diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12855 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9981603153745073,
+  "eval_steps": 500,
+  "global_step": 2139,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0769230769230774e-07,
+      "loss": 1.7419,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.153846153846155e-07,
+      "loss": 2.0083,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.230769230769232e-07,
+      "loss": 2.129,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.230769230769231e-06,
+      "loss": 1.8148,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 2.1526,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8461538461538465e-06,
+      "loss": 1.9496,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.153846153846154e-06,
+      "loss": 1.6757,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.461538461538462e-06,
+      "loss": 1.7259,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7692307692307697e-06,
+      "loss": 1.3623,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1.4647,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.384615384615385e-06,
+      "loss": 1.1653,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.692307692307693e-06,
+      "loss": 1.164,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.0807,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.307692307692308e-06,
+      "loss": 1.0365,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.0156,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.923076923076924e-06,
+      "loss": 0.9507,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.230769230769232e-06,
+      "loss": 0.9915,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.538461538461539e-06,
+      "loss": 0.9226,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.846153846153847e-06,
+      "loss": 1.0317,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 0.9406,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.461538461538463e-06,
+      "loss": 1.0097,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.76923076923077e-06,
+      "loss": 0.9447,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.076923076923078e-06,
+      "loss": 0.9286,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.384615384615386e-06,
+      "loss": 0.9395,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 0.8905,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.9643,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.307692307692309e-06,
+      "loss": 0.9357,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.615384615384617e-06,
+      "loss": 0.8734,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.923076923076925e-06,
+      "loss": 0.9449,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 0.9027,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.53846153846154e-06,
+      "loss": 0.9308,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.846153846153848e-06,
+      "loss": 0.8798,
+      "step": 32
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0153846153846154e-05,
+      "loss": 0.8691,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0461538461538463e-05,
+      "loss": 0.88,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 0.862,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1076923076923079e-05,
+      "loss": 0.8756,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1384615384615385e-05,
+      "loss": 0.78,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1692307692307694e-05,
+      "loss": 0.8229,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2e-05,
+      "loss": 0.8634,
+      "step": 39
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.8198,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2615384615384616e-05,
+      "loss": 0.8879,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2923076923076925e-05,
+      "loss": 0.7818,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3230769230769231e-05,
+      "loss": 0.8714,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.353846153846154e-05,
+      "loss": 0.8908,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.8924,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4153846153846156e-05,
+      "loss": 0.9366,
+      "step": 46
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4461538461538462e-05,
+      "loss": 0.9027,
+      "step": 47
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4769230769230772e-05,
+      "loss": 0.8125,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5076923076923078e-05,
+      "loss": 0.8845,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.8871,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5692307692307693e-05,
+      "loss": 0.8477,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.9008,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.630769230769231e-05,
+      "loss": 0.8595,
+      "step": 53
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6615384615384618e-05,
+      "loss": 0.8407,
+      "step": 54
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.8696,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7230769230769234e-05,
+      "loss": 0.8451,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.753846153846154e-05,
+      "loss": 0.8311,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.784615384615385e-05,
+      "loss": 0.9069,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8153846153846155e-05,
+      "loss": 0.8954,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.9484,
+      "step": 60
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.876923076923077e-05,
+      "loss": 0.8708,
+      "step": 61
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.907692307692308e-05,
+      "loss": 0.8839,
+      "step": 62
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9384615384615386e-05,
+      "loss": 0.8429,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9692307692307696e-05,
+      "loss": 0.9102,
+      "step": 64
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.8719,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999988527655958e-05,
+      "loss": 0.9311,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999995411065015e-05,
+      "loss": 0.9279,
+      "step": 67
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999989674906154e-05,
+      "loss": 0.8833,
+      "step": 68
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999816443021753e-05,
+      "loss": 0.8636,
+      "step": 69
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999971319271504e-05,
+      "loss": 0.9525,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999586998378313e-05,
+      "loss": 0.9452,
+      "step": 71
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999437860301113e-05,
+      "loss": 0.893,
+      "step": 72
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999926577882564e-05,
+      "loss": 0.8425,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999070754346724e-05,
+      "loss": 0.8457,
+      "step": 74
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998852787311844e-05,
+      "loss": 0.8386,
+      "step": 75
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998611878221123e-05,
+      "loss": 0.8182,
+      "step": 76
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998348027627314e-05,
+      "loss": 0.82,
+      "step": 77
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998061236135812e-05,
+      "loss": 0.9117,
+      "step": 78
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999775150440466e-05,
+      "loss": 0.8893,
+      "step": 79
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999741883314452e-05,
+      "loss": 0.8525,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99970632231187e-05,
+      "loss": 0.8071,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996684675143132e-05,
+      "loss": 0.8577,
+      "step": 82
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996283190086388e-05,
+      "loss": 0.9127,
+      "step": 83
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999585876886966e-05,
+      "loss": 0.8566,
+      "step": 84
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995411412466765e-05,
+      "loss": 0.8405,
+      "step": 85
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994941121904158e-05,
+      "loss": 0.8303,
+      "step": 86
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994447898260895e-05,
+      "loss": 0.9155,
+      "step": 87
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993931742668673e-05,
+      "loss": 0.8214,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999339265631179e-05,
+      "loss": 0.8482,
+      "step": 89
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999283064042716e-05,
+      "loss": 0.8267,
+      "step": 90
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992245696304313e-05,
+      "loss": 0.8405,
+      "step": 91
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991637825285385e-05,
+      "loss": 0.8223,
+      "step": 92
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999100702876512e-05,
+      "loss": 0.781,
+      "step": 93
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990353308190857e-05,
+      "loss": 0.8812,
+      "step": 94
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989676665062542e-05,
+      "loss": 0.9635,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988977100932705e-05,
+      "loss": 0.8327,
+      "step": 96
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988254617406483e-05,
+      "loss": 0.893,
+      "step": 97
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998750921614158e-05,
+      "loss": 0.8097,
+      "step": 98
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986740898848306e-05,
+      "loss": 0.9337,
+      "step": 99
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985949667289536e-05,
+      "loss": 0.8509,
+      "step": 100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998513552328073e-05,
+      "loss": 0.7561,
+      "step": 101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984298468689912e-05,
+      "loss": 0.8693,
+      "step": 102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998343850543768e-05,
+      "loss": 0.8464,
+      "step": 103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982555635497193e-05,
+      "loss": 0.868,
+      "step": 104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998164986089417e-05,
+      "loss": 0.8955,
+      "step": 105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980721183706878e-05,
+      "loss": 0.7857,
+      "step": 106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979769606066138e-05,
+      "loss": 0.8483,
+      "step": 107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978795130155322e-05,
+      "loss": 0.7894,
+      "step": 108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997779775821033e-05,
+      "loss": 0.8923,
+      "step": 109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976777492519598e-05,
+      "loss": 0.8046,
+      "step": 110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975734335424094e-05,
+      "loss": 0.9427,
+      "step": 111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974668289317316e-05,
+      "loss": 0.8058,
+      "step": 112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997357935664527e-05,
+      "loss": 0.8256,
+      "step": 113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972467539906477e-05,
+      "loss": 0.8414,
+      "step": 114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971332841651968e-05,
+      "loss": 0.7994,
+      "step": 115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970175264485268e-05,
+      "loss": 0.8641,
+      "step": 116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968994811062407e-05,
+      "loss": 0.8948,
+      "step": 117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967791484091897e-05,
+      "loss": 0.8216,
+      "step": 118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966565286334735e-05,
+      "loss": 0.8463,
+      "step": 119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996531622060439e-05,
+      "loss": 0.807,
+      "step": 120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964044289766804e-05,
+      "loss": 0.8442,
+      "step": 121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962749496740387e-05,
+      "loss": 0.9097,
+      "step": 122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961431844496002e-05,
+      "loss": 0.7962,
+      "step": 123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996009133605696e-05,
+      "loss": 0.8445,
+      "step": 124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995872797449901e-05,
+      "loss": 0.8325,
+      "step": 125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957341762950346e-05,
+      "loss": 0.8395,
+      "step": 126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995593270459159e-05,
+      "loss": 0.8676,
+      "step": 127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954500802655783e-05,
+      "loss": 0.7943,
+      "step": 128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953046060428373e-05,
+      "loss": 0.8747,
+      "step": 129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951568481247227e-05,
+      "loss": 0.7962,
+      "step": 130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99500680685026e-05,
+      "loss": 0.7984,
+      "step": 131
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9948544825637143e-05,
+      "loss": 0.8881,
+      "step": 132
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946998756145894e-05,
+      "loss": 0.7963,
+      "step": 133
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945429863576253e-05,
+      "loss": 0.8265,
+      "step": 134
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943838151528002e-05,
+      "loss": 0.823,
+      "step": 135
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942223623653273e-05,
+      "loss": 0.9136,
+      "step": 136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940586283656547e-05,
+      "loss": 0.7747,
+      "step": 137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938926135294653e-05,
+      "loss": 0.8774,
+      "step": 138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937243182376748e-05,
+      "loss": 0.8523,
+      "step": 139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935537428764314e-05,
+      "loss": 0.8473,
+      "step": 140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993380887837115e-05,
+      "loss": 0.9026,
+      "step": 141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932057535163365e-05,
+      "loss": 0.8723,
+      "step": 142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930283403159356e-05,
+      "loss": 0.82,
+      "step": 143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928486486429818e-05,
+      "loss": 0.9205,
+      "step": 144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926666789097714e-05,
+      "loss": 0.8821,
+      "step": 145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924824315338286e-05,
+      "loss": 0.922,
+      "step": 146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992295906937904e-05,
+      "loss": 0.8821,
+      "step": 147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992107105549971e-05,
+      "loss": 0.791,
+      "step": 148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919160278032298e-05,
+      "loss": 0.9157,
+      "step": 149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917226741361014e-05,
+      "loss": 0.8354,
+      "step": 150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915270449922306e-05,
+      "loss": 0.8276,
+      "step": 151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991329140820482e-05,
+      "loss": 0.8629,
+      "step": 152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99112896207494e-05,
+      "loss": 0.8714,
+      "step": 153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909265092149094e-05,
+      "loss": 0.7928,
+      "step": 154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907217827049114e-05,
+      "loss": 0.8836,
+      "step": 155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990514783014685e-05,
+      "loss": 0.8867,
+      "step": 156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990305510619184e-05,
+      "loss": 0.8422,
+      "step": 157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900939659985776e-05,
+      "loss": 0.876,
+      "step": 158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898801496382486e-05,
+      "loss": 0.7882,
+      "step": 159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896640620287914e-05,
+      "loss": 0.8246,
+      "step": 160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989445703666013e-05,
+      "loss": 0.8197,
+      "step": 161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892250750509293e-05,
+      "loss": 0.8673,
+      "step": 162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890021766897662e-05,
+      "loss": 0.8224,
+      "step": 163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988777009093957e-05,
+      "loss": 0.8626,
+      "step": 164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885495727801412e-05,
+      "loss": 0.8827,
+      "step": 165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883198682701645e-05,
+      "loss": 0.8984,
+      "step": 166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880878960910772e-05,
+      "loss": 0.9401,
+      "step": 167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9878536567751322e-05,
+      "loss": 0.8885,
+      "step": 168
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876171508597835e-05,
+      "loss": 0.8312,
+      "step": 169
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873783788876874e-05,
+      "loss": 0.7857,
+      "step": 170
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987137341406698e-05,
+      "loss": 0.7899,
+      "step": 171
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986894038969869e-05,
+      "loss": 0.8523,
+      "step": 172
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98664847213545e-05,
+      "loss": 0.8223,
+      "step": 173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864006414668863e-05,
+      "loss": 0.8544,
+      "step": 174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9861505475328178e-05,
+      "loss": 0.8771,
+      "step": 175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985898190907077e-05,
+      "loss": 0.8343,
+      "step": 176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856435721686888e-05,
+      "loss": 0.94,
+      "step": 177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853866919018676e-05,
+      "loss": 0.8916,
+      "step": 178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985127550696017e-05,
+      "loss": 0.8973,
+      "step": 179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848661491457286e-05,
+      "loss": 0.8848,
+      "step": 180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846024878507802e-05,
+      "loss": 0.8356,
+      "step": 181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984336567416134e-05,
+      "loss": 0.8447,
+      "step": 182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840683884519368e-05,
+      "loss": 0.8448,
+      "step": 183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837979515735168e-05,
+      "loss": 0.8358,
+      "step": 184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983525257401382e-05,
+      "loss": 0.8274,
+      "step": 185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832503065612217e-05,
+      "loss": 0.7966,
+      "step": 186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982973099683902e-05,
+      "loss": 0.8128,
+      "step": 187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982693637405465e-05,
+      "loss": 0.7749,
+      "step": 188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824119203671282e-05,
+      "loss": 0.8935,
+      "step": 189
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821279492152826e-05,
+      "loss": 0.8417,
+      "step": 190
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818417246014915e-05,
+      "loss": 0.7955,
+      "step": 191
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815532471824878e-05,
+      "loss": 0.8696,
+      "step": 192
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812625176201744e-05,
+      "loss": 0.8145,
+      "step": 193
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980969536581621e-05,
+      "loss": 0.945,
+      "step": 194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980674304739064e-05,
+      "loss": 0.8356,
+      "step": 195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980376822769902e-05,
+      "loss": 0.7595,
+      "step": 196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800770913567002e-05,
+      "loss": 0.8917,
+      "step": 197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979775111187182e-05,
+      "loss": 0.939,
+      "step": 198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979470882954231e-05,
+      "loss": 0.9026,
+      "step": 199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97916440735589e-05,
+      "loss": 0.8686,
+      "step": 200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978855685095358e-05,
+      "loss": 0.7722,
+      "step": 201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978544716880988e-05,
+      "loss": 0.8537,
+      "step": 202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782315034262872e-05,
+      "loss": 0.8428,
+      "step": 203
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779160454499136e-05,
+      "loss": 0.8652,
+      "step": 204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977598343675676e-05,
+      "loss": 0.8208,
+      "step": 205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772783988325315e-05,
+      "loss": 0.8661,
+      "step": 206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769562116545835e-05,
+      "loss": 0.8323,
+      "step": 207
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97663178288108e-05,
+      "loss": 0.8254,
+      "step": 208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976305113256413e-05,
+      "loss": 0.8484,
+      "step": 209
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759762035301158e-05,
+      "loss": 0.7884,
+      "step": 210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756450544568615e-05,
+      "loss": 0.8473,
+      "step": 211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975311666796461e-05,
+      "loss": 0.9278,
+      "step": 212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749760413138625e-05,
+      "loss": 0.881,
+      "step": 213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746381787791476e-05,
+      "loss": 0.8443,
+      "step": 214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742980799675314e-05,
+      "loss": 0.7548,
+      "step": 215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739557456593604e-05,
+      "loss": 0.8749,
+      "step": 216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97361117664011e-05,
+      "loss": 0.7519,
+      "step": 217
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9732643737003827e-05,
+      "loss": 0.8402,
+      "step": 218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729153376359072e-05,
+      "loss": 0.8687,
+      "step": 219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725640692475356e-05,
+      "loss": 0.8254,
+      "step": 220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972210569341243e-05,
+      "loss": 0.7293,
+      "step": 221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971854838728123e-05,
+      "loss": 0.7944,
+      "step": 222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971496878224389e-05,
+      "loss": 0.8271,
+      "step": 223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97113668865137e-05,
+      "loss": 0.7714,
+      "step": 224
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9707742708355096e-05,
+      "loss": 0.7816,
+      "step": 225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970409625608365e-05,
+      "loss": 0.8693,
+      "step": 226
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970042753806602e-05,
+      "loss": 0.7702,
+      "step": 227
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696736562719975e-05,
+      "loss": 0.8853,
+      "step": 228
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693023338514338e-05,
+      "loss": 0.8836,
+      "step": 229
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689287873968984e-05,
+      "loss": 0.7986,
+      "step": 230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685530177654825e-05,
+      "loss": 0.846,
+      "step": 231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968175025819378e-05,
+      "loss": 0.7964,
+      "step": 232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967794812425875e-05,
+      "loss": 0.8677,
+      "step": 233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674123784573617e-05,
+      "loss": 0.769,
+      "step": 234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670277247913205e-05,
+      "loss": 0.7617,
+      "step": 235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666408523103276e-05,
+      "loss": 0.8493,
+      "step": 236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662517619020497e-05,
+      "loss": 0.8552,
+      "step": 237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658604544592426e-05,
+      "loss": 0.8489,
+      "step": 238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965466930879749e-05,
+      "loss": 0.8104,
+      "step": 239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650711920664967e-05,
+      "loss": 0.8376,
+      "step": 240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964673238927496e-05,
+      "loss": 0.7873,
+      "step": 241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642730723758377e-05,
+      "loss": 0.8932,
+      "step": 242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638706933296916e-05,
+      "loss": 0.8605,
+      "step": 243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634661027123042e-05,
+      "loss": 0.8403,
+      "step": 244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630593014519953e-05,
+      "loss": 0.9091,
+      "step": 245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626502904821582e-05,
+      "loss": 0.8359,
+      "step": 246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962239070741256e-05,
+      "loss": 0.777,
+      "step": 247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961825643172819e-05,
+      "loss": 0.8745,
+      "step": 248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961410008725445e-05,
+      "loss": 0.7958,
+      "step": 249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609921683527925e-05,
+      "loss": 0.8818,
+      "step": 250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960572123013585e-05,
+      "loss": 0.8722,
+      "step": 251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960149873671602e-05,
+      "loss": 0.8133,
+      "step": 252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597254212956823e-05,
+      "loss": 0.7926,
+      "step": 253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959298766859718e-05,
+      "loss": 0.8589,
+      "step": 254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588699113426553e-05,
+      "loss": 0.833,
+      "step": 255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584388557284893e-05,
+      "loss": 0.8142,
+      "step": 256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580056010062635e-05,
+      "loss": 0.87,
+      "step": 257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575701481700674e-05,
+      "loss": 0.7855,
+      "step": 258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957132498219034e-05,
+      "loss": 0.8713,
+      "step": 259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566926521573377e-05,
+      "loss": 0.7894,
+      "step": 260
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562506109941913e-05,
+      "loss": 0.843,
+      "step": 261
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558063757438444e-05,
+      "loss": 0.8045,
+      "step": 262
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553599474255813e-05,
+      "loss": 0.8571,
+      "step": 263
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549113270637175e-05,
+      "loss": 0.799,
+      "step": 264
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544605156875987e-05,
+      "loss": 0.9069,
+      "step": 265
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540075143315973e-05,
+      "loss": 0.7588,
+      "step": 266
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535523240351107e-05,
+      "loss": 0.8164,
+      "step": 267
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9530949458425592e-05,
+      "loss": 0.8254,
+      "step": 268
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526353808033827e-05,
+      "loss": 0.81,
+      "step": 269
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521736299720387e-05,
+      "loss": 0.7928,
+      "step": 270
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517096944079998e-05,
+      "loss": 0.8041,
+      "step": 271
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512435751757525e-05,
+      "loss": 0.825,
+      "step": 272
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950775273344792e-05,
+      "loss": 0.8008,
+      "step": 273
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503047899896227e-05,
+      "loss": 0.8583,
+      "step": 274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949832126189754e-05,
+      "loss": 0.8939,
+      "step": 275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949357283029698e-05,
+      "loss": 0.8607,
+      "step": 276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488802615989675e-05,
+      "loss": 0.8406,
+      "step": 277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484010629920736e-05,
+      "loss": 0.8307,
+      "step": 278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9479196883085225e-05,
+      "loss": 0.8718,
+      "step": 279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947436138652813e-05,
+      "loss": 0.7533,
+      "step": 280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9469504151344352e-05,
+      "loss": 0.7761,
+      "step": 281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9464625188678664e-05,
+      "loss": 0.817,
+      "step": 282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459724509725696e-05,
+      "loss": 0.7862,
+      "step": 283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454802125729896e-05,
