diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,24021 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.6297724946862946,
+  "eval_steps": 500,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 1.0099,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 0.8664,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-06,
+      "loss": 1.0055,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 1.0551,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.9447,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.44e-05,
+      "loss": 0.9451,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.9279,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.9709,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.16e-05,
+      "loss": 1.1009,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4e-05,
+      "loss": 1.0055,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.64e-05,
+      "loss": 1.0573,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.88e-05,
+      "loss": 1.0052,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.12e-05,
+      "loss": 0.8902,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 0.7435,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-05,
+      "loss": 0.7892,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 0.9127,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.08e-05,
+      "loss": 0.8227,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.32e-05,
+      "loss": 1.0142,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 0.8347,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8e-05,
+      "loss": 0.8906,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.04e-05,
+      "loss": 0.7521,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.28e-05,
+      "loss": 0.877,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 0.7932,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.76e-05,
+      "loss": 0.7357,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6e-05,
+      "loss": 0.9267,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999977015133e-05,
+      "loss": 0.8972,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999999080605335e-05,
+      "loss": 1.002,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999999793136201e-05,
+      "loss": 0.9171,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999996322421374e-05,
+      "loss": 0.9653,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999994253783456e-05,
+      "loss": 0.8907,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999991725448303e-05,
+      "loss": 0.8956,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999988737415935e-05,
+      "loss": 0.8175,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999998528968639e-05,
+      "loss": 0.9197,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999981382259745e-05,
+      "loss": 0.9345,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999997701513605e-05,
+      "loss": 1.0491,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999972188315365e-05,
+      "loss": 0.8521,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999996690179777e-05,
+      "loss": 0.9119,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999961155583345e-05,
+      "loss": 0.9787,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999954949672174e-05,
+      "loss": 0.727,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999994828406436e-05,
+      "loss": 0.6152,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999994115876e-05,
+      "loss": 0.8991,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999993357375921e-05,
+      "loss": 0.75,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999925529062094e-05,
+      "loss": 1.0197,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999917024668785e-05,
+      "loss": 0.7259,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999990806057941e-05,
+      "loss": 0.9215,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999989863679411e-05,
+      "loss": 0.8737,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999988875331302e-05,
+      "loss": 1.0794,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99998784101363e-05,
+      "loss": 0.7958,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999986760726411e-05,
+      "loss": 0.7886,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99998563446966e-05,
+      "loss": 1.0541,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999844622433963e-05,
+      "loss": 0.7176,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999983244047637e-05,
+      "loss": 0.99,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999819798824006e-05,
+      "loss": 0.8261,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999980669747707e-05,
+      "loss": 0.8775,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999979313643575e-05,
+      "loss": 0.9863,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999779115700266e-05,
+      "loss": 0.9517,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999976463527083e-05,
+      "loss": 0.8547,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999974969514766e-05,
+      "loss": 0.9071,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999734295331e-05,
+      "loss": 0.8818,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999971843582106e-05,
+      "loss": 0.7707,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999970211661811e-05,
+      "loss": 0.8025,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999968533772238e-05,
+      "loss": 0.8487,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999966809913413e-05,
+      "loss": 1.0707,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999965040085364e-05,
+      "loss": 0.7883,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999963224288116e-05,
+      "loss": 0.9995,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999961362521698e-05,
+      "loss": 0.9008,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999959454786139e-05,
+      "loss": 0.7946,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999957501081467e-05,
+      "loss": 0.8892,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999955501407713e-05,
+      "loss": 0.831,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999953455764907e-05,
+      "loss": 0.9009,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99995136415308e-05,
+      "loss": 0.9287,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999492265722643e-05,
+      "loss": 0.8452,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999947043022494e-05,
+      "loss": 1.0281,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999944813503801e-05,
+      "loss": 0.9206,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99994253801622e-05,
+      "loss": 0.9584,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999402165597854e-05,
+      "loss": 0.8812,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999937849134533e-05,
+      "loss": 0.9665,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999935435740499e-05,
+      "loss": 0.8961,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999932976377721e-05,
+      "loss": 0.7666,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999930471046238e-05,
+      "loss": 0.8579,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999927919746084e-05,
+      "loss": 1.0028,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999253224773025e-05,
+      "loss": 1.0429,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999226792399306e-05,
+      "loss": 0.8974,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99991999003401e-05,
+      "loss": 0.8542,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999172548595826e-05,
+      "loss": 0.8322,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999144737166885e-05,
+      "loss": 0.7911,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999911646605372e-05,
+      "loss": 1.0441,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999908773525675e-05,
+      "loss": 0.8627,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999905854477642e-05,
+      "loss": 0.8301,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9999028894613176e-05,
+      "loss": 0.7309,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999899878476749e-05,
+      "loss": 0.8843,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99989682152398e-05,
+      "loss": 0.7874,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999893718603058e-05,
+      "loss": 0.9164,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9998905697140316e-05,
+      "loss": 0.8448,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999887374856947e-05,
+      "loss": 1.0298,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999884134031856e-05,
+      "loss": 0.9022,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999880847238806e-05,
+      "loss": 0.6828,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999877514477848e-05,
+      "loss": 0.8187,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9998741357490334e-05,
+      "loss": 1.0166,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9998707110524136e-05,
+      "loss": 0.7758,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999867240388041e-05,
+      "loss": 0.821,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9998637237559694e-05,
+      "loss": 0.7933,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999860161156252e-05,
+      "loss": 0.9255,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999856552588944e-05,
+      "loss": 0.9009,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9998528980541e-05,
+      "loss": 0.9496,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999849197551777e-05,
+      "loss": 0.8982,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99984545108203e-05,
+      "loss": 0.9588,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999841658644918e-05,
+      "loss": 0.7428,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9998378202404985e-05,
+      "loss": 0.8167,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999833935868831e-05,
+      "loss": 0.8165,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9998300055299736e-05,
+      "loss": 0.8567,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999826029223988e-05,
+      "loss": 1.073,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999822006950934e-05,
+      "loss": 0.8588,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999817938710874e-05,
+      "loss": 0.8485,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99981382450387e-05,
+      "loss": 0.755,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9998096643299846e-05,
+      "loss": 1.1809,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999805458189283e-05,
+      "loss": 1.005,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999801206081828e-05,
+      "loss": 0.9657,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999796908007686e-05,
+      "loss": 0.8831,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999792563966922e-05,
+      "loss": 1.0875,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997881739596025e-05,
+      "loss": 0.8888,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997837379857954e-05,
+      "loss": 0.8166,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999779256045569e-05,
+      "loss": 0.8437,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99977472813899e-05,
+      "loss": 0.8259,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997701542661306e-05,
+      "loss": 0.8322,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999765534427059e-05,
+      "loss": 0.8268,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999760868621846e-05,
+      "loss": 0.6975,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999756156850564e-05,
+      "loss": 0.8667,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9997513991132854e-05,
+      "loss": 0.8607,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999746595410081e-05,
+      "loss": 0.9771,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999741745741027e-05,
+      "loss": 0.7963,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999736850106196e-05,
+      "loss": 0.7803,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999731908505664e-05,
+      "loss": 0.9298,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999726920939506e-05,
+      "loss": 0.8587,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999721887407799e-05,
+      "loss": 0.8066,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99971680791062e-05,
+      "loss": 0.8281,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999711682448046e-05,
+      "loss": 0.7008,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999706511020157e-05,
+      "loss": 1.0162,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99970129362703e-05,
+      "loss": 0.7273,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999696030268748e-05,
+      "loss": 0.9373,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996907209453894e-05,
+      "loss": 0.7207,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996853656570365e-05,
+      "loss": 1.0179,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999679964403771e-05,
+      "loss": 0.7137,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999674517185677e-05,
+      "loss": 0.8013,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999669024002835e-05,
+      "loss": 0.8382,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999663484855332e-05,
+      "loss": 0.8507,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996578997432504e-05,
+      "loss": 0.9923,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999652268666678e-05,
+      "loss": 0.8385,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996465916257004e-05,
+      "loss": 0.7664,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996408686204046e-05,
+      "loss": 0.9129,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999635099650878e-05,
+      "loss": 0.9569,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999629284717209e-05,
+      "loss": 1.1235,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996234238194867e-05,
+      "loss": 0.9859,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999617516957802e-05,
+      "loss": 1.0235,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999611564132243e-05,
+      "loss": 0.7925,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9996055653429035e-05,
+      "loss": 0.8611,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.999599520589874e-05,
+      "loss": 0.7823,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9995934298732467e-05,
+      "loss": 0.7773,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999587293193117e-05,
+      "loss": 0.9664,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999581110549576e-05,
+      "loss": 0.9597,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999574881942722e-05,
+      "loss": 0.8882,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9995686073726467e-05,
+      "loss": 0.9883,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9995622868394494e-05,
+      "loss": 0.8201,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999555920343225e-05,
+      "loss": 1.0745,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999549507884071e-05,
+      "loss": 0.9511,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9995430494620865e-05,
+      "loss": 0.9346,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9995365450773705e-05,
+      "loss": 0.9686,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9995299947300224e-05,
+      "loss": 0.8193,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999523398420143e-05,
+      "loss": 0.7927,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999516756147833e-05,
+      "loss": 0.9249,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999510067913193e-05,
+      "loss": 0.8858,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999503333716328e-05,
+      "loss": 0.9746,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999496553557339e-05,
+      "loss": 0.6803,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994897274363305e-05,
+      "loss": 0.8265,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999482855353408e-05,
+      "loss": 0.9126,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999475937308676e-05,
+      "loss": 1.0128,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999468973302241e-05,
+      "loss": 0.8885,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999461963334208e-05,
+      "loss": 0.9747,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994549074046865e-05,
+      "loss": 0.951,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994478055137836e-05,
+      "loss": 0.7044,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999440657661609e-05,
+      "loss": 0.9,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999433463848271e-05,
+      "loss": 0.8798,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999426224073881e-05,
+      "loss": 0.9441,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994189383385495e-05,
+      "loss": 0.8088,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994116066423876e-05,
+      "loss": 0.7801,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9994042289855085e-05,
+      "loss": 0.8987,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999396805368024e-05,
+      "loss": 0.9816,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999389335790049e-05,
+      "loss": 0.8721,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999381820251699e-05,
+      "loss": 0.9119,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999374258753086e-05,
+      "loss": 0.9875,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9993666512943286e-05,
+      "loss": 0.8161,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999358997875542e-05,
+      "loss": 0.8513,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999351298496844e-05,
+      "loss": 0.7224,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999343553158353e-05,
+      "loss": 0.8395,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999335761860186e-05,
+      "loss": 0.5879,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9993279246024646e-05,
+      "loss": 0.8802,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9993200413853075e-05,
+      "loss": 0.9191,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999312112208835e-05,
+      "loss": 0.9171,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999304137073172e-05,
+      "loss": 1.0182,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992961159784356e-05,
+      "loss": 0.7847,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999288048924753e-05,
+      "loss": 1.0507,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992799359122444e-05,
+      "loss": 0.7339,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999271776941037e-05,
+      "loss": 0.8936,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999263572011254e-05,
+      "loss": 0.9069,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999255321123023e-05,
+      "loss": 0.8425,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992470242764674e-05,
+      "loss": 1.031,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999238681471717e-05,
+      "loss": 0.9838,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999230292708899e-05,
+      "loss": 0.9716,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999221857988141e-05,
+      "loss": 0.8397,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9992133773095734e-05,
+      "loss": 0.8779,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999204850673326e-05,
+      "loss": 0.8194,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999196278079528e-05,
+      "loss": 0.798,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999187659528314e-05,
+      "loss": 0.9654,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9991789950198125e-05,
+      "loss": 1.0165,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9991702845541575e-05,
+      "loss": 0.8978,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999161528131483e-05,
+      "loss": 0.8791,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9991527257519236e-05,
+      "loss": 1.0596,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999143877415612e-05,
+      "loss": 0.9043,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999134983122687e-05,
+      "loss": 1.0067,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999126042873282e-05,
+      "loss": 1.1682,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9991170566675356e-05,
+      "loss": 0.9314,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.999108024505585e-05,
+      "loss": 0.926,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999098946387569e-05,
+      "loss": 0.7746,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999089822313626e-05,
+      "loss": 0.849,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999080652283896e-05,
+      "loss": 1.0146,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9990714362985195e-05,
+      "loss": 0.7924,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999062174357638e-05,
+      "loss": 1.1644,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999052866461393e-05,
+      "loss": 1.1566,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999043512609928e-05,
+      "loss": 0.9494,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999034112803386e-05,
+      "loss": 0.7897,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999024667041911e-05,
+      "loss": 0.9428,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9990151753256466e-05,
+      "loss": 1.0208,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.999005637654739e-05,
+      "loss": 0.9519,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998996054029335e-05,
+      "loss": 0.8827,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998986424449582e-05,
+      "loss": 0.9577,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998976748915625e-05,
+      "loss": 1.1778,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9989670274276143e-05,
+      "loss": 0.9766,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998957259985699e-05,
+      "loss": 0.8869,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9989474465900275e-05,
+      "loss": 0.9489,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998937587240752e-05,
+      "loss": 0.8857,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998927681938021e-05,
+      "loss": 0.8008,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998917730681987e-05,
+      "loss": 1.0753,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998907733472805e-05,
+      "loss": 0.8526,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998897690310625e-05,
+      "loss": 0.8564,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998887601195603e-05,
+      "loss": 0.7733,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998877466127893e-05,
+      "loss": 0.7639,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99886728510765e-05,
+      "loss": 0.9862,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9988570581350304e-05,
+      "loss": 1.0389,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9988467852101904e-05,
+      "loss": 1.0409,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9988364663332865e-05,
+      "loss": 1.0038,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998826101504479e-05,
+      "loss": 1.0068,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998815690723926e-05,
+      "loss": 0.8894,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998805233991787e-05,
+      "loss": 0.8967,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998794731308221e-05,
+      "loss": 0.939,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9987841826733905e-05,
+      "loss": 0.7079,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998773588087457e-05,
+      "loss": 0.9946,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998762947550581e-05,
+      "loss": 0.7007,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998752261062928e-05,
+      "loss": 0.8893,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99874152862466e-05,
+      "loss": 0.9033,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998730750235943e-05,
+      "loss": 0.7834,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998719925896941e-05,
+      "loss": 0.9223,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99870905560782e-05,
+      "loss": 0.9258,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9986981393687465e-05,
+      "loss": 0.7093,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9986871771798885e-05,
+      "loss": 0.9622,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998676169041413e-05,
+      "loss": 0.9463,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99866511495349e-05,
+      "loss": 0.9122,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9986540149162877e-05,
+      "loss": 0.7993,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998642868929976e-05,
+      "loss": 1.023,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998631676994726e-05,
+      "loss": 0.7433,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998620439110711e-05,
+      "loss": 0.8008,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9986091552781e-05,
+      "loss": 1.0025,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9985978254970684e-05,
+      "loss": 0.8545,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9985864497677894e-05,
+      "loss": 0.9042,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998575028090436e-05,
+      "loss": 0.7947,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998563560465184e-05,
+      "loss": 0.8575,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99855204689221e-05,
+      "loss": 0.7848,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9985404873716885e-05,
+      "loss": 0.9065,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998528881903799e-05,
+      "loss": 1.0962,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998517230488717e-05,
+      "loss": 0.7693,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998505533126622e-05,
+      "loss": 0.8369,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998493789817695e-05,
+      "loss": 0.8972,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998482000562113e-05,
+      "loss": 0.7346,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9984701653600585e-05,
+      "loss": 0.8198,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998458284211712e-05,
+      "loss": 0.9731,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.998446357117255e-05,
+      "loss": 0.8813,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998434384076873e-05,
+      "loss": 0.9398,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9984223650907466e-05,
+      "loss": 0.7763,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998410300159061e-05,
+      "loss": 0.8026,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9983981892820006e-05,
+      "loss": 0.7946,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998386032459752e-05,
+      "loss": 0.9739,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9983738296925005e-05,
+      "loss": 1.0146,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998361580980433e-05,
+      "loss": 1.0157,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998349286323739e-05,
+      "loss": 0.7392,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998336945722605e-05,
+      "loss": 0.842,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998324559177221e-05,
+      "loss": 0.8699,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998312126687776e-05,
+      "loss": 0.754,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998299648254461e-05,
+      "loss": 1.1168,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9982871238774677e-05,
+      "loss": 0.9668,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9982745535569867e-05,
+      "loss": 0.8755,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998261937293212e-05,
+      "loss": 0.9887,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9982492750863356e-05,
+      "loss": 0.8281,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998236566936554e-05,
+      "loss": 0.9776,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998223812844058e-05,
+      "loss": 0.8503,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998211012809046e-05,
+      "loss": 0.9252,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998198166831714e-05,
+      "loss": 1.0505,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998185274912258e-05,
+      "loss": 0.9074,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998172337050876e-05,
+      "loss": 1.0526,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9981593532477656e-05,
+      "loss": 0.8298,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998146323503127e-05,
+      "loss": 1.0027,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998133247817157e-05,
+      "loss": 0.9936,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99812012619006e-05,
+      "loss": 0.8453,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998106958622034e-05,
+      "loss": 0.989,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998093745113282e-05,
+      "loss": 0.7751,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998080485664006e-05,
+      "loss": 0.7197,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9980671802744095e-05,
+      "loss": 0.8894,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9980538289446964e-05,
+      "loss": 0.8,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9980404316750725e-05,
+      "loss": 0.9721,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9980269884657406e-05,
+      "loss": 0.7467,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.998013499316908e-05,
+      "loss": 0.9675,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9979999642287814e-05,
+      "loss": 0.6645,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997986383201568e-05,
+      "loss": 0.9146,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997972756235477e-05,
+      "loss": 0.929,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997959083330715e-05,
+      "loss": 1.0047,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997945364487493e-05,
+      "loss": 0.9067,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9979315997060214e-05,
+      "loss": 0.8891,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99791778898651e-05,
+      "loss": 0.7941,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997903932329172e-05,
+      "loss": 0.8755,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9978900297342184e-05,
+      "loss": 1.1208,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997876081201863e-05,
+      "loss": 0.9561,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9978620867323184e-05,
+      "loss": 0.826,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997848046325801e-05,
+      "loss": 0.7728,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997833959982524e-05,
+      "loss": 0.845,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997819827702704e-05,
+      "loss": 1.0724,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9978056494865576e-05,
+      "loss": 0.8317,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997791425334303e-05,
+      "loss": 0.8038,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997777155246156e-05,
+      "loss": 0.6207,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997762839222337e-05,
+      "loss": 0.8596,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9977484772630646e-05,
+      "loss": 0.8097,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9977340693685596e-05,
+      "loss": 0.7331,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997719615539042e-05,
+      "loss": 0.8526,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997705115774733e-05,
+      "loss": 0.897,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997690570075856e-05,
+      "loss": 0.8802,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9976759784426336e-05,
+      "loss": 0.8824,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9976613408752886e-05,
+      "loss": 1.048,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9976466573740455e-05,
+      "loss": 1.0445,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9976319279391306e-05,
+      "loss": 0.9399,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997617152570767e-05,
+      "loss": 0.7753,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9976023312691834e-05,
+      "loss": 0.7903,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.997587464034606e-05,
+      "loss": 0.8665,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997572550867263e-05,
+      "loss": 0.9647,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997557591767382e-05,
+      "loss": 0.8123,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997542586735194e-05,
+      "loss": 0.8578,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997527535770927e-05,
+      "loss": 0.8731,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997512438874812e-05,
+      "loss": 1.0022,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997497296047082e-05,
+      "loss": 0.7565,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9974821072879675e-05,
+      "loss": 0.7337,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997466872597701e-05,
+      "loss": 0.9743,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997451591976517e-05,
+      "loss": 0.9095,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997436265424649e-05,
+      "loss": 0.825,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997420892942333e-05,
+      "loss": 0.8249,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997405474529802e-05,
+      "loss": 0.8442,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997390010187294e-05,
