-05,
+      "loss": 2.4455,
+      "step": 11447
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.26078799249531e-05,
+      "loss": 2.4769,
+      "step": 11448
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.25984990619137e-05,
+      "loss": 2.3859,
+      "step": 11449
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.25891181988743e-05,
+      "loss": 2.487,
+      "step": 11450
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.257973733583491e-05,
+      "loss": 2.322,
+      "step": 11451
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.25703564727955e-05,
+      "loss": 2.3717,
+      "step": 11452
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.25609756097561e-05,
+      "loss": 2.2314,
+      "step": 11453
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.255159474671671e-05,
+      "loss": 2.2863,
+      "step": 11454
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.25422138836773e-05,
+      "loss": 2.3354,
+      "step": 11455
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.25328330206379e-05,
+      "loss": 2.2828,
+      "step": 11456
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.25234521575985e-05,
+      "loss": 2.2482,
+      "step": 11457
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.25140712945591e-05,
+      "loss": 2.5331,
+      "step": 11458
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.25046904315197e-05,
+      "loss": 2.3129,
+      "step": 11459
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.249530956848031e-05,
+      "loss": 2.4818,
+      "step": 11460
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.24859287054409e-05,
+      "loss": 2.3381,
+      "step": 11461
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.24765478424015e-05,
+      "loss": 2.6401,
+      "step": 11462
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.246716697936211e-05,
+      "loss": 2.4537,
+      "step": 11463
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.245778611632271e-05,
+      "loss": 2.4878,
+      "step": 11464
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.244840525328332e-05,
+      "loss": 2.3797,
+      "step": 11465
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.243902439024391e-05,
+      "loss": 2.4741,
+      "step": 11466
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.242964352720451e-05,
+      "loss": 2.4178,
+      "step": 11467
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.24202626641651e-05,
+      "loss": 2.3422,
+      "step": 11468
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.241088180112571e-05,
+      "loss": 2.37,
+      "step": 11469
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.24015009380863e-05,
+      "loss": 2.5113,
+      "step": 11470
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.23921200750469e-05,
+      "loss": 2.3641,
+      "step": 11471
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.238273921200751e-05,
+      "loss": 2.5125,
+      "step": 11472
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.237335834896811e-05,
+      "loss": 2.1178,
+      "step": 11473
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.236397748592872e-05,
+      "loss": 2.2625,
+      "step": 11474
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.235459662288931e-05,
+      "loss": 2.2225,
+      "step": 11475
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.234521575984991e-05,
+      "loss": 2.329,
+      "step": 11476
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.233583489681052e-05,
+      "loss": 2.5348,
+      "step": 11477
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.232645403377112e-05,
+      "loss": 2.2401,
+      "step": 11478
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.231707317073171e-05,
+      "loss": 2.2563,
+      "step": 11479
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 2.4654,
+      "step": 11480
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.229831144465291e-05,
+      "loss": 2.158,
+      "step": 11481
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.228893058161351e-05,
+      "loss": 2.3295,
+      "step": 11482
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.227954971857412e-05,
+      "loss": 2.0677,
+      "step": 11483
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.227016885553471e-05,
+      "loss": 2.2424,
+      "step": 11484
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.226078799249531e-05,
+      "loss": 2.6082,
+      "step": 11485
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.225140712945592e-05,
+      "loss": 2.317,
+      "step": 11486
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.224202626641652e-05,
+      "loss": 2.555,
+      "step": 11487
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.223264540337711e-05,
+      "loss": 2.3747,
+      "step": 11488
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.222326454033772e-05,
+      "loss": 2.3973,
+      "step": 11489
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.221388367729832e-05,
+      "loss": 2.5795,
+      "step": 11490
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.220450281425893e-05,
+      "loss": 2.7107,
+      "step": 11491
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.219512195121952e-05,
+      "loss": 2.1396,
+      "step": 11492
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.218574108818012e-05,
+      "loss": 2.337,
+      "step": 11493
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.217636022514071e-05,
+      "loss": 2.2772,
+      "step": 11494
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.216697936210132e-05,
+      "loss": 2.4825,
+      "step": 11495
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.215759849906192e-05,
+      "loss": 2.3244,
+      "step": 11496
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.214821763602251e-05,
+      "loss": 2.5086,
+      "step": 11497
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.213883677298312e-05,
+      "loss": 2.4115,
+      "step": 11498
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.212945590994372e-05,
+      "loss": 2.3071,
+      "step": 11499
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.212007504690432e-05,
+      "loss": 2.3037,
+      "step": 11500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.211069418386492e-05,
+      "loss": 2.3046,
+      "step": 11501
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.210131332082552e-05,
+      "loss": 2.5414,
+      "step": 11502
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.209193245778612e-05,
+      "loss": 2.4152,
+      "step": 11503
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.208255159474673e-05,
+      "loss": 2.4398,
+      "step": 11504
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.207317073170732e-05,
+      "loss": 2.4798,
+      "step": 11505
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.206378986866792e-05,
+      "loss": 2.4974,
+      "step": 11506
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.205440900562852e-05,
+      "loss": 2.6091,
+      "step": 11507
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.204502814258912e-05,
+      "loss": 2.367,
+      "step": 11508
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.203564727954972e-05,
+      "loss": 2.4839,
+      "step": 11509
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.202626641651032e-05,
+      "loss": 2.3457,
+      "step": 11510
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.201688555347092e-05,
+      "loss": 2.6502,
+      "step": 11511
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.200750469043152e-05,
+      "loss": 2.1668,
+      "step": 11512
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.199812382739213e-05,
+      "loss": 2.3879,
+      "step": 11513
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.198874296435272e-05,
+      "loss": 2.4708,
+      "step": 11514
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.197936210131332e-05,
+      "loss": 2.3872,
+      "step": 11515
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.196998123827393e-05,
+      "loss": 2.1874,
+      "step": 11516
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.196060037523453e-05,