+      "loss": 0.9245,
+      "step": 284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9449858047985524e-05,
+      "loss": 0.8367,
+      "step": 285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444892287836614e-05,
+      "loss": 0.8455,
+      "step": 286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439904856676947e-05,
+      "loss": 0.7466,
+      "step": 287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943489576595003e-05,
+      "loss": 0.8694,
+      "step": 288
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9429865027149062e-05,
+      "loss": 0.8143,
+      "step": 289
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424812651816915e-05,
+      "loss": 0.7842,
+      "step": 290
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9419738651546112e-05,
+      "loss": 0.8762,
+      "step": 291
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414643037978785e-05,
+      "loss": 0.8211,
+      "step": 292
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409525822806663e-05,
+      "loss": 0.7253,
+      "step": 293
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940438701777103e-05,
+      "loss": 0.819,
+      "step": 294
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399226634662722e-05,
+      "loss": 0.8326,
+      "step": 295
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394044685322074e-05,
+      "loss": 0.7938,
+      "step": 296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388841181638906e-05,
+      "loss": 0.7761,
+      "step": 297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383616135552494e-05,
+      "loss": 0.7702,
+      "step": 298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378369559051546e-05,
+      "loss": 0.8285,
+      "step": 299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373101464174166e-05,
+      "loss": 0.8518,
+      "step": 300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367811863007837e-05,
+      "loss": 0.782,
+      "step": 301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936250076768938e-05,
+      "loss": 0.8361,
+      "step": 302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357168190404937e-05,
+      "loss": 0.7931,
+      "step": 303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351814143389944e-05,
+      "loss": 0.8321,
+      "step": 304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346438638929094e-05,
+      "loss": 0.83,
+      "step": 305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934104168935631e-05,
+      "loss": 0.8996,
+      "step": 306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933562330705473e-05,
+      "loss": 0.7373,
+      "step": 307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330183504456662e-05,
+      "loss": 0.8531,
+      "step": 308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932472229404356e-05,
+      "loss": 0.7804,
+      "step": 309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319239688346003e-05,
+      "loss": 0.8637,
+      "step": 310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931373569994366e-05,
+      "loss": 0.8247,
+      "step": 311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308210341465263e-05,
+      "loss": 0.8048,
+      "step": 312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302663625588566e-05,
+      "loss": 0.7787,
+      "step": 313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297095565040343e-05,
+      "loss": 0.8298,
+      "step": 314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9291506172596332e-05,
+      "loss": 0.8295,
+      "step": 315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285895461081223e-05,
+      "loss": 0.8077,
+      "step": 316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9280263443368613e-05,
+      "loss": 0.7756,
+      "step": 317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9274610132381e-05,
+      "loss": 0.8023,
+      "step": 318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926893554108972e-05,
+      "loss": 0.7712,
+      "step": 319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263239682514953e-05,
+      "loss": 0.8152,
+      "step": 320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257522569725663e-05,
+      "loss": 0.8298,
+      "step": 321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251784215839588e-05,
+      "loss": 0.7951,
+      "step": 322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924602463402321e-05,
+      "loss": 0.8145,
+      "step": 323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240243837491697e-05,
+      "loss": 0.8405,
+      "step": 324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9234441839508915e-05,
+      "loss": 0.8133,
+      "step": 325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228618653387365e-05,
+      "loss": 0.7715,
+      "step": 326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222774292488168e-05,
+      "loss": 0.7809,
+      "step": 327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216908770221022e-05,
+      "loss": 0.8215,
+      "step": 328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211022100044193e-05,
+      "loss": 0.8787,
+      "step": 329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205114295464457e-05,
+      "loss": 0.8215,
+      "step": 330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9199185370037086e-05,
+      "loss": 0.7846,
+      "step": 331
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919323533736582e-05,
+      "loss": 0.8143,
+      "step": 332
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918726421110282e-05,
+      "loss": 0.8659,
+      "step": 333
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181272004948647e-05,
+      "loss": 0.7362,
+      "step": 334
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917525873265223e-05,
+      "loss": 0.8696,
+      "step": 335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169224408010843e-05,
+      "loss": 0.7399,
+      "step": 336
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916316904487005e-05,
+      "loss": 0.8419,
+      "step": 337
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157092657123693e-05,
+      "loss": 0.7773,
+      "step": 338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9150995258713854e-05,
+      "loss": 0.7355,
+      "step": 339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144876863630823e-05,
+      "loss": 0.8285,
+      "step": 340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913873748591307e-05,
+      "loss": 0.7926,
+      "step": 341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132577139647202e-05,
+      "loss": 0.771,
+      "step": 342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912639583896794e-05,
+      "loss": 0.7702,
+      "step": 343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120193598058093e-05,
+      "loss": 0.7311,
+      "step": 344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9113970431148506e-05,
+      "loss": 0.852,
+      "step": 345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107726352518034e-05,
+      "loss": 0.915,
+      "step": 346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910146137649353e-05,
+      "loss": 0.6973,
+      "step": 347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095175517449783e-05,
+      "loss": 0.7793,
+      "step": 348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088868789809504e-05,
+      "loss": 0.8411,
+      "step": 349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082541208043273e-05,
+      "loss": 0.7914,
+      "step": 350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076192786669537e-05,
+      "loss": 0.7776,
+      "step": 351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069823540254552e-05,
+      "loss": 0.7067,
+      "step": 352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063433483412347e-05,
+      "loss": 0.8581,
+      "step": 353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057022630804715e-05,
+      "loss": 0.7951,
+      "step": 354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050590997141158e-05,
+      "loss": 0.8299,
+      "step": 355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044138597178855e-05,
+      "loss": 0.8594,
+      "step": 356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9037665445722638e-05,
+      "loss": 0.7745,
+      "step": 357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903117155762495e-05,
+      "loss": 0.7666,
+      "step": 358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902465694778582e-05,
+      "loss": 0.7972,
+      "step": 359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901812163115281e-05,
+      "loss": 0.8283,
+      "step": 360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011565622721e-05,
+      "loss": 0.7955,
+      "step": 361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900498893753295e-05,
+      "loss": 0.7361,
+      "step": 362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998391590678666e-05,
+      "loss": 0.7714,
+      "step": 363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899177359729554e-05,
+      "loss": 0.8323,
+      "step": 364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985134972568365e-05,
+      "loss": 0.8119,
+      "step": 365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897847573172925e-05,
+      "loss": 0.8417,
+      "step": 366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971795890057616e-05,
+      "loss": 0.8199,
+      "step": 367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965095462880153e-05,
+      "loss": 0.8068,
+      "step": 368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8958374465570782e-05,
+      "loss": 0.8884,
+      "step": 369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951632913550625e-05,
+      "loss": 0.8002,
+      "step": 370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944870822287957e-05,
+      "loss": 0.846,
+      "step": 371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893808820729819e-05,
+      "loss": 0.8193,
+      "step": 372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931285084143818e-05,
+      "loss": 0.8478,
+      "step": 373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89244614684344e-05,
+      "loss": 0.8452,
+      "step": 374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917617375826506e-05,
+      "loss": 0.7982,
+      "step": 375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910752822023694e-05,
+      "loss": 0.8358,
+      "step": 376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903867822776468e-05,
+      "loss": 0.7925,
+      "step": 377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896962393882248e-05,
+      "loss": 0.8001,
+      "step": 378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890036551185317e-05,
+      "loss": 0.8402,
+      "step": 379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883090310576812e-05,
+      "loss": 0.7916,
+      "step": 380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8876123687994664e-05,
+      "loss": 0.85,
+      "step": 381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886913669942357e-05,
+      "loss": 0.7984,
+      "step": 382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8862129360894958e-05,
+      "loss": 0.7888,
+      "step": 383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855101688486948e-05,
+      "loss": 0.7698,
+      "step": 384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884805369832431e-05,
+      "loss": 0.7797,
+      "step": 385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840985406578448e-05,
+      "loss": 0.8413,
+      "step": 386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883389682946733e-05,
+      "loss": 0.8943,
+      "step": 387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8826787983255474e-05,
+      "loss": 0.8131,
+      "step": 388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8819658884253913e-05,
+      "loss": 0.7625,
+      "step": 389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812509548820132e-05,
+      "loss": 0.7958,
+      "step": 390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8805339993358066e-05,
+      "loss": 0.7524,
+      "step": 391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798150234318037e-05,
+      "loss": 0.8092,
+      "step": 392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8790940288196715e-05,
+      "loss": 0.7926,
+      "step": 393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783710171537103e-05,
+      "loss": 0.8772,
+      "step": 394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776459900928478e-05,
+      "loss": 0.7391,
+      "step": 395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876918949300636e-05,
+      "loss": 0.7725,
+      "step": 396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8761898964452465e-05,
+      "loss": 0.7899,
+      "step": 397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875458833199469e-05,
+      "loss": 0.7751,
+      "step": 398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874725761240706e-05,
+      "loss": 0.8413,
+      "step": 399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8739906822509673e-05,
+      "loss": 0.8176,
+      "step": 400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8732535979168687e-05,
+      "loss": 0.813,
+      "step": 401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725145099296278e-05,
+      "loss": 0.7481,
+      "step": 402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8717734199850585e-05,
+      "loss": 0.7574,
+      "step": 403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8710303297835684e-05,
+      "loss": 0.8245,
+      "step": 404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8702852410301556e-05,
+      "loss": 0.7973,
+      "step": 405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695381554344018e-05,
+      "loss": 0.8267,
+      "step": 406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687890747104727e-05,
+      "loss": 0.7855,
+      "step": 407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86803800057711e-05,
+      "loss": 0.7758,
+      "step": 408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672849347576303e-05,
+      "loss": 0.8183,
+      "step": 409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665298789799192e-05,
+      "loss": 0.8091,
+      "step": 410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865772834976429e-05,
+      "loss": 0.7902,
+      "step": 411
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650138044841735e-05,
+      "loss": 0.8244,
+      "step": 412
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642527892447242e-05,
+      "loss": 0.7788,
+      "step": 413
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863489791004207e-05,
+      "loss": 0.7548,
+      "step": 414
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8627248115132982e-05,
+      "loss": 0.7651,
+      "step": 415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861957852527218e-05,
+      "loss": 0.7711,
+      "step": 416
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861188915805731e-05,
+      "loss": 0.8192,
+      "step": 417
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860418003113138e-05,
+      "loss": 0.8123,
+      "step": 418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8596451162182743e-05,
+      "loss": 0.7955,
+      "step": 419
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858870256894505e-05,
+      "loss": 0.8166,
+      "step": 420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580934269197204e-05,
+      "loss": 0.7929,
+      "step": 421
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8573146280763327e-05,
+      "loss": 0.8079,
+      "step": 422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565338621512713e-05,
+      "loss": 0.7527,
+      "step": 423
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557511309359793e-05,
+      "loss": 0.8607,
+      "step": 424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549664362264094e-05,
+      "loss": 0.7393,
+      "step": 425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8541797798230193e-05,
+      "loss": 0.7715,
+      "step": 426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533911635307673e-05,
+      "loss": 0.7752,
+      "step": 427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852600589159108e-05,
+      "loss": 0.7814,
+      "step": 428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8518080585219912e-05,
+      "loss": 0.7494,
+      "step": 429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851013573437853e-05,
+      "loss": 0.7996,
+      "step": 430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8502171357296144e-05,
+      "loss": 0.8287,
+      "step": 431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8494187472246768e-05,
+      "loss": 0.7859,
+      "step": 432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8486184097549185e-05,
+      "loss": 0.8064,
+      "step": 433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478161251566887e-05,
+      "loss": 0.7629,
+      "step": 434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470118952708036e-05,
+      "loss": 0.8112,
+      "step": 435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8462057219425443e-05,
+      "loss": 0.849,
+      "step": 436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84539760702165e-05,
+      "loss": 0.7912,
+      "step": 437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8445875523623155e-05,
+      "loss": 0.7941,
+      "step": 438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8437755598231857e-05,
+      "loss": 0.777,
+      "step": 439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429616312673523e-05,
+      "loss": 0.7393,
+      "step": 440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842145768562349e-05,
+      "loss": 0.7934,
+      "step": 441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841327973580147e-05,
+      "loss": 0.7519,
+      "step": 442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8405082481971517e-05,
+      "loss": 0.8013,
+      "step": 443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8396865942941974e-05,
+      "loss": 0.7647,
+      "step": 444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838863013756543e-05,
+      "loss": 0.8566,
+      "step": 445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.838037508473869e-05,
+      "loss": 0.7655,
+      "step": 446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837210080340271e-05,
+      "loss": 0.7628,
+      "step": 447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8363807312542572e-05,
+      "loss": 0.8356,
+      "step": 448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355494631187435e-05,
+      "loss": 0.7765,
+      "step": 449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834716277841048e-05,
+      "loss": 0.8112,
+      "step": 450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8338811773328892e-05,
+      "loss": 0.8132,
+      "step": 451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8330441635103785e-05,
+      "loss": 0.7597,
+      "step": 452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832205238294018e-05,
+      "loss": 0.8402,
+      "step": 453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313644036086962e-05,
+      "loss": 0.7917,
+      "step": 454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830521661383681e-05,
+      "loss": 0.8107,
+      "step": 455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829677013552619e-05,
+      "loss": 0.8728,
+      "step": 456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828830462053528e-05,
+      "loss": 0.7875,
+      "step": 457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279820088287937e-05,
+      "loss": 0.8135,
+      "step": 458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271316558251663e-05,
+      "loss": 0.8987,
+      "step": 459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262794049937537e-05,
+      "loss": 0.7839,
+      "step": 460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254252582900188e-05,
+      "loss": 0.8235,
+      "step": 461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824569217673775e-05,
+      "loss": 0.7602,
+      "step": 462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8237112851091802e-05,
+      "loss": 0.7948,
+      "step": 463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8228514625647346e-05,
+      "loss": 0.8499,
+      "step": 464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8219897520132734e-05,
+      "loss": 0.756,
+      "step": 465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821126155431965e-05,
+      "loss": 0.8174,
+      "step": 466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202606748023057e-05,
+      "loss": 0.802,
+      "step": 467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819393312110112e-05,
+      "loss": 0.8466,
+      "step": 468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818524069345522e-05,
+      "loss": 0.7922,
+      "step": 469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8176529485029853e-05,
+      "loss": 0.8247,
+      "step": 470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816779951581262e-05,
+      "loss": 0.7254,
+      "step": 471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815905080583416e-05,
+      "loss": 0.729,
+      "step": 472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8150283375168112e-05,
+      "loss": 0.7773,
+      "step": 473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141497243931083e-05,
+      "loss": 0.7238,
+      "step": 474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813269243228257e-05,
+      "loss": 0.7468,
+      "step": 475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8123868960424936e-05,
+      "loss": 0.7587,
+      "step": 476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811502684860337e-05,
+      "loss": 0.8257,
+      "step": 477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8106166117105815e-05,
+      "loss": 0.7179,
+      "step": 478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8097286786262943e-05,
+      "loss": 0.8568,
+      "step": 479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8088388876448103e-05,
+      "loss": 0.8039,
+      "step": 480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8079472408077278e-05,
+      "loss": 0.821,
+      "step": 481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8070537401609014e-05,
+      "loss": 0.7648,
+      "step": 482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061583877544416e-05,
+      "loss": 0.7408,
+      "step": 483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805261185642706e-05,
+      "loss": 0.8382,
+      "step": 484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8043621358842966e-05,
+      "loss": 0.751,
+      "step": 485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8034612405420558e-05,
+      "loss": 0.8137,
+      "step": 486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025585016830597e-05,
+      "loss": 0.7585,
+      "step": 487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8016539213786138e-05,
+      "loss": 0.7922,
+      "step": 488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8007475017042498e-05,
+      "loss": 0.8782,
+      "step": 489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7998392447397197e-05,
+      "loss": 0.8258,
+      "step": 490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7989291525689904e-05,
+      "loss": 0.7706,
+      "step": 491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7980172272802398e-05,
+      "loss": 0.7817,
+      "step": 492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7971034709658522e-05,
+      "loss": 0.7161,
+      "step": 493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7961878857224134e-05,
+      "loss": 0.8442,
+      "step": 494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952704736507046e-05,
+      "loss": 0.8244,
+      "step": 495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7943512368556996e-05,
+      "loss": 0.9327,
+      "step": 496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7934301774465583e-05,
+      "loss": 0.7523,
+      "step": 497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792507297536623e-05,
+      "loss": 0.8455,
+      "step": 498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7915825992434125e-05,
+      "loss": 0.7907,
+      "step": 499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7906560846886183e-05,
+      "loss": 0.7975,
+      "step": 500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7897277559980996e-05,
+      "loss": 0.7958,
+      "step": 501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7887976153018772e-05,
+      "loss": 0.7565,
+      "step": 502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7878656647341303e-05,
+      "loss": 0.7721,
+      "step": 503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7869319064331895e-05,
+      "loss": 0.7166,
+      "step": 504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785996342541535e-05,
+      "loss": 0.6955,
+      "step": 505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850589752057885e-05,
+      "loss": 0.734,
+      "step": 506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7841198065767107e-05,
+      "loss": 0.8254,
+      "step": 507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783178838809194e-05,
+      "loss": 0.7955,
+      "step": 508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7822360740622598e-05,
+      "loss": 0.8102,
+      "step": 509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7812915144990527e-05,
+      "loss": 0.7369,
+      "step": 510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780345162286835e-05,
+      "loss": 0.815,
+      "step": 511
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7793970195969825e-05,
+      "loss": 0.7957,
+      "step": 512
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7784470886049785e-05,
+      "loss": 0.7628,
+      "step": 513
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7774953714904105e-05,
+      "loss": 0.7844,
+      "step": 514
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7765418704369634e-05,
+      "loss": 0.8324,
+      "step": 515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7755865876324165e-05,
+      "loss": 0.7448,
+      "step": 516
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774629525268635e-05,
+      "loss": 0.762,
+      "step": 517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7736706855415696e-05,
+      "loss": 0.7627,
+      "step": 518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727100706512477e-05,
+      "loss": 0.7612,
+      "step": 519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7717476828017707e-05,
+      "loss": 0.738,
+      "step": 520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770783524201307e-05,
+      "loss": 0.7498,
+      "step": 521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769817597062089e-05,
+      "loss": 0.8109,
+      "step": 522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768849903600406e-05,
+      "loss": 0.8018,
+      "step": 523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678804460366e-05,
+      "loss": 0.8296,
+      "step": 524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7669092265950624e-05,
+      "loss": 0.7555,
+      "step": 525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659362475042245e-05,
+      "loss": 0.7634,
+      "step": 526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649615109965574e-05,
+      "loss": 0.8262,
+      "step": 527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639850193085636e-05,
+      "loss": 0.8078,
+      "step": 528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630067746807722e-05,
+      "loss": 0.7561,
+      "step": 529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7620267793577356e-05,
+      "loss": 0.7736,
+      "step": 530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7610450355880222e-05,
+      "loss": 0.7919,
+      "step": 531
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600615456242123e-05,
+      "loss": 0.786,
+      "step": 532
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7590763117228936e-05,
+      "loss": 0.7671,
+      "step": 533
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7580893361446537e-05,
+      "loss": 0.8378,
+      "step": 534
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757100621154078e-05,
+      "loss": 0.7684,
+      "step": 535
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756110169019742e-05,
+      "loss": 0.7111,
+      "step": 536
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755117982014207e-05,
+      "loss": 0.7841,
+      "step": 537
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754124062414015e-05,
+      "loss": 0.7517,
+      "step": 538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.753128412499684e-05,
+      "loss": 0.7963,
+      "step": 539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521310345557013e-05,
+      "loss": 0.7421,
+      "step": 540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7511319308705198e-05,
+      "loss": 0.7443,
+      "step": 541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7501311037365518e-05,
+      "loss": 0.773,
+      "step": 542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7491285554501637e-05,
+      "loss": 0.767,
+      "step": 543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481242883116713e-05,
+      "loss": 0.8284,
+      "step": 544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747118304625334e-05,
+      "loss": 0.6967,
+      "step": 545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7461106066993504e-05,
+      "loss": 0.7908,
+      "step": 546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7451011968458518e-05,
+      "loss": 0.8445,
+      "step": 547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7440900773808973e-05,
+      "loss": 0.7614,
+      "step": 548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7430772506244697e-05,
+      "loss": 0.7597,
+      "step": 549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7420627189004676e-05,
+      "loss": 0.8018,
+      "step": 550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7410464845367034e-05,
+      "loss": 0.7231,
+      "step": 551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740028549864894e-05,
+      "loss": 0.7758,
+      "step": 552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7390089172206594e-05,