+      "loss": 0.9672,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997374499915047e-05,
+      "loss": 0.8443,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997358943713297e-05,
+      "loss": 0.8127,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9973433415822813e-05,
+      "loss": 0.8469,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9973276935222425e-05,
+      "loss": 0.9713,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9973119995334175e-05,
+      "loss": 0.9485,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9972962596160477e-05,
+      "loss": 0.7318,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997280473770375e-05,
+      "loss": 1.0088,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99726464199664e-05,
+      "loss": 0.7993,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997248764295085e-05,
+      "loss": 0.8106,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9972328406659554e-05,
+      "loss": 0.7506,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997216871109494e-05,
+      "loss": 0.9545,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997200855625944e-05,
+      "loss": 0.8981,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9971847942155536e-05,
+      "loss": 0.7654,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997168686878567e-05,
+      "loss": 0.7834,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997152533615232e-05,
+      "loss": 0.9168,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997136334425795e-05,
+      "loss": 0.8537,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9971200893105054e-05,
+      "loss": 0.9318,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997103798269612e-05,
+      "loss": 0.8469,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997087461303363e-05,
+      "loss": 0.8766,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9970710784120116e-05,
+      "loss": 0.8578,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997054649595805e-05,
+      "loss": 0.864,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997038174854999e-05,
+      "loss": 0.7321,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.997021654189843e-05,
+      "loss": 0.9426,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9970050876005906e-05,
+      "loss": 0.8383,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9969884750874975e-05,
+      "loss": 0.8792,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996971816650817e-05,
+      "loss": 0.902,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9969551122908027e-05,
+      "loss": 1.0665,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996938362007714e-05,
+      "loss": 0.847,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996921565801805e-05,
+      "loss": 0.931,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996904723673334e-05,
+      "loss": 0.7978,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996887835622559e-05,
+      "loss": 0.9978,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996870901649739e-05,
+      "loss": 0.7933,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9968539217551326e-05,
+      "loss": 0.889,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996836895939001e-05,
+      "loss": 0.7613,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996819824201605e-05,
+      "loss": 0.8652,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9968027065432055e-05,
+      "loss": 0.8189,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996785542964065e-05,
+      "loss": 0.8213,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9967683334644474e-05,
+      "loss": 0.8675,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996751078044615e-05,
+      "loss": 0.7646,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996733776704833e-05,
+      "loss": 1.0524,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996716429445367e-05,
+      "loss": 1.0494,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9966990362664817e-05,
+      "loss": 0.8971,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996681597168445e-05,
+      "loss": 0.9459,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996664112151521e-05,
+      "loss": 0.708,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996646581215982e-05,
+      "loss": 0.7374,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9966290043620945e-05,
+      "loss": 1.1285,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.996611381590128e-05,
+      "loss": 0.7535,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9965937129003513e-05,
+      "loss": 0.9954,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9965759982930373e-05,
+      "loss": 0.8349,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9965582377684556e-05,
+      "loss": 0.8473,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99654043132688e-05,
+      "loss": 0.9003,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9965225789685816e-05,
+      "loss": 0.7927,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996504680693835e-05,
+      "loss": 0.9666,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996486736502914e-05,
+      "loss": 0.7886,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996468746396095e-05,
+      "loss": 0.9025,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9964507103736515e-05,
+      "loss": 0.897,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996432628435862e-05,
+      "loss": 0.8488,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996414500583001e-05,
+      "loss": 0.8962,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996396326815349e-05,
+      "loss": 0.9238,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996378107133183e-05,
+      "loss": 0.9288,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996359841536782e-05,
+      "loss": 0.906,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9963415300264266e-05,
+      "loss": 0.7968,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996323172602397e-05,
+      "loss": 0.7679,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996304769264975e-05,
+      "loss": 0.7994,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996286320014441e-05,
+      "loss": 0.8183,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99626782485108e-05,
+      "loss": 0.9435,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996249283775175e-05,
+      "loss": 0.8431,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996230696787008e-05,
+      "loss": 0.8431,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996212063886866e-05,
+      "loss": 1.0331,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9961933850750334e-05,
+      "loss": 0.9845,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9961746603517966e-05,
+      "loss": 0.8106,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9961558897174435e-05,
+      "loss": 0.8136,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996137073172261e-05,
+      "loss": 0.9669,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996118210716536e-05,
+      "loss": 0.9508,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99609930235056e-05,
+      "loss": 0.9125,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996080348074621e-05,
+      "loss": 0.8248,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9960613478890106e-05,
+      "loss": 0.8647,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99604230179402e-05,
+      "loss": 0.9091,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9960232097899394e-05,
+      "loss": 0.8346,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.996004071877063e-05,
+      "loss": 0.9582,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9959848880556823e-05,
+      "loss": 0.945,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995965658326093e-05,
+      "loss": 1.0963,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9959463826885895e-05,
+      "loss": 0.9732,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9959270611434675e-05,
+      "loss": 0.8366,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995907693691022e-05,
+      "loss": 0.9431,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9958882803315493e-05,
+      "loss": 0.7739,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995868821065348e-05,
+      "loss": 1.1275,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9958493158927165e-05,
+      "loss": 0.717,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9958297648139545e-05,
+      "loss": 0.8208,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995810167829358e-05,
+      "loss": 0.8204,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9957905249392315e-05,
+      "loss": 0.6973,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995770836143873e-05,
+      "loss": 0.9405,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9957511014435854e-05,
+      "loss": 0.7458,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9957313208386705e-05,
+      "loss": 0.7586,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9957114943294326e-05,
+      "loss": 0.7448,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9956916219161744e-05,
+      "loss": 0.9175,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995671703599202e-05,
+      "loss": 0.9379,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995651739378818e-05,
+      "loss": 1.2318,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99563172925533e-05,
+      "loss": 1.1403,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995611673229045e-05,
+      "loss": 0.9391,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995591571300269e-05,
+      "loss": 0.8837,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9955714234693115e-05,
+      "loss": 1.1283,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9955512297364806e-05,
+      "loss": 0.9374,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9955309901020845e-05,
+      "loss": 0.8608,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995510704566434e-05,
+      "loss": 0.8234,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995490373129842e-05,
+      "loss": 0.8104,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995469995792618e-05,
+      "loss": 0.7683,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995449572555074e-05,
+      "loss": 0.928,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9954291034175237e-05,
+      "loss": 0.7577,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995408588380281e-05,
+      "loss": 1.048,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9953880274436595e-05,
+      "loss": 0.8069,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995367420607975e-05,
+      "loss": 0.8644,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.995346767873543e-05,
+      "loss": 0.9724,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9953260692406796e-05,
+      "loss": 0.9837,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9953053247097026e-05,
+      "loss": 0.7652,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9952845342809293e-05,
+      "loss": 0.7171,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.995263697954679e-05,
+      "loss": 0.9328,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9952428157312696e-05,
+      "loss": 0.9102,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.995221887611023e-05,
+      "loss": 0.9028,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9952009135942584e-05,
+      "loss": 0.8373,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.995179893681298e-05,
+      "loss": 0.8648,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9951588278724625e-05,
+      "loss": 0.9203,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.995137716168076e-05,
+      "loss": 0.8595,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9951165585684625e-05,
+      "loss": 0.8139,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.995095355073945e-05,
+      "loss": 0.7564,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9950741056848495e-05,
+      "loss": 0.7668,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.995052810401499e-05,
+      "loss": 0.9176,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9950314692242236e-05,
+      "loss": 0.811,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.995010082153348e-05,
+      "loss": 0.7488,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994988649189201e-05,
+      "loss": 0.7549,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9949671703321096e-05,
+      "loss": 0.8461,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9949456455824044e-05,
+      "loss": 0.914,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994924074940414e-05,
+      "loss": 0.8833,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9949024584064695e-05,
+      "loss": 0.8288,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9948807959809025e-05,
+      "loss": 0.8633,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994859087664045e-05,
+      "loss": 0.9357,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994837333456228e-05,
+      "loss": 0.9197,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994815533357787e-05,
+      "loss": 0.7866,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994793687369056e-05,
+      "loss": 0.7726,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994771795490368e-05,
+      "loss": 1.0708,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994749857722059e-05,
+      "loss": 0.8667,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994727874064466e-05,
+      "loss": 0.8503,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9947058445179246e-05,
+      "loss": 1.0177,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994683769082774e-05,
+      "loss": 0.7325,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994661647759351e-05,
+      "loss": 1.0406,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994639480547995e-05,
+      "loss": 0.8536,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994617267449046e-05,
+      "loss": 0.7854,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994595008462845e-05,
+      "loss": 0.878,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9945727035897315e-05,
+      "loss": 1.0301,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9945503528300486e-05,
+      "loss": 0.8075,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994527956184138e-05,
+      "loss": 0.8733,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994505513652343e-05,
+      "loss": 0.8076,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994483025235008e-05,
+      "loss": 0.8532,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9944604909324766e-05,
+      "loss": 0.8301,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9944379107450945e-05,
+      "loss": 0.7425,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994415284673209e-05,
+      "loss": 0.856,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9943926127171655e-05,
+      "loss": 0.6662,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994369894877311e-05,
+      "loss": 0.8684,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9943471311539945e-05,
+      "loss": 1.1098,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994324321547565e-05,
+      "loss": 0.7029,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994301466058371e-05,
+      "loss": 0.8636,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9942785646867634e-05,
+      "loss": 1.117,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9942556174330945e-05,
+      "loss": 0.983,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994232624297713e-05,
+      "loss": 1.03,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994209585280972e-05,
+      "loss": 0.919,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994186500383227e-05,
+      "loss": 0.9317,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9941633696048294e-05,
+      "loss": 0.8625,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994140192946134e-05,
+      "loss": 1.0562,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994116970407495e-05,
+      "loss": 0.7584,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994093701989271e-05,
+      "loss": 0.9831,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994070387691817e-05,
+      "loss": 0.9831,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994047027515489e-05,
+      "loss": 1.1191,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.994023621460647e-05,
+      "loss": 0.7473,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9940001695276486e-05,
+      "loss": 0.7586,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993976671716853e-05,
+      "loss": 0.8445,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.993953128028621e-05,
+      "loss": 1.0233,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9939295384633125e-05,
+      "loss": 0.7492,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9939059030212904e-05,
+      "loss": 1.1344,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993882221702915e-05,
+      "loss": 0.7594,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993858494508551e-05,
+      "loss": 1.02,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99383472143856e-05,
+      "loss": 0.9762,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993810902493308e-05,
+      "loss": 1.0654,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993787037673159e-05,
+      "loss": 0.9311,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9937631269784796e-05,
+      "loss": 0.9599,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993739170409635e-05,
+      "loss": 0.8435,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9937151679669935e-05,
+      "loss": 0.8389,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993691119650921e-05,
+      "loss": 0.9341,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993667025461788e-05,
+      "loss": 0.8105,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993642885399964e-05,
+      "loss": 0.7921,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9936186994658165e-05,
+      "loss": 0.9274,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993594467659719e-05,
+      "loss": 1.0216,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99357018998204e-05,
+      "loss": 1.0943,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993545866433153e-05,
+      "loss": 0.8688,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993521497013431e-05,
+      "loss": 0.9041,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993497081723247e-05,
+      "loss": 0.7394,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9934726205629755e-05,
+      "loss": 0.9419,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99344811353299e-05,
+      "loss": 0.9816,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993423560633668e-05,
+      "loss": 0.9075,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993398961865383e-05,
+      "loss": 0.8914,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993374317228515e-05,
+      "loss": 0.8079,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9933496267234396e-05,
+      "loss": 0.8676,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9933248903505354e-05,
+      "loss": 0.8758,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993300108110183e-05,
+      "loss": 1.011,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9932752800027605e-05,
+      "loss": 0.7534,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9932504060286486e-05,
+      "loss": 0.9675,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9932254861882285e-05,
+      "loss": 0.912,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993200520481882e-05,
+      "loss": 0.8965,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993175508909993e-05,
+      "loss": 0.8117,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993150451472943e-05,
+      "loss": 0.9548,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9931253481711166e-05,
+      "loss": 0.8034,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993100199004898e-05,
+      "loss": 0.929,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993075003974673e-05,
+      "loss": 0.8431,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.993049763080829e-05,
+      "loss": 0.7135,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99302447632375e-05,
+      "loss": 0.8521,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9929991437038256e-05,
+      "loss": 0.8307,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992973765221443e-05,
+      "loss": 1.0561,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9929483408769924e-05,
+      "loss": 1.077,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9929228706708616e-05,
+      "loss": 1.0287,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992897354603441e-05,
+      "loss": 0.8716,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992871792675123e-05,
+      "loss": 0.76,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992846184886299e-05,
+      "loss": 0.9021,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992820531237361e-05,
+      "loss": 0.882,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9927948317287006e-05,
+      "loss": 0.8994,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992769086360714e-05,
+      "loss": 0.8419,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9927432951337945e-05,
+      "loss": 0.911,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992717458048338e-05,
+      "loss": 0.8271,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9926915751047395e-05,
+      "loss": 0.8864,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992665646303396e-05,
+      "loss": 0.9227,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992639671644705e-05,
+      "loss": 0.9066,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992613651129065e-05,
+      "loss": 0.8884,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992587584756873e-05,
+      "loss": 0.8361,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.99256147252853e-05,
+      "loss": 0.8257,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992535314444436e-05,
+      "loss": 1.008,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9925091105049916e-05,
+      "loss": 0.853,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992482860710598e-05,
+      "loss": 0.8186,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992456565061657e-05,
+      "loss": 0.9968,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992430223558573e-05,
+      "loss": 0.6364,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9924038362017485e-05,
+      "loss": 0.7402,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992377402991589e-05,
+      "loss": 1.068,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992350923928497e-05,
+      "loss": 0.8903,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.992324399012881e-05,
+      "loss": 0.8772,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.992297828245146e-05,
+      "loss": 0.8781,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9922712116257e-05,
+      "loss": 0.9323,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.99224454915495e-05,
+      "loss": 0.9038,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.992217840833305e-05,
+      "loss": 1.1202,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.992191086661175e-05,
+      "loss": 0.9032,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9921642866389684e-05,
+      "loss": 0.8321,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.992137440767096e-05,
+      "loss": 0.7374,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.992110549045971e-05,
+      "loss": 0.7157,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.992083611476004e-05,
+      "loss": 0.7648,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.992056628057608e-05,
+      "loss": 0.8537,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9920295987911964e-05,
+      "loss": 0.8808,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.992002523677183e-05,
+      "loss": 0.9013,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9919754027159836e-05,
+      "loss": 0.9486,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9919482359080135e-05,
+      "loss": 0.9169,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991921023253689e-05,
+      "loss": 0.8833,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991893764753427e-05,
+      "loss": 0.821,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991866460407644e-05,
+      "loss": 0.8764,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991839110216761e-05,
+      "loss": 0.9252,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991811714181194e-05,
+      "loss": 0.8438,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9917842723013655e-05,
+      "loss": 1.1417,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991756784577695e-05,
+      "loss": 0.8777,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991729251010603e-05,
+      "loss": 1.0805,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9917016716005115e-05,
+      "loss": 0.746,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991674046347845e-05,
+      "loss": 0.9589,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991646375253024e-05,
+      "loss": 0.7949,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991618658316476e-05,
+      "loss": 0.8626,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991590895538622e-05,
+      "loss": 0.7865,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9915630869198886e-05,
+      "loss": 1.2073,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991535232460703e-05,
+      "loss": 0.8267,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991507332161492e-05,
+      "loss": 0.7372,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991479386022682e-05,
+      "loss": 1.031,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991451394044701e-05,
+      "loss": 0.8724,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9914233562279796e-05,
+      "loss": 0.9843,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991395272572946e-05,
+      "loss": 0.7353,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9913671430800316e-05,
+      "loss": 0.8012,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9913389677496665e-05,
+      "loss": 0.6741,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9913107465822835e-05,
+      "loss": 0.77,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991282479578313e-05,
+      "loss": 0.7302,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9912541667381906e-05,
+      "loss": 0.8985,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991225808062348e-05,
+      "loss": 0.8717,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991197403551221e-05,
+      "loss": 0.7076,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991168953205245e-05,
+      "loss": 0.8192,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991140457024856e-05,
+      "loss": 0.7473,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991111915010489e-05,
+      "loss": 0.8094,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991083327162585e-05,
+      "loss": 0.7537,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991054693481577e-05,
+      "loss": 0.7281,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.991026013967908e-05,
+      "loss": 0.8761,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990997288622015e-05,
+      "loss": 0.825,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9909685174443385e-05,
+      "loss": 0.8176,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990939700435321e-05,
+      "loss": 0.9537,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990910837595403e-05,
+      "loss": 0.7472,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990881928925027e-05,
+      "loss": 0.811,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990852974424635e-05,
+      "loss": 0.9742,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9908239740946716e-05,
+      "loss": 0.9083,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9907949279355806e-05,
+      "loss": 0.8881,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990765835947807e-05,
+      "loss": 0.8418,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990736698131798e-05,
+      "loss": 0.9111,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990707514487999e-05,
+      "loss": 0.8843,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990678285016857e-05,
+      "loss": 1.0071,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990649009718821e-05,
+      "loss": 1.0337,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990619688594338e-05,
+      "loss": 1.0019,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990590321643858e-05,
+      "loss": 0.8831,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.990560908867832e-05,
+      "loss": 0.7788,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.990531450266709e-05,
+      "loss": 0.9166,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.990501945840941e-05,
+      "loss": 1.0725,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.990472395590981e-05,
+      "loss": 0.7664,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9904427995172805e-05,
+      "loss": 0.7561,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9904131576202944e-05,
+      "loss": 0.7697,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.990383469900475e-05,
+      "loss": 0.7639,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9903537363582795e-05,
+      "loss": 0.927,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.990323956994161e-05,
+      "loss": 0.8465,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9902941318085785e-05,
+      "loss": 1.0513,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9902642608019866e-05,
+      "loss": 0.8759,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.990234343974845e-05,
+      "loss": 0.8522,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9902043813276104e-05,
+      "loss": 0.9943,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9901743728607435e-05,
+      "loss": 0.8627,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.990144318574703e-05,
+      "loss": 0.7221,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.99011421846995e-05,
+      "loss": 0.7463,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.990084072546945e-05,
+      "loss": 0.7501,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.990053880806151e-05,
+      "loss": 0.8836,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.99002364324803e-05,
+      "loss": 0.923,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989993359873045e-05,
+      "loss": 0.8502,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989963030681661e-05,
+      "loss": 0.7621,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989932655674342e-05,
+      "loss": 0.8334,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989902234851554e-05,
+      "loss": 0.9882,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9898717682137624e-05,
+      "loss": 0.6677,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9898412557614334e-05,
+      "loss": 0.7703,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989810697495038e-05,
+      "loss": 1.1109,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98978009341504e-05,
+      "loss": 0.7931,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989749443521912e-05,
+      "loss": 0.7591,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989718747816121e-05,
+      "loss": 1.001,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989688006298138e-05,
+      "loss": 0.9608,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989657218968435e-05,
+      "loss": 0.9758,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989626385827484e-05,
+      "loss": 0.6997,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989595506875757e-05,
+      "loss": 0.9803,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989564582113725e-05,
+      "loss": 0.7584,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989533611541865e-05,
+      "loss": 0.9052,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989502595160651e-05,
+      "loss": 0.9274,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9894715329705564e-05,
+      "loss": 0.8398,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989440424972058e-05,
+      "loss": 0.9017,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989409271165634e-05,
+      "loss": 0.8044,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98937807155176e-05,
+      "loss": 0.8358,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989346826130915e-05,
+      "loss": 0.7417,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989315534903577e-05,
+      "loss": 0.7666,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989284197870226e-05,
+      "loss": 0.7881,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989252815031342e-05,
+      "loss": 0.9064,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989221386387406e-05,
+      "loss": 0.9215,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9891899119389e-05,
+      "loss": 0.9661,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989158391686307e-05,
+      "loss": 0.7942,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989126825630107e-05,
+      "loss": 0.8392,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989095213770787e-05,
+      "loss": 0.909,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989063556108829e-05,
+      "loss": 0.8974,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989031852644719e-05,
+      "loss": 1.0421,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.989000103378944e-05,
+      "loss": 0.756,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988968308311989e-05,