+      "loss": 2.488,
+      "step": 11517
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.195121951219512e-05,
+      "loss": 2.1852,
+      "step": 11518
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.194183864915573e-05,
+      "loss": 2.2476,
+      "step": 11519
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.193245778611632e-05,
+      "loss": 2.1072,
+      "step": 11520
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.192307692307692e-05,
+      "loss": 2.1875,
+      "step": 11521
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.191369606003753e-05,
+      "loss": 2.6716,
+      "step": 11522
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.190431519699812e-05,
+      "loss": 2.3891,
+      "step": 11523
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.189493433395872e-05,
+      "loss": 2.1398,
+      "step": 11524
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.188555347091933e-05,
+      "loss": 2.3421,
+      "step": 11525
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.187617260787993e-05,
+      "loss": 2.4881,
+      "step": 11526
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.186679174484052e-05,
+      "loss": 2.2443,
+      "step": 11527
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.185741088180113e-05,
+      "loss": 2.2714,
+      "step": 11528
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.184803001876173e-05,
+      "loss": 2.4016,
+      "step": 11529
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.183864915572234e-05,
+      "loss": 2.38,
+      "step": 11530
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.182926829268293e-05,
+      "loss": 2.3518,
+      "step": 11531
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.181988742964353e-05,
+      "loss": 2.2168,
+      "step": 11532
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.181050656660414e-05,
+      "loss": 2.3042,
+      "step": 11533
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.180112570356473e-05,
+      "loss": 2.5432,
+      "step": 11534
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.179174484052533e-05,
+      "loss": 2.3601,
+      "step": 11535
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.178236397748592e-05,
+      "loss": 2.2906,
+      "step": 11536
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.177298311444653e-05,
+      "loss": 2.4112,
+      "step": 11537
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.176360225140713e-05,
+      "loss": 2.5317,
+      "step": 11538
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.175422138836774e-05,
+      "loss": 2.3428,
+      "step": 11539
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.174484052532833e-05,
+      "loss": 2.3457,
+      "step": 11540
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.173545966228893e-05,
+      "loss": 2.5744,
+      "step": 11541
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.172607879924954e-05,
+      "loss": 2.3916,
+      "step": 11542
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.171669793621014e-05,
+      "loss": 2.6123,
+      "step": 11543
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.170731707317075e-05,
+      "loss": 2.4561,
+      "step": 11544
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.169793621013134e-05,
+      "loss": 2.1655,
+      "step": 11545
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.168855534709194e-05,
+      "loss": 2.5509,
+      "step": 11546
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.167917448405253e-05,
+      "loss": 2.2713,
+      "step": 11547
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.166979362101314e-05,
+      "loss": 2.2438,
+      "step": 11548
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.166041275797373e-05,
+      "loss": 2.1832,
+      "step": 11549
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.165103189493433e-05,
+      "loss": 2.3573,
+      "step": 11550
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.164165103189494e-05,
+      "loss": 2.4424,
+      "step": 11551
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.163227016885554e-05,
+      "loss": 2.5783,
+      "step": 11552
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.162288930581615e-05,
+      "loss": 2.4011,
+      "step": 11553
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.161350844277674e-05,
+      "loss": 2.4366,
+      "step": 11554
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.160412757973734e-05,
+      "loss": 2.4378,
+      "step": 11555
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.159474671669795e-05,
+      "loss": 2.382,
+      "step": 11556
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.158536585365855e-05,
+      "loss": 2.17,
+      "step": 11557
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.157598499061914e-05,
+      "loss": 2.4414,
+      "step": 11558
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.156660412757975e-05,
+      "loss": 2.5302,
+      "step": 11559
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.155722326454034e-05,
+      "loss": 2.1319,
+      "step": 11560
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.154784240150094e-05,
+      "loss": 2.3238,
+      "step": 11561
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.153846153846155e-05,
+      "loss": 2.3545,
+      "step": 11562
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.152908067542214e-05,
+      "loss": 2.1797,
+      "step": 11563
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.151969981238274e-05,
+      "loss": 2.1865,
+      "step": 11564
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.151031894934335e-05,
+      "loss": 2.1975,
+      "step": 11565
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.150093808630395e-05,
+      "loss": 2.4042,
+      "step": 11566
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.149155722326454e-05,
+      "loss": 2.7283,
+      "step": 11567
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.148217636022515e-05,
+      "loss": 2.348,
+      "step": 11568
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.147279549718575e-05,
+      "loss": 2.5308,
+      "step": 11569
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.146341463414635e-05,
+      "loss": 2.3296,
+      "step": 11570
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.145403377110695e-05,
+      "loss": 2.5072,
+      "step": 11571
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.144465290806755e-05,
+      "loss": 2.4017,
+      "step": 11572
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.143527204502814e-05,
+      "loss": 2.5357,
+      "step": 11573
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.142589118198875e-05,
+      "loss": 2.4543,
+      "step": 11574
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.141651031894935e-05,
+      "loss": 2.3868,
+      "step": 11575
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.140712945590994e-05,
+      "loss": 2.3873,
+      "step": 11576
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.139774859287055e-05,
+      "loss": 2.4037,
+      "step": 11577
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.138836772983115e-05,
+      "loss": 2.3484,
+      "step": 11578
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.137898686679175e-05,
+      "loss": 2.2894,
+      "step": 11579
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.136960600375235e-05,
+      "loss": 2.1854,
+      "step": 11580
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.136022514071295e-05,
+      "loss": 2.4919,
+      "step": 11581
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.135084427767355e-05,
+      "loss": 2.4805,
+      "step": 11582
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.134146341463416e-05,
+      "loss": 2.2246,
+      "step": 11583
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.133208255159475e-05,
+      "loss": 2.492,
+      "step": 11584
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.132270168855535e-05,
+      "loss": 2.4135,
+      "step": 11585
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.131332082551595e-05,
+      "loss": 2.2158,