+      "loss": 0.8212,
+      "step": 553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7379875889435147e-05,
+      "loss": 0.7614,
+      "step": 554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7369645673768662e-05,
+      "loss": 0.7915,
+      "step": 555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735939854868004e-05,
+      "loss": 0.7492,
+      "step": 556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7349134537681003e-05,
+      "loss": 0.7105,
+      "step": 557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7338853664321993e-05,
+      "loss": 0.8146,
+      "step": 558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7328555952192158e-05,
+      "loss": 0.8086,
+      "step": 559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7318241424919277e-05,
+      "loss": 0.7826,
+      "step": 560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7307910106169713e-05,
+      "loss": 0.8096,
+      "step": 561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7297562019648353e-05,
+      "loss": 0.7819,
+      "step": 562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728719718909856e-05,
+      "loss": 0.8163,
+      "step": 563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7276815638302106e-05,
+      "loss": 0.7502,
+      "step": 564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7266417391079148e-05,
+      "loss": 0.7627,
+      "step": 565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725600247128813e-05,
+      "loss": 0.7085,
+      "step": 566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7245570902825765e-05,
+      "loss": 0.7619,
+      "step": 567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7235122709626965e-05,
+      "loss": 0.7129,
+      "step": 568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7224657915664778e-05,
+      "loss": 0.8677,
+      "step": 569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721417654495035e-05,
+      "loss": 0.8156,
+      "step": 570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7203678621532857e-05,
+      "loss": 0.7962,
+      "step": 571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7193164169499453e-05,
+      "loss": 0.8116,
+      "step": 572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7182633212975234e-05,
+      "loss": 0.8222,
+      "step": 573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7172085776123137e-05,
+      "loss": 0.7474,
+      "step": 574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7161521883143936e-05,
+      "loss": 0.8204,
+      "step": 575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7150941558276147e-05,
+      "loss": 0.774,
+      "step": 576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7140344825796e-05,
+      "loss": 0.7835,
+      "step": 577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7129731710017366e-05,
+      "loss": 0.8137,
+      "step": 578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7119102235291708e-05,
+      "loss": 0.7512,
+      "step": 579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7108456426008025e-05,
+      "loss": 0.8211,
+      "step": 580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7097794306592792e-05,
+      "loss": 0.8488,
+      "step": 581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708711590150991e-05,
+      "loss": 0.6901,
+      "step": 582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707642123526065e-05,
+      "loss": 0.8333,
+      "step": 583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065710332383588e-05,
+      "loss": 0.8182,
+      "step": 584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7054983217454552e-05,
+      "loss": 0.7727,
+      "step": 585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704423991508658e-05,
+      "loss": 0.8296,
+      "step": 586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703348044992984e-05,
+      "loss": 0.7815,
+      "step": 587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022704846671593e-05,
+      "loss": 0.7486,
+      "step": 588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7011913130036116e-05,
+      "loss": 0.7989,
+      "step": 589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7001105324784676e-05,
+      "loss": 0.7843,
+      "step": 590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699028145571544e-05,
+      "loss": 0.7209,
+      "step": 591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979441547663434e-05,
+      "loss": 0.7329,
+      "step": 592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69685856255005e-05,
+      "loss": 0.7839,
+      "step": 593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6957713714135207e-05,
+      "loss": 0.7954,
+      "step": 594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946825838512812e-05,
+      "loss": 0.8506,
+      "step": 595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6935922023615213e-05,
+      "loss": 0.7891,
+      "step": 596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692500229446087e-05,
+      "loss": 0.7426,
+      "step": 597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691406667610476e-05,
+      "loss": 0.8326,
+      "step": 598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690311519363832e-05,
+      "loss": 0.7745,
+      "step": 599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6892147872189384e-05,
+      "loss": 0.7764,
+      "step": 600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688116473692213e-05,
+      "loss": 0.8064,
+      "step": 601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687016581303702e-05,
+      "loss": 0.8419,
+      "step": 602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6859151125770737e-05,
+      "loss": 0.7467,
+      "step": 603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6848120700396142e-05,
+      "loss": 0.7953,
+      "step": 604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6837074562222202e-05,
+      "loss": 0.7834,
+      "step": 605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6826012736593936e-05,
+      "loss": 0.8295,
+      "step": 606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6814935248892354e-05,
+      "loss": 0.7351,
+      "step": 607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680384212453441e-05,
+      "loss": 0.8025,
+      "step": 608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679273338897293e-05,
+      "loss": 0.7184,
+      "step": 609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6781609067696567e-05,
+      "loss": 0.8517,
+      "step": 610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6770469186229724e-05,
+      "loss": 0.8025,
+      "step": 611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675931377013251e-05,
+      "loss": 0.7691,
+      "step": 612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674814284500068e-05,
+      "loss": 0.7395,
+      "step": 613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6736956436465573e-05,
+      "loss": 0.8051,
+      "step": 614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6725754570194058e-05,
+      "loss": 0.7497,
+      "step": 615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6714537271888467e-05,
+      "loss": 0.698,
+      "step": 616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670330456728654e-05,
+      "loss": 0.7703,
+      "step": 617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6692056482161368e-05,
+      "loss": 0.7402,
+      "step": 618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6680793042321326e-05,
+      "loss": 0.7869,
+      "step": 619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6669514273610035e-05,
+      "loss": 0.7316,
+      "step": 620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665822020190627e-05,
+      "loss": 0.783,
+      "step": 621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6646910853123933e-05,
+      "loss": 0.7575,
+      "step": 622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6635586253211964e-05,
+      "loss": 0.7212,
+      "step": 623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624246428154313e-05,
+      "loss": 0.7864,
+      "step": 624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6612891403969847e-05,
+      "loss": 0.7228,
+      "step": 625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660152120671232e-05,
+      "loss": 0.7722,
+      "step": 626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659013586247029e-05,
+      "loss": 0.7674,
+      "step": 627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6578735397367083e-05,
+      "loss": 0.7431,
+      "step": 628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65673198375607e-05,
+      "loss": 0.7469,
+      "step": 629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6555889209243798e-05,
+      "loss": 0.7634,
+      "step": 630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544443538643587e-05,
+      "loss": 0.7814,
+      "step": 631
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6532982852021805e-05,
+      "loss": 0.7345,
+      "step": 632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6521507175674644e-05,
+      "loss": 0.8022,
+      "step": 633
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510016535932677e-05,
+      "loss": 0.7448,
+      "step": 634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6498510959160826e-05,
+      "loss": 0.8109,
+      "step": 635
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6486990471758275e-05,
+      "loss": 0.8025,
+      "step": 636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475455100158423e-05,
+      "loss": 0.7895,
+      "step": 637
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646390487082882e-05,
+      "loss": 0.8242,
+      "step": 638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645233981027111e-05,
+      "loss": 0.7983,
+      "step": 639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644075994502096e-05,
+      "loss": 0.774,
+      "step": 640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6429165301648012e-05,
+      "loss": 0.8349,
+      "step": 641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641755590675581e-05,
+      "loss": 0.6944,
+      "step": 642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6405931786981753e-05,
+      "loss": 0.7975,
+      "step": 643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639429296899702e-05,
+      "loss": 0.6836,
+      "step": 644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638263947950651e-05,
+      "loss": 0.7476,
+      "step": 645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6370971345248803e-05,
+      "loss": 0.7795,
+      "step": 646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6359288592996055e-05,
+      "loss": 0.8269,
+      "step": 647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634759124955399e-05,
+      "loss": 0.7509,
+      "step": 648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6335879341761792e-05,
+      "loss": 0.8083,
+      "step": 649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6324152896492066e-05,
+      "loss": 0.7518,
+      "step": 650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6312411940650776e-05,
+      "loss": 0.7333,
+      "step": 651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6300656501177178e-05,
+      "loss": 0.7378,
+      "step": 652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6288886605043767e-05,
+      "loss": 0.7946,
+      "step": 653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6277102279256194e-05,
+      "loss": 0.7108,
+      "step": 654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6265303550853234e-05,
+      "loss": 0.7609,
+      "step": 655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6253490446906698e-05,
+      "loss": 0.8049,
+      "step": 656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6241662994521387e-05,
+      "loss": 0.7801,
+      "step": 657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6229821220835017e-05,
+      "loss": 0.7104,
+      "step": 658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6217965153018173e-05,
+      "loss": 0.77,
+      "step": 659
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6206094818274228e-05,
+      "loss": 0.7243,
+      "step": 660
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61942102438393e-05,
+      "loss": 0.7438,
+      "step": 661
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6182311456982172e-05,
+      "loss": 0.8032,
+      "step": 662
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6170398485004243e-05,
+      "loss": 0.7528,
+      "step": 663
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6158471355239447e-05,
+      "loss": 0.7222,
+      "step": 664
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614653009505422e-05,
+      "loss": 0.7514,
+      "step": 665
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6134574731847402e-05,
+      "loss": 0.7512,
+      "step": 666
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6122605293050212e-05,
+      "loss": 0.7259,
+      "step": 667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6110621806126147e-05,
+      "loss": 0.7051,
+      "step": 668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6098624298570943e-05,
+      "loss": 0.8294,
+      "step": 669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608661279791251e-05,
+      "loss": 0.7783,
+      "step": 670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607458733171086e-05,
+      "loss": 0.8486,
+      "step": 671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606254792755805e-05,
+      "loss": 0.7532,
+      "step": 672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605049461307812e-05,
+      "loss": 0.7755,
+      "step": 673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6038427415927024e-05,
+      "loss": 0.7533,
+      "step": 674
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6026346363792565e-05,
+      "loss": 0.7678,
+      "step": 675
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6014251484394344e-05,
+      "loss": 0.8467,
+      "step": 676
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6002142805483686e-05,
+      "loss": 0.7882,
+      "step": 677
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5990020354843568e-05,
+      "loss": 0.8004,
+      "step": 678
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597788416028859e-05,
+      "loss": 0.823,
+      "step": 679
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965734249664857e-05,
+      "loss": 0.7862,
+      "step": 680
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5953570650849966e-05,
+      "loss": 0.777,
+      "step": 681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5941393391752918e-05,
+      "loss": 0.8052,
+      "step": 682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5929202500314053e-05,
+      "loss": 0.7952,
+      "step": 683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5916998004504987e-05,
+      "loss": 0.7443,
+      "step": 684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5904779932328562e-05,
+      "loss": 0.7398,
+      "step": 685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5892548311818753e-05,
+      "loss": 0.7803,
+      "step": 686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5880303171040644e-05,
+      "loss": 0.7246,
+      "step": 687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5868044538090316e-05,
+      "loss": 0.7336,
+      "step": 688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585577244109483e-05,
+      "loss": 0.7615,
+      "step": 689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5843486908212128e-05,
+      "loss": 0.8167,
+      "step": 690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5831187967630977e-05,
+      "loss": 0.805,
+      "step": 691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5818875647570915e-05,
+      "loss": 0.769,
+      "step": 692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5806549976282185e-05,
+      "loss": 0.7547,
+      "step": 693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5794210982045638e-05,
+      "loss": 0.7964,
+      "step": 694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5781858693172723e-05,
+      "loss": 0.7589,
+      "step": 695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5769493138005384e-05,
+      "loss": 0.7914,
+      "step": 696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5757114344915994e-05,
+      "loss": 0.7761,
+      "step": 697
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574472234230731e-05,
+      "loss": 0.7397,
+      "step": 698
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5732317158612396e-05,
+      "loss": 0.8139,
+      "step": 699
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571989882229456e-05,
+      "loss": 0.7461,
+      "step": 700
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570746736184729e-05,
+      "loss": 0.7429,
+      "step": 701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5695022805794174e-05,
+      "loss": 0.7787,
+      "step": 702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5682565182688868e-05,
+      "loss": 0.7797,
+      "step": 703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5670094521115e-05,
+      "loss": 0.8081,
+      "step": 704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5657610849686103e-05,
+      "loss": 0.7663,
+      "step": 705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5645114197045585e-05,
+      "loss": 0.7283,
+      "step": 706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5632604591866614e-05,
+      "loss": 0.7871,
+      "step": 707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56200820628521e-05,
+      "loss": 0.8166,
+      "step": 708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5607546638734584e-05,
+      "loss": 0.743,
+      "step": 709
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5594998348276214e-05,
+      "loss": 0.8244,
+      "step": 710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5582437220268648e-05,
+      "loss": 0.7914,
+      "step": 711
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5569863283533004e-05,
+      "loss": 0.7938,
+      "step": 712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5557276566919787e-05,
+      "loss": 0.7714,
+      "step": 713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5544677099308824e-05,
+      "loss": 0.6653,
+      "step": 714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55320649096092e-05,
+      "loss": 0.5223,
+      "step": 715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5519440026759192e-05,
+      "loss": 0.4784,
+      "step": 716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5506802479726204e-05,
+      "loss": 0.5026,
+      "step": 717
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549415229750669e-05,
+      "loss": 0.4591,
+      "step": 718
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5481489509126094e-05,
+      "loss": 0.4975,
+      "step": 719
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54688141436388e-05,
+      "loss": 0.4547,
+      "step": 720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5456126230128027e-05,
+      "loss": 0.4453,
+      "step": 721
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54434257977058e-05,
+      "loss": 0.4582,
+      "step": 722
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543071287551287e-05,
+      "loss": 0.4504,
+      "step": 723
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5417987492718637e-05,
+      "loss": 0.4726,
+      "step": 724
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5405249678521096e-05,
+      "loss": 0.4913,
+      "step": 725
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5392499462146762e-05,
+      "loss": 0.4316,
+      "step": 726
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537973687285061e-05,
+      "loss": 0.4453,
+      "step": 727
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536696193991601e-05,
+      "loss": 0.4659,
+      "step": 728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535417469265463e-05,
+      "loss": 0.542,
+      "step": 729
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5341375160406427e-05,
+      "loss": 0.4174,
+      "step": 730
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5328563372539523e-05,
+      "loss": 0.4722,
+      "step": 731
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5315739358450164e-05,
+      "loss": 0.433,
+      "step": 732
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530290314756265e-05,
+      "loss": 0.4389,
+      "step": 733
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290054769329267e-05,
+      "loss": 0.4935,
+      "step": 734
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527719425323022e-05,
+      "loss": 0.4268,
+      "step": 735
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526432162877356e-05,
+      "loss": 0.4416,
+      "step": 736
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5251436925495121e-05,
+      "loss": 0.4595,
+      "step": 737
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5238540172958459e-05,
+      "loss": 0.4157,
+      "step": 738
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5225631400754766e-05,
+      "loss": 0.4971,
+      "step": 739
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5212710638502817e-05,
+      "loss": 0.4388,
+      "step": 740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5199777915848897e-05,
+      "loss": 0.5059,
+      "step": 741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5186833262466734e-05,
+      "loss": 0.4503,
+      "step": 742
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5173876708057438e-05,
+      "loss": 0.4958,
+      "step": 743
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160908282349413e-05,
+      "loss": 0.4491,
+      "step": 744
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5147928015098309e-05,
+      "loss": 0.4821,
+      "step": 745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5134935936086942e-05,
+      "loss": 0.4646,
+      "step": 746
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5121932075125237e-05,
+      "loss": 0.4954,
+      "step": 747
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5108916462050144e-05,
+      "loss": 0.44,
+      "step": 748
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5095889126725584e-05,
+      "loss": 0.4299,
+      "step": 749
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082850099042366e-05,
+      "loss": 0.4649,
+      "step": 750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5069799408918138e-05,
+      "loss": 0.4456,
+      "step": 751
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5056737086297298e-05,
+      "loss": 0.4402,
+      "step": 752
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5043663161150939e-05,
+      "loss": 0.3809,
+      "step": 753
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5030577663476776e-05,
+      "loss": 0.4342,
+      "step": 754
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501748062329907e-05,
+      "loss": 0.4898,
+      "step": 755
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5004372070668579e-05,
+      "loss": 0.4491,
+      "step": 756
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4991252035662462e-05,
+      "loss": 0.4568,
+      "step": 757
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4978120548384232e-05,
+      "loss": 0.4478,
+      "step": 758
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4964977638963678e-05,
+      "loss": 0.4561,
+      "step": 759
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4951823337556795e-05,
+      "loss": 0.4397,
+      "step": 760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4938657674345715e-05,
+      "loss": 0.4635,
+      "step": 761
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4925480679538646e-05,
+      "loss": 0.4909,
+      "step": 762
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4912292383369786e-05,
+      "loss": 0.4362,
+      "step": 763
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4899092816099276e-05,
+      "loss": 0.4049,
+      "step": 764
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4885882008013108e-05,
+      "loss": 0.4219,
+      "step": 765
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4872659989423069e-05,
+      "loss": 0.4466,
+      "step": 766
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4859426790666665e-05,
+      "loss": 0.4718,
+      "step": 767
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4846182442107062e-05,
+      "loss": 0.4879,
+      "step": 768
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832926974133007e-05,
+      "loss": 0.4742,
+      "step": 769
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4819660417158751e-05,
+      "loss": 0.4634,
+      "step": 770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806382801623999e-05,
+      "loss": 0.4672,
+      "step": 771
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4793094157993826e-05,
+      "loss": 0.4618,
+      "step": 772
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477979451675861e-05,
+      "loss": 0.4604,
+      "step": 773
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4766483908433963e-05,
+      "loss": 0.4319,
+      "step": 774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753162363560658e-05,
+      "loss": 0.4564,
+      "step": 775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4739829912704569e-05,
+      "loss": 0.4795,
+      "step": 776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4726486586456585e-05,
+      "loss": 0.5143,
+      "step": 777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713132415432554e-05,
+      "loss": 0.4864,
+      "step": 778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4699767430273202e-05,
+      "loss": 0.5214,
+      "step": 779
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4686391661644074e-05,
+      "loss": 0.4496,
+      "step": 780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4673005140235453e-05,
+      "loss": 0.5031,
+      "step": 781
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465960789676229e-05,
+      "loss": 0.4552,
+      "step": 782
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4646199961964152e-05,
+      "loss": 0.4722,
+      "step": 783
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4632781366605117e-05,
+      "loss": 0.4525,
+      "step": 784
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4619352141473741e-05,
+      "loss": 0.492,
+      "step": 785
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4605912317382957e-05,
+      "loss": 0.4603,
+      "step": 786
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4592461925170027e-05,
+      "loss": 0.4929,
+      "step": 787
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4579000995696456e-05,
+      "loss": 0.4138,
+      "step": 788
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456552955984792e-05,
+      "loss": 0.445,
+      "step": 789
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4552047648534212e-05,
+      "loss": 0.4966,
+      "step": 790
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4538555292689162e-05,
+      "loss": 0.4699,
+      "step": 791
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4525052523270557e-05,
+      "loss": 0.4905,
+      "step": 792
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4511539371260075e-05,
+      "loss": 0.4232,
+      "step": 793
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4498015867663226e-05,
+      "loss": 0.4506,
+      "step": 794
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4484482043509271e-05,
+      "loss": 0.4571,
+      "step": 795
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4470937929851142e-05,
+      "loss": 0.4308,
+      "step": 796
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4457383557765385e-05,
+      "loss": 0.5098,
+      "step": 797
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4443818958352085e-05,
+      "loss": 0.4582,
+      "step": 798
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4430244162734794e-05,
+      "loss": 0.4718,
+      "step": 799
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4416659202060457e-05,
+      "loss": 0.4573,
+      "step": 800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4403064107499343e-05,
+      "loss": 0.4689,
+      "step": 801
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4389458910244965e-05,
+      "loss": 0.5016,
+      "step": 802
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4375843641514036e-05,
+      "loss": 0.4254,
+      "step": 803
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4362218332546355e-05,
+      "loss": 0.4953,
+      "step": 804
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4348583014604776e-05,
+      "loss": 0.5001,
+      "step": 805
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4334937718975106e-05,
+      "loss": 0.4449,
+      "step": 806
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4321282476966052e-05,
+      "loss": 0.4093,
+      "step": 807
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4307617319909138e-05,
+      "loss": 0.4599,
+      "step": 808
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4293942279158648e-05,
+      "loss": 0.4765,
+      "step": 809
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4280257386091526e-05,
+      "loss": 0.448,
+      "step": 810
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426656267210734e-05,
+      "loss": 0.473,
+      "step": 811
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4252858168628184e-05,
+      "loss": 0.4836,
+      "step": 812
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423914390709861e-05,
+      "loss": 0.4596,
+      "step": 813
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4225419918985563e-05,
+      "loss": 0.4698,
+      "step": 814
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421168623577831e-05,
+      "loss": 0.4362,
+      "step": 815
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4197942888988354e-05,
+      "loss": 0.4808,
+      "step": 816
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4184189910149382e-05,
+      "loss": 0.399,
+      "step": 817
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4170427330817169e-05,
+      "loss": 0.4128,
+      "step": 818
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156655182569524e-05,
+      "loss": 0.5083,
+      "step": 819
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4142873497006215e-05,
+      "loss": 0.4561,