+      "loss": 0.7742,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988936467444341e-05,
+      "loss": 0.9328,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988904580776489e-05,
+      "loss": 0.9459,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988872648308921e-05,
+      "loss": 0.7283,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988840670042126e-05,
+      "loss": 0.8277,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988808645976595e-05,
+      "loss": 0.784,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9887765761128184e-05,
+      "loss": 0.7785,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9887444604512865e-05,
+      "loss": 0.8355,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988712298992492e-05,
+      "loss": 0.911,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988680091736929e-05,
+      "loss": 0.8178,
+      "step": 727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9886478386850894e-05,
+      "loss": 0.9822,
+      "step": 728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988615539837469e-05,
+      "loss": 0.7367,
+      "step": 729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.988583195194561e-05,
+      "loss": 0.8836,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988550804756861e-05,
+      "loss": 1.021,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988518368524867e-05,
+      "loss": 0.7187,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9884858864990756e-05,
+      "loss": 0.9138,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9884533586799824e-05,
+      "loss": 0.8025,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988420785068089e-05,
+      "loss": 0.8667,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988388165663893e-05,
+      "loss": 0.9239,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9883555004678936e-05,
+      "loss": 0.8557,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988322789480592e-05,
+      "loss": 0.8387,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9882900327024895e-05,
+      "loss": 0.7696,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988257230134089e-05,
+      "loss": 1.0313,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9882243817758904e-05,
+      "loss": 0.864,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9881914876284e-05,
+      "loss": 0.9093,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988158547692121e-05,
+      "loss": 0.8476,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988125561967557e-05,
+      "loss": 1.0155,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988092530455213e-05,
+      "loss": 0.8061,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988059453155598e-05,
+      "loss": 0.8285,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.988026330069217e-05,
+      "loss": 0.8105,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9879931611965774e-05,
+      "loss": 0.9237,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9879599465381876e-05,
+      "loss": 0.9502,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987926686094557e-05,
+      "loss": 0.8415,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9878933798661956e-05,
+      "loss": 0.7335,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9878600278536125e-05,
+      "loss": 1.0758,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987826630057319e-05,
+      "loss": 0.7395,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9877931864778284e-05,
+      "loss": 0.7736,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987759697115652e-05,
+      "loss": 0.7338,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9877261619713025e-05,
+      "loss": 0.8759,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987692581045294e-05,
+      "loss": 0.7644,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987658954338142e-05,
+      "loss": 0.8815,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9876252818503614e-05,
+      "loss": 1.0604,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9875915635824673e-05,
+      "loss": 0.933,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987557799534977e-05,
+      "loss": 0.7447,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987523989708409e-05,
+      "loss": 0.8687,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9874901341032786e-05,
+      "loss": 1.02,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987456232720107e-05,
+      "loss": 0.8493,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987422285559413e-05,
+      "loss": 0.9866,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987388292621716e-05,
+      "loss": 0.8756,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987354253907538e-05,
+      "loss": 0.8083,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9873201694174004e-05,
+      "loss": 0.7893,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987286039151825e-05,
+      "loss": 1.0375,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987251863111335e-05,
+      "loss": 0.9659,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9872176412964544e-05,
+      "loss": 1.0739,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987183373707707e-05,
+      "loss": 0.8364,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987149060345618e-05,
+      "loss": 0.6627,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9871147012107135e-05,
+      "loss": 0.7412,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9870802963035195e-05,
+      "loss": 0.8583,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.987045845624565e-05,
+      "loss": 1.0115,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9870113491743755e-05,
+      "loss": 0.8234,
+      "step": 777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.98697680695348e-05,
+      "loss": 0.8871,
+      "step": 778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986942218962409e-05,
+      "loss": 1.1474,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9869075852016914e-05,
+      "loss": 0.8196,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9868729056718584e-05,
+      "loss": 0.8978,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986838180373442e-05,
+      "loss": 1.096,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9868034093069734e-05,
+      "loss": 0.6886,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986768592472986e-05,
+      "loss": 0.9357,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9867337298720124e-05,
+      "loss": 0.9279,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986698821504588e-05,
+      "loss": 0.9726,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986663867371246e-05,
+      "loss": 0.9913,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9866288674725236e-05,
+      "loss": 0.8175,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986593821808957e-05,
+      "loss": 1.0469,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9865587303810826e-05,
+      "loss": 0.7551,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9865235931894386e-05,
+      "loss": 0.8113,
+      "step": 791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986488410234563e-05,
+      "loss": 0.6639,
+      "step": 792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986453181516995e-05,
+      "loss": 0.6726,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.986417907037274e-05,
+      "loss": 0.9369,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9863825867959414e-05,
+      "loss": 0.8101,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9863472207935384e-05,
+      "loss": 0.6088,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.986311809030606e-05,
+      "loss": 1.0567,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.986276351507688e-05,
+      "loss": 0.9965,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.986240848225327e-05,
+      "loss": 0.803,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.986205299184066e-05,
+      "loss": 0.7703,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.986169704384452e-05,
+      "loss": 0.8331,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.986134063827029e-05,
+      "loss": 0.7779,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9860983775123437e-05,
+      "loss": 0.8846,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9860626454409424e-05,
+      "loss": 0.7158,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.986026867613373e-05,
+      "loss": 0.9338,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985991044030184e-05,
+      "loss": 0.7266,
+      "step": 806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985955174691923e-05,
+      "loss": 0.7591,
+      "step": 807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9859192595991405e-05,
+      "loss": 0.905,
+      "step": 808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985883298752388e-05,
+      "loss": 0.9269,
+      "step": 809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985847292152215e-05,
+      "loss": 0.7954,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9858112397991734e-05,
+      "loss": 0.7699,
+      "step": 811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985775141693816e-05,
+      "loss": 1.0019,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9857389978366955e-05,
+      "loss": 0.8266,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985702808228367e-05,
+      "loss": 0.893,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9856665728693835e-05,
+      "loss": 0.8974,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985630291760301e-05,
+      "loss": 0.7962,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9855939649016753e-05,
+      "loss": 0.7538,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985557592294063e-05,
+      "loss": 0.8232,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985521173938023e-05,
+      "loss": 1.0387,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9854847098341105e-05,
+      "loss": 0.9673,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985448199982886e-05,
+      "loss": 0.8115,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9854116443849077e-05,
+      "loss": 0.8857,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9853750430407374e-05,
+      "loss": 0.9108,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9853383959509347e-05,
+      "loss": 0.7428,
+      "step": 824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985301703116062e-05,
+      "loss": 0.823,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9852649645366814e-05,
+      "loss": 0.7972,
+      "step": 826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985228180213355e-05,
+      "loss": 0.8704,
+      "step": 827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9851913501466474e-05,
+      "loss": 1.1143,
+      "step": 828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985154474337123e-05,
+      "loss": 0.8172,
+      "step": 829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985117552785346e-05,
+      "loss": 0.8835,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9850805854918824e-05,
+      "loss": 0.8295,
+      "step": 831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9850435724573e-05,
+      "loss": 0.7639,
+      "step": 832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.985006513682164e-05,
+      "loss": 0.8652,
+      "step": 833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984969409167043e-05,
+      "loss": 1.0861,
+      "step": 834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9849322589125066e-05,
+      "loss": 0.953,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9848950629191224e-05,
+      "loss": 0.8387,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984857821187462e-05,
+      "loss": 0.9582,
+      "step": 837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984820533718094e-05,
+      "loss": 0.831,
+      "step": 838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9847832005115924e-05,
+      "loss": 0.8556,
+      "step": 839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984745821568526e-05,
+      "loss": 0.9013,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984708396889471e-05,
+      "loss": 0.8172,
+      "step": 841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984670926474998e-05,
+      "loss": 0.5962,
+      "step": 842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9846334103256836e-05,
+      "loss": 0.9005,
+      "step": 843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9845958484421e-05,
+      "loss": 0.8582,
+      "step": 844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9845582408248263e-05,
+      "loss": 0.7971,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9845205874744356e-05,
+      "loss": 0.9033,
+      "step": 846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9844828883915055e-05,
+      "loss": 0.8592,
+      "step": 847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984445143576615e-05,
+      "loss": 0.9385,
+      "step": 848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984407353030341e-05,
+      "loss": 0.875,
+      "step": 849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9843695167532645e-05,
+      "loss": 0.9173,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9843316347459624e-05,
+      "loss": 0.7719,
+      "step": 851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9842937070090185e-05,
+      "loss": 0.7226,
+      "step": 852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9842557335430107e-05,
+      "loss": 0.7658,
+      "step": 853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984217714348523e-05,
+      "loss": 0.8783,
+      "step": 854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984179649426137e-05,
+      "loss": 0.9692,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984141538776437e-05,
+      "loss": 0.7021,
+      "step": 856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984103382400006e-05,
+      "loss": 0.9538,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9840651802974295e-05,
+      "loss": 1.0153,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9840269324692915e-05,
+      "loss": 0.8221,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9839886389161804e-05,
+      "loss": 0.9816,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98395029963868e-05,
+      "loss": 1.1447,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9839119146373805e-05,
+      "loss": 0.894,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9838734839128686e-05,
+      "loss": 0.9561,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983835007465734e-05,
+      "loss": 0.7769,
+      "step": 864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983796485296564e-05,
+      "loss": 1.0024,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9837579174059525e-05,
+      "loss": 0.8658,
+      "step": 866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9837193037944885e-05,
+      "loss": 0.8068,
+      "step": 867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983680644462764e-05,
+      "loss": 0.9669,
+      "step": 868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9836419394113704e-05,
+      "loss": 0.7925,
+      "step": 869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983603188640903e-05,
+      "loss": 0.8498,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9835643921519525e-05,
+      "loss": 0.8575,
+      "step": 871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983525549945116e-05,
+      "loss": 0.8363,
+      "step": 872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983486662020988e-05,
+      "loss": 0.9318,
+      "step": 873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9834477283801636e-05,
+      "loss": 0.8276,
+      "step": 874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983408749023241e-05,
+      "loss": 0.898,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9833697239508155e-05,
+      "loss": 0.8422,
+      "step": 876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983330653163486e-05,
+      "loss": 1.0477,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983291536661852e-05,
+      "loss": 0.8668,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9832523744465114e-05,
+      "loss": 0.7804,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9832131665180656e-05,
+      "loss": 1.028,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983173912877115e-05,
+      "loss": 0.8649,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983134613524261e-05,
+      "loss": 0.8768,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9830952684601056e-05,
+      "loss": 0.8076,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9830558776852515e-05,
+      "loss": 1.0186,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9830164412003016e-05,
+      "loss": 0.882,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9829769590058624e-05,
+      "loss": 0.8239,
+      "step": 886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9829374311025385e-05,
+      "loss": 0.8136,
+      "step": 887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982897857490934e-05,
+      "loss": 0.8128,
+      "step": 888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982858238171655e-05,
+      "loss": 0.8216,
+      "step": 889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9828185731453116e-05,
+      "loss": 0.7389,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9827788624125084e-05,
+      "loss": 0.836,
+      "step": 891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982739105973856e-05,
+      "loss": 0.7635,
+      "step": 892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982699303829963e-05,
+      "loss": 0.9204,
+      "step": 893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982659455981438e-05,
+      "loss": 1.044,
+      "step": 894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982619562428892e-05,
+      "loss": 0.7724,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982579623172939e-05,
+      "loss": 0.9329,
+      "step": 896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9825396382141876e-05,
+      "loss": 0.8432,
+      "step": 897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982499607553253e-05,
+      "loss": 0.8897,
+      "step": 898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9824595311907465e-05,
+      "loss": 0.9274,
+      "step": 899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9824194091272834e-05,
+      "loss": 0.7944,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9823792413634783e-05,
+      "loss": 0.902,
+      "step": 901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9823390278999466e-05,
+      "loss": 0.8976,
+      "step": 902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982298768737305e-05,
+      "loss": 0.76,
+      "step": 903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9822584638761696e-05,
+      "loss": 0.7936,
+      "step": 904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982218113317159e-05,
+      "loss": 0.7879,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9821777170608896e-05,
+      "loss": 0.7945,
+      "step": 906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982137275107983e-05,
+      "loss": 0.9249,
+      "step": 907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9820967874590566e-05,
+      "loss": 0.8274,
+      "step": 908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9820562541147324e-05,
+      "loss": 0.882,
+      "step": 909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982015675075631e-05,
+      "loss": 0.7507,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981975050342374e-05,
+      "loss": 0.9393,
+      "step": 911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9819343799155836e-05,
+      "loss": 0.9018,
+      "step": 912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981893663795884e-05,
+      "loss": 0.7534,
+      "step": 913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981852901983898e-05,
+      "loss": 0.6655,
+      "step": 914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98181209448025e-05,
+      "loss": 0.7613,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.981771241285567e-05,
+      "loss": 0.7969,
+      "step": 916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9817303424004746e-05,
+      "loss": 0.7559,
+      "step": 917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9816893978255977e-05,
+      "loss": 0.7587,
+      "step": 918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9816484075615656e-05,
+      "loss": 0.8534,
+      "step": 919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9816073716090054e-05,
+      "loss": 0.7129,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9815662899685464e-05,
+      "loss": 0.7965,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9815251626408175e-05,
+      "loss": 0.7879,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9814839896264496e-05,
+      "loss": 0.7877,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9814427709260736e-05,
+      "loss": 0.8908,
+      "step": 924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981401506540321e-05,
+      "loss": 0.8852,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981360196469824e-05,
+      "loss": 0.8655,
+      "step": 926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981318840715215e-05,
+      "loss": 1.055,
+      "step": 927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981277439277128e-05,
+      "loss": 0.8767,
+      "step": 928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981235992156198e-05,
+      "loss": 0.7783,
+      "step": 929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9811944993530605e-05,
+      "loss": 0.6938,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9811529608683494e-05,
+      "loss": 0.6072,
+      "step": 931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981111376702703e-05,
+      "loss": 0.9558,
+      "step": 932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981069746856758e-05,
+      "loss": 0.9704,
+      "step": 933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9810280713311515e-05,
+      "loss": 0.8251,
+      "step": 934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980986350126523e-05,
+      "loss": 0.8155,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980944583243512e-05,
+      "loss": 0.9107,
+      "step": 936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9809027706827585e-05,
+      "loss": 1.0305,
+      "step": 937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980860912444902e-05,
+      "loss": 0.8768,
+      "step": 938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9808190085305854e-05,
+      "loss": 0.9857,
+      "step": 939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9807770589404486e-05,
+      "loss": 0.8706,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980735063675137e-05,
+      "loss": 0.8727,
+      "step": 941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980693022735293e-05,
+      "loss": 0.8388,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980650936121562e-05,
+      "loss": 0.7987,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9806088038345856e-05,
+      "loss": 0.7027,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980566625875012e-05,
+      "loss": 0.9746,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9805244022434875e-05,
+      "loss": 0.9148,
+      "step": 946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980482132940659e-05,
+      "loss": 0.9275,
+      "step": 947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9804398179671735e-05,
+      "loss": 0.9503,
+      "step": 948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980397457323679e-05,
+      "loss": 0.9835,
+      "step": 949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980355051010826e-05,
+      "loss": 0.8661,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9803125990292644e-05,
+      "loss": 0.7379,
+      "step": 951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980270101379643e-05,
+      "loss": 0.733,
+      "step": 952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9802275580626144e-05,
+      "loss": 0.9211,
+      "step": 953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9801849690788296e-05,
+      "loss": 0.9228,
+      "step": 954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980142334428941e-05,
+      "loss": 0.8266,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980099654113603e-05,
+      "loss": 1.028,
+      "step": 956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.98005692813347e-05,
+      "loss": 0.7947,
+      "step": 957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.980014156489196e-05,
+      "loss": 0.7236,
+      "step": 958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979971339181435e-05,
+      "loss": 0.9455,
+      "step": 959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979928476210846e-05,
+      "loss": 0.7984,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979885567578083e-05,
+      "loss": 0.7904,
+      "step": 961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979842613283804e-05,
+      "loss": 1.0285,
+      "step": 962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979799613328669e-05,
+      "loss": 0.6664,
+      "step": 963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979756567713335e-05,
+      "loss": 0.6963,
+      "step": 964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9797134764384626e-05,
+      "loss": 0.7891,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9796703395047125e-05,
+      "loss": 1.0428,
+      "step": 966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9796271569127444e-05,
+      "loss": 1.0194,
+      "step": 967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979583928663221e-05,
+      "loss": 0.9265,
+      "step": 968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9795406547568043e-05,
+      "loss": 0.9694,
+      "step": 969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9794973351941565e-05,
+      "loss": 1.1393,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979453969975943e-05,
+      "loss": 0.7714,
+      "step": 971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979410559102828e-05,
+      "loss": 0.9637,
+      "step": 972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979367102575476e-05,
+      "loss": 1.0506,
+      "step": 973
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979323600394553e-05,
+      "loss": 0.9357,
+      "step": 974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979280052560726e-05,
+      "loss": 0.8839,
+      "step": 975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979236459074662e-05,
+      "loss": 0.8171,
+      "step": 976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979192819937029e-05,
+      "loss": 1.0136,
+      "step": 977
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979149135148496e-05,
+      "loss": 0.9451,
+      "step": 978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.979105404709732e-05,
+      "loss": 0.8546,
+      "step": 979
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9790616286214086e-05,
+      "loss": 0.8665,
+      "step": 980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9790178068841934e-05,
+      "loss": 0.881,
+      "step": 981
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9789739394987606e-05,
+      "loss": 0.9077,
+      "step": 982
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9789300264657816e-05,
+      "loss": 1.0465,
+      "step": 983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.978886067785929e-05,
+      "loss": 0.9498,
+      "step": 984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9788420634598764e-05,
+      "loss": 1.0453,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978798013488299e-05,
+      "loss": 0.7222,
+      "step": 986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9787539178718703e-05,
+      "loss": 0.9895,
+      "step": 987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978709776611268e-05,
+      "loss": 0.7401,
+      "step": 988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978665589707167e-05,
+      "loss": 0.9339,
+      "step": 989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978621357160243e-05,
+      "loss": 0.7115,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978577078971177e-05,
+      "loss": 0.7897,
+      "step": 991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9785327551406454e-05,
+      "loss": 1.0782,
+      "step": 992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978488385669329e-05,
+      "loss": 0.9501,
+      "step": 993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978443970557905e-05,
+      "loss": 1.0111,
+      "step": 994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978399509807056e-05,
+      "loss": 0.8385,
+      "step": 995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978355003417464e-05,
+      "loss": 0.8583,
+      "step": 996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9783104513898085e-05,
+      "loss": 0.7796,
+      "step": 997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978265853724775e-05,
+      "loss": 0.8436,
+      "step": 998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9782212104230444e-05,
+      "loss": 0.786,
+      "step": 999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978176521485301e-05,
+      "loss": 1.0197,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978131786912232e-05,
+      "loss": 0.748,
+      "step": 1001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9780870067045206e-05,
+      "loss": 0.8145,
+      "step": 1002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.978042180862853e-05,
+      "loss": 1.0008,
+      "step": 1003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9779973093879175e-05,
+      "loss": 0.778,
+      "step": 1004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977952392280401e-05,
+      "loss": 0.9933,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977907429540991e-05,
+      "loss": 0.8687,
+      "step": 1006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977862421170378e-05,
+      "loss": 0.8621,
+      "step": 1007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977817367169251e-05,
+      "loss": 1.1291,
+      "step": 1008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977772267538299e-05,
+      "loss": 0.8729,
+      "step": 1009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977727122278215e-05,
+      "loss": 1.1699,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97768193138969e-05,
+      "loss": 0.906,
+      "step": 1011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977636694873416e-05,
+      "loss": 0.9539,
+      "step": 1012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977591412730087e-05,
+      "loss": 0.8883,
+      "step": 1013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9775460849603964e-05,
+      "loss": 0.8311,
+      "step": 1014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97750071156504e-05,
+      "loss": 0.7805,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97745529254471e-05,
+      "loss": 0.7459,
+      "step": 1016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9774098279001054e-05,
+      "loss": 0.8449,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977364317631922e-05,
+      "loss": 0.9166,
+      "step": 1018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9773187617408566e-05,
+      "loss": 1.1148,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977273160227608e-05,
+      "loss": 0.8874,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9772275130928744e-05,
+      "loss": 0.8364,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977181820337355e-05,
+      "loss": 1.0594,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977136081961752e-05,
+      "loss": 0.8626,
+      "step": 1023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977090297966764e-05,
+      "loss": 0.9608,
+      "step": 1024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.977044468353092e-05,
+      "loss": 0.799,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976998593121441e-05,
+      "loss": 0.8722,
+      "step": 1026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976952672272512e-05,
+      "loss": 0.9192,
+      "step": 1027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976906705807009e-05,
+      "loss": 0.7561,
+      "step": 1028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976860693725636e-05,
+      "loss": 0.6473,
+      "step": 1029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9768146360290994e-05,
+      "loss": 0.8606,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976768532718104e-05,
+      "loss": 0.8529,
+      "step": 1031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976722383793355e-05,
+      "loss": 0.9543,
+      "step": 1032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976676189255563e-05,
+      "loss": 0.9164,
+      "step": 1033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9766299491054315e-05,
+      "loss": 0.7367,
+      "step": 1034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976583663343673e-05,
+      "loss": 0.8902,
+      "step": 1035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9765373319709945e-05,
+      "loss": 0.9454,
+      "step": 1036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976490954988107e-05,
+      "loss": 0.9793,
+      "step": 1037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.97644453239572e-05,
+      "loss": 0.9826,
+      "step": 1038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976398064194545e-05,
+      "loss": 1.0553,
+      "step": 1039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976351550385296e-05,
+      "loss": 0.7749,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9763049909686834e-05,
+      "loss": 0.9769,