+      "step": 11586
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.130393996247655e-05,
+      "loss": 2.3432,
+      "step": 11587
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.129455909943715e-05,
+      "loss": 2.4963,
+      "step": 11588
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.128517823639775e-05,
+      "loss": 2.1208,
+      "step": 11589
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.127579737335835e-05,
+      "loss": 2.1388,
+      "step": 11590
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.126641651031895e-05,
+      "loss": 2.2549,
+      "step": 11591
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.125703564727956e-05,
+      "loss": 2.3691,
+      "step": 11592
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.124765478424015e-05,
+      "loss": 2.232,
+      "step": 11593
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.123827392120075e-05,
+      "loss": 2.4154,
+      "step": 11594
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.122889305816136e-05,
+      "loss": 2.3114,
+      "step": 11595
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.121951219512196e-05,
+      "loss": 2.4379,
+      "step": 11596
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.121013133208255e-05,
+      "loss": 2.3123,
+      "step": 11597
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.120075046904316e-05,
+      "loss": 2.2688,
+      "step": 11598
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.119136960600375e-05,
+      "loss": 2.4087,
+      "step": 11599
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.118198874296435e-05,
+      "loss": 2.4278,
+      "step": 11600
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.117260787992496e-05,
+      "loss": 2.3878,
+      "step": 11601
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.116322701688555e-05,
+      "loss": 2.5186,
+      "step": 11602
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.115384615384615e-05,
+      "loss": 2.3941,
+      "step": 11603
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.114446529080676e-05,
+      "loss": 2.2995,
+      "step": 11604
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.113508442776736e-05,
+      "loss": 2.3198,
+      "step": 11605
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.112570356472795e-05,
+      "loss": 2.2571,
+      "step": 11606
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.111632270168856e-05,
+      "loss": 2.3564,
+      "step": 11607
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.110694183864916e-05,
+      "loss": 2.4815,
+      "step": 11608
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.109756097560977e-05,
+      "loss": 2.3755,
+      "step": 11609
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.108818011257036e-05,
+      "loss": 2.3139,
+      "step": 11610
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.107879924953096e-05,
+      "loss": 2.5167,
+      "step": 11611
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.106941838649155e-05,
+      "loss": 2.2458,
+      "step": 11612
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.106003752345216e-05,
+      "loss": 2.1966,
+      "step": 11613
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.105065666041276e-05,
+      "loss": 2.2638,
+      "step": 11614
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.104127579737335e-05,
+      "loss": 2.2672,
+      "step": 11615
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.103189493433396e-05,
+      "loss": 2.3123,
+      "step": 11616
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.102251407129456e-05,
+      "loss": 2.1513,
+      "step": 11617
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.101313320825517e-05,
+      "loss": 2.3174,
+      "step": 11618
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.100375234521576e-05,
+      "loss": 2.5534,
+      "step": 11619
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.099437148217636e-05,
+      "loss": 2.7855,
+      "step": 11620
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.098499061913697e-05,
+      "loss": 2.1869,
+      "step": 11621
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.097560975609757e-05,
+      "loss": 2.4468,
+      "step": 11622
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.096622889305818e-05,
+      "loss": 2.5349,
+      "step": 11623
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.095684803001877e-05,
+      "loss": 2.217,
+      "step": 11624
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.094746716697936e-05,
+      "loss": 2.4739,
+      "step": 11625
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.093808630393996e-05,
+      "loss": 2.3291,
+      "step": 11626
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.092870544090057e-05,
+      "loss": 2.5612,
+      "step": 11627
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.091932457786116e-05,
+      "loss": 2.4976,
+      "step": 11628
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.090994371482176e-05,
+      "loss": 2.6325,
+      "step": 11629
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.090056285178237e-05,
+      "loss": 2.3428,
+      "step": 11630
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.089118198874297e-05,
+      "loss": 2.3826,
+      "step": 11631
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.088180112570358e-05,
+      "loss": 2.3753,
+      "step": 11632
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.087242026266417e-05,
+      "loss": 2.2887,
+      "step": 11633
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.086303939962477e-05,
+      "loss": 2.3176,
+      "step": 11634
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.085365853658538e-05,
+      "loss": 2.5296,
+      "step": 11635
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.084427767354598e-05,
+      "loss": 2.6819,
+      "step": 11636
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.083489681050657e-05,
+      "loss": 2.5366,
+      "step": 11637
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.082551594746718e-05,
+      "loss": 2.4309,
+      "step": 11638
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.081613508442777e-05,
+      "loss": 2.1644,
+      "step": 11639
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.080675422138837e-05,
+      "loss": 2.6494,
+      "step": 11640
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.079737335834898e-05,
+      "loss": 2.3415,
+      "step": 11641
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.078799249530957e-05,
+      "loss": 2.4731,
+      "step": 11642
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.077861163227017e-05,
+      "loss": 2.2525,
+      "step": 11643
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.076923076923078e-05,
+      "loss": 2.3397,
+      "step": 11644
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.075984990619138e-05,
+      "loss": 2.139,
+      "step": 11645
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.075046904315197e-05,
+      "loss": 2.2661,
+      "step": 11646
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.074108818011258e-05,
+      "loss": 2.4363,
+      "step": 11647
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.073170731707318e-05,
+      "loss": 2.2654,
+      "step": 11648
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.072232645403378e-05,
+      "loss": 2.5398,
+      "step": 11649
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.071294559099438e-05,
+      "loss": 2.4265,
+      "step": 11650
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.070356472795498e-05,
+      "loss": 2.4082,
+      "step": 11651
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.069418386491557e-05,
+      "loss": 2.3344,
+      "step": 11652
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.068480300187618e-05,
+      "loss": 2.3863,
+      "step": 11653
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.067542213883678e-05,
+      "loss": 2.5334,
+      "step": 11654
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.066604127579737e-05,
+      "loss": 2.8169,
+      "step": 11655