+      "step": 820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412908230574889e-05,
+      "loss": 0.4518,
+      "step": 821
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4115281640441003e-05,
+      "loss": 0.4712,
+      "step": 822
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4101471532747753e-05,
+      "loss": 0.4659,
+      "step": 823
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4087652014356e-05,
+      "loss": 0.4672,
+      "step": 824
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4073823116974202e-05,
+      "loss": 0.4719,
+      "step": 825
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4059984872332324e-05,
+      "loss": 0.4006,
+      "step": 826
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4046137312181796e-05,
+      "loss": 0.4447,
+      "step": 827
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4032280468295407e-05,
+      "loss": 0.465,
+      "step": 828
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4018414372467258e-05,
+      "loss": 0.4819,
+      "step": 829
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4004539056512667e-05,
+      "loss": 0.459,
+      "step": 830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3990654552268116e-05,
+      "loss": 0.4624,
+      "step": 831
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3976760891591172e-05,
+      "loss": 0.5239,
+      "step": 832
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3962858106360399e-05,
+      "loss": 0.489,
+      "step": 833
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394894622847531e-05,
+      "loss": 0.5272,
+      "step": 834
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393502528985627e-05,
+      "loss": 0.433,
+      "step": 835
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921095322444443e-05,
+      "loss": 0.4326,
+      "step": 836
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.39071563582017e-05,
+      "loss": 0.4875,
+      "step": 837
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3893208429110558e-05,
+      "loss": 0.4233,
+      "step": 838
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3879251567174117e-05,
+      "loss": 0.4744,
+      "step": 839
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3865285804415946e-05,
+      "loss": 0.4505,
+      "step": 840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3851311172880064e-05,
+      "loss": 0.4866,
+      "step": 841
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3837327704630823e-05,
+      "loss": 0.4551,
+      "step": 842
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3823335431752854e-05,
+      "loss": 0.4903,
+      "step": 843
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3809334386350987e-05,
+      "loss": 0.4729,
+      "step": 844
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3795324600550192e-05,
+      "loss": 0.5031,
+      "step": 845
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378130610649548e-05,
+      "loss": 0.4476,
+      "step": 846
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3767278936351853e-05,
+      "loss": 0.4389,
+      "step": 847
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753243122304209e-05,
+      "loss": 0.4672,
+      "step": 848
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373919869655729e-05,
+      "loss": 0.416,
+      "step": 849
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3725145691335593e-05,
+      "loss": 0.4792,
+      "step": 850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3711084138883298e-05,
+      "loss": 0.4939,
+      "step": 851
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3697014071464198e-05,
+      "loss": 0.4259,
+      "step": 852
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3682935521361627e-05,
+      "loss": 0.4636,
+      "step": 853
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3668848520878374e-05,
+      "loss": 0.4288,
+      "step": 854
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3654753102336628e-05,
+      "loss": 0.4425,
+      "step": 855
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640649298077887e-05,
+      "loss": 0.4793,
+      "step": 856
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3626537140462883e-05,
+      "loss": 0.4464,
+      "step": 857
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3612416661871532e-05,
+      "loss": 0.3892,
+      "step": 858
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3598287894702822e-05,
+      "loss": 0.4897,
+      "step": 859
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3584150871374773e-05,
+      "loss": 0.5036,
+      "step": 860
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3570005624324342e-05,
+      "loss": 0.4948,
+      "step": 861
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3555852186007358e-05,
+      "loss": 0.4168,
+      "step": 862
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.354169058889845e-05,
+      "loss": 0.4288,
+      "step": 863
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352752086549095e-05,
+      "loss": 0.5082,
+      "step": 864
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3513343048296852e-05,
+      "loss": 0.4731,
+      "step": 865
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3499157169846715e-05,
+      "loss": 0.484,
+      "step": 866
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3484963262689597e-05,
+      "loss": 0.3899,
+      "step": 867
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3470761359392972e-05,
+      "loss": 0.4279,
+      "step": 868
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3456551492542663e-05,
+      "loss": 0.4755,
+      "step": 869
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3442333694742768e-05,
+      "loss": 0.481,
+      "step": 870
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3428107998615582e-05,
+      "loss": 0.3909,
+      "step": 871
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3413874436801524e-05,
+      "loss": 0.4596,
+      "step": 872
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3399633041959047e-05,
+      "loss": 0.3956,
+      "step": 873
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3385383846764594e-05,
+      "loss": 0.4681,
+      "step": 874
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3371126883912503e-05,
+      "loss": 0.4573,
+      "step": 875
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3356862186114924e-05,
+      "loss": 0.4781,
+      "step": 876
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342589786101762e-05,
+      "loss": 0.4847,
+      "step": 877
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3328309716620597e-05,
+      "loss": 0.4127,
+      "step": 878
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3314022010436603e-05,
+      "loss": 0.4505,
+      "step": 879
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3299726700332474e-05,
+      "loss": 0.4731,
+      "step": 880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3285423819108349e-05,
+      "loss": 0.4767,
+      "step": 881
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.327111339958175e-05,
+      "loss": 0.4215,
+      "step": 882
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3256795474587484e-05,
+      "loss": 0.4358,
+      "step": 883
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3242470076977587e-05,
+      "loss": 0.4907,
+      "step": 884
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3228137239621232e-05,
+      "loss": 0.4405,
+      "step": 885
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3213796995404672e-05,
+      "loss": 0.4836,
+      "step": 886
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3199449377231148e-05,
+      "loss": 0.4629,
+      "step": 887
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3185094418020823e-05,
+      "loss": 0.4677,
+      "step": 888
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31707321507107e-05,
+      "loss": 0.4765,
+      "step": 889
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.315636260825456e-05,
+      "loss": 0.4539,
+      "step": 890
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3141985823622862e-05,
+      "loss": 0.4731,
+      "step": 891
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3127601829802697e-05,
+      "loss": 0.4593,
+      "step": 892
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3113210659797686e-05,
+      "loss": 0.4562,
+      "step": 893
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3098812346627922e-05,
+      "loss": 0.5033,
+      "step": 894
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.308440692332988e-05,
+      "loss": 0.4588,
+      "step": 895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3069994422956362e-05,
+      "loss": 0.4509,
+      "step": 896
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055574878576398e-05,
+      "loss": 0.4801,
+      "step": 897
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3041148323275182e-05,
+      "loss": 0.4321,
+      "step": 898
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026714790153992e-05,
+      "loss": 0.4721,
+      "step": 899
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3012274312330124e-05,
+      "loss": 0.4774,
+      "step": 900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2997826922936805e-05,
+      "loss": 0.4532,
+      "step": 901
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2983372655123115e-05,
+      "loss": 0.4984,
+      "step": 902
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2968911542053924e-05,
+      "loss": 0.4586,
+      "step": 903
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29544436169098e-05,
+      "loss": 0.4388,
+      "step": 904
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2939968912886952e-05,
+      "loss": 0.45,
+      "step": 905
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2925487463197138e-05,
+      "loss": 0.4885,
+      "step": 906
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2910999301067584e-05,
+      "loss": 0.406,
+      "step": 907
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2896504459740932e-05,
+      "loss": 0.4668,
+      "step": 908
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2882002972475142e-05,
+      "loss": 0.4927,
+      "step": 909
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2867494872543427e-05,
+      "loss": 0.424,
+      "step": 910
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2852980193234167e-05,
+      "loss": 0.4413,
+      "step": 911
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2838458967850843e-05,
+      "loss": 0.4892,
+      "step": 912
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2823931229711946e-05,
+      "loss": 0.408,
+      "step": 913
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280939701215093e-05,
+      "loss": 0.4499,
+      "step": 914
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2794856348516095e-05,
+      "loss": 0.4817,
+      "step": 915
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2780309272170546e-05,
+      "loss": 0.4681,
+      "step": 916
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276575581649209e-05,
+      "loss": 0.4862,
+      "step": 917
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2751196014873182e-05,
+      "loss": 0.4605,
+      "step": 918
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2736629900720832e-05,
+      "loss": 0.4561,
+      "step": 919
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2722057507456531e-05,
+      "loss": 0.4451,
+      "step": 920
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2707478868516181e-05,
+      "loss": 0.4434,
+      "step": 921
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2692894017350022e-05,
+      "loss": 0.4757,
+      "step": 922
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2678302987422533e-05,
+      "loss": 0.4886,
+      "step": 923
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2663705812212377e-05,
+      "loss": 0.4996,
+      "step": 924
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264910252521232e-05,
+      "loss": 0.4205,
+      "step": 925
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2634493159929146e-05,
+      "loss": 0.4343,
+      "step": 926
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2619877749883594e-05,
+      "loss": 0.4535,
+      "step": 927
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2605256328610258e-05,
+      "loss": 0.4707,
+      "step": 928
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2590628929657538e-05,
+      "loss": 0.484,
+      "step": 929
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2575995586587544e-05,
+      "loss": 0.4165,
+      "step": 930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2561356332976022e-05,
+      "loss": 0.4737,
+      "step": 931
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2546711202412287e-05,
+      "loss": 0.4525,
+      "step": 932
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2532060228499137e-05,
+      "loss": 0.5194,
+      "step": 933
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2517403444852764e-05,
+      "loss": 0.4415,
+      "step": 934
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2502740885102713e-05,
+      "loss": 0.4822,
+      "step": 935
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2488072582891764e-05,
+      "loss": 0.4679,
+      "step": 936
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247339857187588e-05,
+      "loss": 0.4776,
+      "step": 937
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2458718885724117e-05,
+      "loss": 0.471,
+      "step": 938
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2444033558118562e-05,
+      "loss": 0.4499,
+      "step": 939
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2429342622754242e-05,
+      "loss": 0.4569,
+      "step": 940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414646113339048e-05,
+      "loss": 0.4801,
+      "step": 941
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2399944063593662e-05,
+      "loss": 0.456,
+      "step": 942
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2385236507251477e-05,
+      "loss": 0.5,
+      "step": 943
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2370523478058527e-05,
+      "loss": 0.4465,
+      "step": 944
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2355805009773396e-05,
+      "loss": 0.4376,
+      "step": 945
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.234108113616715e-05,
+      "loss": 0.5045,
+      "step": 946
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2326351891023256e-05,
+      "loss": 0.4546,
+      "step": 947
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.231161730813751e-05,
+      "loss": 0.4831,
+      "step": 948
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2296877421317958e-05,
+      "loss": 0.4605,
+      "step": 949
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.22821322643848e-05,
+      "loss": 0.4971,
+      "step": 950
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2267381871170342e-05,
+      "loss": 0.4466,
+      "step": 951
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225262627551891e-05,
+      "loss": 0.4439,
+      "step": 952
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2237865511286745e-05,
+      "loss": 0.5009,
+      "step": 953
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2223099612341968e-05,
+      "loss": 0.4099,
+      "step": 954
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2208328612564471e-05,
+      "loss": 0.45,
+      "step": 955
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2193552545845853e-05,
+      "loss": 0.4238,
+      "step": 956
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2178771446089338e-05,
+      "loss": 0.4222,
+      "step": 957
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.21639853472097e-05,
+      "loss": 0.4341,
+      "step": 958
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2149194283133176e-05,
+      "loss": 0.4989,
+      "step": 959
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213439828779741e-05,
+      "loss": 0.5231,
+      "step": 960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2119597395151343e-05,
+      "loss": 0.4912,
+      "step": 961
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2104791639155167e-05,
+      "loss": 0.4539,
+      "step": 962
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2089981053780225e-05,
+      "loss": 0.4587,
+      "step": 963
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2075165673008944e-05,
+      "loss": 0.4834,
+      "step": 964
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2060345530834754e-05,
+      "loss": 0.4966,
+      "step": 965
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2045520661262011e-05,
+      "loss": 0.4239,
+      "step": 966
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2030691098305909e-05,
+      "loss": 0.4129,
+      "step": 967
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.201585687599242e-05,
+      "loss": 0.469,
+      "step": 968
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200101802835821e-05,
+      "loss": 0.5114,
+      "step": 969
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1986174589450548e-05,
+      "loss": 0.4275,
+      "step": 970
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1971326593327241e-05,
+      "loss": 0.4458,
+      "step": 971
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.195647407405655e-05,
+      "loss": 0.5078,
+      "step": 972
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1941617065717123e-05,
+      "loss": 0.4337,
+      "step": 973
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1926755602397903e-05,
+      "loss": 0.45,
+      "step": 974
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1911889718198049e-05,
+      "loss": 0.4356,
+      "step": 975
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1897019447226869e-05,
+      "loss": 0.4699,
+      "step": 976
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1882144823603738e-05,
+      "loss": 0.4878,
+      "step": 977
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1867265881458016e-05,
+      "loss": 0.4503,
+      "step": 978
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852382654928973e-05,
+      "loss": 0.4612,
+      "step": 979
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1837495178165706e-05,
+      "loss": 0.4139,
+      "step": 980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1822603485327062e-05,
+      "loss": 0.4072,
+      "step": 981
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1807707610581574e-05,
+      "loss": 0.4421,
+      "step": 982
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1792807588107358e-05,
+      "loss": 0.4568,
+      "step": 983
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.177790345209205e-05,
+      "loss": 0.4506,
+      "step": 984
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1762995236732726e-05,
+      "loss": 0.4991,
+      "step": 985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1748082976235822e-05,
+      "loss": 0.4917,
+      "step": 986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1733166704817052e-05,
+      "loss": 0.4614,
+      "step": 987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1718246456701339e-05,
+      "loss": 0.49,
+      "step": 988
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1703322266122723e-05,
+      "loss": 0.5475,
+      "step": 989
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1688394167324298e-05,
+      "loss": 0.497,
+      "step": 990
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.167346219455812e-05,
+      "loss": 0.4455,
+      "step": 991
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165852638208513e-05,
+      "loss": 0.4344,
+      "step": 992
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1643586764175092e-05,
+      "loss": 0.4481,
+      "step": 993
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1628643375106486e-05,
+      "loss": 0.433,
+      "step": 994
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1613696249166455e-05,
+      "loss": 0.4261,
+      "step": 995
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1598745420650717e-05,
+      "loss": 0.497,
+      "step": 996
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1583790923863476e-05,
+      "loss": 0.4572,
+      "step": 997
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1568832793117356e-05,
+      "loss": 0.5209,
+      "step": 998
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1553871062733332e-05,
+      "loss": 0.4467,
+      "step": 999
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.153890576704062e-05,
+      "loss": 0.4649,
+      "step": 1000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523936940376625e-05,
+      "loss": 0.4021,
+      "step": 1001
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1508964617086852e-05,
+      "loss": 0.4317,
+      "step": 1002
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1493988831524834e-05,
+      "loss": 0.4902,
+      "step": 1003
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147900961805204e-05,
+      "loss": 0.4945,
+      "step": 1004
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1464027011037813e-05,
+      "loss": 0.4658,
+      "step": 1005
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.144904104485927e-05,
+      "loss": 0.4443,
+      "step": 1006
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434051753901248e-05,
+      "loss": 0.4146,
+      "step": 1007
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1419059172556207e-05,
+      "loss": 0.4289,
+      "step": 1008
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1404063335224153e-05,
+      "loss": 0.4004,
+      "step": 1009
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1389064276312575e-05,
+      "loss": 0.4808,
+      "step": 1010
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137406203023634e-05,
+      "loss": 0.3859,
+      "step": 1011
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1359056631417635e-05,
+      "loss": 0.4706,
+      "step": 1012
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1344048114285881e-05,
+      "loss": 0.4637,
+      "step": 1013
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.132903651327765e-05,
+      "loss": 0.4971,
+      "step": 1014
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1314021862836593e-05,
+      "loss": 0.4469,
+      "step": 1015
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1299004197413358e-05,
+      "loss": 0.4703,
+      "step": 1016
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1283983551465512e-05,
+      "loss": 0.4883,
+      "step": 1017
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1268959959457455e-05,
+      "loss": 0.4206,
+      "step": 1018
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1253933455860348e-05,
+      "loss": 0.4515,
+      "step": 1019
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1238904075152039e-05,
+      "loss": 0.4754,
+      "step": 1020
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1223871851816976e-05,
+      "loss": 0.444,
+      "step": 1021
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.120883682034612e-05,
+      "loss": 0.4525,
+      "step": 1022
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1193799015236886e-05,
+      "loss": 0.463,
+      "step": 1023
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1178758470993047e-05,
+      "loss": 0.4793,
+      "step": 1024
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1163715222124662e-05,
+      "loss": 0.4492,
+      "step": 1025
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1148669303148e-05,
+      "loss": 0.4446,
+      "step": 1026
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.113362074858545e-05,
+      "loss": 0.469,
+      "step": 1027
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.111856959296545e-05,
+      "loss": 0.4302,
+      "step": 1028
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110351587082241e-05,
+      "loss": 0.4454,
+      "step": 1029
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1088459616696623e-05,
+      "loss": 0.4538,
+      "step": 1030
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1073400865134194e-05,
+      "loss": 0.448,
+      "step": 1031
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1058339650686962e-05,
+      "loss": 0.4669,
+      "step": 1032
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1043276007912413e-05,
+      "loss": 0.4629,
+      "step": 1033
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1028209971373605e-05,
+      "loss": 0.4686,
+      "step": 1034
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.101314157563909e-05,
+      "loss": 0.4662,
+      "step": 1035
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0998070855282828e-05,
+      "loss": 0.4296,
+      "step": 1036
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.098299784488412e-05,
+      "loss": 0.413,
+      "step": 1037
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0967922579027519e-05,
+      "loss": 0.4194,
+      "step": 1038
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095284509230275e-05,
+      "loss": 0.4336,
+      "step": 1039
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0937765419304638e-05,
+      "loss": 0.3807,
+      "step": 1040
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092268359463302e-05,
+      "loss": 0.4504,
+      "step": 1041
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0907599652892675e-05,
+      "loss": 0.444,
+      "step": 1042
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0892513628693237e-05,
+      "loss": 0.4299,
+      "step": 1043
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0877425556649115e-05,
+      "loss": 0.486,
+      "step": 1044
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0862335471379419e-05,
+      "loss": 0.4887,
+      "step": 1045
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0847243407507884e-05,
+      "loss": 0.5102,
+      "step": 1046
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0832149399662777e-05,
+      "loss": 0.4445,
+      "step": 1047
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0817053482476828e-05,
+      "loss": 0.4618,
+      "step": 1048
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0801955690587146e-05,
+      "loss": 0.4441,
+      "step": 1049
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0786856058635147e-05,
+      "loss": 0.4625,
+      "step": 1050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0771754621266466e-05,
+      "loss": 0.4766,
+      "step": 1051
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0756651413130876e-05,
+      "loss": 0.4558,
+      "step": 1052
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0741546468882223e-05,
+      "loss": 0.4505,
+      "step": 1053
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0726439823178322e-05,
+      "loss": 0.4666,
+      "step": 1054
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0711331510680909e-05,
+      "loss": 0.4594,
+      "step": 1055
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0696221566055529e-05,
+      "loss": 0.4944,
+      "step": 1056
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0681110023971484e-05,
+      "loss": 0.4787,
+      "step": 1057
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0665996919101732e-05,
+      "loss": 0.4449,
+      "step": 1058
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0650882286122823e-05,
+      "loss": 0.4834,
+      "step": 1059
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0635766159714812e-05,
+      "loss": 0.4557,
+      "step": 1060
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0620648574561177e-05,
+      "loss": 0.4796,
+      "step": 1061
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0605529565348744e-05,
+      "loss": 0.493,
+      "step": 1062
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0590409166767615e-05,
+      "loss": 0.4572,
+      "step": 1063
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0575287413511064e-05,
+      "loss": 0.4819,
+      "step": 1064
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0560164340275489e-05,
+      "loss": 0.4308,
+      "step": 1065
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545039981760305e-05,
+      "loss": 0.4844,
+      "step": 1066
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0529914372667884e-05,
+      "loss": 0.4818,
+      "step": 1067
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0514787547703466e-05,
+      "loss": 0.4612,
+      "step": 1068
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0499659541575075e-05,
+      "loss": 0.5058,
+      "step": 1069
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0484530388993452e-05,
+      "loss": 0.4894,
+      "step": 1070
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0469400124671962e-05,
+      "loss": 0.4785,
+      "step": 1071
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0454268783326529e-05,
+      "loss": 0.4645,
+      "step": 1072
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0439136399675542e-05,
+      "loss": 0.4886,
+      "step": 1073
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0424003008439784e-05,
+      "loss": 0.3948,
+      "step": 1074
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0408868644342344e-05,
+      "loss": 0.4333,
+      "step": 1075
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.039373334210856e-05,
+      "loss": 0.443,
+      "step": 1076
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0378597136465896e-05,
+      "loss": 0.4727,
+      "step": 1077
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0363460062143914e-05,
+      "loss": 0.4721,
+      "step": 1078
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0348322153874157e-05,
+      "loss": 0.4483,
+      "step": 1079
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033318344639008e-05,
+      "loss": 0.4374,
+      "step": 1080
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0318043974426981e-05,
+      "loss": 0.4728,
+      "step": 1081
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0302903772721904e-05,
+      "loss": 0.4454,
+      "step": 1082
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0287762876013562e-05,
+      "loss": 0.4206,
+      "step": 1083
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.027262131904228e-05,
+      "loss": 0.4579,
+      "step": 1084
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0257479136549889e-05,
+      "loss": 0.4407,
+      "step": 1085
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0242336363279643e-05,