+      "step": 1041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9762583859454214e-05,
+      "loss": 0.9582,
+      "step": 1042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9762117353162244e-05,
+      "loss": 0.8001,
+      "step": 1043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9761650390818065e-05,
+      "loss": 0.8592,
+      "step": 1044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976118297242884e-05,
+      "loss": 0.7934,
+      "step": 1045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976071509800173e-05,
+      "loss": 0.8841,
+      "step": 1046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.976024676754391e-05,
+      "loss": 0.8831,
+      "step": 1047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975977798106255e-05,
+      "loss": 0.857,
+      "step": 1048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9759308738564826e-05,
+      "loss": 0.729,
+      "step": 1049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975883904005794e-05,
+      "loss": 0.8461,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9758368885549087e-05,
+      "loss": 0.8665,
+      "step": 1051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975789827504547e-05,
+      "loss": 0.8269,
+      "step": 1052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97574272085543e-05,
+      "loss": 0.9697,
+      "step": 1053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975695568608279e-05,
+      "loss": 1.1683,
+      "step": 1054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9756483707638175e-05,
+      "loss": 0.8755,
+      "step": 1055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975601127322768e-05,
+      "loss": 0.6962,
+      "step": 1056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9755538382858545e-05,
+      "loss": 0.8464,
+      "step": 1057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975506503653802e-05,
+      "loss": 0.8519,
+      "step": 1058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9754591234273356e-05,
+      "loss": 0.9124,
+      "step": 1059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975411697607182e-05,
+      "loss": 0.9246,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975364226194067e-05,
+      "loss": 0.7844,
+      "step": 1061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9753167091887176e-05,
+      "loss": 0.8842,
+      "step": 1062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975269146591863e-05,
+      "loss": 0.8765,
+      "step": 1063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975221538404231e-05,
+      "loss": 1.018,
+      "step": 1064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9751738846265525e-05,
+      "loss": 1.0125,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975126185259556e-05,
+      "loss": 0.7131,
+      "step": 1066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975078440303974e-05,
+      "loss": 0.8449,
+      "step": 1067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.975030649760538e-05,
+      "loss": 0.8284,
+      "step": 1068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974982813629978e-05,
+      "loss": 0.7269,
+      "step": 1069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974934931913031e-05,
+      "loss": 0.9882,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974887004610427e-05,
+      "loss": 0.8423,
+      "step": 1071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974839031722902e-05,
+      "loss": 0.8231,
+      "step": 1072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974791013251191e-05,
+      "loss": 0.9403,
+      "step": 1073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9747429491960296e-05,
+      "loss": 0.7093,
+      "step": 1074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974694839558155e-05,
+      "loss": 0.8607,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974646684338304e-05,
+      "loss": 1.0001,
+      "step": 1076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9745984835372146e-05,
+      "loss": 0.835,
+      "step": 1077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974550237155624e-05,
+      "loss": 0.8862,
+      "step": 1078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974501945194273e-05,
+      "loss": 0.9405,
+      "step": 1079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974453607653902e-05,
+      "loss": 0.9753,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97440522453525e-05,
+      "loss": 0.8341,
+      "step": 1081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9743567958390585e-05,
+      "loss": 0.9724,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974308321566072e-05,
+      "loss": 0.9773,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974259801717031e-05,
+      "loss": 0.7474,
+      "step": 1084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9742112362926795e-05,
+      "loss": 0.8137,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974162625293762e-05,
+      "loss": 0.5769,
+      "step": 1086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9741139687210234e-05,
+      "loss": 0.805,
+      "step": 1087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9740652665752085e-05,
+      "loss": 0.7965,
+      "step": 1088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.974016518857065e-05,
+      "loss": 1.0236,
+      "step": 1089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973967725567338e-05,
+      "loss": 0.7625,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973918886706777e-05,
+      "loss": 0.8378,
+      "step": 1091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97387000227613e-05,
+      "loss": 1.0023,
+      "step": 1092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973821072276144e-05,
+      "loss": 0.7123,
+      "step": 1093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9737720967075714e-05,
+      "loss": 1.0749,
+      "step": 1094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973723075571162e-05,
+      "loss": 0.8796,
+      "step": 1095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9736740088676655e-05,
+      "loss": 0.9707,
+      "step": 1096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973624896597836e-05,
+      "loss": 0.8163,
+      "step": 1097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973575738762424e-05,
+      "loss": 0.9002,
+      "step": 1098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9735265353621844e-05,
+      "loss": 0.7687,
+      "step": 1099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.97347728639787e-05,
+      "loss": 0.6943,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973427991870236e-05,
+      "loss": 0.9981,
+      "step": 1101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973378651780038e-05,
+      "loss": 0.954,
+      "step": 1102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9733292661280314e-05,
+      "loss": 1.1031,
+      "step": 1103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973279834914974e-05,
+      "loss": 0.7892,
+      "step": 1104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9732303581416214e-05,
+      "loss": 0.8691,
+      "step": 1105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973180835808733e-05,
+      "loss": 0.8869,
+      "step": 1106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.973131267917068e-05,
+      "loss": 0.8572,
+      "step": 1107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9730816544673855e-05,
+      "loss": 0.8481,
+      "step": 1108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9730319954604456e-05,
+      "loss": 0.7045,
+      "step": 1109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.972982290897009e-05,
+      "loss": 0.879,
+      "step": 1110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9729325407778384e-05,
+      "loss": 0.9144,
+      "step": 1111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9728827451036944e-05,
+      "loss": 0.9204,
+      "step": 1112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972832903875341e-05,
+      "loss": 0.9748,
+      "step": 1113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9727830170935426e-05,
+      "loss": 0.9954,
+      "step": 1114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972733084759063e-05,
+      "loss": 0.811,
+      "step": 1115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972683106872667e-05,
+      "loss": 0.8801,
+      "step": 1116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9726330834351204e-05,
+      "loss": 0.7299,
+      "step": 1117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972583014447191e-05,
+      "loss": 0.8624,
+      "step": 1118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972532899909645e-05,
+      "loss": 0.784,
+      "step": 1119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9724827398232494e-05,
+      "loss": 0.6405,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972432534188774e-05,
+      "loss": 0.8506,
+      "step": 1121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972382283006988e-05,
+      "loss": 0.9067,
+      "step": 1122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972331986278662e-05,
+      "loss": 0.8447,
+      "step": 1123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972281644004565e-05,
+      "loss": 0.9632,
+      "step": 1124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9722312561854696e-05,
+      "loss": 0.692,
+      "step": 1125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9721808228221484e-05,
+      "loss": 0.7574,
+      "step": 1126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.972130343915373e-05,
+      "loss": 0.8349,
+      "step": 1127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9720798194659176e-05,
+      "loss": 1.135,
+      "step": 1128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9720292494745564e-05,
+      "loss": 1.077,
+      "step": 1129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9719786339420636e-05,
+      "loss": 0.8246,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971927972869216e-05,
+      "loss": 0.8678,
+      "step": 1131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9718772662567886e-05,
+      "loss": 0.8268,
+      "step": 1132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9718265141055595e-05,
+      "loss": 1.1035,
+      "step": 1133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971775716416306e-05,
+      "loss": 0.8371,
+      "step": 1134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971724873189806e-05,
+      "loss": 0.7949,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971673984426839e-05,
+      "loss": 1.0459,
+      "step": 1136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971623050128184e-05,
+      "loss": 0.6803,
+      "step": 1137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9715720702946236e-05,
+      "loss": 0.8911,
+      "step": 1138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971521044926937e-05,
+      "loss": 0.9266,
+      "step": 1139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971469974025907e-05,
+      "loss": 0.8469,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971418857592315e-05,
+      "loss": 0.9847,
+      "step": 1141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971367695626946e-05,
+      "loss": 0.8731,
+      "step": 1142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971316488130583e-05,
+      "loss": 0.8558,
+      "step": 1143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.97126523510401e-05,
+      "loss": 0.8551,
+      "step": 1144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9712139365480135e-05,
+      "loss": 0.8839,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.97116259246338e-05,
+      "loss": 0.9231,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971111202850895e-05,
+      "loss": 0.8687,
+      "step": 1147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971059767711346e-05,
+      "loss": 0.9955,
+      "step": 1148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.971008287045521e-05,
+      "loss": 0.9291,
+      "step": 1149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9709567608542105e-05,
+      "loss": 0.765,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970905189138203e-05,
+      "loss": 0.8366,
+      "step": 1151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970853571898287e-05,
+      "loss": 0.9868,
+      "step": 1152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9708019091352575e-05,
+      "loss": 1.0038,
+      "step": 1153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970750200849902e-05,
+      "loss": 0.8568,
+      "step": 1154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970698447043016e-05,
+      "loss": 1.1827,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9706466477153895e-05,
+      "loss": 0.9209,
+      "step": 1156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970594802867819e-05,
+      "loss": 0.8104,
+      "step": 1157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9705429125010976e-05,
+      "loss": 0.7765,
+      "step": 1158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.97049097661602e-05,
+      "loss": 1.0098,
+      "step": 1159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970438995213384e-05,
+      "loss": 0.7813,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970386968293984e-05,
+      "loss": 1.0445,
+      "step": 1161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9703348958586176e-05,
+      "loss": 0.7753,
+      "step": 1162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9702827779080834e-05,
+      "loss": 0.7541,
+      "step": 1163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.97023061444318e-05,
+      "loss": 0.8339,
+      "step": 1164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9701784054647064e-05,
+      "loss": 0.8985,
+      "step": 1165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9701261509734626e-05,
+      "loss": 0.7796,
+      "step": 1166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9700738509702496e-05,
+      "loss": 0.8548,
+      "step": 1167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.970021505455869e-05,
+      "loss": 0.8969,
+      "step": 1168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969969114431121e-05,
+      "loss": 0.9256,
+      "step": 1169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969916677896811e-05,
+      "loss": 0.8969,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969864195853742e-05,
+      "loss": 0.9286,
+      "step": 1171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.969811668302716e-05,
+      "loss": 0.8449,
+      "step": 1172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9697590952445407e-05,
+      "loss": 0.8968,
+      "step": 1173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9697064766800194e-05,
+      "loss": 1.0222,
+      "step": 1174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.96965381260996e-05,
+      "loss": 1.0171,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9696011030351693e-05,
+      "loss": 0.9706,
+      "step": 1176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.969548347956455e-05,
+      "loss": 0.6992,
+      "step": 1177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.969495547374624e-05,
+      "loss": 0.8039,
+      "step": 1178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.969442701290486e-05,
+      "loss": 0.8648,
+      "step": 1179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9693898097048525e-05,
+      "loss": 0.8368,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.969336872618532e-05,
+      "loss": 0.78,
+      "step": 1181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.969283890032337e-05,
+      "loss": 0.815,
+      "step": 1182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.969230861947079e-05,
+      "loss": 0.8545,
+      "step": 1183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9691777883635694e-05,
+      "loss": 0.9259,
+      "step": 1184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.969124669282622e-05,
+      "loss": 0.8333,
+      "step": 1185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.969071504705053e-05,
+      "loss": 1.01,
+      "step": 1186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.969018294631673e-05,
+      "loss": 0.7809,
+      "step": 1187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9689650390633e-05,
+      "loss": 0.8697,
+      "step": 1188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968911738000752e-05,
+      "loss": 0.9749,
+      "step": 1189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9688583914448404e-05,
+      "loss": 0.9205,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968804999396387e-05,
+      "loss": 0.8444,
+      "step": 1191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9687515618562075e-05,
+      "loss": 0.7265,
+      "step": 1192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9686980788251224e-05,
+      "loss": 0.9527,
+      "step": 1193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9686445503039506e-05,
+      "loss": 0.8123,
+      "step": 1194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968590976293513e-05,
+      "loss": 0.8102,
+      "step": 1195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9685373567946286e-05,
+      "loss": 1.119,
+      "step": 1196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968483691808121e-05,
+      "loss": 0.852,
+      "step": 1197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968429981334812e-05,
+      "loss": 0.9904,
+      "step": 1198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968376225375524e-05,
+      "loss": 0.8018,
+      "step": 1199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9683224239310806e-05,
+      "loss": 0.9976,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968268577002307e-05,
+      "loss": 0.8959,
+      "step": 1201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968214684590028e-05,
+      "loss": 0.805,
+      "step": 1202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9681607466950704e-05,
+      "loss": 0.7849,
+      "step": 1203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.968106763318259e-05,
+      "loss": 0.92,
+      "step": 1204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9680527344604214e-05,
+      "loss": 0.709,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967998660122387e-05,
+      "loss": 0.967,
+      "step": 1206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967944540304982e-05,
+      "loss": 0.9301,
+      "step": 1207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9678903750090376e-05,
+      "loss": 0.8613,
+      "step": 1208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967836164235382e-05,
+      "loss": 0.9167,
+      "step": 1209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9677819079848486e-05,
+      "loss": 1.0151,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967727606258266e-05,
+      "loss": 0.8027,
+      "step": 1211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9676732590564684e-05,
+      "loss": 0.7892,
+      "step": 1212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967618866380287e-05,
+      "loss": 0.9314,
+      "step": 1213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967564428230557e-05,
+      "loss": 0.899,
+      "step": 1214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.96750994460811e-05,
+      "loss": 0.7722,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9674554155137825e-05,
+      "loss": 0.9156,
+      "step": 1216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9674008409484106e-05,
+      "loss": 0.7931,
+      "step": 1217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9673462209128294e-05,
+      "loss": 0.6611,
+      "step": 1218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9672915554078766e-05,
+      "loss": 0.9852,
+      "step": 1219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9672368444343896e-05,
+      "loss": 0.7498,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9671820879932074e-05,
+      "loss": 1.1311,
+      "step": 1221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967127286085167e-05,
+      "loss": 1.0355,
+      "step": 1222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.967072438711111e-05,
+      "loss": 0.986,
+      "step": 1223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9670175458718765e-05,
+      "loss": 0.8446,
+      "step": 1224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966962607568307e-05,
+      "loss": 0.9718,
+      "step": 1225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966907623801245e-05,
+      "loss": 0.8574,
+      "step": 1226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966852594571531e-05,
+      "loss": 0.9148,
+      "step": 1227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.96679751988001e-05,
+      "loss": 0.7127,
+      "step": 1228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966742399727524e-05,
+      "loss": 0.8697,
+      "step": 1229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966687234114919e-05,
+      "loss": 0.7854,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.96663202304304e-05,
+      "loss": 0.9273,
+      "step": 1231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9665767665127326e-05,
+      "loss": 0.8742,
+      "step": 1232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966521464524844e-05,
+      "loss": 0.9368,
+      "step": 1233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966466117080221e-05,
+      "loss": 0.9189,
+      "step": 1234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966410724179713e-05,
+      "loss": 0.8322,
+      "step": 1235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966355285824168e-05,
+      "loss": 1.008,
+      "step": 1236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.966299802014435e-05,
+      "loss": 0.8657,
+      "step": 1237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9662442727513645e-05,
+      "loss": 0.8877,
+      "step": 1238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.966188698035809e-05,
+      "loss": 0.677,
+      "step": 1239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.966133077868617e-05,
+      "loss": 0.9546,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.966077412250643e-05,
+      "loss": 1.1987,
+      "step": 1241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.966021701182739e-05,
+      "loss": 1.1926,
+      "step": 1242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965965944665759e-05,
+      "loss": 0.9451,
+      "step": 1243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965910142700559e-05,
+      "loss": 0.9633,
+      "step": 1244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965854295287991e-05,
+      "loss": 0.8723,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9657984024289124e-05,
+      "loss": 0.9519,
+      "step": 1246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9657424641241795e-05,
+      "loss": 0.8629,
+      "step": 1247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96568648037465e-05,
+      "loss": 0.6903,
+      "step": 1248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96563045118118e-05,
+      "loss": 0.8045,
+      "step": 1249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96557437654463e-05,
+      "loss": 0.8837,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96551825646586e-05,
+      "loss": 0.7649,
+      "step": 1251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965462090945727e-05,
+      "loss": 0.8421,
+      "step": 1252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9654058799850936e-05,
+      "loss": 1.0024,
+      "step": 1253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96534962358482e-05,
+      "loss": 0.8861,
+      "step": 1254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965293321745769e-05,
+      "loss": 0.7674,
+      "step": 1255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965236974468804e-05,
+      "loss": 0.9583,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965180581754787e-05,
+      "loss": 0.7053,
+      "step": 1257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.965124143604583e-05,
+      "loss": 1.0184,
+      "step": 1258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9650676600190554e-05,
+      "loss": 0.7713,
+      "step": 1259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9650111309990725e-05,
+      "loss": 0.8015,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964954556545498e-05,
+      "loss": 0.9837,
+      "step": 1261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9648979366592e-05,
+      "loss": 0.7961,
+      "step": 1262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964841271341046e-05,
+      "loss": 0.8687,
+      "step": 1263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9647845605919033e-05,
+      "loss": 0.835,
+      "step": 1264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9647278044126426e-05,
+      "loss": 0.8561,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9646710028041324e-05,
+      "loss": 0.7153,
+      "step": 1266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964614155767242e-05,
+      "loss": 0.8813,
+      "step": 1267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9645572633028454e-05,
+      "loss": 0.8805,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964500325411813e-05,
+      "loss": 0.9287,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9644433420950165e-05,
+      "loss": 0.7393,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964386313353329e-05,
+      "loss": 0.9028,
+      "step": 1271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9643292391876266e-05,
+      "loss": 0.9361,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964272119598781e-05,
+      "loss": 0.7541,
+      "step": 1273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964214954587669e-05,
+      "loss": 0.8979,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964157744155166e-05,
+      "loss": 0.8537,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.96410048830215e-05,
+      "loss": 0.7555,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.964043187029496e-05,
+      "loss": 1.0224,
+      "step": 1277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963985840338084e-05,
+      "loss": 1.0368,
+      "step": 1278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963928448228792e-05,
+      "loss": 0.8544,
+      "step": 1279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9638710107024994e-05,
+      "loss": 0.7535,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963813527760086e-05,
+      "loss": 0.761,
+      "step": 1281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9637559994024335e-05,
+      "loss": 0.7117,
+      "step": 1282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9636984256304236e-05,
+      "loss": 0.7436,
+      "step": 1283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963640806444937e-05,
+      "loss": 0.8704,
+      "step": 1284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963583141846857e-05,
+      "loss": 0.8949,
+      "step": 1285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963525431837069e-05,
+      "loss": 1.008,
+      "step": 1286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963467676416456e-05,
+      "loss": 0.8667,
+      "step": 1287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9634098755859026e-05,
+      "loss": 1.0061,
+      "step": 1288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963352029346294e-05,
+      "loss": 0.7559,
+      "step": 1289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963294137698519e-05,
+      "loss": 0.9717,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9632362006434616e-05,
+      "loss": 0.7884,
+      "step": 1291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963178218182012e-05,
+      "loss": 0.9679,
+      "step": 1292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963120190315058e-05,
+      "loss": 0.9181,
+      "step": 1293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.963062117043489e-05,
+      "loss": 0.8486,
+      "step": 1294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9630039983681944e-05,
+      "loss": 1.0931,
+      "step": 1295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962945834290064e-05,
+      "loss": 0.8307,
+      "step": 1296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962887624809991e-05,
+      "loss": 0.8484,
+      "step": 1297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962829369928865e-05,
+      "loss": 0.9812,
+      "step": 1298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9627710696475815e-05,
+      "loss": 0.8565,
+      "step": 1299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962712723967031e-05,
+      "loss": 0.754,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.962654332888109e-05,
+      "loss": 1.049,
+      "step": 1301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9625958964117105e-05,
+      "loss": 0.8452,
+      "step": 1302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.96253741453873e-05,
+      "loss": 0.9293,
+      "step": 1303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.962478887270065e-05,
+      "loss": 0.8398,
+      "step": 1304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.962420314606611e-05,
+      "loss": 1.0392,
+      "step": 1305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.962361696549266e-05,
+      "loss": 0.8179,
+      "step": 1306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9623030330989275e-05,
+      "loss": 0.8757,
+      "step": 1307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.962244324256496e-05,
+      "loss": 0.7345,
+      "step": 1308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9621855700228696e-05,
+      "loss": 0.9775,
+      "step": 1309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.962126770398951e-05,
+      "loss": 0.7009,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9620679253856377e-05,
+      "loss": 0.8647,
+      "step": 1311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.962009034983834e-05,
+      "loss": 0.929,
+      "step": 1312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961950099194441e-05,
+      "loss": 0.8497,
+      "step": 1313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9618911180183616e-05,
+      "loss": 0.8487,
+      "step": 1314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9618320914565007e-05,
+      "loss": 0.9315,
+      "step": 1315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9617730195097626e-05,
+      "loss": 0.994,
+      "step": 1316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9617139021790514e-05,
+      "loss": 0.8679,
+      "step": 1317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9616547394652745e-05,
+      "loss": 0.9106,
+      "step": 1318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961595531369338e-05,
+      "loss": 0.9631,
+      "step": 1319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961536277892147e-05,
+      "loss": 0.813,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961476979034614e-05,
+      "loss": 0.8712,
+      "step": 1321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9614176347976426e-05,
+      "loss": 0.9522,
+      "step": 1322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961358245182146e-05,
+      "loss": 0.8637,
+      "step": 1323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961298810189032e-05,
+      "loss": 0.8268,
+      "step": 1324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961239329819213e-05,
+      "loss": 0.8125,
+      "step": 1325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961179804073599e-05,
+      "loss": 0.7978,
+      "step": 1326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961120232953102e-05,
+      "loss": 0.8183,
+      "step": 1327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.961060616458635e-05,
+      "loss": 0.9051,
+      "step": 1328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9610009545911133e-05,
+      "loss": 0.9256,
+      "step": 1329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9609412473514496e-05,
+      "loss": 0.7858,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960881494740559e-05,
+      "loss": 0.8147,
+      "step": 1331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9608216967593566e-05,
+      "loss": 1.026,
+      "step": 1332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9607618534087596e-05,
+      "loss": 0.9934,
+      "step": 1333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960701964689684e-05,
+      "loss": 1.081,
+      "step": 1334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960642030603049e-05,
+      "loss": 0.7643,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960582051149772e-05,
+      "loss": 0.7356,
+      "step": 1336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960522026330771e-05,
+      "loss": 0.8632,
+      "step": 1337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960461956146968e-05,
+      "loss": 0.7867,
+      "step": 1338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960401840599283e-05,
+      "loss": 0.9578,
+      "step": 1339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960341679688635e-05,
+      "loss": 0.7903,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960281473415948e-05,
+      "loss": 0.8677,
+      "step": 1341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9602212217821444e-05,
+      "loss": 0.874,
+      "step": 1342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960160924788146e-05,
+      "loss": 0.8447,
+      "step": 1343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.960100582434879e-05,
+      "loss": 0.8253,
+      "step": 1344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9600401947232665e-05,
+      "loss": 0.8057,