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.065666041275798e-05,
+      "loss": 2.5091,
+      "step": 11656
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.064727954971858e-05,
+      "loss": 2.7787,
+      "step": 11657
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.063789868667918e-05,
+      "loss": 2.508,
+      "step": 11658
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.062851782363978e-05,
+      "loss": 2.4842,
+      "step": 11659
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.061913696060038e-05,
+      "loss": 2.3352,
+      "step": 11660
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.060975609756098e-05,
+      "loss": 2.4047,
+      "step": 11661
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.060037523452159e-05,
+      "loss": 2.5151,
+      "step": 11662
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.059099437148218e-05,
+      "loss": 2.5888,
+      "step": 11663
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.058161350844278e-05,
+      "loss": 2.4803,
+      "step": 11664
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.057223264540338e-05,
+      "loss": 2.3692,
+      "step": 11665
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.056285178236398e-05,
+      "loss": 2.4339,
+      "step": 11666
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.055347091932458e-05,
+      "loss": 2.4404,
+      "step": 11667
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.054409005628518e-05,
+      "loss": 2.3963,
+      "step": 11668
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.053470919324578e-05,
+      "loss": 2.397,
+      "step": 11669
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.052532833020638e-05,
+      "loss": 2.5555,
+      "step": 11670
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.051594746716699e-05,
+      "loss": 2.2238,
+      "step": 11671
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.050656660412758e-05,
+      "loss": 2.4527,
+      "step": 11672
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.049718574108818e-05,
+      "loss": 2.2739,
+      "step": 11673
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.048780487804879e-05,
+      "loss": 2.4429,
+      "step": 11674
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.04784240150094e-05,
+      "loss": 2.2067,
+      "step": 11675
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.046904315196998e-05,
+      "loss": 2.4408,
+      "step": 11676
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.045966228893059e-05,
+      "loss": 2.6165,
+      "step": 11677
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.045028142589118e-05,
+      "loss": 2.3407,
+      "step": 11678
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.044090056285178e-05,
+      "loss": 2.6189,
+      "step": 11679
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.043151969981239e-05,
+      "loss": 2.3789,
+      "step": 11680
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.042213883677298e-05,
+      "loss": 2.408,
+      "step": 11681
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.041275797373358e-05,
+      "loss": 2.472,
+      "step": 11682
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.040337711069419e-05,
+      "loss": 2.3139,
+      "step": 11683
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.039399624765479e-05,
+      "loss": 2.2577,
+      "step": 11684
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.038461538461538e-05,
+      "loss": 2.4791,
+      "step": 11685
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.037523452157599e-05,
+      "loss": 2.2876,
+      "step": 11686
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.036585365853659e-05,
+      "loss": 2.5494,
+      "step": 11687
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.03564727954972e-05,
+      "loss": 2.5201,
+      "step": 11688
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.034709193245779e-05,
+      "loss": 2.5093,
+      "step": 11689
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.033771106941839e-05,
+      "loss": 2.5508,
+      "step": 11690
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.032833020637898e-05,
+      "loss": 2.3485,
+      "step": 11691
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.031894934333959e-05,
+      "loss": 2.3249,
+      "step": 11692
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.030956848030019e-05,
+      "loss": 2.4907,
+      "step": 11693
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.030018761726078e-05,
+      "loss": 2.7121,
+      "step": 11694
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.029080675422139e-05,
+      "loss": 2.2153,
+      "step": 11695
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.028142589118199e-05,
+      "loss": 2.2845,
+      "step": 11696
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.02720450281426e-05,
+      "loss": 2.2562,
+      "step": 11697
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.026266416510319e-05,
+      "loss": 2.4616,
+      "step": 11698
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.025328330206379e-05,
+      "loss": 2.2663,
+      "step": 11699
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.02439024390244e-05,
+      "loss": 2.3451,
+      "step": 11700
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.0234521575985e-05,
+      "loss": 2.2192,
+      "step": 11701
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.02251407129456e-05,
+      "loss": 2.259,
+      "step": 11702
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.02157598499062e-05,
+      "loss": 2.3352,
+      "step": 11703
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.020637898686679e-05,
+      "loss": 2.4643,
+      "step": 11704
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.019699812382739e-05,
+      "loss": 2.1644,
+      "step": 11705
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.0187617260788e-05,
+      "loss": 2.3611,
+      "step": 11706
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.017823639774859e-05,
+      "loss": 2.3427,
+      "step": 11707
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.016885553470919e-05,
+      "loss": 2.255,
+      "step": 11708
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.01594746716698e-05,
+      "loss": 2.3539,
+      "step": 11709
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.01500938086304e-05,
+      "loss": 2.2348,
+      "step": 11710
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.0140712945591e-05,
+      "loss": 2.2683,
+      "step": 11711
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.01313320825516e-05,
+      "loss": 2.4839,
+      "step": 11712
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.01219512195122e-05,
+      "loss": 2.243,
+      "step": 11713
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.01125703564728e-05,
+      "loss": 2.3409,
+      "step": 11714
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.010318949343341e-05,
+      "loss": 2.4841,
+      "step": 11715
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.0093808630394e-05,
+      "loss": 2.3528,
+      "step": 11716
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.008442776735459e-05,
+      "loss": 2.3753,
+      "step": 11717
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.00750469043152e-05,
+      "loss": 2.4438,
+      "step": 11718
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.00656660412758e-05,
+      "loss": 2.0532,
+      "step": 11719
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.00562851782364e-05,
+      "loss": 2.4358,
+      "step": 11720
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.0046904315197e-05,
+      "loss": 2.4959,
+      "step": 11721
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.00375234521576e-05,
+      "loss": 2.4607,
+      "step": 11722
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.00281425891182e-05,
+      "loss": 2.747,
+      "step": 11723
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.001876172607881e-05,
+      "loss": 2.4773,
+      "step": 11724
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.00093808630394e-05,
+      "loss": 2.1032,