+      "loss": 0.4574,
+      "step": 1086
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0227193033976176e-05,
+      "loss": 0.4322,
+      "step": 1087
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0212049183385375e-05,
+      "loss": 0.4944,
+      "step": 1088
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019690484625434e-05,
+      "loss": 0.4792,
+      "step": 1089
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0181760057331276e-05,
+      "loss": 0.4859,
+      "step": 1090
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0166614851365428e-05,
+      "loss": 0.4041,
+      "step": 1091
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0151469263107002e-05,
+      "loss": 0.49,
+      "step": 1092
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0136323327307076e-05,
+      "loss": 0.462,
+      "step": 1093
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0121177078717527e-05,
+      "loss": 0.4231,
+      "step": 1094
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.010603055209095e-05,
+      "loss": 0.4355,
+      "step": 1095
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.009088378218058e-05,
+      "loss": 0.4259,
+      "step": 1096
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0075736803740207e-05,
+      "loss": 0.5245,
+      "step": 1097
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00605896515241e-05,
+      "loss": 0.4494,
+      "step": 1098
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0045442360286925e-05,
+      "loss": 0.449,
+      "step": 1099
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0030294964783676e-05,
+      "loss": 0.4474,
+      "step": 1100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0015147499769571e-05,
+      "loss": 0.4867,
+      "step": 1101
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.48,
+      "step": 1102
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.984852500230432e-06,
+      "loss": 0.4408,
+      "step": 1103
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.969705035216329e-06,
+      "loss": 0.5114,
+      "step": 1104
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.954557639713079e-06,
+      "loss": 0.4753,
+      "step": 1105
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.939410348475905e-06,
+      "loss": 0.4561,
+      "step": 1106
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.924263196259797e-06,
+      "loss": 0.4677,
+      "step": 1107
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.909116217819422e-06,
+      "loss": 0.4353,
+      "step": 1108
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.89396944790905e-06,
+      "loss": 0.4619,
+      "step": 1109
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.878822921282474e-06,
+      "loss": 0.4871,
+      "step": 1110
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.863676672692929e-06,
+      "loss": 0.4612,
+      "step": 1111
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.848530736893003e-06,
+      "loss": 0.4382,
+      "step": 1112
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.833385148634574e-06,
+      "loss": 0.4843,
+      "step": 1113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.818239942668727e-06,
+      "loss": 0.4745,
+      "step": 1114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.803095153745662e-06,
+      "loss": 0.5149,
+      "step": 1115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.78795081661463e-06,
+      "loss": 0.5083,
+      "step": 1116
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.772806966023828e-06,
+      "loss": 0.4833,
+      "step": 1117
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.757663636720359e-06,
+      "loss": 0.4823,
+      "step": 1118
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.742520863450116e-06,
+      "loss": 0.4718,
+      "step": 1119
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72737868095772e-06,
+      "loss": 0.4936,
+      "step": 1120
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71223712398644e-06,
+      "loss": 0.4199,
+      "step": 1121
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.697096227278103e-06,
+      "loss": 0.4165,
+      "step": 1122
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.681956025573024e-06,
+      "loss": 0.4271,
+      "step": 1123
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.666816553609922e-06,
+      "loss": 0.4427,
+      "step": 1124
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.651677846125846e-06,
+      "loss": 0.421,
+      "step": 1125
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.636539937856089e-06,
+      "loss": 0.489,
+      "step": 1126
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.621402863534109e-06,
+      "loss": 0.4721,
+      "step": 1127
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.606266657891447e-06,
+      "loss": 0.4879,
+      "step": 1128
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.591131355657657e-06,
+      "loss": 0.4781,
+      "step": 1129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57599699156022e-06,
+      "loss": 0.4326,
+      "step": 1130
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56086360032446e-06,
+      "loss": 0.4835,
+      "step": 1131
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.545731216673471e-06,
+      "loss": 0.4503,
+      "step": 1132
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.530599875328041e-06,
+      "loss": 0.3931,
+      "step": 1133
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51546961100655e-06,
+      "loss": 0.4773,
+      "step": 1134
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.500340458424926e-06,
+      "loss": 0.4372,
+      "step": 1135
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.485212452296535e-06,
+      "loss": 0.4383,
+      "step": 1136
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.470085627332115e-06,
+      "loss": 0.4602,
+      "step": 1137
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.454960018239696e-06,
+      "loss": 0.455,
+      "step": 1138
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.439835659724514e-06,
+      "loss": 0.4326,
+      "step": 1139
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.424712586488939e-06,
+      "loss": 0.4327,
+      "step": 1140
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40959083323239e-06,
+      "loss": 0.4982,
+      "step": 1141
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.394470434651258e-06,
+      "loss": 0.4441,
+      "step": 1142
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.379351425438826e-06,
+      "loss": 0.4947,
+      "step": 1143
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.364233840285193e-06,
+      "loss": 0.4526,
+      "step": 1144
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.349117713877179e-06,
+      "loss": 0.4432,
+      "step": 1145
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.334003080898271e-06,
+      "loss": 0.409,
+      "step": 1146
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.31888997602852e-06,
+      "loss": 0.4571,
+      "step": 1147
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.303778433944471e-06,
+      "loss": 0.4632,
+      "step": 1148
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.288668489319098e-06,
+      "loss": 0.4693,
+      "step": 1149
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.273560176821681e-06,
+      "loss": 0.513,
+      "step": 1150
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.258453531117782e-06,
+      "loss": 0.4398,
+      "step": 1151
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.243348586869125e-06,
+      "loss": 0.4926,
+      "step": 1152
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.228245378733537e-06,
+      "loss": 0.4691,
+      "step": 1153
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.213143941364854e-06,
+      "loss": 0.4485,
+      "step": 1154
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.198044309412858e-06,
+      "loss": 0.4433,
+      "step": 1155
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182946517523177e-06,
+      "loss": 0.4381,
+      "step": 1156
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.167850600337226e-06,
+      "loss": 0.4439,
+      "step": 1157
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.152756592492118e-06,
+      "loss": 0.4569,
+      "step": 1158
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.137664528620583e-06,
+      "loss": 0.4569,
+      "step": 1159
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12257444335089e-06,
+      "loss": 0.4329,
+      "step": 1160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.107486371306768e-06,
+      "loss": 0.4336,
+      "step": 1161
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.092400347107327e-06,
+      "loss": 0.4029,
+      "step": 1162
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07731640536698e-06,
+      "loss": 0.4207,
+      "step": 1163
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.062234580695366e-06,
+      "loss": 0.4384,
+      "step": 1164
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.04715490769725e-06,
+      "loss": 0.4199,
+      "step": 1165
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.032077420972485e-06,
+      "loss": 0.506,
+      "step": 1166
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.017002155115884e-06,
+      "loss": 0.536,
+      "step": 1167
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.001929144717177e-06,
+      "loss": 0.4375,
+      "step": 1168
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.986858424360912e-06,
+      "loss": 0.4573,
+      "step": 1169
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.971790028626395e-06,
+      "loss": 0.511,
+      "step": 1170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.956723992087592e-06,
+      "loss": 0.4498,
+      "step": 1171
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94166034931304e-06,
+      "loss": 0.4216,
+      "step": 1172
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.926599134865808e-06,
+      "loss": 0.4635,
+      "step": 1173
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.911540383303382e-06,
+      "loss": 0.4734,
+      "step": 1174
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.896484129177595e-06,
+      "loss": 0.4597,
+      "step": 1175
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.881430407034552e-06,
+      "loss": 0.4595,
+      "step": 1176
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.866379251414554e-06,
+      "loss": 0.479,
+      "step": 1177
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.851330696852003e-06,
+      "loss": 0.4569,
+      "step": 1178
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.83628477787534e-06,
+      "loss": 0.4619,
+      "step": 1179
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.821241529006955e-06,
+      "loss": 0.4497,
+      "step": 1180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.806200984763117e-06,
+      "loss": 0.4011,
+      "step": 1181
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.791163179653884e-06,
+      "loss": 0.4203,
+      "step": 1182
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77612814818303e-06,
+      "loss": 0.4654,
+      "step": 1183
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.761095924847963e-06,
+      "loss": 0.4556,
+      "step": 1184
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.746066544139655e-06,
+      "loss": 0.4676,
+      "step": 1185
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.731040040542549e-06,
+      "loss": 0.451,
+      "step": 1186
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.71601644853449e-06,
+      "loss": 0.4352,
+      "step": 1187
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.700995802586647e-06,
+      "loss": 0.4926,
+      "step": 1188
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68597813716341e-06,
+      "loss": 0.3712,
+      "step": 1189
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.670963486722353e-06,
+      "loss": 0.3897,
+      "step": 1190
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65595188571412e-06,
+      "loss": 0.4258,
+      "step": 1191
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.640943368582367e-06,
+      "loss": 0.4198,
+      "step": 1192
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.625937969763662e-06,
+      "loss": 0.4328,
+      "step": 1193
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.610935723687428e-06,
+      "loss": 0.4987,
+      "step": 1194
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.595936664775848e-06,
+      "loss": 0.4462,
+      "step": 1195
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580940827443798e-06,
+      "loss": 0.4393,
+      "step": 1196
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565948246098755e-06,
+      "loss": 0.4885,
+      "step": 1197
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.550958955140733e-06,
+      "loss": 0.3738,
+      "step": 1198
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.535972988962192e-06,
+      "loss": 0.4508,
+      "step": 1199
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.520990381947962e-06,
+      "loss": 0.358,
+      "step": 1200
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50601116847517e-06,
+      "loss": 0.4357,
+      "step": 1201
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.49103538291315e-06,
+      "loss": 0.4875,
+      "step": 1202
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.476063059623376e-06,
+      "loss": 0.4735,
+      "step": 1203
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.461094232959381e-06,
+      "loss": 0.3856,
+      "step": 1204
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.446128937266673e-06,
+      "loss": 0.441,
+      "step": 1205
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.431167206882647e-06,
+      "loss": 0.5007,
+      "step": 1206
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.416209076136528e-06,
+      "loss": 0.4764,
+      "step": 1207
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.401254579349287e-06,
+      "loss": 0.4168,
+      "step": 1208
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.386303750833545e-06,
+      "loss": 0.5315,
+      "step": 1209
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.37135662489352e-06,
+      "loss": 0.5191,
+      "step": 1210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.356413235824912e-06,
+      "loss": 0.4828,
+      "step": 1211
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.341473617914871e-06,
+      "loss": 0.4655,
+      "step": 1212
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.326537805441883e-06,
+      "loss": 0.4577,
+      "step": 1213
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.311605832675704e-06,
+      "loss": 0.4604,
+      "step": 1214
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.296677733877277e-06,
+      "loss": 0.4621,
+      "step": 1215
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.281753543298664e-06,
+      "loss": 0.3936,
+      "step": 1216
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266833295182951e-06,
+      "loss": 0.4391,
+      "step": 1217
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.251917023764181e-06,
+      "loss": 0.4454,
+      "step": 1218
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.237004763267278e-06,
+      "loss": 0.4519,
+      "step": 1219
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.222096547907952e-06,
+      "loss": 0.4188,
+      "step": 1220
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.207192411892645e-06,
+      "loss": 0.4652,
+      "step": 1221
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.19229238941843e-06,
+      "loss": 0.4866,
+      "step": 1222
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.17739651467294e-06,
+      "loss": 0.4741,
+      "step": 1223
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.162504821834296e-06,
+      "loss": 0.4314,
+      "step": 1224
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.147617345071029e-06,
+      "loss": 0.4046,
+      "step": 1225
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.132734118541982e-06,
+      "loss": 0.5037,
+      "step": 1226
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.117855176396266e-06,
+      "loss": 0.4507,
+      "step": 1227
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.102980552773133e-06,
+      "loss": 0.4997,
+      "step": 1228
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.088110281801955e-06,
+      "loss": 0.3952,
+      "step": 1229
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.0732443976021e-06,
+      "loss": 0.4263,
+      "step": 1230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.058382934282877e-06,
+      "loss": 0.4339,
+      "step": 1231
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.043525925943452e-06,
+      "loss": 0.4357,
+      "step": 1232
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.028673406672764e-06,
+      "loss": 0.4593,
+      "step": 1233
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.013825410549456e-06,
+      "loss": 0.4582,
+      "step": 1234
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.998981971641791e-06,
+      "loss": 0.4449,
+      "step": 1235
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98414312400758e-06,
+      "loss": 0.4964,
+      "step": 1236
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.969308901694093e-06,
+      "loss": 0.4523,
+      "step": 1237
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.954479338737995e-06,
+      "loss": 0.4379,
+      "step": 1238
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.939654469165248e-06,
+      "loss": 0.4436,
+      "step": 1239
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.924834326991057e-06,
+      "loss": 0.401,
+      "step": 1240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.910018946219777e-06,
+      "loss": 0.4549,
+      "step": 1241
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.895208360844835e-06,
+      "loss": 0.4504,
+      "step": 1242
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.880402604848662e-06,
+      "loss": 0.4443,
+      "step": 1243
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.865601712202596e-06,
+      "loss": 0.4477,
+      "step": 1244
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.850805716866826e-06,
+      "loss": 0.4367,
+      "step": 1245
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.836014652790304e-06,
+      "loss": 0.4413,
+      "step": 1246
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.821228553910664e-06,
+      "loss": 0.4634,
+      "step": 1247
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.806447454154149e-06,
+      "loss": 0.4646,
+      "step": 1248
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.791671387435532e-06,
+      "loss": 0.4692,
+      "step": 1249
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.776900387658035e-06,
+      "loss": 0.4787,
+      "step": 1250
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.762134488713258e-06,
+      "loss": 0.4789,
+      "step": 1251
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.747373724481092e-06,
+      "loss": 0.453,
+      "step": 1252
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.732618128829656e-06,
+      "loss": 0.3994,
+      "step": 1253
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.717867735615203e-06,
+      "loss": 0.4349,
+      "step": 1254
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.703122578682047e-06,
+      "loss": 0.4392,
+      "step": 1255
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.688382691862491e-06,
+      "loss": 0.4626,
+      "step": 1256
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.673648108976745e-06,
+      "loss": 0.4587,
+      "step": 1257
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.658918863832854e-06,
+      "loss": 0.4621,
+      "step": 1258
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.644194990226604e-06,
+      "loss": 0.4472,
+      "step": 1259
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.629476521941477e-06,
+      "loss": 0.4199,
+      "step": 1260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.614763492748526e-06,
+      "loss": 0.4077,
+      "step": 1261
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6000559364063434e-06,
+      "loss": 0.4456,
+      "step": 1262
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.585353886660954e-06,
+      "loss": 0.4774,
+      "step": 1263
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.570657377245759e-06,
+      "loss": 0.4208,
+      "step": 1264
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.555966441881443e-06,
+      "loss": 0.4713,
+      "step": 1265
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.541281114275887e-06,
+      "loss": 0.4455,
+      "step": 1266
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.526601428124125e-06,
+      "loss": 0.419,
+      "step": 1267
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.511927417108239e-06,
+      "loss": 0.4496,
+      "step": 1268
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.497259114897288e-06,
+      "loss": 0.4745,
+      "step": 1269
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.482596555147234e-06,
+      "loss": 0.4346,
+      "step": 1270
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.467939771500867e-06,
+      "loss": 0.4639,
+      "step": 1271
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.453288797587714e-06,
+      "loss": 0.4094,
+      "step": 1272
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.438643667023979e-06,
+      "loss": 0.4461,
+      "step": 1273
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.42400441341246e-06,
+      "loss": 0.4668,
+      "step": 1274
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.409371070342466e-06,
+      "loss": 0.4447,
+      "step": 1275
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.394743671389747e-06,
+      "loss": 0.4011,
+      "step": 1276
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.380122250116412e-06,
+      "loss": 0.4327,
+      "step": 1277
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.365506840070856e-06,
+      "loss": 0.4667,
+      "step": 1278
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.350897474787684e-06,
+      "loss": 0.4884,
+      "step": 1279
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.336294187787625e-06,
+      "loss": 0.421,
+      "step": 1280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.321697012577469e-06,
+      "loss": 0.4197,
+      "step": 1281
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.307105982649982e-06,
+      "loss": 0.4348,
+      "step": 1282
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2925211314838216e-06,
+      "loss": 0.4728,
+      "step": 1283
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.277942492543473e-06,
+      "loss": 0.4643,
+      "step": 1284
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.263370099279173e-06,
+      "loss": 0.4872,
+      "step": 1285
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.24880398512682e-06,
+      "loss": 0.4223,
+      "step": 1286
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.234244183507912e-06,
+      "loss": 0.4581,
+      "step": 1287
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.219690727829458e-06,
+      "loss": 0.4126,
+      "step": 1288
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2051436514839064e-06,
+      "loss": 0.4046,
+      "step": 1289
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.190602987849073e-06,
+      "loss": 0.4386,
+      "step": 1290
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.176068770288054e-06,
+      "loss": 0.4236,
+      "step": 1291
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1615410321491605e-06,
+      "loss": 0.4571,
+      "step": 1292
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.147019806765836e-06,
+      "loss": 0.4536,
+      "step": 1293
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.132505127456575e-06,
+      "loss": 0.4627,
+      "step": 1294
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.117997027524859e-06,
+      "loss": 0.4354,
+      "step": 1295
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.10349554025907e-06,
+      "loss": 0.4669,
+      "step": 1296
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.089000698932416e-06,
+      "loss": 0.4671,
+      "step": 1297
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.074512536802869e-06,
+      "loss": 0.5102,
+      "step": 1298
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.060031087113051e-06,
+      "loss": 0.4664,
+      "step": 1299
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.045556383090202e-06,
+      "loss": 0.4365,
+      "step": 1300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0310884579460785e-06,
+      "loss": 0.4605,
+      "step": 1301
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0166273448768874e-06,
+      "loss": 0.4415,
+      "step": 1302
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.002173077063196e-06,
+      "loss": 0.4813,
+      "step": 1303
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.98772568766988e-06,
+      "loss": 0.4461,
+      "step": 1304
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.97328520984601e-06,
+      "loss": 0.4594,
+      "step": 1305
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.958851676724823e-06,
+      "loss": 0.4509,
+      "step": 1306
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9444251214236045e-06,
+      "loss": 0.4285,
+      "step": 1307
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.930005577043639e-06,
+      "loss": 0.4362,
+      "step": 1308
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.915593076670122e-06,
+      "loss": 0.4417,
+      "step": 1309
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.901187653372083e-06,
+      "loss": 0.4503,
+      "step": 1310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.886789340202318e-06,
+      "loss": 0.4524,
+      "step": 1311
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.872398170197305e-06,
+      "loss": 0.4573,
+      "step": 1312
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.858014176377139e-06,
+      "loss": 0.4124,
+      "step": 1313
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.843637391745444e-06,
+      "loss": 0.4785,
+      "step": 1314
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.829267849289303e-06,
+      "loss": 0.4491,
+      "step": 1315
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.814905581979181e-06,
+      "loss": 0.4534,
+      "step": 1316
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.800550622768856e-06,
+      "loss": 0.4458,
+      "step": 1317
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7862030045953284e-06,
+      "loss": 0.5001,
+      "step": 1318
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.771862760378768e-06,
+      "loss": 0.5796,
+      "step": 1319
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.757529923022419e-06,
+      "loss": 0.4636,
+      "step": 1320
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.74320452541252e-06,
+      "loss": 0.4654,
+      "step": 1321
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.728886600418253e-06,
+      "loss": 0.4909,
+      "step": 1322
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.714576180891653e-06,
+      "loss": 0.4695,
+      "step": 1323
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.70027329966753e-06,
+      "loss": 0.4286,
+      "step": 1324
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.685977989563399e-06,
+      "loss": 0.4467,
+      "step": 1325
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.671690283379404e-06,
+      "loss": 0.438,
+      "step": 1326
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657410213898239e-06,
+      "loss": 0.4032,
+      "step": 1327
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.643137813885079e-06,
+      "loss": 0.4001,
+      "step": 1328
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6288731160875e-06,
+      "loss": 0.4793,
+      "step": 1329
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6146161532354055e-06,
+      "loss": 0.4421,
+      "step": 1330
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.600366958040957e-06,
+      "loss": 0.451,
+      "step": 1331
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.586125563198482e-06,
+      "loss": 0.433,
+      "step": 1332
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.57189200138442e-06,
+      "loss": 0.4561,
+      "step": 1333
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.557666305257233e-06,
+      "loss": 0.4311,
+      "step": 1334
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.543448507457339e-06,
+      "loss": 0.4613,
+      "step": 1335
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.529238640607029e-06,
+      "loss": 0.457,
+      "step": 1336
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.515036737310407e-06,
+      "loss": 0.43,
+      "step": 1337
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.500842830153289e-06,
+      "loss": 0.5096,
+      "step": 1338
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4866569517031495e-06,
+      "loss": 0.4201,
+      "step": 1339
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.472479134509052e-06,
+      "loss": 0.4411,
+      "step": 1340
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.458309411101553e-06,
+      "loss": 0.4571,
+      "step": 1341
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.444147813992645e-06,
+      "loss": 0.44,
+      "step": 1342
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.429994375675661e-06,
+      "loss": 0.4576,
+      "step": 1343
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.415849128625231e-06,
+      "loss": 0.4721,
+      "step": 1344
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.401712105297181e-06,
+      "loss": 0.4311,
+      "step": 1345
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.387583338128471e-06,
+      "loss": 0.4625,
+      "step": 1346
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.373462859537116e-06,
+      "loss": 0.4238,
+      "step": 1347
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359350701922117e-06,
+      "loss": 0.4739,
+      "step": 1348
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.345246897663373e-06,
+      "loss": 0.4678,
+      "step": 1349
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.331151479121627e-06,
+      "loss": 0.45,