+      "step": 1345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959979761654234e-05,
+      "loss": 0.9457,
+      "step": 1346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9599192832287074e-05,
+      "loss": 0.8449,
+      "step": 1347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959858759447614e-05,
+      "loss": 0.8395,
+      "step": 1348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959798190311881e-05,
+      "loss": 0.8574,
+      "step": 1349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959737575822436e-05,
+      "loss": 0.7122,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.95967691598021e-05,
+      "loss": 0.7029,
+      "step": 1351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959616210786129e-05,
+      "loss": 0.906,
+      "step": 1352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959555460241125e-05,
+      "loss": 0.8122,
+      "step": 1353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959494664346129e-05,
+      "loss": 0.8475,
+      "step": 1354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959433823102073e-05,
+      "loss": 0.7493,
+      "step": 1355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959372936509888e-05,
+      "loss": 0.7489,
+      "step": 1356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959312004570508e-05,
+      "loss": 0.8439,
+      "step": 1357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959251027284866e-05,
+      "loss": 0.8218,
+      "step": 1358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959190004653898e-05,
+      "loss": 0.9164,
+      "step": 1359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9591289366785374e-05,
+      "loss": 0.9643,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.95906782335972e-05,
+      "loss": 0.9477,
+      "step": 1361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.959006664698382e-05,
+      "loss": 0.8991,
+      "step": 1362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958945460695462e-05,
+      "loss": 0.7962,
+      "step": 1363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9588842113518965e-05,
+      "loss": 1.0116,
+      "step": 1364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.958822916668626e-05,
+      "loss": 0.9243,
+      "step": 1365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958761576646587e-05,
+      "loss": 0.8864,
+      "step": 1366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958700191286721e-05,
+      "loss": 0.7426,
+      "step": 1367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9586387605899676e-05,
+      "loss": 0.8451,
+      "step": 1368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.95857728455727e-05,
+      "loss": 0.9681,
+      "step": 1369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958515763189568e-05,
+      "loss": 1.0094,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9584541964878063e-05,
+      "loss": 0.8805,
+      "step": 1371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958392584452928e-05,
+      "loss": 1.2219,
+      "step": 1372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958330927085875e-05,
+      "loss": 0.8964,
+      "step": 1373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958269224387595e-05,
+      "loss": 0.7325,
+      "step": 1374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958207476359031e-05,
+      "loss": 0.8937,
+      "step": 1375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958145683001132e-05,
+      "loss": 0.7646,
+      "step": 1376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.958083844314841e-05,
+      "loss": 1.1983,
+      "step": 1377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.95802196030111e-05,
+      "loss": 0.8832,
+      "step": 1378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957960030960884e-05,
+      "loss": 0.9967,
+      "step": 1379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9578980562951126e-05,
+      "loss": 0.8713,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9578360363047465e-05,
+      "loss": 0.8355,
+      "step": 1381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957773970990735e-05,
+      "loss": 0.9032,
+      "step": 1382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9577118603540294e-05,
+      "loss": 0.9565,
+      "step": 1383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957649704395582e-05,
+      "loss": 0.8337,
+      "step": 1384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957587503116345e-05,
+      "loss": 0.7541,
+      "step": 1385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957525256517271e-05,
+      "loss": 0.717,
+      "step": 1386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9574629645993136e-05,
+      "loss": 0.9598,
+      "step": 1387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957400627363429e-05,
+      "loss": 0.9757,
+      "step": 1388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.95733824481057e-05,
+      "loss": 0.8622,
+      "step": 1389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9572758169416955e-05,
+      "loss": 0.7992,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9572133437577595e-05,
+      "loss": 0.8461,
+      "step": 1391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9571508252597204e-05,
+      "loss": 0.8923,
+      "step": 1392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957088261448536e-05,
+      "loss": 0.9033,
+      "step": 1393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.957025652325165e-05,
+      "loss": 0.7764,
+      "step": 1394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9569629978905656e-05,
+      "loss": 0.7568,
+      "step": 1395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9569002981457005e-05,
+      "loss": 0.8902,
+      "step": 1396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956837553091528e-05,
+      "loss": 0.7621,
+      "step": 1397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956774762729011e-05,
+      "loss": 0.7336,
+      "step": 1398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956711927059111e-05,
+      "loss": 0.7241,
+      "step": 1399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956649046082791e-05,
+      "loss": 0.9181,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9565861198010145e-05,
+      "loss": 0.9901,
+      "step": 1401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956523148214745e-05,
+      "loss": 1.0194,
+      "step": 1402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.95646013132495e-05,
+      "loss": 0.9934,
+      "step": 1403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956397069132592e-05,
+      "loss": 0.8509,
+      "step": 1404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956333961638639e-05,
+      "loss": 0.8359,
+      "step": 1405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956270808844058e-05,
+      "loss": 0.8903,
+      "step": 1406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956207610749816e-05,
+      "loss": 0.8688,
+      "step": 1407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956144367356882e-05,
+      "loss": 0.7972,
+      "step": 1408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956081078666225e-05,
+      "loss": 0.8465,
+      "step": 1409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.956017744678814e-05,
+      "loss": 0.8494,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.95595436539562e-05,
+      "loss": 0.7427,
+      "step": 1411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9558909408176146e-05,
+      "loss": 0.8724,
+      "step": 1412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.95582747094577e-05,
+      "loss": 1.0721,
+      "step": 1413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955763955781058e-05,
+      "loss": 0.8009,
+      "step": 1414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955700395324452e-05,
+      "loss": 0.7669,
+      "step": 1415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955636789576925e-05,
+      "loss": 0.7156,
+      "step": 1416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955573138539454e-05,
+      "loss": 0.8271,
+      "step": 1417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9555094422130116e-05,
+      "loss": 0.9501,
+      "step": 1418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955445700598576e-05,
+      "loss": 0.8764,
+      "step": 1419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955381913697123e-05,
+      "loss": 0.846,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.95531808150963e-05,
+      "loss": 0.8359,
+      "step": 1421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955254204037076e-05,
+      "loss": 0.7808,
+      "step": 1422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955190281280438e-05,
+      "loss": 0.7532,
+      "step": 1423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9551263132406976e-05,
+      "loss": 0.9638,
+      "step": 1424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.955062299918833e-05,
+      "loss": 0.9006,
+      "step": 1425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954998241315826e-05,
+      "loss": 0.8507,
+      "step": 1426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954934137432658e-05,
+      "loss": 0.893,
+      "step": 1427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954869988270313e-05,
+      "loss": 0.8773,
+      "step": 1428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.954805793829772e-05,
+      "loss": 0.8911,
+      "step": 1429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.954741554112019e-05,
+      "loss": 0.7253,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9546772691180376e-05,
+      "loss": 0.7429,
+      "step": 1431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9546129388488146e-05,
+      "loss": 0.8137,
+      "step": 1432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.954548563305335e-05,
+      "loss": 0.7277,
+      "step": 1433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9544841424885846e-05,
+      "loss": 0.7935,
+      "step": 1434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9544196763995516e-05,
+      "loss": 0.8477,
+      "step": 1435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.954355165039223e-05,
+      "loss": 0.8571,
+      "step": 1436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.954290608408588e-05,
+      "loss": 0.8635,
+      "step": 1437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9542260065086355e-05,
+      "loss": 1.0289,
+      "step": 1438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.954161359340355e-05,
+      "loss": 0.6924,
+      "step": 1439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.954096666904738e-05,
+      "loss": 1.0471,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.954031929202775e-05,
+      "loss": 0.9146,
+      "step": 1441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953967146235458e-05,
+      "loss": 0.8199,
+      "step": 1442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9539023180037805e-05,
+      "loss": 0.7975,
+      "step": 1443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953837444508735e-05,
+      "loss": 0.7861,
+      "step": 1444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953772525751316e-05,
+      "loss": 0.7949,
+      "step": 1445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953707561732518e-05,
+      "loss": 0.7676,
+      "step": 1446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953642552453337e-05,
+      "loss": 0.8829,
+      "step": 1447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953577497914768e-05,
+      "loss": 0.7468,
+      "step": 1448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953512398117809e-05,
+      "loss": 0.8472,
+      "step": 1449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953447253063458e-05,
+      "loss": 0.9939,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9533820627527114e-05,
+      "loss": 0.926,
+      "step": 1451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9533168271865694e-05,
+      "loss": 0.9013,
+      "step": 1452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953251546366031e-05,
+      "loss": 0.9668,
+      "step": 1453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953186220292098e-05,
+      "loss": 0.9837,
+      "step": 1454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953120848965769e-05,
+      "loss": 0.689,
+      "step": 1455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.953055432388048e-05,
+      "loss": 0.8297,
+      "step": 1456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952989970559936e-05,
+      "loss": 1.0156,
+      "step": 1457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9529244634824366e-05,
+      "loss": 0.8336,
+      "step": 1458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952858911156553e-05,
+      "loss": 0.9739,
+      "step": 1459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.95279331358329e-05,
+      "loss": 0.9615,
+      "step": 1460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9527276707636535e-05,
+      "loss": 0.7727,
+      "step": 1461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952661982698649e-05,
+      "loss": 0.97,
+      "step": 1462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952596249389282e-05,
+      "loss": 0.6725,
+      "step": 1463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952530470836561e-05,
+      "loss": 0.9493,
+      "step": 1464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9524646470414935e-05,
+      "loss": 1.0004,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952398778005088e-05,
+      "loss": 0.8324,
+      "step": 1466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9523328637283544e-05,
+      "loss": 0.8516,
+      "step": 1467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952266904212301e-05,
+      "loss": 0.7342,
+      "step": 1468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.952200899457941e-05,
+      "loss": 0.8706,
+      "step": 1469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9521348494662844e-05,
+      "loss": 0.706,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9520687542383425e-05,
+      "loss": 0.8255,
+      "step": 1471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9520026137751305e-05,
+      "loss": 0.7822,
+      "step": 1472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9519364280776595e-05,
+      "loss": 0.8819,
+      "step": 1473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9518701971469445e-05,
+      "loss": 0.9012,
+      "step": 1474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9518039209840014e-05,
+      "loss": 0.9176,
+      "step": 1475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951737599589845e-05,
+      "loss": 0.8285,
+      "step": 1476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.95167123296549e-05,
+      "loss": 0.8141,
+      "step": 1477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951604821111956e-05,
+      "loss": 0.7918,
+      "step": 1478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951538364030259e-05,
+      "loss": 1.0418,
+      "step": 1479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951471861721419e-05,
+      "loss": 0.8536,
+      "step": 1480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9514053141864525e-05,
+      "loss": 0.9003,
+      "step": 1481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951338721426381e-05,
+      "loss": 1.0091,
+      "step": 1482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951272083442224e-05,
+      "loss": 1.0028,
+      "step": 1483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951205400235003e-05,
+      "loss": 0.8474,
+      "step": 1484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9511386718057405e-05,
+      "loss": 0.8276,
+      "step": 1485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.951071898155458e-05,
+      "loss": 0.9617,
+      "step": 1486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9510050792851795e-05,
+      "loss": 0.769,
+      "step": 1487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950938215195927e-05,
+      "loss": 0.9199,
+      "step": 1488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9508713058887286e-05,
+      "loss": 0.8194,
+      "step": 1489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950804351364606e-05,
+      "loss": 0.9477,
+      "step": 1490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.950737351624588e-05,
+      "loss": 0.7588,
+      "step": 1491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9506703066696986e-05,
+      "loss": 0.8482,
+      "step": 1492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9506032165009665e-05,
+      "loss": 0.9557,
+      "step": 1493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9505360811194196e-05,
+      "loss": 0.8484,
+      "step": 1494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.950468900526088e-05,
+      "loss": 0.8724,
+      "step": 1495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.950401674721999e-05,
+      "loss": 0.9232,
+      "step": 1496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.950334403708183e-05,
+      "loss": 0.8488,
+      "step": 1497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.950267087485672e-05,
+      "loss": 0.8353,
+      "step": 1498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9501997260554964e-05,
+      "loss": 0.7388,
+      "step": 1499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.950132319418689e-05,
+      "loss": 0.8793,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9500648675762827e-05,
+      "loss": 0.7498,
+      "step": 1501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949997370529311e-05,
+      "loss": 0.8402,
+      "step": 1502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9499298282788085e-05,
+      "loss": 0.7964,
+      "step": 1503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.94986224082581e-05,
+      "loss": 0.9365,
+      "step": 1504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9497946081713496e-05,
+      "loss": 0.7882,
+      "step": 1505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949726930316466e-05,
+      "loss": 0.9133,
+      "step": 1506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949659207262195e-05,
+      "loss": 0.739,
+      "step": 1507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949591439009574e-05,
+      "loss": 1.0923,
+      "step": 1508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9495236255596425e-05,
+      "loss": 0.8895,
+      "step": 1509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9494557669134385e-05,
+      "loss": 0.8392,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9493878630720034e-05,
+      "loss": 0.734,
+      "step": 1511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9493199140363766e-05,
+      "loss": 0.7633,
+      "step": 1512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949251919807599e-05,
+      "loss": 0.9014,
+      "step": 1513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.949183880386713e-05,
+      "loss": 0.8216,
+      "step": 1514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9491157957747614e-05,
+      "loss": 0.9079,
+      "step": 1515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9490476659727864e-05,
+      "loss": 0.934,
+      "step": 1516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948979490981834e-05,
+      "loss": 0.7736,
+      "step": 1517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9489112708029466e-05,
+      "loss": 0.7081,
+      "step": 1518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.94884300543717e-05,
+      "loss": 0.9572,
+      "step": 1519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9487746948855525e-05,
+      "loss": 0.8165,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9487063391491385e-05,
+      "loss": 0.9191,
+      "step": 1521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948637938228976e-05,
+      "loss": 1.1383,
+      "step": 1522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948569492126113e-05,
+      "loss": 1.007,
+      "step": 1523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948501000841599e-05,
+      "loss": 0.8165,
+      "step": 1524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948432464376483e-05,
+      "loss": 0.947,
+      "step": 1525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948363882731815e-05,
+      "loss": 0.8936,
+      "step": 1526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948295255908646e-05,
+      "loss": 0.972,
+      "step": 1527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948226583908028e-05,
+      "loss": 0.9629,
+      "step": 1528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948157866731013e-05,
+      "loss": 0.7225,
+      "step": 1529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948089104378653e-05,
+      "loss": 0.7261,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.948020296852004e-05,
+      "loss": 0.7297,
+      "step": 1531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9479514441521186e-05,
+      "loss": 0.7706,
+      "step": 1532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947882546280052e-05,
+      "loss": 1.011,
+      "step": 1533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.94781360323686e-05,
+      "loss": 0.9034,
+      "step": 1534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9477446150236004e-05,
+      "loss": 0.8346,
+      "step": 1535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947675581641328e-05,
+      "loss": 0.7594,
+      "step": 1536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947606503091102e-05,
+      "loss": 0.9954,
+      "step": 1537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9475373793739805e-05,
+      "loss": 0.7584,
+      "step": 1538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947468210491023e-05,
+      "loss": 0.9605,
+      "step": 1539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9473989964432896e-05,
+      "loss": 0.9162,
+      "step": 1540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.94732973723184e-05,
+      "loss": 0.8218,
+      "step": 1541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947260432857736e-05,
+      "loss": 0.8348,
+      "step": 1542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947191083322039e-05,
+      "loss": 1.0188,
+      "step": 1543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947121688625814e-05,
+      "loss": 0.8854,
+      "step": 1544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.947052248770121e-05,
+      "loss": 0.78,
+      "step": 1545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946982763756026e-05,
+      "loss": 0.9691,
+      "step": 1546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946913233584594e-05,
+      "loss": 0.7559,
+      "step": 1547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946843658256889e-05,
+      "loss": 0.7434,
+      "step": 1548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946774037773978e-05,
+      "loss": 0.9898,
+      "step": 1549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946704372136928e-05,
+      "loss": 0.813,
+      "step": 1550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946634661346807e-05,
+      "loss": 0.9302,
+      "step": 1551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946564905404681e-05,
+      "loss": 0.8421,
+      "step": 1552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9464951043116206e-05,
+      "loss": 0.9702,
+      "step": 1553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946425258068696e-05,
+      "loss": 0.8577,
+      "step": 1554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.946355366676975e-05,
+      "loss": 0.8786,
+      "step": 1555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9462854301375316e-05,
+      "loss": 0.8083,
+      "step": 1556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.946215448451435e-05,
+      "loss": 1.0123,
+      "step": 1557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.946145421619759e-05,
+      "loss": 1.0708,
+      "step": 1558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.946075349643576e-05,
+      "loss": 0.8922,
+      "step": 1559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9460052325239603e-05,
+      "loss": 0.7108,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.945935070261985e-05,
+      "loss": 0.7878,
+      "step": 1561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9458648628587275e-05,
+      "loss": 0.7835,
+      "step": 1562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.945794610315261e-05,
+      "loss": 0.7269,
+      "step": 1563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.945724312632664e-05,
+      "loss": 0.7928,
+      "step": 1564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9456539698120126e-05,
+      "loss": 0.9166,
+      "step": 1565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9455835818543844e-05,
+      "loss": 0.851,
+      "step": 1566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9455131487608595e-05,
+      "loss": 0.8411,
+      "step": 1567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9454426705325164e-05,
+      "loss": 0.9737,
+      "step": 1568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.945372147170434e-05,
+      "loss": 0.9304,
+      "step": 1569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.945301578675695e-05,
+      "loss": 0.8593,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.945230965049378e-05,
+      "loss": 0.8989,
+      "step": 1571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.945160306292567e-05,
+      "loss": 1.0739,
+      "step": 1572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9450896024063446e-05,
+      "loss": 0.8526,
+      "step": 1573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9450188533917944e-05,
+      "loss": 0.8743,
+      "step": 1574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.94494805925e-05,
+      "loss": 0.8203,
+      "step": 1575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9448772199820455e-05,
+      "loss": 0.9906,
+      "step": 1576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944806335589018e-05,
+      "loss": 0.8941,
+      "step": 1577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944735406072002e-05,
+      "loss": 0.894,
+      "step": 1578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944664431432085e-05,
+      "loss": 0.9618,
+      "step": 1579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9445934116703555e-05,
+      "loss": 0.7822,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944522346787901e-05,
+      "loss": 0.7442,
+      "step": 1581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.94445123678581e-05,
+      "loss": 0.9754,
+      "step": 1582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944380081665173e-05,
+      "loss": 0.8521,
+      "step": 1583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944308881427079e-05,
+      "loss": 0.9799,
+      "step": 1584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9442376360726205e-05,
+      "loss": 0.8622,
+      "step": 1585
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9441663456028894e-05,
+      "loss": 0.7874,
+      "step": 1586
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944095010018976e-05,
+      "loss": 0.7297,
+      "step": 1587
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.944023629321976e-05,
+      "loss": 0.8255,
+      "step": 1588
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943952203512981e-05,
+      "loss": 0.9515,
+      "step": 1589
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943880732593086e-05,
+      "loss": 0.8443,
+      "step": 1590
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943809216563387e-05,
+      "loss": 0.8766,
+      "step": 1591
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.94373765542498e-05,
+      "loss": 0.8866,
+      "step": 1592
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.94366604917896e-05,
+      "loss": 1.0827,
+      "step": 1593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943594397826426e-05,
+      "loss": 0.8424,
+      "step": 1594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943522701368475e-05,
+      "loss": 0.8016,
+      "step": 1595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943450959806205e-05,
+      "loss": 0.7317,
+      "step": 1596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943379173140717e-05,
+      "loss": 0.9322,
+      "step": 1597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.94330734137311e-05,
+      "loss": 0.8586,
+      "step": 1598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9432354645044834e-05,
+      "loss": 0.8379,
+      "step": 1599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943163542535941e-05,
+      "loss": 0.7141,
+      "step": 1600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943091575468584e-05,
+      "loss": 1.1297,
+      "step": 1601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.943019563303515e-05,
+      "loss": 1.0197,
+      "step": 1602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942947506041837e-05,
+      "loss": 0.8862,
+      "step": 1603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9428754036846547e-05,
+      "loss": 0.8046,
+      "step": 1604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9428032562330735e-05,
+      "loss": 0.7186,
+      "step": 1605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942731063688197e-05,
+      "loss": 0.9266,
+      "step": 1606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9426588260511344e-05,
+      "loss": 0.7681,
+      "step": 1607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9425865433229896e-05,
+      "loss": 0.9048,
+      "step": 1608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942514215504872e-05,
+      "loss": 0.8321,
+      "step": 1609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942441842597889e-05,
+      "loss": 0.9239,
+      "step": 1610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.94236942460315e-05,
+      "loss": 0.6775,
+      "step": 1611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9422969615217646e-05,
+      "loss": 0.8314,
+      "step": 1612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942224453354844e-05,
+      "loss": 0.8078,
+      "step": 1613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942151900103498e-05,
+      "loss": 0.9713,
+      "step": 1614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.942079301768839e-05,
+      "loss": 1.0187,
+      "step": 1615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9420066583519795e-05,
+      "loss": 0.7435,
+      "step": 1616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941933969854033e-05,
+      "loss": 0.8961,
+      "step": 1617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941861236276111e-05,
+      "loss": 0.7906,
+      "step": 1618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.941788457619331e-05,
+      "loss": 0.9544,
+      "step": 1619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9417156338848074e-05,
+      "loss": 0.783,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.941642765073654e-05,
+      "loss": 1.0384,
+      "step": 1621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9415698511869913e-05,
+      "loss": 0.9996,
+      "step": 1622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.941496892225933e-05,
+      "loss": 0.8459,
+      "step": 1623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.941423888191598e-05,
+      "loss": 0.8425,
+      "step": 1624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9413508390851066e-05,
+      "loss": 0.8197,
+      "step": 1625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.941277744907576e-05,
+      "loss": 0.8238,
+      "step": 1626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9412046056601275e-05,
+      "loss": 0.7688,
+      "step": 1627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.941131421343881e-05,
+      "loss": 0.8197,
+      "step": 1628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9410581919599586e-05,
+      "loss": 0.8487,
+      "step": 1629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940984917509483e-05,
+      "loss": 0.799,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940911597993576e-05,
+      "loss": 0.6639,
+      "step": 1631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940838233413361e-05,
+      "loss": 0.7401,
+      "step": 1632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940764823769962e-05,
+      "loss": 0.7757,
+      "step": 1633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940691369064505e-05,
+      "loss": 0.794,
+      "step": 1634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940617869298115e-05,
+      "loss": 0.8553,
+      "step": 1635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940544324471918e-05,
+      "loss": 0.8344,
+      "step": 1636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940470734587041e-05,
+      "loss": 0.8629,
+      "step": 1637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9403970996446124e-05,
+      "loss": 0.869,
+      "step": 1638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.94032341964576e-05,
+      "loss": 0.8447,
+      "step": 1639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940249694591611e-05,
+      "loss": 0.8369,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940175924483298e-05,
+      "loss": 0.7628,
+      "step": 1641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.94010210932195e-05,
+      "loss": 0.9632,
+      "step": 1642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.940028249108699e-05,
+      "loss": 0.7368,
+      "step": 1643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9399543438446754e-05,