+      "step": 11725
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9e-05,
+      "loss": 2.297,
+      "step": 11726
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.999061913696061e-05,
+      "loss": 2.2331,
+      "step": 11727
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.998123827392121e-05,
+      "loss": 2.1984,
+      "step": 11728
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.99718574108818e-05,
+      "loss": 2.1098,
+      "step": 11729
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.99624765478424e-05,
+      "loss": 2.3012,
+      "step": 11730
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.9953095684803e-05,
+      "loss": 2.3731,
+      "step": 11731
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.99437148217636e-05,
+      "loss": 2.159,
+      "step": 11732
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.993433395872421e-05,
+      "loss": 2.2947,
+      "step": 11733
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.99249530956848e-05,
+      "loss": 2.3565,
+      "step": 11734
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.99155722326454e-05,
+      "loss": 2.6074,
+      "step": 11735
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.990619136960601e-05,
+      "loss": 2.3458,
+      "step": 11736
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.989681050656661e-05,
+      "loss": 2.4388,
+      "step": 11737
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.98874296435272e-05,
+      "loss": 2.427,
+      "step": 11738
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.987804878048781e-05,
+      "loss": 2.511,
+      "step": 11739
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.986866791744841e-05,
+      "loss": 2.2959,
+      "step": 11740
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.985928705440902e-05,
+      "loss": 2.024,
+      "step": 11741
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.984990619136961e-05,
+      "loss": 2.3496,
+      "step": 11742
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.984052532833021e-05,
+      "loss": 2.2781,
+      "step": 11743
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.98311444652908e-05,
+      "loss": 2.5372,
+      "step": 11744
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.982176360225141e-05,
+      "loss": 2.5631,
+      "step": 11745
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.981238273921201e-05,
+      "loss": 2.358,
+      "step": 11746
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.98030018761726e-05,
+      "loss": 2.4845,
+      "step": 11747
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.979362101313321e-05,
+      "loss": 2.416,
+      "step": 11748
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.978424015009381e-05,
+      "loss": 2.3728,
+      "step": 11749
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.977485928705442e-05,
+      "loss": 2.2015,
+      "step": 11750
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.976547842401501e-05,
+      "loss": 2.3565,
+      "step": 11751
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.975609756097561e-05,
+      "loss": 2.4163,
+      "step": 11752
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.974671669793622e-05,
+      "loss": 2.351,
+      "step": 11753
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.973733583489682e-05,
+      "loss": 2.402,
+      "step": 11754
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.972795497185741e-05,
+      "loss": 2.6801,
+      "step": 11755
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.971857410881802e-05,
+      "loss": 2.2537,
+      "step": 11756
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.970919324577861e-05,
+      "loss": 2.2743,
+      "step": 11757
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.969981238273921e-05,
+      "loss": 2.3316,
+      "step": 11758
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.969043151969982e-05,
+      "loss": 2.4888,
+      "step": 11759
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.968105065666041e-05,
+      "loss": 2.4043,
+      "step": 11760
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.967166979362101e-05,
+      "loss": 2.4715,
+      "step": 11761
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.966228893058162e-05,
+      "loss": 2.1954,
+      "step": 11762
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.965290806754222e-05,
+      "loss": 2.5015,
+      "step": 11763
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.964352720450281e-05,
+      "loss": 2.3379,
+      "step": 11764
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.963414634146342e-05,
+      "loss": 2.1986,
+      "step": 11765
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.962476547842402e-05,
+      "loss": 2.2513,
+      "step": 11766
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.961538461538463e-05,
+      "loss": 2.7822,
+      "step": 11767
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.960600375234522e-05,
+      "loss": 2.1225,
+      "step": 11768
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.959662288930582e-05,
+      "loss": 2.4286,
+      "step": 11769
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.958724202626641e-05,
+      "loss": 2.264,
+      "step": 11770
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.957786116322702e-05,
+      "loss": 2.2175,
+      "step": 11771
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.956848030018762e-05,
+      "loss": 2.3519,
+      "step": 11772
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.955909943714821e-05,
+      "loss": 2.3141,
+      "step": 11773
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.954971857410882e-05,
+      "loss": 2.3406,
+      "step": 11774
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.954033771106942e-05,
+      "loss": 2.4214,
+      "step": 11775
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.953095684803003e-05,
+      "loss": 2.4526,
+      "step": 11776
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.952157598499062e-05,
+      "loss": 2.3205,
+      "step": 11777
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.951219512195122e-05,
+      "loss": 2.3533,
+      "step": 11778
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.950281425891183e-05,
+      "loss": 2.2619,
+      "step": 11779
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.949343339587243e-05,
+      "loss": 2.4563,
+      "step": 11780
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.948405253283302e-05,
+      "loss": 2.3547,
+      "step": 11781
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.947467166979363e-05,
+      "loss": 2.4181,
+      "step": 11782
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.946529080675422e-05,
+      "loss": 2.2777,
+      "step": 11783
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.945590994371482e-05,
+      "loss": 2.4177,
+      "step": 11784
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.944652908067543e-05,
+      "loss": 2.3058,
+      "step": 11785
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.943714821763602e-05,
+      "loss": 2.1339,
+      "step": 11786
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.942776735459662e-05,
+      "loss": 2.5474,
+      "step": 11787
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.941838649155723e-05,
+      "loss": 2.457,
+      "step": 11788
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.940900562851783e-05,
+      "loss": 2.4709,
+      "step": 11789
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.939962476547844e-05,
+      "loss": 2.3544,
+      "step": 11790
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.939024390243903e-05,
+      "loss": 2.2567,
+      "step": 11791
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.938086303939963e-05,
+      "loss": 2.4672,
+      "step": 11792
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.937148217636024e-05,
+      "loss": 2.474,
+      "step": 11793
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.936210131332084e-05,
+      "loss": 2.4125,
+      "step": 11794
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.935272045028143e-05,