+      "step": 1350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.317064478638376e-06,
+      "loss": 0.3998,
+      "step": 1351
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3029859285358045e-06,
+      "loss": 0.4337,
+      "step": 1352
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.288915861116707e-06,
+      "loss": 0.4154,
+      "step": 1353
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.274854308664411e-06,
+      "loss": 0.5063,
+      "step": 1354
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.260801303442713e-06,
+      "loss": 0.4964,
+      "step": 1355
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.246756877695795e-06,
+      "loss": 0.5134,
+      "step": 1356
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.232721063648148e-06,
+      "loss": 0.4749,
+      "step": 1357
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.218693893504519e-06,
+      "loss": 0.4288,
+      "step": 1358
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.204675399449812e-06,
+      "loss": 0.3662,
+      "step": 1359
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.190665613649014e-06,
+      "loss": 0.439,
+      "step": 1360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1766645682471505e-06,
+      "loss": 0.4569,
+      "step": 1361
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.16267229536918e-06,
+      "loss": 0.4437,
+      "step": 1362
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.148688827119938e-06,
+      "loss": 0.4213,
+      "step": 1363
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.134714195584055e-06,
+      "loss": 0.4303,
+      "step": 1364
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1207484328258885e-06,
+      "loss": 0.4558,
+      "step": 1365
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.106791570889443e-06,
+      "loss": 0.4213,
+      "step": 1366
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.092843641798305e-06,
+      "loss": 0.4293,
+      "step": 1367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0789046775555616e-06,
+      "loss": 0.4554,
+      "step": 1368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.064974710143732e-06,
+      "loss": 0.4414,
+      "step": 1369
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.051053771524694e-06,
+      "loss": 0.3927,
+      "step": 1370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.037141893639604e-06,
+      "loss": 0.4623,
+      "step": 1371
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.023239108408832e-06,
+      "loss": 0.4312,
+      "step": 1372
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.009345447731886e-06,
+      "loss": 0.4436,
+      "step": 1373
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.995460943487334e-06,
+      "loss": 0.4255,
+      "step": 1374
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.981585627532744e-06,
+      "loss": 0.4523,
+      "step": 1375
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.967719531704597e-06,
+      "loss": 0.4005,
+      "step": 1376
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.953862687818206e-06,
+      "loss": 0.4047,
+      "step": 1377
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.940015127667677e-06,
+      "loss": 0.3678,
+      "step": 1378
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9261768830258025e-06,
+      "loss": 0.4082,
+      "step": 1379
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.912347985644002e-06,
+      "loss": 0.5143,
+      "step": 1380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.898528467252251e-06,
+      "loss": 0.4647,
+      "step": 1381
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.884718359559001e-06,
+      "loss": 0.4045,
+      "step": 1382
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.870917694251114e-06,
+      "loss": 0.4493,
+      "step": 1383
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8571265029937865e-06,
+      "loss": 0.4124,
+      "step": 1384
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843344817430477e-06,
+      "loss": 0.4412,
+      "step": 1385
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8295726691828345e-06,
+      "loss": 0.4233,
+      "step": 1386
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.815810089850622e-06,
+      "loss": 0.4878,
+      "step": 1387
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.802057111011645e-06,
+      "loss": 0.4613,
+      "step": 1388
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788313764221694e-06,
+      "loss": 0.3939,
+      "step": 1389
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.774580081014438e-06,
+      "loss": 0.4686,
+      "step": 1390
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.760856092901394e-06,
+      "loss": 0.3953,
+      "step": 1391
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.747141831371822e-06,
+      "loss": 0.4118,
+      "step": 1392
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.733437327892661e-06,
+      "loss": 0.4521,
+      "step": 1393
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7197426139084775e-06,
+      "loss": 0.4656,
+      "step": 1394
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7060577208413555e-06,
+      "loss": 0.4227,
+      "step": 1395
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.692382680090864e-06,
+      "loss": 0.3981,
+      "step": 1396
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.67871752303395e-06,
+      "loss": 0.4186,
+      "step": 1397
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.665062281024898e-06,
+      "loss": 0.4378,
+      "step": 1398
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651416985395228e-06,
+      "loss": 0.4345,
+      "step": 1399
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6377816674536456e-06,
+      "loss": 0.4351,
+      "step": 1400
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.624156358485968e-06,
+      "loss": 0.4397,
+      "step": 1401
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.610541089755035e-06,
+      "loss": 0.3898,
+      "step": 1402
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.596935892500663e-06,
+      "loss": 0.4515,
+      "step": 1403
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5833407979395485e-06,
+      "loss": 0.4423,
+      "step": 1404
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.569755837265208e-06,
+      "loss": 0.4438,
+      "step": 1405
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5561810416479146e-06,
+      "loss": 0.4556,
+      "step": 1406
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.542616442234618e-06,
+      "loss": 0.4232,
+      "step": 1407
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.529062070148859e-06,
+      "loss": 0.4272,
+      "step": 1408
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.515517956490731e-06,
+      "loss": 0.4017,
+      "step": 1409
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.501984132336775e-06,
+      "loss": 0.4342,
+      "step": 1410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.488460628739926e-06,
+      "loss": 0.4452,
+      "step": 1411
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.474947476729447e-06,
+      "loss": 0.4335,
+      "step": 1412
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.461444707310838e-06,
+      "loss": 0.4564,
+      "step": 1413
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.447952351465788e-06,
+      "loss": 0.4082,
+      "step": 1414
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4344704401520845e-06,
+      "loss": 0.4808,
+      "step": 1415
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.420999004303549e-06,
+      "loss": 0.4062,
+      "step": 1416
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.407538074829976e-06,
+      "loss": 0.3874,
+      "step": 1417
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.394087682617043e-06,
+      "loss": 0.3978,
+      "step": 1418
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.380647858526258e-06,
+      "loss": 0.3859,
+      "step": 1419
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.367218633394887e-06,
+      "loss": 0.443,
+      "step": 1420
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.353800038035852e-06,
+      "loss": 0.3835,
+      "step": 1421
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.340392103237713e-06,
+      "loss": 0.4383,
+      "step": 1422
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3269948597645515e-06,
+      "loss": 0.4262,
+      "step": 1423
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3136083383559264e-06,
+      "loss": 0.4129,
+      "step": 1424
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.300232569726805e-06,
+      "loss": 0.4797,
+      "step": 1425
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.28686758456745e-06,
+      "loss": 0.4084,
+      "step": 1426
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.273513413543416e-06,
+      "loss": 0.4063,
+      "step": 1427
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2601700872954346e-06,
+      "loss": 0.1919,
+      "step": 1428
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246837636439342e-06,
+      "loss": 0.2393,
+      "step": 1429
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.23351609156604e-06,
+      "loss": 0.1775,
+      "step": 1430
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.220205483241394e-06,
+      "loss": 0.1895,
+      "step": 1431
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.206905842006176e-06,
+      "loss": 0.1826,
+      "step": 1432
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.193617198376004e-06,
+      "loss": 0.1762,
+      "step": 1433
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.18033958284125e-06,
+      "loss": 0.1759,
+      "step": 1434
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1670730258669974e-06,
+      "loss": 0.1662,
+      "step": 1435
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.153817557892942e-06,
+      "loss": 0.1681,
+      "step": 1436
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.140573209333337e-06,
+      "loss": 0.1731,
+      "step": 1437
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.127340010576938e-06,
+      "loss": 0.1608,
+      "step": 1438
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.114117991986895e-06,
+      "loss": 0.17,
+      "step": 1439
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.100907183900723e-06,
+      "loss": 0.1562,
+      "step": 1440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.087707616630213e-06,
+      "loss": 0.1672,
+      "step": 1441
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.074519320461358e-06,
+      "loss": 0.1578,
+      "step": 1442
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.06134232565429e-06,
+      "loss": 0.1644,
+      "step": 1443
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.04817666244321e-06,
+      "loss": 0.1602,
+      "step": 1444
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0350223610363235e-06,
+      "loss": 0.1653,
+      "step": 1445
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0218794516157695e-06,
+      "loss": 0.1878,
+      "step": 1446
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0087479643375426e-06,
+      "loss": 0.1357,
+      "step": 1447
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.995627929331424e-06,
+      "loss": 0.1269,
+      "step": 1448
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.982519376700933e-06,
+      "loss": 0.1469,
+      "step": 1449
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.969422336523227e-06,
+      "loss": 0.1887,
+      "step": 1450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9563368388490644e-06,
+      "loss": 0.1385,
+      "step": 1451
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.943262913702703e-06,
+      "loss": 0.1453,
+      "step": 1452
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.930200591081865e-06,
+      "loss": 0.1246,
+      "step": 1453
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.917149900957638e-06,
+      "loss": 0.1664,
+      "step": 1454
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.904110873274419e-06,
+      "loss": 0.2001,
+      "step": 1455
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.891083537949859e-06,
+      "loss": 0.1499,
+      "step": 1456
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.878067924874763e-06,
+      "loss": 0.1615,
+      "step": 1457
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.86506406391306e-06,
+      "loss": 0.1222,
+      "step": 1458
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.852071984901696e-06,
+      "loss": 0.1443,
+      "step": 1459
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.83909171765059e-06,
+      "loss": 0.1713,
+      "step": 1460
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.826123291942562e-06,
+      "loss": 0.1555,
+      "step": 1461
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.813166737533267e-06,
+      "loss": 0.1413,
+      "step": 1462
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.800222084151105e-06,
+      "loss": 0.1405,
+      "step": 1463
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.78728936149719e-06,
+      "loss": 0.1416,
+      "step": 1464
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.774368599245239e-06,
+      "loss": 0.131,
+      "step": 1465
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.761459827041542e-06,
+      "loss": 0.1437,
+      "step": 1466
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.74856307450488e-06,
+      "loss": 0.1428,
+      "step": 1467
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7356783712264405e-06,
+      "loss": 0.1541,
+      "step": 1468
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.722805746769783e-06,
+      "loss": 0.1211,
+      "step": 1469
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.709945230670737e-06,
+      "loss": 0.1673,
+      "step": 1470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697096852437353e-06,
+      "loss": 0.1438,
+      "step": 1471
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.684260641549841e-06,
+      "loss": 0.1251,
+      "step": 1472
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.671436627460479e-06,
+      "loss": 0.124,
+      "step": 1473
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658624839593575e-06,
+      "loss": 0.1546,
+      "step": 1474
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.645825307345373e-06,
+      "loss": 0.1467,
+      "step": 1475
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.633038060083996e-06,
+      "loss": 0.1547,
+      "step": 1476
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.620263127149392e-06,
+      "loss": 0.1559,
+      "step": 1477
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.60750053785324e-06,
+      "loss": 0.1593,
+      "step": 1478
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.594750321478905e-06,
+      "loss": 0.1505,
+      "step": 1479
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.582012507281368e-06,
+      "loss": 0.1321,
+      "step": 1480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.569287124487134e-06,
+      "loss": 0.1239,
+      "step": 1481
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.556574202294202e-06,
+      "loss": 0.1796,
+      "step": 1482
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.543873769871978e-06,
+      "loss": 0.131,
+      "step": 1483
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531185856361204e-06,
+      "loss": 0.1584,
+      "step": 1484
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518510490873907e-06,
+      "loss": 0.1341,
+      "step": 1485
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.505847702493316e-06,
+      "loss": 0.1642,
+      "step": 1486
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.493197520273797e-06,
+      "loss": 0.1394,
+      "step": 1487
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.480559973240809e-06,
+      "loss": 0.1581,
+      "step": 1488
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4679350903908016e-06,
+      "loss": 0.1671,
+      "step": 1489
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4553229006911806e-06,
+      "loss": 0.2066,
+      "step": 1490
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.442723433080218e-06,
+      "loss": 0.1651,
+      "step": 1491
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.430136716466998e-06,
+      "loss": 0.1507,
+      "step": 1492
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.417562779731355e-06,
+      "loss": 0.1644,
+      "step": 1493
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.405001651723787e-06,
+      "loss": 0.1535,
+      "step": 1494
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.39245336126542e-06,
+      "loss": 0.1432,
+      "step": 1495
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.379917937147904e-06,
+      "loss": 0.1764,
+      "step": 1496
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.367395408133388e-06,
+      "loss": 0.1436,
+      "step": 1497
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.354885802954422e-06,
+      "loss": 0.1407,
+      "step": 1498
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.342389150313899e-06,
+      "loss": 0.1588,
+      "step": 1499
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3299054788850025e-06,
+      "loss": 0.1583,
+      "step": 1500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.317434817311132e-06,
+      "loss": 0.1505,
+      "step": 1501
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3049771942058285e-06,
+      "loss": 0.1472,
+      "step": 1502
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.292532638152713e-06,
+      "loss": 0.1699,
+      "step": 1503
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.280101177705442e-06,
+      "loss": 0.1504,
+      "step": 1504
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.267682841387604e-06,
+      "loss": 0.1799,
+      "step": 1505
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.255277657692692e-06,
+      "loss": 0.162,
+      "step": 1506
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.242885655084007e-06,
+      "loss": 0.166,
+      "step": 1507
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.230506861994619e-06,
+      "loss": 0.1432,
+      "step": 1508
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.218141306827279e-06,
+      "loss": 0.1358,
+      "step": 1509
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.205789017954364e-06,
+      "loss": 0.1511,
+      "step": 1510
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1934500237178226e-06,
+      "loss": 0.1384,
+      "step": 1511
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.181124352429085e-06,
+      "loss": 0.1762,
+      "step": 1512
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.168812032369026e-06,
+      "loss": 0.1595,
+      "step": 1513
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.156513091787878e-06,
+      "loss": 0.1891,
+      "step": 1514
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144227558905173e-06,
+      "loss": 0.1948,
+      "step": 1515
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.131955461909688e-06,
+      "loss": 0.1611,
+      "step": 1516
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.119696828959361e-06,
+      "loss": 0.1629,
+      "step": 1517
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1074516881812466e-06,
+      "loss": 0.1622,
+      "step": 1518
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.095220067671446e-06,
+      "loss": 0.1555,
+      "step": 1519
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0830019954950165e-06,
+      "loss": 0.1441,
+      "step": 1520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.070797499685949e-06,
+      "loss": 0.1516,
+      "step": 1521
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0586066082470836e-06,
+      "loss": 0.1283,
+      "step": 1522
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.046429349150033e-06,
+      "loss": 0.1691,
+      "step": 1523
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034265750335146e-06,
+      "loss": 0.156,
+      "step": 1524
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.022115839711417e-06,
+      "loss": 0.169,
+      "step": 1525
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.009979645156432e-06,
+      "loss": 0.1409,
+      "step": 1526
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.997857194516319e-06,
+      "loss": 0.1395,
+      "step": 1527
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.985748515605657e-06,
+      "loss": 0.1635,
+      "step": 1528
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.973653636207437e-06,
+      "loss": 0.1983,
+      "step": 1529
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.961572584072981e-06,
+      "loss": 0.1671,
+      "step": 1530
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.949505386921882e-06,
+      "loss": 0.1766,
+      "step": 1531
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.937452072441953e-06,
+      "loss": 0.1235,
+      "step": 1532
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.925412668289143e-06,
+      "loss": 0.1434,
+      "step": 1533
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9133872020874905e-06,
+      "loss": 0.1822,
+      "step": 1534
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.90137570142906e-06,
+      "loss": 0.1096,
+      "step": 1535
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.889378193873858e-06,
+      "loss": 0.151,
+      "step": 1536
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8773947069497946e-06,
+      "loss": 0.152,
+      "step": 1537
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8654252681526e-06,
+      "loss": 0.1841,
+      "step": 1538
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.853469904945782e-06,
+      "loss": 0.1256,
+      "step": 1539
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.841528644760555e-06,
+      "loss": 0.1837,
+      "step": 1540
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8296015149957625e-06,
+      "loss": 0.187,
+      "step": 1541
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.817688543017828e-06,
+      "loss": 0.1748,
+      "step": 1542
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8057897561607014e-06,
+      "loss": 0.1739,
+      "step": 1543
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.793905181725772e-06,
+      "loss": 0.1447,
+      "step": 1544
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.78203484698183e-06,
+      "loss": 0.1384,
+      "step": 1545
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7701787791649835e-06,
+      "loss": 0.2001,
+      "step": 1546
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.758337005478616e-06,
+      "loss": 0.1539,
+      "step": 1547
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.746509553093306e-06,
+      "loss": 0.1534,
+      "step": 1548
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7346964491467672e-06,
+      "loss": 0.1351,
+      "step": 1549
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.722897720743809e-06,
+      "loss": 0.1331,
+      "step": 1550
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7111133949562363e-06,
+      "loss": 0.166,
+      "step": 1551
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.699343498822824e-06,
+      "loss": 0.1502,
+      "step": 1552
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6875880593492296e-06,
+      "loss": 0.1467,
+      "step": 1553
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.675847103507938e-06,
+      "loss": 0.162,
+      "step": 1554
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6641206582382094e-06,
+      "loss": 0.1568,
+      "step": 1555
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6524087504460104e-06,
+      "loss": 0.1595,
+      "step": 1556
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.640711407003942e-06,
+      "loss": 0.1434,
+      "step": 1557
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6290286547512033e-06,
+      "loss": 0.1407,
+      "step": 1558
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6173605204934914e-06,
+      "loss": 0.1425,
+      "step": 1559
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.605707031002983e-06,
+      "loss": 0.1875,
+      "step": 1560
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.594068213018249e-06,
+      "loss": 0.1452,
+      "step": 1561
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5824440932441894e-06,
+      "loss": 0.1518,
+      "step": 1562
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5708346983519905e-06,
+      "loss": 0.1592,
+      "step": 1563
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.559240054979043e-06,
+      "loss": 0.1394,
+      "step": 1564
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5476601897288907e-06,
+      "loss": 0.1442,
+      "step": 1565
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.536095129171181e-06,
+      "loss": 0.1708,
+      "step": 1566
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.524544899841578e-06,
+      "loss": 0.1897,
+      "step": 1567
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5130095282417274e-06,
+      "loss": 0.1547,
+      "step": 1568
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.501489040839178e-06,
+      "loss": 0.1381,
+      "step": 1569
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4899834640673245e-06,
+      "loss": 0.1438,
+      "step": 1570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4784928243253615e-06,
+      "loss": 0.1443,
+      "step": 1571
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4670171479781965e-06,
+      "loss": 0.1286,
+      "step": 1572
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.455556461356413e-06,
+      "loss": 0.1842,
+      "step": 1573
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.444110790756208e-06,
+      "loss": 0.1338,
+      "step": 1574
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4326801624393014e-06,
+      "loss": 0.1759,
+      "step": 1575
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.421264602632919e-06,
+      "loss": 0.1587,
+      "step": 1576
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4098641375297103e-06,
+      "loss": 0.1436,
+      "step": 1577
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.398478793287682e-06,
+      "loss": 0.1295,
+      "step": 1578
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.387108596030155e-06,
+      "loss": 0.1415,
+      "step": 1579
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3757535718456923e-06,
+      "loss": 0.163,
+      "step": 1580
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.364413746788038e-06,
+      "loss": 0.1727,
+      "step": 1581
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3530891468760717e-06,
+      "loss": 0.1524,
+      "step": 1582
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3417797980937307e-06,
+      "loss": 0.1547,
+      "step": 1583
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3304857263899683e-06,
+      "loss": 0.1533,
+      "step": 1584
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3192069576786767e-06,
+      "loss": 0.1331,
+      "step": 1585
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3079435178386366e-06,
+      "loss": 0.158,
+      "step": 1586
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2966954327134636e-06,
+      "loss": 0.1516,
+      "step": 1587
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.285462728111535e-06,
+      "loss": 0.1307,
+      "step": 1588
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2742454298059445e-06,
+      "loss": 0.1706,
+      "step": 1589
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2630435635344283e-06,
+      "loss": 0.152,
+      "step": 1590
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2518571549993248e-06,
+      "loss": 0.1654,
+      "step": 1591
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.240686229867497e-06,
+      "loss": 0.137,
+      "step": 1592
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2295308137702808e-06,
+      "loss": 0.1575,
+      "step": 1593
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2183909323034336e-06,
+      "loss": 0.1508,
+      "step": 1594
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.207266611027069e-06,
+      "loss": 0.1447,
+      "step": 1595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.196157875465593e-06,
+      "loss": 0.1443,
+      "step": 1596
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.185064751107648e-06,
+      "loss": 0.1584,
+      "step": 1597
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1739872634060685e-06,
+      "loss": 0.1682,
+      "step": 1598
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1629254377777995e-06,
+      "loss": 0.1572,
+      "step": 1599
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1518792996038606e-06,
+      "loss": 0.1454,
+      "step": 1600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1408488742292644e-06,
+      "loss": 0.158,
+      "step": 1601
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1298341869629844e-06,
+      "loss": 0.157,
+      "step": 1602
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.118835263077874e-06,
+      "loss": 0.1139,
+      "step": 1603
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1078521278106177e-06,
+      "loss": 0.1292,
+      "step": 1604
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.096884806361684e-06,
+      "loss": 0.151,
+      "step": 1605
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0859333238952417e-06,
+      "loss": 0.1736,
+      "step": 1606
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0749977055391333e-06,
+      "loss": 0.1487,
+      "step": 1607
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0640779763847916e-06,
+      "loss": 0.1625,
+      "step": 1608
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0531741614871892e-06,
+      "loss": 0.1699,
+      "step": 1609
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0422862858647984e-06,
+      "loss": 0.1605,
+      "step": 1610
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0314143744995006e-06,
+      "loss": 0.1694,
+      "step": 1611
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0205584523365626e-06,
+      "loss": 0.1847,
+      "step": 1612
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0097185442845655e-06,
+      "loss": 0.1828,
+      "step": 1613
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.998894675215328e-06,