+      "loss": 0.8653,
+      "step": 1644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9398803935310124e-05,
+      "loss": 0.8025,
+      "step": 1645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939806398168843e-05,
+      "loss": 0.8184,
+      "step": 1646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939732357759302e-05,
+      "loss": 0.7713,
+      "step": 1647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939658272303523e-05,
+      "loss": 0.7814,
+      "step": 1648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.93958414180264e-05,
+      "loss": 0.8206,
+      "step": 1649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939509966257792e-05,
+      "loss": 0.8935,
+      "step": 1650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939435745670112e-05,
+      "loss": 0.9629,
+      "step": 1651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939361480040741e-05,
+      "loss": 0.7134,
+      "step": 1652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939287169370814e-05,
+      "loss": 1.0987,
+      "step": 1653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9392128136614705e-05,
+      "loss": 0.8187,
+      "step": 1654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939138412913852e-05,
+      "loss": 0.8936,
+      "step": 1655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.939063967129095e-05,
+      "loss": 0.8197,
+      "step": 1656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9389894763083426e-05,
+      "loss": 0.8427,
+      "step": 1657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938914940452736e-05,
+      "loss": 0.7513,
+      "step": 1658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938840359563417e-05,
+      "loss": 0.6746,
+      "step": 1659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938765733641527e-05,
+      "loss": 0.9526,
+      "step": 1660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938691062688213e-05,
+      "loss": 0.7924,
+      "step": 1661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938616346704616e-05,
+      "loss": 0.8883,
+      "step": 1662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9385415856918814e-05,
+      "loss": 0.9964,
+      "step": 1663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938466779651157e-05,
+      "loss": 0.7549,
+      "step": 1664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938391928583587e-05,
+      "loss": 0.8851,
+      "step": 1665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.938317032490318e-05,
+      "loss": 0.9252,
+      "step": 1666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9382420913724985e-05,
+      "loss": 0.6933,
+      "step": 1667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9381671052312784e-05,
+      "loss": 0.7529,
+      "step": 1668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9380920740678034e-05,
+      "loss": 0.9777,
+      "step": 1669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9380169978832256e-05,
+      "loss": 0.8232,
+      "step": 1670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937941876678695e-05,
+      "loss": 1.0248,
+      "step": 1671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937866710455362e-05,
+      "loss": 1.1087,
+      "step": 1672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937791499214379e-05,
+      "loss": 0.9724,
+      "step": 1673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9377162429568986e-05,
+      "loss": 0.8139,
+      "step": 1674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937640941684074e-05,
+      "loss": 0.9652,
+      "step": 1675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937565595397058e-05,
+      "loss": 0.7024,
+      "step": 1676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937490204097006e-05,
+      "loss": 0.8868,
+      "step": 1677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937414767785074e-05,
+      "loss": 0.9893,
+      "step": 1678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937339286462416e-05,
+      "loss": 0.7577,
+      "step": 1679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937263760130191e-05,
+      "loss": 0.8469,
+      "step": 1680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937188188789554e-05,
+      "loss": 0.8718,
+      "step": 1681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.937112572441665e-05,
+      "loss": 0.7616,
+      "step": 1682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.93703691108768e-05,
+      "loss": 0.9511,
+      "step": 1683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.936961204728762e-05,
+      "loss": 0.9294,
+      "step": 1684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9368854533660675e-05,
+      "loss": 0.9508,
+      "step": 1685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.936809657000761e-05,
+      "loss": 0.8661,
+      "step": 1686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.936733815634e-05,
+      "loss": 0.638,
+      "step": 1687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.936657929266949e-05,
+      "loss": 0.899,
+      "step": 1688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9365819979007706e-05,
+      "loss": 0.7888,
+      "step": 1689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.936506021536628e-05,
+      "loss": 0.8674,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.936430000175686e-05,
+      "loss": 0.9212,
+      "step": 1691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.936353933819108e-05,
+      "loss": 0.6614,
+      "step": 1692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.936277822468061e-05,
+      "loss": 0.9698,
+      "step": 1693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.93620166612371e-05,
+      "loss": 0.7024,
+      "step": 1694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9361254647872236e-05,
+      "loss": 0.8899,
+      "step": 1695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9360492184597685e-05,
+      "loss": 0.8614,
+      "step": 1696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935972927142513e-05,
+      "loss": 0.8241,
+      "step": 1697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9358965908366254e-05,
+      "loss": 0.7737,
+      "step": 1698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935820209543278e-05,
+      "loss": 0.787,
+      "step": 1699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9357437832636385e-05,
+      "loss": 0.8534,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9356673119988797e-05,
+      "loss": 0.7663,
+      "step": 1701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9355907957501715e-05,
+      "loss": 0.7888,
+      "step": 1702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935514234518689e-05,
+      "loss": 0.7609,
+      "step": 1703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9354376283056025e-05,
+      "loss": 0.853,
+      "step": 1704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9353609771120884e-05,
+      "loss": 0.8508,
+      "step": 1705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9352842809393195e-05,
+      "loss": 0.97,
+      "step": 1706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9352075397884723e-05,
+      "loss": 0.7553,
+      "step": 1707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935130753660722e-05,
+      "loss": 0.8751,
+      "step": 1708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.935053922557244e-05,
+      "loss": 0.7972,
+      "step": 1709
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934977046479218e-05,
+      "loss": 0.7648,
+      "step": 1710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934900125427821e-05,
+      "loss": 0.9353,
+      "step": 1711
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934823159404232e-05,
+      "loss": 1.0442,
+      "step": 1712
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934746148409629e-05,
+      "loss": 0.8734,
+      "step": 1713
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934669092445194e-05,
+      "loss": 0.958,
+      "step": 1714
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9345919915121056e-05,
+      "loss": 0.7572,
+      "step": 1715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9345148456115465e-05,
+      "loss": 0.9979,
+      "step": 1716
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9344376547447e-05,
+      "loss": 0.932,
+      "step": 1717
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9343604189127464e-05,
+      "loss": 0.8805,
+      "step": 1718
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934283138116871e-05,
+      "loss": 0.7399,
+      "step": 1719
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9342058123582575e-05,
+      "loss": 0.8617,
+      "step": 1720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.93412844163809e-05,
+      "loss": 0.9247,
+      "step": 1721
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.934051025957556e-05,
+      "loss": 0.8696,
+      "step": 1722
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933973565317839e-05,
+      "loss": 0.871,
+      "step": 1723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9338960597201284e-05,
+      "loss": 0.8098,
+      "step": 1724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9338185091656107e-05,
+      "loss": 0.9235,
+      "step": 1725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933740913655475e-05,
+      "loss": 0.7692,
+      "step": 1726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9336632731909086e-05,
+      "loss": 0.8541,
+      "step": 1727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933585587773103e-05,
+      "loss": 0.8312,
+      "step": 1728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933507857403249e-05,
+      "loss": 0.8509,
+      "step": 1729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933430082082536e-05,
+      "loss": 0.7486,
+      "step": 1730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933352261812155e-05,
+      "loss": 0.973,
+      "step": 1731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9332743965933015e-05,
+      "loss": 0.8361,
+      "step": 1732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933196486427167e-05,
+      "loss": 0.9279,
+      "step": 1733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.933118531314945e-05,
+      "loss": 0.8541,
+      "step": 1734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.93304053125783e-05,
+      "loss": 0.8604,
+      "step": 1735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932962486257018e-05,
+      "loss": 0.8765,
+      "step": 1736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9328843963137045e-05,
+      "loss": 0.9059,
+      "step": 1737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932806261429085e-05,
+      "loss": 0.725,
+      "step": 1738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932728081604359e-05,
+      "loss": 0.8822,
+      "step": 1739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9326498568407236e-05,
+      "loss": 0.8744,
+      "step": 1740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9325715871393765e-05,
+      "loss": 0.8172,
+      "step": 1741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932493272501519e-05,
+      "loss": 0.8253,
+      "step": 1742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932414912928348e-05,
+      "loss": 0.844,
+      "step": 1743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932336508421068e-05,
+      "loss": 0.9447,
+      "step": 1744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932258058980877e-05,
+      "loss": 0.8719,
+      "step": 1745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.932179564608979e-05,
+      "loss": 0.8857,
+      "step": 1746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9321010253065775e-05,
+      "loss": 1.0214,
+      "step": 1747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.932022441074873e-05,
+      "loss": 1.1603,
+      "step": 1748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9319438119150726e-05,
+      "loss": 0.7237,
+      "step": 1749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.93186513782838e-05,
+      "loss": 0.8136,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.931786418816001e-05,
+      "loss": 0.7679,
+      "step": 1751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9317076548791416e-05,
+      "loss": 0.9641,
+      "step": 1752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.931628846019009e-05,
+      "loss": 0.8308,
+      "step": 1753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.93154999223681e-05,
+      "loss": 1.0631,
+      "step": 1754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.931471093533753e-05,
+      "loss": 0.7635,
+      "step": 1755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.931392149911049e-05,
+      "loss": 0.8152,
+      "step": 1756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.931313161369905e-05,
+      "loss": 0.8479,
+      "step": 1757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.931234127911533e-05,
+      "loss": 0.8968,
+      "step": 1758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.931155049537144e-05,
+      "loss": 0.8239,
+      "step": 1759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.931075926247949e-05,
+      "loss": 0.9128,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.93099675804516e-05,
+      "loss": 0.8194,
+      "step": 1761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9309175449299916e-05,
+      "loss": 1.021,
+      "step": 1762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930838286903656e-05,
+      "loss": 0.908,
+      "step": 1763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930758983967369e-05,
+      "loss": 0.8723,
+      "step": 1764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930679636122346e-05,
+      "loss": 0.9779,
+      "step": 1765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9306002433698016e-05,
+      "loss": 0.944,
+      "step": 1766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930520805710953e-05,
+      "loss": 0.9847,
+      "step": 1767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930441323147017e-05,
+      "loss": 0.8441,
+      "step": 1768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930361795679211e-05,
+      "loss": 0.8444,
+      "step": 1769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930282223308756e-05,
+      "loss": 0.8339,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.93020260603687e-05,
+      "loss": 0.9577,
+      "step": 1771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930122943864773e-05,
+      "loss": 0.9993,
+      "step": 1772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.930043236793684e-05,
+      "loss": 0.9153,
+      "step": 1773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929963484824827e-05,
+      "loss": 0.899,
+      "step": 1774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929883687959422e-05,
+      "loss": 0.8238,
+      "step": 1775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929803846198694e-05,
+      "loss": 0.9337,
+      "step": 1776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929723959543864e-05,
+      "loss": 0.8386,
+      "step": 1777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929644027996158e-05,
+      "loss": 0.8369,
+      "step": 1778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9295640515567994e-05,
+      "loss": 0.7405,
+      "step": 1779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9294840302270154e-05,
+      "loss": 0.7299,
+      "step": 1780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.92940396400803e-05,
+      "loss": 0.8752,
+      "step": 1781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9293238529010715e-05,
+      "loss": 0.7521,
+      "step": 1782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929243696907367e-05,
+      "loss": 0.988,
+      "step": 1783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9291634960281455e-05,
+      "loss": 0.8075,
+      "step": 1784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929083250264635e-05,
+      "loss": 0.942,
+      "step": 1785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.929002959618065e-05,
+      "loss": 0.7825,
+      "step": 1786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928922624089668e-05,
+      "loss": 0.9834,
+      "step": 1787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928842243680672e-05,
+      "loss": 0.8425,
+      "step": 1788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9287618183923095e-05,
+      "loss": 0.8883,
+      "step": 1789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928681348225814e-05,
+      "loss": 0.9892,
+      "step": 1790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9286008331824184e-05,
+      "loss": 0.954,
+      "step": 1791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9285202732633557e-05,
+      "loss": 0.8941,
+      "step": 1792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9284396684698595e-05,
+      "loss": 1.0013,
+      "step": 1793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928359018803168e-05,
+      "loss": 0.9179,
+      "step": 1794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928278324264514e-05,
+      "loss": 0.9207,
+      "step": 1795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928197584855135e-05,
+      "loss": 0.9696,
+      "step": 1796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928116800576268e-05,
+      "loss": 0.8595,
+      "step": 1797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.928035971429152e-05,
+      "loss": 0.8432,
+      "step": 1798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9279550974150235e-05,
+      "loss": 1.0155,
+      "step": 1799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927874178535124e-05,
+      "loss": 0.8274,
+      "step": 1800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927793214790692e-05,
+      "loss": 0.876,
+      "step": 1801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927712206182968e-05,
+      "loss": 1.025,
+      "step": 1802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9276311527131944e-05,
+      "loss": 0.8272,
+      "step": 1803
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927550054382613e-05,
+      "loss": 0.8747,
+      "step": 1804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927468911192466e-05,
+      "loss": 0.819,
+      "step": 1805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927387723143996e-05,
+      "loss": 0.9605,
+      "step": 1806
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9273064902384485e-05,
+      "loss": 0.7923,
+      "step": 1807
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927225212477068e-05,
+      "loss": 1.0319,
+      "step": 1808
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.927143889861099e-05,
+      "loss": 0.7804,
+      "step": 1809
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9270625223917886e-05,
+      "loss": 0.7562,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.926981110070384e-05,
+      "loss": 0.8734,
+      "step": 1811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.926899652898131e-05,
+      "loss": 0.8279,
+      "step": 1812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.926818150876279e-05,
+      "loss": 0.8526,
+      "step": 1813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9267366040060763e-05,
+      "loss": 0.9017,
+      "step": 1814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.926655012288774e-05,
+      "loss": 0.728,
+      "step": 1815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9265733757256204e-05,
+      "loss": 0.9321,
+      "step": 1816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.926491694317867e-05,
+      "loss": 0.8529,
+      "step": 1817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9264099680667665e-05,
+      "loss": 0.8445,
+      "step": 1818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.926328196973569e-05,
+      "loss": 0.7637,
+      "step": 1819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.92624638103953e-05,
+      "loss": 0.8078,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9261645202659014e-05,
+      "loss": 0.8815,
+      "step": 1821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.926082614653938e-05,
+      "loss": 0.7787,
+      "step": 1822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.926000664204896e-05,
+      "loss": 1.1219,
+      "step": 1823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9259186689200296e-05,
+      "loss": 1.2435,
+      "step": 1824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9258366288005956e-05,
+      "loss": 0.9374,
+      "step": 1825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925754543847852e-05,
+      "loss": 0.8967,
+      "step": 1826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925672414063055e-05,
+      "loss": 0.754,
+      "step": 1827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9255902394474657e-05,
+      "loss": 0.7262,
+      "step": 1828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925508020002341e-05,
+      "loss": 0.7775,
+      "step": 1829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9254257557289405e-05,
+      "loss": 0.7478,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9253434466285265e-05,
+      "loss": 0.9745,
+      "step": 1831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9252610927023584e-05,
+      "loss": 0.875,
+      "step": 1832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9251786939517e-05,
+      "loss": 0.688,
+      "step": 1833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925096250377814e-05,
+      "loss": 1.0855,
+      "step": 1834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.925013761981961e-05,
+      "loss": 0.8758,
+      "step": 1835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924931228765407e-05,
+      "loss": 0.7229,
+      "step": 1836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9248486507294173e-05,
+      "loss": 1.018,
+      "step": 1837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9247660278752555e-05,
+      "loss": 1.0313,
+      "step": 1838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9246833602041895e-05,
+      "loss": 0.8285,
+      "step": 1839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924600647717484e-05,
+      "loss": 0.9709,
+      "step": 1840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924517890416408e-05,
+      "loss": 0.8414,
+      "step": 1841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9244350883022294e-05,
+      "loss": 0.8935,
+      "step": 1842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924352241376217e-05,
+      "loss": 0.8971,
+      "step": 1843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.92426934963964e-05,
+      "loss": 1.0875,
+      "step": 1844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924186413093767e-05,
+      "loss": 0.8655,
+      "step": 1845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.924103431739871e-05,
+      "loss": 1.1521,
+      "step": 1846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9240204055792244e-05,
+      "loss": 0.8362,
+      "step": 1847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923937334613097e-05,
+      "loss": 0.9117,
+      "step": 1848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923854218842763e-05,
+      "loss": 0.7497,
+      "step": 1849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923771058269495e-05,
+      "loss": 0.8275,
+      "step": 1850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923687852894568e-05,
+      "loss": 0.8505,
+      "step": 1851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923604602719257e-05,
+      "loss": 0.8031,
+      "step": 1852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9235213077448384e-05,
+      "loss": 0.787,
+      "step": 1853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923437967972588e-05,
+      "loss": 0.9471,
+      "step": 1854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923354583403781e-05,
+      "loss": 0.6952,
+      "step": 1855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923271154039698e-05,
+      "loss": 0.8086,
+      "step": 1856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9231876798816165e-05,
+      "loss": 0.7337,
+      "step": 1857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9231041609308144e-05,
+      "loss": 1.0653,
+      "step": 1858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.923020597188573e-05,
+      "loss": 1.1022,
+      "step": 1859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922936988656171e-05,
+      "loss": 0.8246,
+      "step": 1860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922853335334892e-05,
+      "loss": 0.8048,
+      "step": 1861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9227696372260156e-05,
+      "loss": 0.836,
+      "step": 1862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922685894330826e-05,
+      "loss": 0.9686,
+      "step": 1863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922602106650605e-05,
+      "loss": 0.8278,
+      "step": 1864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9225182741866374e-05,
+      "loss": 0.6868,
+      "step": 1865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922434396940207e-05,
+      "loss": 0.8802,
+      "step": 1866
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9223504749126005e-05,
+      "loss": 0.8694,
+      "step": 1867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922266508105101e-05,
+      "loss": 0.8648,
+      "step": 1868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9221824965189986e-05,
+      "loss": 0.8858,
+      "step": 1869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.92209844015558e-05,
+      "loss": 0.7806,
+      "step": 1870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.922014339016131e-05,
+      "loss": 0.9028,
+      "step": 1871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921930193101942e-05,
+      "loss": 0.92,
+      "step": 1872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.921846002414302e-05,
+      "loss": 0.77,
+      "step": 1873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.921761766954501e-05,
+      "loss": 0.8928,
+      "step": 1874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.92167748672383e-05,
+      "loss": 0.9829,
+      "step": 1875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.921593161723581e-05,
+      "loss": 0.8523,
+      "step": 1876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.921508791955044e-05,
+      "loss": 0.8626,
+      "step": 1877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9214243774195145e-05,
+      "loss": 0.8623,
+      "step": 1878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.921339918118284e-05,
+      "loss": 0.8083,
+      "step": 1879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9212554140526476e-05,
+      "loss": 0.8363,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.921170865223901e-05,
+      "loss": 0.7484,
+      "step": 1881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9210862716333373e-05,
+      "loss": 1.0696,
+      "step": 1882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.921001633282255e-05,
+      "loss": 0.7779,
+      "step": 1883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.92091695017195e-05,
+      "loss": 0.7639,
+      "step": 1884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9208322223037206e-05,
+      "loss": 0.8338,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9207474496788646e-05,
+      "loss": 0.9109,
+      "step": 1886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.92066263229868e-05,
+      "loss": 0.7239,
+      "step": 1887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9205777701644694e-05,
+      "loss": 0.6164,
+      "step": 1888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.92049286327753e-05,
+      "loss": 0.8423,
+      "step": 1889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9204079116391645e-05,
+      "loss": 0.9428,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9203229152506746e-05,
+      "loss": 0.7623,
+      "step": 1891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.920237874113362e-05,
+      "loss": 0.8284,
+      "step": 1892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.920152788228531e-05,
+      "loss": 0.752,
+      "step": 1893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.920067657597484e-05,
+      "loss": 0.7598,
+      "step": 1894
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919982482221526e-05,
+      "loss": 1.0424,
+      "step": 1895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919897262101963e-05,
+      "loss": 0.7662,
+      "step": 1896
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9198119972400996e-05,
+      "loss": 1.1342,
+      "step": 1897
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919726687637243e-05,
+      "loss": 0.808,
+      "step": 1898
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9196413332947005e-05,
+      "loss": 0.8114,
+      "step": 1899
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91955593421378e-05,
+      "loss": 1.083,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919470490395789e-05,
+      "loss": 0.6214,
+      "step": 1901
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919385001842039e-05,
+      "loss": 0.746,
+      "step": 1902
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9192994685538375e-05,
+      "loss": 0.8056,
+      "step": 1903
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919213890532496e-05,
+      "loss": 0.8316,
+      "step": 1904
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919128267779327e-05,
+      "loss": 0.8019,
+      "step": 1905
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.919042600295641e-05,
+      "loss": 0.7612,
+      "step": 1906
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9189568880827514e-05,
+      "loss": 1.1439,
+      "step": 1907
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918871131141971e-05,
+      "loss": 0.9033,
+      "step": 1908
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9187853294746155e-05,
+      "loss": 0.8544,
+      "step": 1909
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918699483081998e-05,
+      "loss": 1.0071,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918613591965434e-05,
+      "loss": 0.7693,
+      "step": 1911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9185276561262406e-05,
+      "loss": 0.9433,
+      "step": 1912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918441675565734e-05,
+      "loss": 0.8287,
+      "step": 1913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9183556502852324e-05,
+      "loss": 0.9332,
+      "step": 1914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9182695802860536e-05,
+      "loss": 0.8551,
+      "step": 1915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918183465569515e-05,
+      "loss": 0.8728,
+      "step": 1916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.918097306136939e-05,
+      "loss": 0.6393,
+      "step": 1917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9180111019896434e-05,
+      "loss": 0.7511,
+      "step": 1918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91792485312895e-05,
+      "loss": 0.7499,
+      "step": 1919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91783855955618e-05,
+      "loss": 0.9655,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917752221272657e-05,
+      "loss": 0.9849,
+      "step": 1921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917665838279703e-05,
+      "loss": 0.9746,
+      "step": 1922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917579410578641e-05,
+      "loss": 0.8675,
+      "step": 1923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9174929381707965e-05,
+      "loss": 0.9271,
+      "step": 1924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917406421057495e-05,
+      "loss": 0.9002,
+      "step": 1925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9173198592400605e-05,
+      "loss": 0.8026,
+      "step": 1926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91723325271982e-05,
+      "loss": 1.0287,
+      "step": 1927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917146601498102e-05,
+      "loss": 0.8981,
+      "step": 1928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.917059905576233e-05,
+      "loss": 0.9411,
+      "step": 1929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916973164955541e-05,
+      "loss": 0.9139,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9168863796373554e-05,
+      "loss": 0.8826,
+      "step": 1931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916799549623007e-05,
+      "loss": 0.8811,
+      "step": 1932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9167126749138255e-05,
+      "loss": 0.9443,
+      "step": 1933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916625755511143e-05,
+      "loss": 0.8747,
+      "step": 1934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.91653879141629e-05,
+      "loss": 0.9773,
+      "step": 1935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916451782630601e-05,
+      "loss": 0.787,
+      "step": 1936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.916364729155407e-05,
+      "loss": 0.7377,
+      "step": 1937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.916277630992043e-05,
+      "loss": 0.7927,
+      "step": 1938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.916190488141844e-05,
+      "loss": 0.8765,
+      "step": 1939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9161033006061444e-05,
+      "loss": 1.0182,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.916016068386282e-05,
+      "loss": 0.8987,