+      "loss": 2.2936,
+      "step": 11795
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.934333958724202e-05,
+      "loss": 2.4827,
+      "step": 11796
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.933395872420263e-05,
+      "loss": 2.455,
+      "step": 11797
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.932457786116323e-05,
+      "loss": 2.3037,
+      "step": 11798
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.931519699812384e-05,
+      "loss": 2.3337,
+      "step": 11799
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.930581613508443e-05,
+      "loss": 2.2216,
+      "step": 11800
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.929643527204503e-05,
+      "loss": 2.2691,
+      "step": 11801
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.928705440900564e-05,
+      "loss": 2.2726,
+      "step": 11802
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.927767354596624e-05,
+      "loss": 2.3888,
+      "step": 11803
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.926829268292683e-05,
+      "loss": 2.2653,
+      "step": 11804
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.925891181988744e-05,
+      "loss": 2.5845,
+      "step": 11805
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.924953095684804e-05,
+      "loss": 2.3285,
+      "step": 11806
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.924015009380864e-05,
+      "loss": 2.541,
+      "step": 11807
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.923076923076924e-05,
+      "loss": 2.446,
+      "step": 11808
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.922138836772983e-05,
+      "loss": 2.3882,
+      "step": 11809
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.921200750469043e-05,
+      "loss": 2.1517,
+      "step": 11810
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.920262664165104e-05,
+      "loss": 2.5333,
+      "step": 11811
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.919324577861164e-05,
+      "loss": 2.137,
+      "step": 11812
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.918386491557223e-05,
+      "loss": 2.4177,
+      "step": 11813
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.917448405253284e-05,
+      "loss": 2.2257,
+      "step": 11814
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.916510318949344e-05,
+      "loss": 2.2022,
+      "step": 11815
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.915572232645404e-05,
+      "loss": 2.3568,
+      "step": 11816
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.914634146341464e-05,
+      "loss": 2.4571,
+      "step": 11817
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.913696060037524e-05,
+      "loss": 2.2784,
+      "step": 11818
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.912757973733584e-05,
+      "loss": 2.3171,
+      "step": 11819
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.911819887429645e-05,
+      "loss": 2.3093,
+      "step": 11820
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.910881801125704e-05,
+      "loss": 2.4278,
+      "step": 11821
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.909943714821763e-05,
+      "loss": 2.3959,
+      "step": 11822
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.909005628517824e-05,
+      "loss": 2.6684,
+      "step": 11823
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.908067542213884e-05,
+      "loss": 2.4873,
+      "step": 11824
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.907129455909944e-05,
+      "loss": 2.7199,
+      "step": 11825
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.906191369606004e-05,
+      "loss": 2.4254,
+      "step": 11826
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.905253283302064e-05,
+      "loss": 2.2703,
+      "step": 11827
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.904315196998124e-05,
+      "loss": 2.2894,
+      "step": 11828
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.903377110694185e-05,
+      "loss": 2.4974,
+      "step": 11829
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.902439024390244e-05,
+      "loss": 2.2832,
+      "step": 11830
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.901500938086304e-05,
+      "loss": 2.6274,
+      "step": 11831
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.900562851782365e-05,
+      "loss": 2.4132,
+      "step": 11832
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.899624765478425e-05,
+      "loss": 2.2467,
+      "step": 11833
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.898686679174484e-05,
+      "loss": 2.4337,
+      "step": 11834
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.897748592870543e-05,
+      "loss": 2.5912,
+      "step": 11835
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.896810506566604e-05,
+      "loss": 2.5057,
+      "step": 11836
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.895872420262664e-05,
+      "loss": 2.7,
+      "step": 11837
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.894934333958725e-05,
+      "loss": 2.5511,
+      "step": 11838
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.893996247654784e-05,
+      "loss": 2.4864,
+      "step": 11839
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.893058161350844e-05,
+      "loss": 2.4767,
+      "step": 11840
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.892120075046905e-05,
+      "loss": 2.3229,
+      "step": 11841
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.891181988742965e-05,
+      "loss": 2.4604,
+      "step": 11842
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.890243902439024e-05,
+      "loss": 2.5043,
+      "step": 11843
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.889305816135085e-05,
+      "loss": 2.3982,
+      "step": 11844
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.888367729831145e-05,
+      "loss": 2.0519,
+      "step": 11845
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.887429643527206e-05,
+      "loss": 2.5666,
+      "step": 11846
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.886491557223265e-05,
+      "loss": 2.306,
+      "step": 11847
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.885553470919325e-05,
+      "loss": 2.3274,
+      "step": 11848
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.884615384615384e-05,
+      "loss": 2.4395,
+      "step": 11849
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.883677298311445e-05,
+      "loss": 2.1608,
+      "step": 11850
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.882739212007505e-05,
+      "loss": 2.4614,
+      "step": 11851
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.881801125703564e-05,
+      "loss": 2.163,
+      "step": 11852
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.880863039399625e-05,
+      "loss": 2.3187,
+      "step": 11853
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.879924953095685e-05,
+      "loss": 2.4832,
+      "step": 11854
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.878986866791746e-05,
+      "loss": 2.2568,
+      "step": 11855
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.878048780487805e-05,
+      "loss": 2.1428,
+      "step": 11856
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.877110694183865e-05,
+      "loss": 2.319,
+      "step": 11857
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.876172607879926e-05,
+      "loss": 2.4307,
+      "step": 11858
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.875234521575986e-05,
+      "loss": 2.5802,
+      "step": 11859
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.874296435272045e-05,
+      "loss": 2.2457,
+      "step": 11860
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.873358348968106e-05,
+      "loss": 2.2844,
+      "step": 11861
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.872420262664165e-05,
+      "loss": 2.4817,
+      "step": 11862
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.871482176360225e-05,
+      "loss": 2.2863,
+      "step": 11863
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.870544090056286e-05,
+      "loss": 2.4293,
+      "step": 11864