+      "loss": 0.1432,
+      "step": 1614
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9880868699638845e-06,
+      "loss": 0.1739,
+      "step": 1615
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9772951533284124e-06,
+      "loss": 0.1362,
+      "step": 1616
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9665195500701605e-06,
+      "loss": 0.1411,
+      "step": 1617
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9557600849134215e-06,
+      "loss": 0.1627,
+      "step": 1618
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.94501678254545e-06,
+      "loss": 0.1265,
+      "step": 1619
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9342896676164146e-06,
+      "loss": 0.1352,
+      "step": 1620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9235787647393533e-06,
+      "loss": 0.1882,
+      "step": 1621
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9128840984900897e-06,
+      "loss": 0.1891,
+      "step": 1622
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.902205693407211e-06,
+      "loss": 0.1695,
+      "step": 1623
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8915435739919795e-06,
+      "loss": 0.1428,
+      "step": 1624
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8808977647082946e-06,
+      "loss": 0.1909,
+      "step": 1625
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8702682899826373e-06,
+      "loss": 0.1496,
+      "step": 1626
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8596551742040013e-06,
+      "loss": 0.1395,
+      "step": 1627
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.849058441723853e-06,
+      "loss": 0.1502,
+      "step": 1628
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8384781168560693e-06,
+      "loss": 0.1571,
+      "step": 1629
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.827914223876865e-06,
+      "loss": 0.1514,
+      "step": 1630
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.817366787024768e-06,
+      "loss": 0.144,
+      "step": 1631
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.806835830500546e-06,
+      "loss": 0.1451,
+      "step": 1632
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.796321378467146e-06,
+      "loss": 0.1519,
+      "step": 1633
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7858234550496545e-06,
+      "loss": 0.1383,
+      "step": 1634
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7753420843352265e-06,
+      "loss": 0.1388,
+      "step": 1635
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.764877290373037e-06,
+      "loss": 0.1754,
+      "step": 1636
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.754429097174236e-06,
+      "loss": 0.1575,
+      "step": 1637
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7439975287118713e-06,
+      "loss": 0.1908,
+      "step": 1638
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7335826089208574e-06,
+      "loss": 0.1395,
+      "step": 1639
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.723184361697898e-06,
+      "loss": 0.1634,
+      "step": 1640
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.712802810901446e-06,
+      "loss": 0.1228,
+      "step": 1641
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.702437980351651e-06,
+      "loss": 0.1356,
+      "step": 1642
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6920898938302885e-06,
+      "loss": 0.1443,
+      "step": 1643
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.681758575080724e-06,
+      "loss": 0.1705,
+      "step": 1644
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.671444047807843e-06,
+      "loss": 0.1594,
+      "step": 1645
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.66114633567801e-06,
+      "loss": 0.1246,
+      "step": 1646
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6508654623190034e-06,
+      "loss": 0.1364,
+      "step": 1647
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.640601451319961e-06,
+      "loss": 0.1753,
+      "step": 1648
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.630354326231341e-06,
+      "loss": 0.1741,
+      "step": 1649
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6201241105648555e-06,
+      "loss": 0.1623,
+      "step": 1650
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6099108277934105e-06,
+      "loss": 0.1442,
+      "step": 1651
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.599714501351063e-06,
+      "loss": 0.1377,
+      "step": 1652
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.589535154632972e-06,
+      "loss": 0.143,
+      "step": 1653
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5793728109953244e-06,
+      "loss": 0.1399,
+      "step": 1654
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5692274937553065e-06,
+      "loss": 0.1444,
+      "step": 1655
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5590992261910264e-06,
+      "loss": 0.189,
+      "step": 1656
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.548988031541485e-06,
+      "loss": 0.1389,
+      "step": 1657
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5388939330065e-06,
+      "loss": 0.1579,
+      "step": 1658
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5288169537466623e-06,
+      "loss": 0.1455,
+      "step": 1659
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5187571168832913e-06,
+      "loss": 0.1544,
+      "step": 1660
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5087144454983645e-06,
+      "loss": 0.1295,
+      "step": 1661
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4986889626344835e-06,
+      "loss": 0.142,
+      "step": 1662
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4886806912948034e-06,
+      "loss": 0.148,
+      "step": 1663
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4786896544429873e-06,
+      "loss": 0.1516,
+      "step": 1664
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4687158750031635e-06,
+      "loss": 0.1655,
+      "step": 1665
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4587593758598517e-06,
+      "loss": 0.1401,
+      "step": 1666
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4488201798579323e-06,
+      "loss": 0.1246,
+      "step": 1667
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4388983098025863e-06,
+      "loss": 0.151,
+      "step": 1668
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.428993788459223e-06,
+      "loss": 0.1582,
+      "step": 1669
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4191066385534636e-06,
+      "loss": 0.1477,
+      "step": 1670
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4092368827710675e-06,
+      "loss": 0.1328,
+      "step": 1671
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3993845437578766e-06,
+      "loss": 0.1378,
+      "step": 1672
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.389549644119781e-06,
+      "loss": 0.1687,
+      "step": 1673
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.379732206422648e-06,
+      "loss": 0.1556,
+      "step": 1674
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.36993225319228e-06,
+      "loss": 0.1727,
+      "step": 1675
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.360149806914368e-06,
+      "loss": 0.1653,
+      "step": 1676
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.350384890034426e-06,
+      "loss": 0.1735,
+      "step": 1677
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.340637524957756e-06,
+      "loss": 0.1386,
+      "step": 1678
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3309077340493815e-06,
+      "loss": 0.1526,
+      "step": 1679
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3211955396340003e-06,
+      "loss": 0.1602,
+      "step": 1680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.311500963995945e-06,
+      "loss": 0.1413,
+      "step": 1681
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3018240293791116e-06,
+      "loss": 0.1287,
+      "step": 1682
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.292164757986928e-06,
+      "loss": 0.1349,
+      "step": 1683
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.282523171982297e-06,
+      "loss": 0.1489,
+      "step": 1684
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.272899293487524e-06,
+      "loss": 0.1365,
+      "step": 1685
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.263293144584309e-06,
+      "loss": 0.1905,
+      "step": 1686
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.253704747313653e-06,
+      "loss": 0.1653,
+      "step": 1687
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2441341236758383e-06,
+      "loss": 0.1313,
+      "step": 1688
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.234581295630366e-06,
+      "loss": 0.1462,
+      "step": 1689
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2250462850958976e-06,
+      "loss": 0.1286,
+      "step": 1690
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2155291139502166e-06,
+      "loss": 0.1288,
+      "step": 1691
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2060298040301788e-06,
+      "loss": 0.1579,
+      "step": 1692
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1965483771316498e-06,
+      "loss": 0.1899,
+      "step": 1693
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1870848550094736e-06,
+      "loss": 0.1353,
+      "step": 1694
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1776392593774043e-06,
+      "loss": 0.1412,
+      "step": 1695
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1682116119080622e-06,
+      "loss": 0.153,
+      "step": 1696
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.158801934232897e-06,
+      "loss": 0.1313,
+      "step": 1697
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1494102479421155e-06,
+      "loss": 0.1392,
+      "step": 1698
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.140036574584653e-06,
+      "loss": 0.175,
+      "step": 1699
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1306809356681067e-06,
+      "loss": 0.162,
+      "step": 1700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1213433526587024e-06,
+      "loss": 0.1589,
+      "step": 1701
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1120238469812315e-06,
+      "loss": 0.1373,
+      "step": 1702
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.102722440019006e-06,
+      "loss": 0.1522,
+      "step": 1703
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0934391531138156e-06,
+      "loss": 0.1463,
+      "step": 1704
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0841740075658766e-06,
+      "loss": 0.1658,
+      "step": 1705
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0749270246337738e-06,
+      "loss": 0.1395,
+      "step": 1706
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0656982255344204e-06,
+      "loss": 0.1851,
+      "step": 1707
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.056487631443006e-06,
+      "loss": 0.161,
+      "step": 1708
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0472952634929543e-06,
+      "loss": 0.1373,
+      "step": 1709
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0381211427758685e-06,
+      "loss": 0.1329,
+      "step": 1710
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.028965290341478e-06,
+      "loss": 0.1722,
+      "step": 1711
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.019827727197605e-06,
+      "loss": 0.1479,
+      "step": 1712
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0107084743101023e-06,
+      "loss": 0.133,
+      "step": 1713
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0016075526028066e-06,
+      "loss": 0.1449,
+      "step": 1714
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9925249829575044e-06,
+      "loss": 0.1446,
+      "step": 1715
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9834607862138644e-06,
+      "loss": 0.139,
+      "step": 1716
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.974414983169408e-06,
+      "loss": 0.1578,
+      "step": 1717
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9653875945794455e-06,
+      "loss": 0.1393,
+      "step": 1718
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.956378641157035e-06,
+      "loss": 0.1264,
+      "step": 1719
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9473881435729446e-06,
+      "loss": 0.1392,
+      "step": 1720
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9384161224555874e-06,
+      "loss": 0.1286,
+      "step": 1721
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9294625983909866e-06,
+      "loss": 0.1137,
+      "step": 1722
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9205275919227284e-06,
+      "loss": 0.1146,
+      "step": 1723
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9116111235518986e-06,
+      "loss": 0.1754,
+      "step": 1724
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9027132137370586e-06,
+      "loss": 0.1672,
+      "step": 1725
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8938338828941892e-06,
+      "loss": 0.1391,
+      "step": 1726
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.884973151396632e-06,
+      "loss": 0.1539,
+      "step": 1727
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8761310395750642e-06,
+      "loss": 0.1571,
+      "step": 1728
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8673075677174334e-06,
+      "loss": 0.136,
+      "step": 1729
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8585027560689173e-06,
+      "loss": 0.1578,
+      "step": 1730
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8497166248318876e-06,
+      "loss": 0.1639,
+      "step": 1731
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.840949194165843e-06,
+      "loss": 0.1466,
+      "step": 1732
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.832200484187384e-06,
+      "loss": 0.1526,
+      "step": 1733
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8234705149701515e-06,
+      "loss": 0.1542,
+      "step": 1734
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8147593065447844e-06,
+      "loss": 0.1572,
+      "step": 1735
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8060668788988833e-06,
+      "loss": 0.1578,
+      "step": 1736
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7973932519769476e-06,
+      "loss": 0.1159,
+      "step": 1737
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7887384456803503e-06,
+      "loss": 0.1401,
+      "step": 1738
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7801024798672672e-06,
+      "loss": 0.1462,
+      "step": 1739
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7714853743526583e-06,
+      "loss": 0.1439,
+      "step": 1740
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.762887148908201e-06,
+      "loss": 0.1462,
+      "step": 1741
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7543078232622524e-06,
+      "loss": 0.1376,
+      "step": 1742
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7457474170998112e-06,
+      "loss": 0.1445,
+      "step": 1743
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.737205950062465e-06,
+      "loss": 0.1406,
+      "step": 1744
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7286834417483388e-06,
+      "loss": 0.1176,
+      "step": 1745
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7201799117120622e-06,
+      "loss": 0.1634,
+      "step": 1746
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7116953794647229e-06,
+      "loss": 0.1648,
+      "step": 1747
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.703229864473811e-06,
+      "loss": 0.1347,
+      "step": 1748
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6947833861631912e-06,
+      "loss": 0.1538,
+      "step": 1749
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.686355963913041e-06,
+      "loss": 0.1626,
+      "step": 1750
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6779476170598207e-06,
+      "loss": 0.1383,
+      "step": 1751
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.66955836489622e-06,
+      "loss": 0.1378,
+      "step": 1752
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.661188226671111e-06,
+      "loss": 0.1342,
+      "step": 1753
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6528372215895228e-06,
+      "loss": 0.1843,
+      "step": 1754
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.644505368812568e-06,
+      "loss": 0.1614,
+      "step": 1755
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6361926874574297e-06,
+      "loss": 0.1572,
+      "step": 1756
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6278991965972935e-06,
+      "loss": 0.153,
+      "step": 1757
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.619624915261313e-06,
+      "loss": 0.1249,
+      "step": 1758
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.611369862434572e-06,
+      "loss": 0.132,
+      "step": 1759
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6031340570580289e-06,
+      "loss": 0.1361,
+      "step": 1760
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5949175180284838e-06,
+      "loss": 0.1307,
+      "step": 1761
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5867202641985346e-06,
+      "loss": 0.1372,
+      "step": 1762
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5785423143765144e-06,
+      "loss": 0.1731,
+      "step": 1763
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5703836873264777e-06,
+      "loss": 0.1291,
+      "step": 1764
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5622444017681438e-06,
+      "loss": 0.1596,
+      "step": 1765
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5541244763768449e-06,
+      "loss": 0.164,
+      "step": 1766
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5460239297835e-06,
+      "loss": 0.1375,
+      "step": 1767
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5379427805745595e-06,
+      "loss": 0.1285,
+      "step": 1768
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5298810472919646e-06,
+      "loss": 0.1732,
+      "step": 1769
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5218387484331165e-06,
+      "loss": 0.157,
+      "step": 1770
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5138159024508137e-06,
+      "loss": 0.1583,
+      "step": 1771
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.505812527753231e-06,
+      "loss": 0.1319,
+      "step": 1772
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4978286427038602e-06,
+      "loss": 0.1387,
+      "step": 1773
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4898642656214734e-06,
+      "loss": 0.1285,
+      "step": 1774
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4819194147800907e-06,
+      "loss": 0.138,
+      "step": 1775
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4739941084089204e-06,
+      "loss": 0.1372,
+      "step": 1776
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4660883646923308e-06,
+      "loss": 0.1407,
+      "step": 1777
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4582022017698116e-06,
+      "loss": 0.1563,
+      "step": 1778
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4503356377359068e-06,
+      "loss": 0.1416,
+      "step": 1779
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4424886906402091e-06,
+      "loss": 0.1623,
+      "step": 1780
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.434661378487291e-06,
+      "loss": 0.1436,
+      "step": 1781
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.426853719236676e-06,
+      "loss": 0.1331,
+      "step": 1782
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4190657308027989e-06,
+      "loss": 0.1575,
+      "step": 1783
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4112974310549522e-06,
+      "loss": 0.1679,
+      "step": 1784
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4035488378172568e-06,
+      "loss": 0.1241,
+      "step": 1785
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3958199688686213e-06,
+      "loss": 0.1143,
+      "step": 1786
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3881108419426915e-06,
+      "loss": 0.14,
+      "step": 1787
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.380421474727821e-06,
+      "loss": 0.1755,
+      "step": 1788
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3727518848670229e-06,
+      "loss": 0.1532,
+      "step": 1789
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3651020899579292e-06,
+      "loss": 0.1869,
+      "step": 1790
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3574721075527585e-06,
+      "loss": 0.1592,
+      "step": 1791
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.349861955158266e-06,
+      "loss": 0.1189,
+      "step": 1792
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3422716502357103e-06,
+      "loss": 0.1617,
+      "step": 1793
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3347012102008072e-06,
+      "loss": 0.1265,
+      "step": 1794
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3271506524236987e-06,
+      "loss": 0.126,
+      "step": 1795
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3196199942289023e-06,
+      "loss": 0.1453,
+      "step": 1796
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.312109252895275e-06,
+      "loss": 0.1263,
+      "step": 1797
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3046184456559818e-06,
+      "loss": 0.1654,
+      "step": 1798
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2971475896984475e-06,
+      "loss": 0.1485,
+      "step": 1799
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.289696702164318e-06,
+      "loss": 0.1726,
+      "step": 1800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2822658001494204e-06,
+      "loss": 0.1547,
+      "step": 1801
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2748549007037247e-06,
+      "loss": 0.1352,
+      "step": 1802
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2674640208313138e-06,
+      "loss": 0.1413,
+      "step": 1803
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2600931774903302e-06,
+      "loss": 0.2183,
+      "step": 1804
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2527423875929413e-06,
+      "loss": 0.1689,
+      "step": 1805
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2454116680053085e-06,
+      "loss": 0.1655,
+      "step": 1806
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2381010355475377e-06,
+      "loss": 0.1413,
+      "step": 1807
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.230810506993645e-06,
+      "loss": 0.1367,
+      "step": 1808
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2235400990715253e-06,
+      "loss": 0.1504,
+      "step": 1809
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2162898284628976e-06,
+      "loss": 0.1353,
+      "step": 1810
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2090597118032876e-06,
+      "loss": 0.1649,
+      "step": 1811
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2018497656819683e-06,
+      "loss": 0.1533,
+      "step": 1812
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1946600066419345e-06,
+      "loss": 0.1722,
+      "step": 1813
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1874904511798702e-06,
+      "loss": 0.1325,
+      "step": 1814
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1803411157460898e-06,
+      "loss": 0.1494,
+      "step": 1815
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1732120167445248e-06,
+      "loss": 0.1428,
+      "step": 1816
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1661031705326742e-06,
+      "loss": 0.1208,
+      "step": 1817
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1590145934215536e-06,
+      "loss": 0.1327,
+      "step": 1818
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1519463016756893e-06,
+      "loss": 0.1346,
+      "step": 1819
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1448983115130552e-06,
+      "loss": 0.1711,
+      "step": 1820
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1378706391050442e-06,
+      "loss": 0.133,
+      "step": 1821
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1308633005764346e-06,
+      "loss": 0.1449,
+      "step": 1822
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1238763120053386e-06,
+      "loss": 0.1183,
+      "step": 1823
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.116909689423189e-06,
+      "loss": 0.1405,
+      "step": 1824
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.109963448814685e-06,
+      "loss": 0.1538,
+      "step": 1825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.103037606117755e-06,
+      "loss": 0.1882,
+      "step": 1826
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0961321772235334e-06,
+      "loss": 0.134,
+      "step": 1827
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0892471779763093e-06,
+      "loss": 0.1234,
+      "step": 1828
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0823826241734957e-06,
+      "loss": 0.1291,
+      "step": 1829
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.075538531565603e-06,
+      "loss": 0.161,
+      "step": 1830
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0687149158561826e-06,
+      "loss": 0.1786,
+      "step": 1831
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.061911792701813e-06,
+      "loss": 0.1197,
+      "step": 1832
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0551291777120465e-06,
+      "loss": 0.1627,
+      "step": 1833
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0483670864493777e-06,
+      "loss": 0.1407,
+      "step": 1834
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0416255344292191e-06,
+      "loss": 0.1798,
+      "step": 1835
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0349045371198485e-06,
+      "loss": 0.1348,
+      "step": 1836
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0282041099423844e-06,
+      "loss": 0.1488,
+      "step": 1837
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0215242682707526e-06,
+      "loss": 0.1538,
+      "step": 1838
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0148650274316373e-06,
+      "loss": 0.1583,
+      "step": 1839
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0082264027044587e-06,
+      "loss": 0.1409,
+      "step": 1840
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0016084093213363e-06,
+      "loss": 0.1509,
+      "step": 1841
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.950110624670473e-07,
+      "loss": 0.1921,
+      "step": 1842
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.884343772790006e-07,
+      "loss": 0.1551,
+      "step": 1843
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.818783688471945e-07,
+      "loss": 0.1477,
+      "step": 1844
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.753430522141828e-07,
+      "loss": 0.1712,
+      "step": 1845
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.688284423750505e-07,
+      "loss": 0.1218,
+      "step": 1846
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.623345542773631e-07,
+      "loss": 0.1555,
+      "step": 1847
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.558614028211465e-07,
+      "loss": 0.1643,
+      "step": 1848
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.494090028588431e-07,
+      "loss": 0.1139,
+      "step": 1849
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.42977369195286e-07,
+      "loss": 0.1303,
+      "step": 1850
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.365665165876558e-07,
+      "loss": 0.153,
+      "step": 1851
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.301764597454522e-07,
+      "loss": 0.1225,
+      "step": 1852
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.238072133304654e-07,
+      "loss": 0.1509,
+      "step": 1853
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.17458791956729e-07,
+      "loss": 0.1596,
+      "step": 1854
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.111312101905012e-07,
+      "loss": 0.1603,
+      "step": 1855
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.048244825502184e-07,
+      "loss": 0.1328,
+      "step": 1856
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.985386235064708e-07,
+      "loss": 0.1243,
+      "step": 1857
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.92273647481966e-07,
+      "loss": 0.1477,
+      "step": 1858
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.860295688514975e-07,
+      "loss": 0.1444,
+      "step": 1859
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.798064019419073e-07,
+      "loss": 0.1516,
+      "step": 1860
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.736041610320589e-07,
+      "loss": 0.1495,
+      "step": 1861
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.674228603528013e-07,
+      "loss": 0.18,
+      "step": 1862
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.612625140869324e-07,
+      "loss": 0.1652,
+      "step": 1863
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.551231363691792e-07,
+      "loss": 0.1542,
+      "step": 1864
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.490047412861491e-07,
+      "loss": 0.1372,
+      "step": 1865
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.429073428763102e-07,
+      "loss": 0.1549,
+      "step": 1866
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.368309551299536e-07,
+      "loss": 0.1422,
+      "step": 1867
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.307755919891592e-07,
+      "loss": 0.1377,
+      "step": 1868
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.247412673477717e-07,
+      "loss": 0.1388,
+      "step": 1869
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.187279950513571e-07,
+      "loss": 0.1286,
+      "step": 1870
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.127357888971832e-07,
+      "loss": 0.11,
+      "step": 1871
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.06764662634184e-07,
+      "loss": 0.1637,
+      "step": 1872
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.00814629962916e-07,
+      "loss": 0.1645,
+      "step": 1873
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.948857045355474e-07,
+      "loss": 0.1391,
+      "step": 1874
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.889778999558095e-07,
+      "loss": 0.1346,
+      "step": 1875
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.830912297789773e-07,
+      "loss": 0.1574,
+      "step": 1876
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.772257075118373e-07,
+      "loss": 0.1531,