+      "step": 1941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.91592879148359e-05,
+      "loss": 0.8447,
+      "step": 1942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.915841469899409e-05,
+      "loss": 0.761,
+      "step": 1943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.915754103635077e-05,
+      "loss": 0.8432,
+      "step": 1944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9156666926919304e-05,
+      "loss": 0.6789,
+      "step": 1945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.915579237071311e-05,
+      "loss": 1.0052,
+      "step": 1946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.915491736774557e-05,
+      "loss": 0.8503,
+      "step": 1947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9154041918030106e-05,
+      "loss": 0.786,
+      "step": 1948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.915316602158013e-05,
+      "loss": 0.8888,
+      "step": 1949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.915228967840906e-05,
+      "loss": 1.0508,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.915141288853032e-05,
+      "loss": 0.9768,
+      "step": 1951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9150535651957354e-05,
+      "loss": 0.8153,
+      "step": 1952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.91496579687036e-05,
+      "loss": 0.9496,
+      "step": 1953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914877983878251e-05,
+      "loss": 0.8338,
+      "step": 1954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914790126220754e-05,
+      "loss": 0.864,
+      "step": 1955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914702223899214e-05,
+      "loss": 0.8981,
+      "step": 1956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.91461427691498e-05,
+      "loss": 0.6556,
+      "step": 1957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914526285269398e-05,
+      "loss": 0.8803,
+      "step": 1958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914438248963817e-05,
+      "loss": 0.7582,
+      "step": 1959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9143501679995865e-05,
+      "loss": 1.066,
+      "step": 1960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9142620423780545e-05,
+      "loss": 0.9801,
+      "step": 1961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9141738721005735e-05,
+      "loss": 1.068,
+      "step": 1962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.914085657168493e-05,
+      "loss": 0.7596,
+      "step": 1963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913997397583165e-05,
+      "loss": 1.0615,
+      "step": 1964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913909093345942e-05,
+      "loss": 0.7453,
+      "step": 1965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913820744458179e-05,
+      "loss": 0.819,
+      "step": 1966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913732350921227e-05,
+      "loss": 0.8761,
+      "step": 1967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913643912736441e-05,
+      "loss": 0.7714,
+      "step": 1968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9135554299051776e-05,
+      "loss": 0.7755,
+      "step": 1969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913466902428792e-05,
+      "loss": 0.8549,
+      "step": 1970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913378330308639e-05,
+      "loss": 0.8371,
+      "step": 1971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913289713546079e-05,
+      "loss": 0.824,
+      "step": 1972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913201052142467e-05,
+      "loss": 0.9345,
+      "step": 1973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9131123460991634e-05,
+      "loss": 0.9539,
+      "step": 1974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.913023595417527e-05,
+      "loss": 1.0337,
+      "step": 1975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9129348000989166e-05,
+      "loss": 0.983,
+      "step": 1976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912845960144695e-05,
+      "loss": 0.9664,
+      "step": 1977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912757075556222e-05,
+      "loss": 0.8496,
+      "step": 1978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.91266814633486e-05,
+      "loss": 0.8043,
+      "step": 1979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912579172481972e-05,
+      "loss": 1.0203,
+      "step": 1980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.91249015399892e-05,
+      "loss": 0.9101,
+      "step": 1981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912401090887069e-05,
+      "loss": 0.8793,
+      "step": 1982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9123119831477846e-05,
+      "loss": 0.9389,
+      "step": 1983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912222830782432e-05,
+      "loss": 0.8306,
+      "step": 1984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.912133633792375e-05,
+      "loss": 0.9349,
+      "step": 1985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9120443921789834e-05,
+      "loss": 0.845,
+      "step": 1986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911955105943623e-05,
+      "loss": 1.1432,
+      "step": 1987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9118657750876625e-05,
+      "loss": 0.7923,
+      "step": 1988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9117763996124695e-05,
+      "loss": 0.8291,
+      "step": 1989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9116869795194156e-05,
+      "loss": 0.7867,
+      "step": 1990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.91159751480987e-05,
+      "loss": 0.754,
+      "step": 1991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911508005485204e-05,
+      "loss": 0.9894,
+      "step": 1992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911418451546788e-05,
+      "loss": 0.8279,
+      "step": 1993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911328852995995e-05,
+      "loss": 0.5989,
+      "step": 1994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.911239209834198e-05,
+      "loss": 0.8934,
+      "step": 1995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9111495220627706e-05,
+      "loss": 0.9616,
+      "step": 1996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9110597896830876e-05,
+      "loss": 0.6693,
+      "step": 1997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910970012696523e-05,
+      "loss": 1.1799,
+      "step": 1998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910880191104453e-05,
+      "loss": 0.9585,
+      "step": 1999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.910790324908254e-05,
+      "loss": 0.8492,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.910700414109303e-05,
+      "loss": 0.7437,
+      "step": 2001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.910610458708977e-05,
+      "loss": 0.9517,
+      "step": 2002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.910520458708656e-05,
+      "loss": 0.8672,
+      "step": 2003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.910430414109718e-05,
+      "loss": 0.8561,
+      "step": 2004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.910340324913543e-05,
+      "loss": 0.879,
+      "step": 2005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.910250191121511e-05,
+      "loss": 0.7723,
+      "step": 2006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.910160012735004e-05,
+      "loss": 0.9224,
+      "step": 2007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.910069789755403e-05,
+      "loss": 0.986,
+      "step": 2008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.909979522184091e-05,
+      "loss": 0.92,
+      "step": 2009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.909889210022451e-05,
+      "loss": 0.884,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.909798853271867e-05,
+      "loss": 0.9713,
+      "step": 2011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9097084519337235e-05,
+      "loss": 0.8681,
+      "step": 2012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.909618006009406e-05,
+      "loss": 0.9526,
+      "step": 2013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9095275155003e-05,
+      "loss": 1.056,
+      "step": 2014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.909436980407793e-05,
+      "loss": 0.8909,
+      "step": 2015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9093464007332704e-05,
+      "loss": 0.8661,
+      "step": 2016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9092557764781216e-05,
+      "loss": 0.8323,
+      "step": 2017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.909165107643735e-05,
+      "loss": 1.0096,
+      "step": 2018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9090743942315e-05,
+      "loss": 0.9299,
+      "step": 2019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9089836362428074e-05,
+      "loss": 0.6653,
+      "step": 2020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9088928336790463e-05,
+      "loss": 0.7023,
+      "step": 2021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9088019865416096e-05,
+      "loss": 1.0906,
+      "step": 2022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908711094831888e-05,
+      "loss": 0.9412,
+      "step": 2023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9086201585512744e-05,
+      "loss": 0.8536,
+      "step": 2024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908529177701164e-05,
+      "loss": 0.8023,
+      "step": 2025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908438152282948e-05,
+      "loss": 0.9277,
+      "step": 2026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908347082298024e-05,
+      "loss": 0.7818,
+      "step": 2027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9082559677477866e-05,
+      "loss": 0.8971,
+      "step": 2028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908164808633632e-05,
+      "loss": 1.1794,
+      "step": 2029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.908073604956956e-05,
+      "loss": 1.0033,
+      "step": 2030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9079823567191563e-05,
+      "loss": 0.9846,
+      "step": 2031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9078910639216336e-05,
+      "loss": 0.8694,
+      "step": 2032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907799726565783e-05,
+      "loss": 0.9894,
+      "step": 2033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907708344653008e-05,
+      "loss": 1.0082,
+      "step": 2034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907616918184705e-05,
+      "loss": 0.9771,
+      "step": 2035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9075254471622776e-05,
+      "loss": 0.8234,
+      "step": 2036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907433931587126e-05,
+      "loss": 0.9701,
+      "step": 2037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9073423714606555e-05,
+      "loss": 0.9882,
+      "step": 2038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907250766784264e-05,
+      "loss": 0.6428,
+      "step": 2039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.90715911755936e-05,
+      "loss": 0.8171,
+      "step": 2040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.907067423787344e-05,
+      "loss": 0.8463,
+      "step": 2041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906975685469624e-05,
+      "loss": 0.8823,
+      "step": 2042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9068839026076044e-05,
+      "loss": 0.7846,
+      "step": 2043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906792075202692e-05,
+      "loss": 0.9222,
+      "step": 2044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9067002032562936e-05,
+      "loss": 0.8019,
+      "step": 2045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906608286769817e-05,
+      "loss": 0.7602,
+      "step": 2046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906516325744671e-05,
+      "loss": 1.1666,
+      "step": 2047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9064243201822636e-05,
+      "loss": 0.8368,
+      "step": 2048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906332270084007e-05,
+      "loss": 1.0404,
+      "step": 2049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.90624017545131e-05,
+      "loss": 0.7511,
+      "step": 2050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906148036285583e-05,
+      "loss": 0.8501,
+      "step": 2051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.906055852588239e-05,
+      "loss": 1.0803,
+      "step": 2052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905963624360691e-05,
+      "loss": 0.6931,
+      "step": 2053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905871351604351e-05,
+      "loss": 0.7788,
+      "step": 2054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905779034320634e-05,
+      "loss": 0.8238,
+      "step": 2055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905686672510954e-05,
+      "loss": 1.0648,
+      "step": 2056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9055942661767265e-05,
+      "loss": 0.8354,
+      "step": 2057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905501815319368e-05,
+      "loss": 0.823,
+      "step": 2058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905409319940294e-05,
+      "loss": 0.7879,
+      "step": 2059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905316780040923e-05,
+      "loss": 0.8163,
+      "step": 2060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.905224195622672e-05,
+      "loss": 0.9419,
+      "step": 2061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.90513156668696e-05,
+      "loss": 0.9295,
+      "step": 2062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9050388932352055e-05,
+      "loss": 0.6654,
+      "step": 2063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9049461752688315e-05,
+      "loss": 0.8729,
+      "step": 2064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.904853412789256e-05,
+      "loss": 0.7881,
+      "step": 2065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.904760605797901e-05,
+      "loss": 0.8054,
+      "step": 2066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.904667754296189e-05,
+      "loss": 0.7881,
+      "step": 2067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.904574858285542e-05,
+      "loss": 0.7893,
+      "step": 2068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.904481917767385e-05,
+      "loss": 0.7654,
+      "step": 2069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.904388932743141e-05,
+      "loss": 0.9983,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9042959032142344e-05,
+      "loss": 0.7033,
+      "step": 2071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.904202829182092e-05,
+      "loss": 0.9616,
+      "step": 2072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.90410971064814e-05,
+      "loss": 0.8722,
+      "step": 2073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.904016547613803e-05,
+      "loss": 0.9709,
+      "step": 2074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9039233400805114e-05,
+      "loss": 0.9356,
+      "step": 2075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9038300880496915e-05,
+      "loss": 0.7852,
+      "step": 2076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9037367915227734e-05,
+      "loss": 0.8144,
+      "step": 2077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.903643450501187e-05,
+      "loss": 0.8702,
+      "step": 2078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.90355006498636e-05,
+      "loss": 0.8004,
+      "step": 2079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9034566349797264e-05,
+      "loss": 0.8031,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9033631604827164e-05,
+      "loss": 0.9508,
+      "step": 2081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.903269641496763e-05,
+      "loss": 0.8699,
+      "step": 2082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9031760780232985e-05,
+      "loss": 0.9774,
+      "step": 2083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.903082470063757e-05,
+      "loss": 0.7113,
+      "step": 2084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902988817619573e-05,
+      "loss": 0.8369,
+      "step": 2085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9028951206921814e-05,
+      "loss": 0.8323,
+      "step": 2086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902801379283017e-05,
+      "loss": 0.7961,
+      "step": 2087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902707593393518e-05,
+      "loss": 0.8946,
+      "step": 2088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9026137630251204e-05,
+      "loss": 0.8038,
+      "step": 2089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9025198881792625e-05,
+      "loss": 0.8152,
+      "step": 2090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902425968857382e-05,
+      "loss": 0.9142,
+      "step": 2091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9023320050609184e-05,
+      "loss": 0.8566,
+      "step": 2092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902237996791312e-05,
+      "loss": 0.73,
+      "step": 2093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.902143944050003e-05,
+      "loss": 0.9969,
+      "step": 2094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9020498468384334e-05,
+      "loss": 0.9484,
+      "step": 2095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9019557051580426e-05,
+      "loss": 0.7739,
+      "step": 2096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9018615190102754e-05,
+      "loss": 0.8593,
+      "step": 2097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901767288396575e-05,
+      "loss": 1.0336,
+      "step": 2098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9016730133183844e-05,
+      "loss": 0.7744,
+      "step": 2099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901578693777148e-05,
+      "loss": 0.8561,
+      "step": 2100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901484329774312e-05,
+      "loss": 0.8934,
+      "step": 2101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9013899213113223e-05,
+      "loss": 0.8216,
+      "step": 2102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9012954683896245e-05,
+      "loss": 0.8538,
+      "step": 2103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9012009710106685e-05,
+      "loss": 1.092,
+      "step": 2104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901106429175898e-05,
+      "loss": 0.7682,
+      "step": 2105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.901011842886766e-05,
+      "loss": 0.9317,
+      "step": 2106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900917212144719e-05,
+      "loss": 0.745,
+      "step": 2107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900822536951208e-05,
+      "loss": 0.9681,
+      "step": 2108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900727817307685e-05,
+      "loss": 0.6404,
+      "step": 2109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9006330532156e-05,
+      "loss": 0.7598,
+      "step": 2110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9005382446764045e-05,
+      "loss": 0.8853,
+      "step": 2111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9004433916915514e-05,
+      "loss": 0.9518,
+      "step": 2112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900348494262496e-05,
+      "loss": 0.8008,
+      "step": 2113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900253552390691e-05,
+      "loss": 0.9209,
+      "step": 2114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.900158566077591e-05,
+      "loss": 0.834,
+      "step": 2115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9000635353246534e-05,
+      "loss": 0.8014,
+      "step": 2116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899968460133332e-05,
+      "loss": 0.7834,
+      "step": 2117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899873340505085e-05,
+      "loss": 0.9427,
+      "step": 2118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8997781764413686e-05,
+      "loss": 0.7833,
+      "step": 2119
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899682967943643e-05,
+      "loss": 0.8088,
+      "step": 2120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899587715013366e-05,
+      "loss": 0.7539,
+      "step": 2121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899492417651997e-05,
+      "loss": 0.9071,
+      "step": 2122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899397075860996e-05,
+      "loss": 0.7208,
+      "step": 2123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.8993016896418254e-05,
+      "loss": 0.8979,
+      "step": 2124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.899206258995946e-05,
+      "loss": 1.0991,
+      "step": 2125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.89911078392482e-05,
+      "loss": 0.8769,
+      "step": 2126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.89901526442991e-05,
+      "loss": 0.7562,
+      "step": 2127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.89891970051268e-05,
+      "loss": 0.9414,
+      "step": 2128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8988240921745944e-05,
+      "loss": 0.8033,
+      "step": 2129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8987284394171186e-05,
+      "loss": 1.1078,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.898632742241718e-05,
+      "loss": 0.8749,
+      "step": 2131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.898537000649859e-05,
+      "loss": 1.1465,
+      "step": 2132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8984412146430086e-05,
+      "loss": 0.8443,
+      "step": 2133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.898345384222635e-05,
+      "loss": 0.8053,
+      "step": 2134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.898249509390205e-05,
+      "loss": 0.9535,
+      "step": 2135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.898153590147189e-05,
+      "loss": 0.7133,
+      "step": 2136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.898057626495058e-05,
+      "loss": 0.7125,
+      "step": 2137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.89796161843528e-05,
+      "loss": 0.726,
+      "step": 2138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8978655659693276e-05,
+      "loss": 0.7092,
+      "step": 2139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.897769469098674e-05,
+      "loss": 1.0573,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.897673327824788e-05,
+      "loss": 0.9625,
+      "step": 2141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.897577142149146e-05,
+      "loss": 0.8234,
+      "step": 2142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.897480912073221e-05,
+      "loss": 0.9383,
+      "step": 2143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.897384637598486e-05,
+      "loss": 0.8651,
+      "step": 2144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8972883187264185e-05,
+      "loss": 0.7731,
+      "step": 2145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.897191955458493e-05,
+      "loss": 1.0301,
+      "step": 2146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.897095547796188e-05,
+      "loss": 0.9874,
+      "step": 2147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896999095740978e-05,
+      "loss": 0.8246,
+      "step": 2148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8969025992943425e-05,
+      "loss": 0.8433,
+      "step": 2149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8968060584577606e-05,
+      "loss": 0.9366,
+      "step": 2150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8967094732327105e-05,
+      "loss": 0.9838,
+      "step": 2151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896612843620673e-05,
+      "loss": 0.7424,
+      "step": 2152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896516169623128e-05,
+      "loss": 0.7352,
+      "step": 2153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.896419451241557e-05,
+      "loss": 0.8945,
+      "step": 2154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8963226884774434e-05,
+      "loss": 0.8269,
+      "step": 2155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8962258813322684e-05,
+      "loss": 0.8016,
+      "step": 2156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8961290298075156e-05,
+      "loss": 0.9004,
+      "step": 2157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.89603213390467e-05,
+      "loss": 1.0081,
+      "step": 2158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895935193625216e-05,
+      "loss": 0.8506,
+      "step": 2159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895838208970638e-05,
+      "loss": 0.7242,
+      "step": 2160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895741179942422e-05,
+      "loss": 0.9338,
+      "step": 2161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895644106542057e-05,
+      "loss": 0.8135,
+      "step": 2162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895546988771029e-05,
+      "loss": 0.8748,
+      "step": 2163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8954498266308264e-05,
+      "loss": 0.8233,
+      "step": 2164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895352620122938e-05,
+      "loss": 0.8035,
+      "step": 2165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895255369248852e-05,
+      "loss": 0.815,
+      "step": 2166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895158074010061e-05,
+      "loss": 0.7493,
+      "step": 2167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.895060734408055e-05,
+      "loss": 0.9084,
+      "step": 2168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8949633504443254e-05,
+      "loss": 0.8946,
+      "step": 2169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894865922120364e-05,
+      "loss": 0.8136,
+      "step": 2170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8947684494376635e-05,
+      "loss": 0.9347,
+      "step": 2171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894670932397719e-05,
+      "loss": 1.0443,
+      "step": 2172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894573371002023e-05,
+      "loss": 0.7645,
+      "step": 2173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894475765252072e-05,
+      "loss": 0.746,
+      "step": 2174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.89437811514936e-05,
+      "loss": 0.8203,
+      "step": 2175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894280420695386e-05,
+      "loss": 0.7548,
+      "step": 2176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894182681891643e-05,
+      "loss": 1.0423,
+      "step": 2177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.894084898739633e-05,
+      "loss": 0.9136,
+      "step": 2178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893987071240851e-05,
+      "loss": 0.8468,
+      "step": 2179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893889199396798e-05,
+      "loss": 0.9634,
+      "step": 2180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8937912832089724e-05,
+      "loss": 0.8482,
+      "step": 2181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8936933226788754e-05,
+      "loss": 1.3464,
+      "step": 2182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893595317808008e-05,
+      "loss": 0.7381,
+      "step": 2183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893497268597872e-05,
+      "loss": 0.8791,
+      "step": 2184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.8933991750499696e-05,
+      "loss": 0.9684,
+      "step": 2185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893301037165804e-05,
+      "loss": 0.7993,
+      "step": 2186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893202854946879e-05,
+      "loss": 0.9931,
+      "step": 2187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893104628394699e-05,
+      "loss": 0.8173,
+      "step": 2188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.893006357510769e-05,
+      "loss": 0.8949,
+      "step": 2189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892908042296596e-05,
+      "loss": 0.7506,
+      "step": 2190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.892809682753684e-05,
+      "loss": 0.8427,
+      "step": 2191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.892711278883543e-05,
+      "loss": 0.7396,
+      "step": 2192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.892612830687679e-05,
+      "loss": 0.952,
+      "step": 2193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8925143381676014e-05,
+      "loss": 0.9773,
+      "step": 2194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.892415801324819e-05,
+      "loss": 0.9234,
+      "step": 2195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.892317220160842e-05,
+      "loss": 0.9944,
+      "step": 2196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8922185946771804e-05,
+      "loss": 0.89,
+      "step": 2197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.892119924875346e-05,
+      "loss": 0.9193,
+      "step": 2198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.892021210756851e-05,
+      "loss": 0.7417,
+      "step": 2199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.891922452323208e-05,
+      "loss": 0.8768,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.89182364957593e-05,
+      "loss": 0.8526,
+      "step": 2201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.89172480251653e-05,
+      "loss": 1.0373,
+      "step": 2202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8916259111465245e-05,
+      "loss": 0.9676,
+      "step": 2203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8915269754674274e-05,
+      "loss": 0.9895,
+      "step": 2204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.891427995480756e-05,
+      "loss": 0.8865,
+      "step": 2205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8913289711880255e-05,
+      "loss": 1.0256,
+      "step": 2206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8912299025907545e-05,
+      "loss": 0.8437,
+      "step": 2207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.89113078969046e-05,
+      "loss": 0.9435,
+      "step": 2208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.891031632488662e-05,
+      "loss": 0.9974,
+      "step": 2209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890932430986879e-05,
+      "loss": 0.9492,
+      "step": 2210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890833185186631e-05,
+      "loss": 0.9731,
+      "step": 2211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8907338950894394e-05,
+      "loss": 0.9152,
+      "step": 2212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890634560696825e-05,
+      "loss": 0.8248,
+      "step": 2213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8905351820103106e-05,
+      "loss": 0.8928,
+      "step": 2214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890435759031419e-05,
+      "loss": 0.7993,
+      "step": 2215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890336291761673e-05,
+      "loss": 0.7443,
+      "step": 2216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8902367802025964e-05,
+      "loss": 0.9216,
+      "step": 2217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890137224355716e-05,
+      "loss": 0.7917,
+      "step": 2218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.890037624222555e-05,
+      "loss": 1.0332,
+      "step": 2219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8899379798046406e-05,
+      "loss": 0.8232,
+      "step": 2220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889838291103501e-05,
+      "loss": 0.9738,
+      "step": 2221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8897385581206614e-05,
+      "loss": 0.735,
+      "step": 2222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889638780857652e-05,
+      "loss": 0.9976,
+      "step": 2223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889538959316e-05,
+      "loss": 0.9309,
+      "step": 2224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889439093497236e-05,
+      "loss": 0.8735,
+      "step": 2225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.88933918340289e-05,
+      "loss": 0.9156,
+      "step": 2226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8892392290344934e-05,
+      "loss": 0.7457,
+      "step": 2227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889139230393577e-05,
+      "loss": 0.756,
+      "step": 2228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.889039187481675e-05,
+      "loss": 0.8891,
+      "step": 2229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888939100300317e-05,
+      "loss": 1.219,
+      "step": 2230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.88883896885104e-05,
+      "loss": 0.8593,
+      "step": 2231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8887387931353755e-05,
+      "loss": 0.7541,
+      "step": 2232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8886385731548616e-05,
+      "loss": 0.8636,
+      "step": 2233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888538308911031e-05,
+      "loss": 0.897,
+      "step": 2234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8884380004054225e-05,
+      "loss": 0.9706,
+      "step": 2235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888337647639572e-05,
+      "loss": 0.7886,
+      "step": 2236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888237250615017e-05,
+      "loss": 0.772,
+      "step": 2237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.888136809333296e-05,