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.869606003752345e-05,
+      "loss": 2.4747,
+      "step": 11865
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.868667917448405e-05,
+      "loss": 2.3431,
+      "step": 11866
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.867729831144466e-05,
+      "loss": 2.228,
+      "step": 11867
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.866791744840526e-05,
+      "loss": 2.622,
+      "step": 11868
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.865853658536585e-05,
+      "loss": 2.2802,
+      "step": 11869
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.864915572232646e-05,
+      "loss": 2.5105,
+      "step": 11870
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.863977485928706e-05,
+      "loss": 2.278,
+      "step": 11871
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.863039399624767e-05,
+      "loss": 2.6722,
+      "step": 11872
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.862101313320827e-05,
+      "loss": 2.2583,
+      "step": 11873
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.861163227016886e-05,
+      "loss": 2.2424,
+      "step": 11874
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.860225140712945e-05,
+      "loss": 2.2774,
+      "step": 11875
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.859287054409006e-05,
+      "loss": 2.367,
+      "step": 11876
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.858348968105066e-05,
+      "loss": 2.3112,
+      "step": 11877
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.857410881801127e-05,
+      "loss": 2.4386,
+      "step": 11878
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.856472795497186e-05,
+      "loss": 2.231,
+      "step": 11879
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.855534709193246e-05,
+      "loss": 2.3293,
+      "step": 11880
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.854596622889307e-05,
+      "loss": 2.3738,
+      "step": 11881
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.853658536585367e-05,
+      "loss": 2.527,
+      "step": 11882
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.852720450281426e-05,
+      "loss": 2.5256,
+      "step": 11883
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.851782363977487e-05,
+      "loss": 2.7993,
+      "step": 11884
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.850844277673547e-05,
+      "loss": 2.5224,
+      "step": 11885
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.849906191369607e-05,
+      "loss": 2.2559,
+      "step": 11886
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.848968105065667e-05,
+      "loss": 2.4354,
+      "step": 11887
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.848030018761726e-05,
+      "loss": 2.1936,
+      "step": 11888
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.847091932457786e-05,
+      "loss": 2.6436,
+      "step": 11889
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.846153846153847e-05,
+      "loss": 2.4314,
+      "step": 11890
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.845215759849907e-05,
+      "loss": 2.227,
+      "step": 11891
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.844277673545966e-05,
+      "loss": 2.2346,
+      "step": 11892
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.843339587242027e-05,
+      "loss": 2.3014,
+      "step": 11893
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.842401500938087e-05,
+      "loss": 2.3224,
+      "step": 11894
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.841463414634147e-05,
+      "loss": 2.2139,
+      "step": 11895
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.840525328330207e-05,
+      "loss": 2.2527,
+      "step": 11896
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.839587242026267e-05,
+      "loss": 2.2641,
+      "step": 11897
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.838649155722327e-05,
+      "loss": 2.559,
+      "step": 11898
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.837711069418388e-05,
+      "loss": 2.6693,
+      "step": 11899
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.836772983114447e-05,
+      "loss": 2.4638,
+      "step": 11900
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.835834896810506e-05,
+      "loss": 2.5325,
+      "step": 11901
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.834896810506566e-05,
+      "loss": 2.2835,
+      "step": 11902
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.833958724202627e-05,
+      "loss": 2.3979,
+      "step": 11903
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.833020637898687e-05,
+      "loss": 2.3432,
+      "step": 11904
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.832082551594746e-05,
+      "loss": 2.2657,
+      "step": 11905
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.831144465290807e-05,
+      "loss": 2.1237,
+      "step": 11906
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.830206378986867e-05,
+      "loss": 2.3198,
+      "step": 11907
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.829268292682928e-05,
+      "loss": 2.532,
+      "step": 11908
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.828330206378987e-05,
+      "loss": 2.3275,
+      "step": 11909
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.827392120075047e-05,
+      "loss": 2.5127,
+      "step": 11910
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.826454033771108e-05,
+      "loss": 2.3272,
+      "step": 11911
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.825515947467168e-05,
+      "loss": 2.3796,
+      "step": 11912
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.824577861163227e-05,
+      "loss": 2.7251,
+      "step": 11913
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.823639774859286e-05,
+      "loss": 2.3412,
+      "step": 11914
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.822701688555347e-05,
+      "loss": 2.4196,
+      "step": 11915
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.821763602251407e-05,
+      "loss": 2.2033,
+      "step": 11916
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.820825515947468e-05,
+      "loss": 2.2433,
+      "step": 11917
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.819887429643527e-05,
+      "loss": 2.494,
+      "step": 11918
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.818949343339587e-05,
+      "loss": 2.6641,
+      "step": 11919
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.818011257035648e-05,
+      "loss": 2.3586,
+      "step": 11920
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.817073170731708e-05,
+      "loss": 2.2398,
+      "step": 11921
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.816135084427767e-05,
+      "loss": 2.3758,
+      "step": 11922
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.815196998123828e-05,
+      "loss": 2.4348,
+      "step": 11923
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.814258911819888e-05,
+      "loss": 2.3222,
+      "step": 11924
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.813320825515949e-05,
+      "loss": 2.2446,
+      "step": 11925
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.812382739212008e-05,
+      "loss": 2.2521,
+      "step": 11926
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.811444652908067e-05,
+      "loss": 2.4731,
+      "step": 11927
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.810506566604127e-05,
+      "loss": 2.5359,
+      "step": 11928
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.809568480300188e-05,
+      "loss": 2.183,
+      "step": 11929
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.808630393996248e-05,
+      "loss": 2.5861,
+      "step": 11930
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.807692307692307e-05,
+      "loss": 2.4719,
+      "step": 11931
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.806754221388368e-05,
+      "loss": 2.311,
+      "step": 11932
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.805816135084428e-05,
+      "loss": 2.1141,
+      "step": 11933
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.804878048780489e-05,