+      "step": 1877
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.713813466126375e-07,
+      "loss": 0.1357,
+      "step": 1878
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.655581604910867e-07,
+      "loss": 0.141,
+      "step": 1879
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.597561625083061e-07,
+      "loss": 0.1556,
+      "step": 1880
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.539753659767946e-07,
+      "loss": 0.1416,
+      "step": 1881
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.482157841604132e-07,
+      "loss": 0.148,
+      "step": 1882
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.424774302743409e-07,
+      "loss": 0.1451,
+      "step": 1883
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.367603174850502e-07,
+      "loss": 0.1492,
+      "step": 1884
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.310644589102822e-07,
+      "loss": 0.1267,
+      "step": 1885
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.253898676190019e-07,
+      "loss": 0.1456,
+      "step": 1886
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.197365566313874e-07,
+      "loss": 0.1457,
+      "step": 1887
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.141045389187784e-07,
+      "loss": 0.1549,
+      "step": 1888
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.08493827403669e-07,
+      "loss": 0.1321,
+      "step": 1889
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.029044349596603e-07,
+      "loss": 0.1714,
+      "step": 1890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.973363744114369e-07,
+      "loss": 0.1467,
+      "step": 1891
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.917896585347405e-07,
+      "loss": 0.1432,
+      "step": 1892
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.862643000563407e-07,
+      "loss": 0.1665,
+      "step": 1893
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.807603116539985e-07,
+      "loss": 0.1292,
+      "step": 1894
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.752777059564431e-07,
+      "loss": 0.1303,
+      "step": 1895
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.698164955433417e-07,
+      "loss": 0.1658,
+      "step": 1896
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.643766929452711e-07,
+      "loss": 0.1655,
+      "step": 1897
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.589583106436914e-07,
+      "loss": 0.1582,
+      "step": 1898
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.53561361070908e-07,
+      "loss": 0.1667,
+      "step": 1899
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.481858566100574e-07,
+      "loss": 0.1359,
+      "step": 1900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.428318095950648e-07,
+      "loss": 0.1358,
+      "step": 1901
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.374992323106221e-07,
+      "loss": 0.1182,
+      "step": 1902
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.321881369921656e-07,
+      "loss": 0.1224,
+      "step": 1903
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.268985358258339e-07,
+      "loss": 0.1454,
+      "step": 1904
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.216304409484553e-07,
+      "loss": 0.1251,
+      "step": 1905
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.163838644475084e-07,
+      "loss": 0.1327,
+      "step": 1906
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.111588183610962e-07,
+      "loss": 0.1374,
+      "step": 1907
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.059553146779284e-07,
+      "loss": 0.1632,
+      "step": 1908
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.007733653372794e-07,
+      "loss": 0.1424,
+      "step": 1909
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.956129822289691e-07,
+      "loss": 0.1368,
+      "step": 1910
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.904741771933409e-07,
+      "loss": 0.1474,
+      "step": 1911
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.853569620212163e-07,
+      "loss": 0.1623,
+      "step": 1912
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.802613484538888e-07,
+      "loss": 0.1109,
+      "step": 1913
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.75187348183086e-07,
+      "loss": 0.1324,
+      "step": 1914
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.701349728509409e-07,
+      "loss": 0.1922,
+      "step": 1915
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.651042340499746e-07,
+      "loss": 0.1416,
+      "step": 1916
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.600951433230551e-07,
+      "loss": 0.187,
+      "step": 1917
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.551077121633875e-07,
+      "loss": 0.1232,
+      "step": 1918
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.50141952014479e-07,
+      "loss": 0.1634,
+      "step": 1919
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.45197874270107e-07,
+      "loss": 0.1362,
+      "step": 1920
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.40275490274309e-07,
+      "loss": 0.1598,
+      "step": 1921
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.353748113213375e-07,
+      "loss": 0.1161,
+      "step": 1922
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.304958486556488e-07,
+      "loss": 0.1393,
+      "step": 1923
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.256386134718705e-07,
+      "loss": 0.1393,
+      "step": 1924
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.208031169147765e-07,
+      "loss": 0.1783,
+      "step": 1925
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.159893700792651e-07,
+      "loss": 0.1165,
+      "step": 1926
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.111973840103268e-07,
+      "loss": 0.1473,
+      "step": 1927
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.064271697030232e-07,
+      "loss": 0.1578,
+      "step": 1928
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.016787381024635e-07,
+      "loss": 0.141,
+      "step": 1929
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.969521001037747e-07,
+      "loss": 0.1462,
+      "step": 1930
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.922472665520806e-07,
+      "loss": 0.158,
+      "step": 1931
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.87564248242477e-07,
+      "loss": 0.1415,
+      "step": 1932
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.829030559200032e-07,
+      "loss": 0.1179,
+      "step": 1933
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.782637002796153e-07,
+      "loss": 0.1451,
+      "step": 1934
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.73646191966175e-07,
+      "loss": 0.1335,
+      "step": 1935
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6905054157440733e-07,
+      "loss": 0.1534,
+      "step": 1936
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.64476759648893e-07,
+      "loss": 0.1507,
+      "step": 1937
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5992485668402865e-07,
+      "loss": 0.1622,
+      "step": 1938
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5539484312401404e-07,
+      "loss": 0.1453,
+      "step": 1939
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5088672936282606e-07,
+      "loss": 0.1293,
+      "step": 1940
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4640052574418724e-07,
+      "loss": 0.2171,
+      "step": 1941
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4193624256155587e-07,
+      "loss": 0.1439,
+      "step": 1942
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.374938900580883e-07,
+      "loss": 0.1781,
+      "step": 1943
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.330734784266244e-07,
+      "loss": 0.1562,
+      "step": 1944
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2867501780966215e-07,
+      "loss": 0.1734,
+      "step": 1945
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2429851829932866e-07,
+      "loss": 0.1246,
+      "step": 1946
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.199439899373692e-07,
+      "loss": 0.1777,
+      "step": 1947
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1561144271510925e-07,
+      "loss": 0.1414,
+      "step": 1948
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.11300886573448e-07,
+      "loss": 0.1713,
+      "step": 1949
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.070123314028207e-07,
+      "loss": 0.1284,
+      "step": 1950
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0274578704318057e-07,
+      "loss": 0.1575,
+      "step": 1951
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.985012632839824e-07,
+      "loss": 0.1441,
+      "step": 1952
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.942787698641548e-07,
+      "loss": 0.1217,
+      "step": 1953
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.900783164720756e-07,
+      "loss": 0.1523,
+      "step": 1954
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8589991274555536e-07,
+      "loss": 0.1553,
+      "step": 1955
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.817435682718096e-07,
+      "loss": 0.133,
+      "step": 1956
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7760929258744194e-07,
+      "loss": 0.1563,
+      "step": 1957
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.734970951784189e-07,
+      "loss": 0.1387,
+      "step": 1958
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.694069854800486e-07,
+      "loss": 0.1339,
+      "step": 1959
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6533897287696295e-07,
+      "loss": 0.1576,
+      "step": 1960
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.612930667030867e-07,
+      "loss": 0.1671,
+      "step": 1961
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5726927624162524e-07,
+      "loss": 0.1373,
+      "step": 1962
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.532676107250421e-07,
+      "loss": 0.1449,
+      "step": 1963
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.492880793350328e-07,
+      "loss": 0.1451,
+      "step": 1964
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4533069120250896e-07,
+      "loss": 0.1476,
+      "step": 1965
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.41395455407576e-07,
+      "loss": 0.1676,
+      "step": 1966
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.374823809795047e-07,
+      "loss": 0.1368,
+      "step": 1967
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.335914768967263e-07,
+      "loss": 0.1248,
+      "step": 1968
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2972275208679625e-07,
+      "loss": 0.1439,
+      "step": 1969
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2587621542638504e-07,
+      "loss": 0.1549,
+      "step": 1970
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2205187574125275e-07,
+      "loss": 0.1507,
+      "step": 1971
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1824974180622247e-07,
+      "loss": 0.153,
+      "step": 1972
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1446982234517474e-07,
+      "loss": 0.159,
+      "step": 1973
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1071212603101754e-07,
+      "loss": 0.1601,
+      "step": 1974
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0697666148566616e-07,
+      "loss": 0.1503,
+      "step": 1975
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0326343728002915e-07,
+      "loss": 0.1328,
+      "step": 1976
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9957246193398237e-07,
+      "loss": 0.1518,
+      "step": 1977
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9590374391635367e-07,
+      "loss": 0.1341,
+      "step": 1978
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.92257291644904e-07,
+      "loss": 0.123,
+      "step": 1979
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8863311348630075e-07,
+      "loss": 0.1407,
+      "step": 1980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.85031217756111e-07,
+      "loss": 0.1572,
+      "step": 1981
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8145161271877166e-07,
+      "loss": 0.1442,
+      "step": 1982
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.778943065875728e-07,
+      "loss": 0.1315,
+      "step": 1983
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.743593075246442e-07,
+      "loss": 0.1572,
+      "step": 1984
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.708466236409302e-07,
+      "loss": 0.1336,
+      "step": 1985
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6735626299617456e-07,
+      "loss": 0.1461,
+      "step": 1986
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.638882335989024e-07,
+      "loss": 0.1186,
+      "step": 1987
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6044254340639617e-07,
+      "loss": 0.136,
+      "step": 1988
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.570192003246874e-07,
+      "loss": 0.1566,
+      "step": 1989
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.536182122085262e-07,
+      "loss": 0.1518,
+      "step": 1990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5023958686137716e-07,
+      "loss": 0.1343,
+      "step": 1991
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4688333203539027e-07,
+      "loss": 0.1703,
+      "step": 1992
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4354945543138775e-07,
+      "loss": 0.1787,
+      "step": 1993
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4023796469884285e-07,
+      "loss": 0.1385,
+      "step": 1994
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.369488674358711e-07,
+      "loss": 0.143,
+      "step": 1995
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3368217118920122e-07,
+      "loss": 0.1275,
+      "step": 1996
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3043788345416762e-07,
+      "loss": 0.1331,
+      "step": 1997
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2721601167468467e-07,
+      "loss": 0.1664,
+      "step": 1998
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2401656324324006e-07,
+      "loss": 0.1593,
+      "step": 1999
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2083954550086718e-07,
+      "loss": 0.134,
+      "step": 2000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.176849657371316e-07,
+      "loss": 0.1754,
+      "step": 2001
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1455283119012237e-07,
+      "loss": 0.1303,
+      "step": 2002
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1144314904642194e-07,
+      "loss": 0.1038,
+      "step": 2003
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.083559264410995e-07,
+      "loss": 0.1132,
+      "step": 2004
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.05291170457691e-07,
+      "loss": 0.1631,
+      "step": 2005
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0224888812818366e-07,
+      "loss": 0.138,
+      "step": 2006
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.992290864329982e-07,
+      "loss": 0.1179,
+      "step": 2007
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9623177230097767e-07,
+      "loss": 0.1365,
+      "step": 2008
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9325695260936416e-07,
+      "loss": 0.1363,
+      "step": 2009
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.903046341837911e-07,
+      "loss": 0.154,
+      "step": 2010
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8737482379825644e-07,
+      "loss": 0.1569,
+      "step": 2011
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.844675281751218e-07,
+      "loss": 0.1466,
+      "step": 2012
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8158275398508785e-07,
+      "loss": 0.1443,
+      "step": 2013
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.787205078471743e-07,
+      "loss": 0.1616,
+      "step": 2014
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7588079632872013e-07,
+      "loss": 0.1495,
+      "step": 2015
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.730636259453533e-07,
+      "loss": 0.1168,
+      "step": 2016
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7026900316098217e-07,
+      "loss": 0.156,
+      "step": 2017
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6749693438778302e-07,
+      "loss": 0.1307,
+      "step": 2018
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.647474259861803e-07,
+      "loss": 0.1251,
+      "step": 2019
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6202048426483652e-07,
+      "loss": 0.1294,
+      "step": 2020
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.593161154806333e-07,
+      "loss": 0.1494,
+      "step": 2021
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5663432583866046e-07,
+      "loss": 0.1635,
+      "step": 2022
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5397512149220139e-07,
+      "loss": 0.1377,
+      "step": 2023
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5133850854271658e-07,
+      "loss": 0.1522,
+      "step": 2024
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4872449303983238e-07,
+      "loss": 0.1542,
+      "step": 2025
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4613308098132772e-07,
+      "loss": 0.1514,
+      "step": 2026
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4356427831311415e-07,
+      "loss": 0.1401,
+      "step": 2027
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4101809092922914e-07,
+      "loss": 0.1378,
+      "step": 2028
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3849452467182388e-07,
+      "loss": 0.1243,
+      "step": 2029
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.359935853311378e-07,
+      "loss": 0.1474,
+      "step": 2030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3351527864550186e-07,
+      "loss": 0.1354,
+      "step": 2031
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3105961030131175e-07,
+      "loss": 0.1356,
+      "step": 2032
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2862658593302046e-07,
+      "loss": 0.1306,
+      "step": 2033
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.262162111231302e-07,
+      "loss": 0.134,
+      "step": 2034
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.23828491402167e-07,
+      "loss": 0.1319,
+      "step": 2035
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2146343224868184e-07,
+      "loss": 0.131,
+      "step": 2036
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1912103908922945e-07,
+      "loss": 0.1779,
+      "step": 2037
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1680131729835509e-07,
+      "loss": 0.122,
+      "step": 2038
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1450427219859006e-07,
+      "loss": 0.1494,
+      "step": 2039
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1222990906043286e-07,
+      "loss": 0.1582,
+      "step": 2040
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0997823310233802e-07,
+      "loss": 0.1482,
+      "step": 2041
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.077492494907062e-07,
+      "loss": 0.1736,
+      "step": 2042
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.055429633398708e-07,
+      "loss": 0.132,
+      "step": 2043
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0335937971208577e-07,
+      "loss": 0.1656,
+      "step": 2044
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.011985036175167e-07,
+      "loss": 0.1203,
+      "step": 2045
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.906034001422538e-08,
+      "loss": 0.1499,
+      "step": 2046
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.694489380816297e-08,
+      "loss": 0.1372,
+      "step": 2047
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.485216985315349e-08,
+      "loss": 0.1284,
+      "step": 2048
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.278217295088709e-08,
+      "loss": 0.1597,
+      "step": 2049
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.073490785090678e-08,
+      "loss": 0.1724,
+      "step": 2050
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.871037925059944e-08,
+      "loss": 0.1318,
+      "step": 2051
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.670859179518266e-08,
+      "loss": 0.1582,
+      "step": 2052
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.472955007769456e-08,
+      "loss": 0.1155,
+      "step": 2053
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.27732586389851e-08,
+      "loss": 0.134,
+      "step": 2054
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.083972196770373e-08,
+      "loss": 0.1824,
+      "step": 2055
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.892894450029053e-08,
+      "loss": 0.1578,
+      "step": 2056
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.704093062096407e-08,
+      "loss": 0.1341,
+      "step": 2057
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.517568466171355e-08,
+      "loss": 0.1362,
+      "step": 2058
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.333321090228773e-08,
+      "loss": 0.1598,
+      "step": 2059
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.15135135701861e-08,
+      "loss": 0.188,
+      "step": 2060
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.971659684064546e-08,
+      "loss": 0.1304,
+      "step": 2061
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.794246483663669e-08,
+      "loss": 0.1355,
+      "step": 2062
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.619112162885022e-08,
+      "loss": 0.125,
+      "step": 2063
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.446257123568723e-08,
+      "loss": 0.1517,
+      "step": 2064
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.275681762325514e-08,
+      "loss": 0.1642,
+      "step": 2065
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.107386470534882e-08,
+      "loss": 0.1184,
+      "step": 2066
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.941371634345494e-08,
+      "loss": 0.159,
+      "step": 2067
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7776376346729835e-08,
+      "loss": 0.1747,
+      "step": 2068
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6161848471999456e-08,
+      "loss": 0.1226,
+      "step": 2069
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.457013642374831e-08,
+      "loss": 0.1342,
+      "step": 2070
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.300124385410943e-08,
+      "loss": 0.1505,
+      "step": 2071
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.145517436285774e-08,
+      "loss": 0.1575,
+      "step": 2072
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9931931497403384e-08,
+      "loss": 0.1661,
+      "step": 2073
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8431518752776185e-08,
+      "loss": 0.1505,
+      "step": 2074
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.695393957162897e-08,
+      "loss": 0.1579,
+      "step": 2075
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5499197344220924e-08,
+      "loss": 0.1549,
+      "step": 2076
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.406729540841092e-08,
+      "loss": 0.1275,
+      "step": 2077
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2658237049655325e-08,
+      "loss": 0.1395,
+      "step": 2078
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.1272025500992405e-08,
+      "loss": 0.1544,
+      "step": 2079
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.990866394304349e-08,
+      "loss": 0.151,
+      "step": 2080
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.856815550399962e-08,
+      "loss": 0.1355,
+      "step": 2081
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.725050325961266e-08,
+      "loss": 0.1289,
+      "step": 2082
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.595571023319755e-08,
+      "loss": 0.1757,
+      "step": 2083
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4683779395614514e-08,
+      "loss": 0.1366,
+      "step": 2084
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.343471366526907e-08,
+      "loss": 0.1514,
+      "step": 2085
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.220851590810536e-08,
+      "loss": 0.1327,
+      "step": 2086
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1005188937595075e-08,
+      "loss": 0.1306,
+      "step": 2087
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.982473551473297e-08,
+      "loss": 0.1385,
+      "step": 2088
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8667158348034686e-08,
+      "loss": 0.1332,
+      "step": 2089
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7532460093524503e-08,
+      "loss": 0.14,
+      "step": 2090
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6420643354732046e-08,
+      "loss": 0.159,
+      "step": 2091
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5331710682684473e-08,
+      "loss": 0.1577,
+      "step": 2092
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.426566457590651e-08,
+      "loss": 0.16,
+      "step": 2093
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3222507480406e-08,
+      "loss": 0.1228,
+      "step": 2094
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2202241789673894e-08,
+      "loss": 0.1168,
+      "step": 2095
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.120486984467984e-08,
+      "loss": 0.1521,
+      "step": 2096
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0230393933862168e-08,
+      "loss": 0.1427,
+      "step": 2097
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9278816293124557e-08,
+      "loss": 0.1148,
+      "step": 2098
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.835013910583272e-08,
+      "loss": 0.1666,
+      "step": 2099
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.744436450280662e-08,
+      "loss": 0.1563,
+      "step": 2100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.656149456231937e-08,
+      "loss": 0.1638,
+      "step": 2101
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5701531310087224e-08,
+      "loss": 0.1515,
+      "step": 2102
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4864476719270716e-08,
+      "loss": 0.1234,
+      "step": 2103
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.405033271046352e-08,
+      "loss": 0.1315,
+      "step": 2104
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.325910115169471e-08,
+      "loss": 0.1577,
+      "step": 2105
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2490783858420952e-08,
+      "loss": 0.1134,
+      "step": 2106
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1745382593520982e-08,
+      "loss": 0.1382,
+      "step": 2107
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.102289906729448e-08,
+      "loss": 0.1203,
+      "step": 2108
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.032333493745985e-08,
+      "loss": 0.1359,
+      "step": 2109
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.646691809143126e-09,
+      "loss": 0.14,
+      "step": 2110
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.992971234881299e-09,
+      "loss": 0.1364,
+      "step": 2111
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.362174714615645e-09,
+      "loss": 0.1434,
+      "step": 2112
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.754303695688414e-09,
+      "loss": 0.1502,
+      "step": 2113
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.169359572842816e-09,
+      "loss": 0.1283,
+      "step": 2114
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.607343688214141e-09,
+      "loss": 0.1372,
+      "step": 2115
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.068257331328653e-09,
+      "loss": 0.1685,
+      "step": 2116
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.552101739104698e-09,
+      "loss": 0.1553,
+      "step": 2117
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.058878095844932e-09,
+      "loss": 0.1178,
+      "step": 2118
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.588587533235211e-09,
+      "loss": 0.1316,
+      "step": 2119
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.141231130343482e-09,
+      "loss": 0.1465,
+      "step": 2120
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.716809913614228e-09,
+      "loss": 0.1246,
+      "step": 2121
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.315324856869584e-09,
+      "loss": 0.1671,
+      "step": 2122
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.936776881302672e-09,
+      "loss": 0.128,
+      "step": 2123
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5811668554820422e-09,
+      "loss": 0.1547,
+      "step": 2124
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.248495595342792e-09,
+      "loss": 0.1554,
+      "step": 2125
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9387638641887862e-09,
+      "loss": 0.1523,
+      "step": 2126
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6519723726893255e-09,
+      "loss": 0.1366,
+      "step": 2127
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.388121778880258e-09,
+      "loss": 0.1548,
+      "step": 2128
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1472126881562074e-09,
+      "loss": 0.1418,
+      "step": 2129
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.292456532783433e-10,
+      "loss": 0.1401,
+      "step": 2130
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.342211743621708e-10,
+      "loss": 0.1334,
+      "step": 2131
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.621396988875205e-10,
+      "loss": 0.1508,
+      "step": 2132
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.130016216896682e-10,
+      "loss": 0.1312,
+      "step": 2133
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.868072849604442e-10,
+      "loss": 0.1187,
+      "step": 2134
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8355697824934404e-10,
+      "loss": 0.1271,
+      "step": 2135
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0325093846019762e-10,
+      "loss": 0.1258,
+      "step": 2136
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.5889349853389664e-11,
+      "loss": 0.1462,
+      "step": 2137
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1472344044749506e-11,
+      "loss": 0.1354,
+      "step": 2138
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.1318,
+      "step": 2139
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2139,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 2139,
+  "total_flos": 1.0721490884613898e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}