+      "loss": 2.4828,
+      "step": 11934
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.803939962476548e-05,
+      "loss": 2.3918,
+      "step": 11935
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.803001876172608e-05,
+      "loss": 2.3584,
+      "step": 11936
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.802063789868669e-05,
+      "loss": 2.1391,
+      "step": 11937
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.801125703564729e-05,
+      "loss": 2.4635,
+      "step": 11938
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.800187617260788e-05,
+      "loss": 2.431,
+      "step": 11939
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.799249530956847e-05,
+      "loss": 2.4376,
+      "step": 11940
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.798311444652908e-05,
+      "loss": 2.2801,
+      "step": 11941
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.797373358348968e-05,
+      "loss": 2.275,
+      "step": 11942
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.796435272045029e-05,
+      "loss": 2.3849,
+      "step": 11943
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.795497185741088e-05,
+      "loss": 2.5137,
+      "step": 11944
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.794559099437148e-05,
+      "loss": 2.3892,
+      "step": 11945
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.793621013133209e-05,
+      "loss": 2.2867,
+      "step": 11946
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.792682926829269e-05,
+      "loss": 1.975,
+      "step": 11947
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.791744840525328e-05,
+      "loss": 2.5581,
+      "step": 11948
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.790806754221389e-05,
+      "loss": 2.3095,
+      "step": 11949
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.789868667917449e-05,
+      "loss": 2.3758,
+      "step": 11950
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.78893058161351e-05,
+      "loss": 2.5871,
+      "step": 11951
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.78799249530957e-05,
+      "loss": 2.6304,
+      "step": 11952
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.787054409005629e-05,
+      "loss": 2.4111,
+      "step": 11953
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.786116322701688e-05,
+      "loss": 2.1409,
+      "step": 11954
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.785178236397749e-05,
+      "loss": 2.3861,
+      "step": 11955
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.784240150093809e-05,
+      "loss": 2.5998,
+      "step": 11956
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.78330206378987e-05,
+      "loss": 2.2655,
+      "step": 11957
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.782363977485929e-05,
+      "loss": 2.2547,
+      "step": 11958
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.781425891181989e-05,
+      "loss": 2.2757,
+      "step": 11959
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.78048780487805e-05,
+      "loss": 2.3812,
+      "step": 11960
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.77954971857411e-05,
+      "loss": 2.468,
+      "step": 11961
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.778611632270169e-05,
+      "loss": 2.3846,
+      "step": 11962
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.77767354596623e-05,
+      "loss": 2.5897,
+      "step": 11963
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.77673545966229e-05,
+      "loss": 2.4174,
+      "step": 11964
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.77579737335835e-05,
+      "loss": 2.3729,
+      "step": 11965
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.77485928705441e-05,
+      "loss": 2.6579,
+      "step": 11966
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.773921200750469e-05,
+      "loss": 2.1689,
+      "step": 11967
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.772983114446529e-05,
+      "loss": 2.2857,
+      "step": 11968
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.77204502814259e-05,
+      "loss": 2.4771,
+      "step": 11969
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.77110694183865e-05,
+      "loss": 2.4875,
+      "step": 11970
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.770168855534709e-05,
+      "loss": 2.4716,
+      "step": 11971
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.76923076923077e-05,
+      "loss": 2.3502,
+      "step": 11972
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.76829268292683e-05,
+      "loss": 2.3673,
+      "step": 11973
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.76735459662289e-05,
+      "loss": 2.4411,
+      "step": 11974
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.76641651031895e-05,
+      "loss": 2.3696,
+      "step": 11975
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.76547842401501e-05,
+      "loss": 2.4911,
+      "step": 11976
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.76454033771107e-05,
+      "loss": 2.3658,
+      "step": 11977
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.763602251407131e-05,
+      "loss": 2.2792,
+      "step": 11978
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.76266416510319e-05,
+      "loss": 2.279,
+      "step": 11979
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.761726078799249e-05,
+      "loss": 2.434,
+      "step": 11980
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.76078799249531e-05,
+      "loss": 2.2425,
+      "step": 11981
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.75984990619137e-05,
+      "loss": 2.2886,
+      "step": 11982
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.75891181988743e-05,
+      "loss": 2.5853,
+      "step": 11983
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.75797373358349e-05,
+      "loss": 2.6657,
+      "step": 11984
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.75703564727955e-05,
+      "loss": 2.3073,
+      "step": 11985
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.75609756097561e-05,
+      "loss": 2.6005,
+      "step": 11986
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.755159474671671e-05,
+      "loss": 2.546,
+      "step": 11987
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.75422138836773e-05,
+      "loss": 2.2047,
+      "step": 11988
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.75328330206379e-05,
+      "loss": 2.4895,
+      "step": 11989
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.752345215759851e-05,
+      "loss": 2.3102,
+      "step": 11990
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.751407129455911e-05,
+      "loss": 2.2082,
+      "step": 11991
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.75046904315197e-05,
+      "loss": 2.3822,
+      "step": 11992
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.74953095684803e-05,
+      "loss": 2.3869,
+      "step": 11993
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.74859287054409e-05,
+      "loss": 2.3884,
+      "step": 11994
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.74765478424015e-05,
+      "loss": 2.2573,
+      "step": 11995
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.746716697936211e-05,
+      "loss": 2.232,
+      "step": 11996
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.74577861163227e-05,
+      "loss": 2.6031,
+      "step": 11997
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.74484052532833e-05,
+      "loss": 2.3479,
+      "step": 11998
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.743902439024391e-05,
+      "loss": 2.5476,
+      "step": 11999
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.742964352720451e-05,
+      "loss": 2.4236,
+      "step": 12000
+    },
+    {
+      "epoch": 5.63,
+      "eval_loss": 2.597888231277466,
+      "eval_runtime": 68.0281,
+      "eval_samples_per_second": 27.856,
+      "eval_steps_per_second": 3.484,
+      "step": 12000
+    }
+  ],
+  "max_steps": 21320,
+  "num_train_epochs": 10,
+  "total_flos": 3.4627136117007974e+17,
+  "trial_name": null,
+  "trial_params": null
+}