+      "loss": 1.1017,
+      "step": 2238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8880363237959485e-05,
+      "loss": 0.8776,
+      "step": 2239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8879357940045134e-05,
+      "loss": 0.7707,
+      "step": 2240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8878352199605324e-05,
+      "loss": 1.1625,
+      "step": 2241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8877346016655465e-05,
+      "loss": 1.0017,
+      "step": 2242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.8876339391210966e-05,
+      "loss": 0.7707,
+      "step": 2243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887533232328726e-05,
+      "loss": 0.9898,
+      "step": 2244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887432481289977e-05,
+      "loss": 0.9184,
+      "step": 2245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887331686006395e-05,
+      "loss": 1.0888,
+      "step": 2246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887230846479522e-05,
+      "loss": 0.9065,
+      "step": 2247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.887129962710905e-05,
+      "loss": 1.0075,
+      "step": 2248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.88702903470209e-05,
+      "loss": 0.6752,
+      "step": 2249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886928062454622e-05,
+      "loss": 0.9293,
+      "step": 2250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886827045970049e-05,
+      "loss": 0.8536,
+      "step": 2251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886725985249921e-05,
+      "loss": 1.0592,
+      "step": 2252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886624880295783e-05,
+      "loss": 1.0162,
+      "step": 2253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.886523731109187e-05,
+      "loss": 0.8241,
+      "step": 2254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8864225376916813e-05,
+      "loss": 1.0316,
+      "step": 2255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.886321300044817e-05,
+      "loss": 0.8595,
+      "step": 2256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.886220018170146e-05,
+      "loss": 1.0196,
+      "step": 2257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8861186920692206e-05,
+      "loss": 0.8605,
+      "step": 2258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.886017321743591e-05,
+      "loss": 0.8047,
+      "step": 2259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.885915907194813e-05,
+      "loss": 0.7811,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.88581444842444e-05,
+      "loss": 0.7867,
+      "step": 2261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8857129454340255e-05,
+      "loss": 0.8244,
+      "step": 2262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.885611398225126e-05,
+      "loss": 0.9867,
+      "step": 2263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.885509806799298e-05,
+      "loss": 0.8149,
+      "step": 2264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8854081711580965e-05,
+      "loss": 0.8005,
+      "step": 2265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.88530649130308e-05,
+      "loss": 0.7149,
+      "step": 2266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.885204767235808e-05,
+      "loss": 0.8401,
+      "step": 2267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.885102998957836e-05,
+      "loss": 0.839,
+      "step": 2268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.885001186470726e-05,
+      "loss": 0.9119,
+      "step": 2269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884899329776037e-05,
+      "loss": 0.7314,
+      "step": 2270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884797428875331e-05,
+      "loss": 0.6562,
+      "step": 2271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884695483770166e-05,
+      "loss": 0.8265,
+      "step": 2272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884593494462108e-05,
+      "loss": 0.7731,
+      "step": 2273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884491460952719e-05,
+      "loss": 0.9405,
+      "step": 2274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884389383243561e-05,
+      "loss": 0.9549,
+      "step": 2275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884287261336199e-05,
+      "loss": 0.711,
+      "step": 2276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884185095232198e-05,
+      "loss": 0.7997,
+      "step": 2277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.884082884933124e-05,
+      "loss": 0.9389,
+      "step": 2278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883980630440543e-05,
+      "loss": 0.8194,
+      "step": 2279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.88387833175602e-05,
+      "loss": 1.0232,
+      "step": 2280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883775988881125e-05,
+      "loss": 0.7843,
+      "step": 2281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883673601817424e-05,
+      "loss": 0.89,
+      "step": 2282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883571170566488e-05,
+      "loss": 0.7501,
+      "step": 2283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883468695129887e-05,
+      "loss": 0.9551,
+      "step": 2284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8833661755091884e-05,
+      "loss": 0.8277,
+      "step": 2285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883263611705965e-05,
+      "loss": 1.306,
+      "step": 2286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.883161003721788e-05,
+      "loss": 0.8481,
+      "step": 2287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.88305835155823e-05,
+      "loss": 0.8139,
+      "step": 2288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882955655216864e-05,
+      "loss": 0.8717,
+      "step": 2289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882852914699262e-05,
+      "loss": 0.8984,
+      "step": 2290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8827501300070015e-05,
+      "loss": 0.9613,
+      "step": 2291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882647301141654e-05,
+      "loss": 0.6467,
+      "step": 2292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882544428104798e-05,
+      "loss": 0.7941,
+      "step": 2293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882441510898009e-05,
+      "loss": 0.7254,
+      "step": 2294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8823385495228637e-05,
+      "loss": 0.7738,
+      "step": 2295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8822355439809394e-05,
+      "loss": 0.7758,
+      "step": 2296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.882132494273815e-05,
+      "loss": 0.8429,
+      "step": 2297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.88202940040307e-05,
+      "loss": 0.8153,
+      "step": 2298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881926262370283e-05,
+      "loss": 0.9911,
+      "step": 2299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8818230801770353e-05,
+      "loss": 0.8752,
+      "step": 2300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8817198538249076e-05,
+      "loss": 0.7561,
+      "step": 2301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881616583315482e-05,
+      "loss": 0.7626,
+      "step": 2302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8815132686503405e-05,
+      "loss": 0.781,
+      "step": 2303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881409909831067e-05,
+      "loss": 0.754,
+      "step": 2304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881306506859244e-05,
+      "loss": 0.7119,
+      "step": 2305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881203059736457e-05,
+      "loss": 0.8088,
+      "step": 2306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.881099568464291e-05,
+      "loss": 0.7664,
+      "step": 2307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880996033044332e-05,
+      "loss": 0.7086,
+      "step": 2308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880892453478165e-05,
+      "loss": 0.8227,
+      "step": 2309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8807888297673795e-05,
+      "loss": 0.9485,
+      "step": 2310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8806851619135616e-05,
+      "loss": 0.8996,
+      "step": 2311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8805814499183015e-05,
+      "loss": 0.9736,
+      "step": 2312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8804776937831854e-05,
+      "loss": 0.7645,
+      "step": 2313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8803738935098066e-05,
+      "loss": 0.9697,
+      "step": 2314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880270049099754e-05,
+      "loss": 0.7565,
+      "step": 2315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.880166160554618e-05,
+      "loss": 0.8996,
+      "step": 2316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.8800622278759926e-05,
+      "loss": 0.807,
+      "step": 2317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.879958251065469e-05,
+      "loss": 0.7822,
+      "step": 2318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8798542301246405e-05,
+      "loss": 0.7005,
+      "step": 2319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8797501650551014e-05,
+      "loss": 1.0612,
+      "step": 2320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.879646055858447e-05,
+      "loss": 0.8035,
+      "step": 2321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.879541902536271e-05,
+      "loss": 0.8476,
+      "step": 2322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8794377050901704e-05,
+      "loss": 0.9652,
+      "step": 2323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.879333463521741e-05,
+      "loss": 0.8487,
+      "step": 2324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8792291778325814e-05,
+      "loss": 0.6872,
+      "step": 2325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8791248480242885e-05,
+      "loss": 1.0472,
+      "step": 2326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.879020474098462e-05,
+      "loss": 0.7508,
+      "step": 2327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8789160560567e-05,
+      "loss": 0.7364,
+      "step": 2328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8788115939006036e-05,
+      "loss": 0.7619,
+      "step": 2329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.878707087631773e-05,
+      "loss": 1.0406,
+      "step": 2330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87860253725181e-05,
+      "loss": 0.8796,
+      "step": 2331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.878497942762316e-05,
+      "loss": 0.8045,
+      "step": 2332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.878393304164894e-05,
+      "loss": 0.7601,
+      "step": 2333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.878288621461147e-05,
+      "loss": 0.7577,
+      "step": 2334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87818389465268e-05,
+      "loss": 0.7996,
+      "step": 2335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.878079123741097e-05,
+      "loss": 0.7366,
+      "step": 2336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877974308728004e-05,
+      "loss": 0.8644,
+      "step": 2337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877869449615006e-05,
+      "loss": 0.8732,
+      "step": 2338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8777645464037115e-05,
+      "loss": 0.9526,
+      "step": 2339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877659599095727e-05,
+      "loss": 0.8289,
+      "step": 2340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87755460769266e-05,
+      "loss": 1.0003,
+      "step": 2341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8774495721961205e-05,
+      "loss": 0.8313,
+      "step": 2342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877344492607717e-05,
+      "loss": 0.8524,
+      "step": 2343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8772393689290605e-05,
+      "loss": 0.8291,
+      "step": 2344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877134201161761e-05,
+      "loss": 1.024,
+      "step": 2345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.877028989307431e-05,
+      "loss": 0.8923,
+      "step": 2346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876923733367682e-05,
+      "loss": 0.9712,
+      "step": 2347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8768184333441276e-05,
+      "loss": 0.79,
+      "step": 2348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876713089238379e-05,
+      "loss": 1.0251,
+      "step": 2349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876607701052054e-05,
+      "loss": 0.8787,
+      "step": 2350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876502268786765e-05,
+      "loss": 0.7583,
+      "step": 2351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876396792444129e-05,
+      "loss": 0.8313,
+      "step": 2352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8762912720257606e-05,
+      "loss": 0.8136,
+      "step": 2353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.876185707533277e-05,
+      "loss": 0.8395,
+      "step": 2354
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8760800989682974e-05,
+      "loss": 0.7742,
+      "step": 2355
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8759744463324386e-05,
+      "loss": 0.8708,
+      "step": 2356
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87586874962732e-05,
+      "loss": 1.0409,
+      "step": 2357
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8757630088545616e-05,
+      "loss": 0.9728,
+      "step": 2358
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8756572240157824e-05,
+      "loss": 0.7854,
+      "step": 2359
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875551395112605e-05,
+      "loss": 0.7197,
+      "step": 2360
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87544552214665e-05,
+      "loss": 0.8701,
+      "step": 2361
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87533960511954e-05,
+      "loss": 0.8418,
+      "step": 2362
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875233644032898e-05,
+      "loss": 0.8522,
+      "step": 2363
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875127638888348e-05,
+      "loss": 0.8431,
+      "step": 2364
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.875021589687514e-05,
+      "loss": 0.8438,
+      "step": 2365
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87491549643202e-05,
+      "loss": 0.796,
+      "step": 2366
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874809359123494e-05,
+      "loss": 0.8065,
+      "step": 2367
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87470317776356e-05,
+      "loss": 0.7283,
+      "step": 2368
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874596952353847e-05,
+      "loss": 0.7335,
+      "step": 2369
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874490682895982e-05,
+      "loss": 0.8432,
+      "step": 2370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874384369391593e-05,
+      "loss": 0.9137,
+      "step": 2371
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8742780118423095e-05,
+      "loss": 0.7557,
+      "step": 2372
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87417161024976e-05,
+      "loss": 0.9979,
+      "step": 2373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.874065164615577e-05,
+      "loss": 0.8639,
+      "step": 2374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8739586749413895e-05,
+      "loss": 0.9285,
+      "step": 2375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873852141228831e-05,
+      "loss": 0.9555,
+      "step": 2376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873745563479534e-05,
+      "loss": 0.9973,
+      "step": 2377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87363894169513e-05,
+      "loss": 0.8016,
+      "step": 2378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.873532275877254e-05,
+      "loss": 0.8151,
+      "step": 2379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.87342556602754e-05,
+      "loss": 0.7471,
+      "step": 2380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.8733188121476237e-05,
+      "loss": 0.5844,
+      "step": 2381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.87321201423914e-05,
+      "loss": 0.9546,
+      "step": 2382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.873105172303726e-05,
+      "loss": 0.8162,
+      "step": 2383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.872998286343019e-05,
+      "loss": 1.0588,
+      "step": 2384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.872891356358656e-05,
+      "loss": 1.0111,
+      "step": 2385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.872784382352277e-05,
+      "loss": 1.0627,
+      "step": 2386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.87267736432552e-05,
+      "loss": 0.9821,
+      "step": 2387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.872570302280025e-05,
+      "loss": 1.0406,
+      "step": 2388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8724631962174325e-05,
+      "loss": 1.1221,
+      "step": 2389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.872356046139384e-05,
+      "loss": 1.0468,
+      "step": 2390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8722488520475216e-05,
+      "loss": 0.9308,
+      "step": 2391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.872141613943487e-05,
+      "loss": 0.8211,
+      "step": 2392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8720343318289244e-05,
+      "loss": 0.7643,
+      "step": 2393
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.871927005705478e-05,
+      "loss": 0.843,
+      "step": 2394
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.871819635574791e-05,
+      "loss": 0.8618,
+      "step": 2395
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.87171222143851e-05,
+      "loss": 1.0477,
+      "step": 2396
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8716047632982795e-05,
+      "loss": 1.144,
+      "step": 2397
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.871497261155747e-05,
+      "loss": 0.7512,
+      "step": 2398
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.871389715012559e-05,
+      "loss": 0.9384,
+      "step": 2399
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8712821248703655e-05,
+      "loss": 0.9693,
+      "step": 2400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.871174490730813e-05,
+      "loss": 0.8787,
+      "step": 2401
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.871066812595552e-05,
+      "loss": 0.8771,
+      "step": 2402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8709590904662315e-05,
+      "loss": 0.9817,
+      "step": 2403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870851324344503e-05,
+      "loss": 1.0066,
+      "step": 2404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870743514232018e-05,
+      "loss": 0.8936,
+      "step": 2405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870635660130427e-05,
+      "loss": 1.0549,
+      "step": 2406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870527762041385e-05,
+      "loss": 0.7309,
+      "step": 2407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8704198199665426e-05,
+      "loss": 0.7891,
+      "step": 2408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870311833907556e-05,
+      "loss": 0.7273,
+      "step": 2409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870203803866079e-05,
+      "loss": 0.9375,
+      "step": 2410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.870095729843768e-05,
+      "loss": 1.065,
+      "step": 2411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869987611842277e-05,
+      "loss": 0.7292,
+      "step": 2412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869879449863264e-05,
+      "loss": 0.8149,
+      "step": 2413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869771243908387e-05,
+      "loss": 0.7868,
+      "step": 2414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869662993979303e-05,
+      "loss": 0.8454,
+      "step": 2415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869554700077671e-05,
+      "loss": 1.1952,
+      "step": 2416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869446362205151e-05,
+      "loss": 1.0268,
+      "step": 2417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869337980363402e-05,
+      "loss": 0.9091,
+      "step": 2418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869229554554086e-05,
+      "loss": 0.7741,
+      "step": 2419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8691210847788635e-05,
+      "loss": 0.7814,
+      "step": 2420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.869012571039397e-05,
+      "loss": 0.8835,
+      "step": 2421
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8689040133373485e-05,
+      "loss": 0.881,
+      "step": 2422
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868795411674383e-05,
+      "loss": 0.8196,
+      "step": 2423
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868686766052163e-05,
+      "loss": 0.7542,
+      "step": 2424
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868578076472355e-05,
+      "loss": 0.7663,
+      "step": 2425
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868469342936623e-05,
+      "loss": 0.8549,
+      "step": 2426
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868360565446633e-05,
+      "loss": 1.0225,
+      "step": 2427
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8682517440040534e-05,
+      "loss": 0.7412,
+      "step": 2428
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868142878610551e-05,
+      "loss": 0.8115,
+      "step": 2429
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.868033969267794e-05,
+      "loss": 0.834,
+      "step": 2430
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.86792501597745e-05,
+      "loss": 0.7312,
+      "step": 2431
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867816018741191e-05,
+      "loss": 0.835,
+      "step": 2432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867706977560685e-05,
+      "loss": 0.7142,
+      "step": 2433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867597892437603e-05,
+      "loss": 0.7188,
+      "step": 2434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867488763373618e-05,
+      "loss": 0.8002,
+      "step": 2435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867379590370401e-05,
+      "loss": 0.7956,
+      "step": 2436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867270373429625e-05,
+      "loss": 0.9242,
+      "step": 2437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8671611125529645e-05,
+      "loss": 0.7352,
+      "step": 2438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.867051807742092e-05,
+      "loss": 0.7924,
+      "step": 2439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866942458998685e-05,
+      "loss": 0.824,
+      "step": 2440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866833066324417e-05,
+      "loss": 0.7635,
+      "step": 2441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866723629720964e-05,
+      "loss": 0.9788,
+      "step": 2442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.866614149190004e-05,
+      "loss": 0.8275,
+      "step": 2443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8665046247332144e-05,
+      "loss": 0.9862,
+      "step": 2444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.8663950563522735e-05,
+      "loss": 0.9354,
+      "step": 2445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8662854440488606e-05,
+      "loss": 1.2451,
+      "step": 2446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.866175787824654e-05,
+      "loss": 0.6867,
+      "step": 2447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.866066087681335e-05,
+      "loss": 0.8611,
+      "step": 2448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.865956343620585e-05,
+      "loss": 0.8604,
+      "step": 2449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8658465556440844e-05,
+      "loss": 0.8054,
+      "step": 2450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.865736723753516e-05,
+      "loss": 0.8754,
+      "step": 2451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.865626847950563e-05,
+      "loss": 1.0385,
+      "step": 2452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8655169282369094e-05,
+      "loss": 0.9903,
+      "step": 2453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.865406964614238e-05,
+      "loss": 0.7883,
+      "step": 2454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.865296957084236e-05,
+      "loss": 0.85,
+      "step": 2455
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.865186905648588e-05,
+      "loss": 0.9685,
+      "step": 2456
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.865076810308979e-05,
+      "loss": 0.9911,
+      "step": 2457
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8649666710670986e-05,
+      "loss": 0.877,
+      "step": 2458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.864856487924633e-05,
+      "loss": 0.948,
+      "step": 2459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8647462608832707e-05,
+      "loss": 0.8355,
+      "step": 2460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8646359899447e-05,
+      "loss": 0.8496,
+      "step": 2461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.864525675110612e-05,
+      "loss": 1.0739,
+      "step": 2462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8644153163826975e-05,
+      "loss": 0.8614,
+      "step": 2463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.864304913762645e-05,
+      "loss": 0.836,
+      "step": 2464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8641944672521485e-05,
+      "loss": 0.7468,
+      "step": 2465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8640839768529e-05,
+      "loss": 0.8117,
+      "step": 2466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863973442566591e-05,
+      "loss": 0.8533,
+      "step": 2467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863862864394918e-05,
+      "loss": 0.8283,
+      "step": 2468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863752242339573e-05,
+      "loss": 1.0154,
+      "step": 2469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863641576402252e-05,
+      "loss": 0.7589,
+      "step": 2470
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863530866584651e-05,
+      "loss": 0.7258,
+      "step": 2471
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8634201128884667e-05,
+      "loss": 0.8727,
+      "step": 2472
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863309315315395e-05,
+      "loss": 0.8287,
+      "step": 2473
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8631984738671345e-05,
+      "loss": 0.7395,
+      "step": 2474
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.863087588545384e-05,
+      "loss": 0.9024,
+      "step": 2475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862976659351841e-05,
+      "loss": 0.7865,
+      "step": 2476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862865686288208e-05,
+      "loss": 0.7473,
+      "step": 2477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8627546693561826e-05,
+      "loss": 0.9119,
+      "step": 2478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862643608557468e-05,
+      "loss": 0.7844,
+      "step": 2479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862532503893765e-05,
+      "loss": 0.8626,
+      "step": 2480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862421355366776e-05,
+      "loss": 0.7962,
+      "step": 2481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862310162978205e-05,
+      "loss": 1.0602,
+      "step": 2482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.862198926729755e-05,
+      "loss": 0.8472,
+      "step": 2483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.86208764662313e-05,
+      "loss": 0.9069,
+      "step": 2484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861976322660038e-05,
+      "loss": 0.9367,
+      "step": 2485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861864954842182e-05,
+      "loss": 0.9127,
+      "step": 2486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8617535431712684e-05,
+      "loss": 0.9545,
+      "step": 2487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861642087649007e-05,
+      "loss": 1.0416,
+      "step": 2488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861530588277103e-05,
+      "loss": 0.9006,
+      "step": 2489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861419045057267e-05,
+      "loss": 0.7987,
+      "step": 2490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861307457991206e-05,
+      "loss": 0.8129,
+      "step": 2491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8611958270806305e-05,
+      "loss": 1.0658,
+      "step": 2492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.861084152327253e-05,
+      "loss": 0.9217,
+      "step": 2493
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860972433732782e-05,
+      "loss": 0.7951,
+      "step": 2494
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8608606712989316e-05,
+      "loss": 0.9314,
+      "step": 2495
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860748865027413e-05,
+      "loss": 0.976,
+      "step": 2496
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8606370149199405e-05,
+      "loss": 0.7751,
+      "step": 2497
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860525120978227e-05,
+      "loss": 0.7023,
+      "step": 2498
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8604131832039876e-05,
+      "loss": 0.9595,
+      "step": 2499
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860301201598937e-05,
+      "loss": 0.7549,
+      "step": 2500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8601891761647914e-05,
+      "loss": 0.7618,
+      "step": 2501
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.860077106903269e-05,
+      "loss": 0.8093,
+      "step": 2502
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.859964993816084e-05,
+      "loss": 0.8576,
+      "step": 2503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.859852836904956e-05,
+      "loss": 0.8398,
+      "step": 2504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.859740636171604e-05,
+      "loss": 0.9751,
+      "step": 2505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8596283916177473e-05,
+      "loss": 0.8754,
+      "step": 2506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.859516103245104e-05,
+      "loss": 0.819,
+      "step": 2507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.8594037710553974e-05,
+      "loss": 1.1095,
+      "step": 2508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.859291395050348e-05,
+      "loss": 0.9808,
+      "step": 2509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.859178975231676e-05,
+      "loss": 0.6962,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.859066511601105e-05,
+      "loss": 0.9126,
+      "step": 2511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.858954004160359e-05,
+      "loss": 0.8346,
+      "step": 2512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.858841452911162e-05,
+      "loss": 0.8349,
+      "step": 2513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.858728857855238e-05,
+      "loss": 0.7564,
+      "step": 2514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.858616218994313e-05,
+      "loss": 0.743,
+      "step": 2515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8585035363301116e-05,
+      "loss": 0.9131,
+      "step": 2516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.858390809864362e-05,
+      "loss": 0.9119,
+      "step": 2517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.858278039598791e-05,
+      "loss": 1.0454,
+      "step": 2518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.858165225535126e-05,
+      "loss": 0.8513,
+      "step": 2519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.858052367675097e-05,
+      "loss": 0.8921,
+      "step": 2520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.857939466020432e-05,
+      "loss": 0.8295,
+      "step": 2521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8578265205728615e-05,
+      "loss": 0.7715,
+      "step": 2522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8577135313341164e-05,
+      "loss": 0.7773,
+      "step": 2523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.857600498305928e-05,
+      "loss": 0.8654,
+      "step": 2524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8574874214900285e-05,
+      "loss": 0.8831,
+      "step": 2525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.857374300888151e-05,
+      "loss": 0.9626,
+      "step": 2526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.857261136502027e-05,
+      "loss": 0.59,
+      "step": 2527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.857147928333392e-05,
+      "loss": 0.9458,
+      "step": 2528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.857034676383981e-05,
+      "loss": 0.8741,
+      "step": 2529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856921380655528e-05,
+      "loss": 0.842,
+      "step": 2530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856808041149771e-05,
+      "loss": 0.877,
+      "step": 2531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856694657868445e-05,
+      "loss": 0.8188,
+      "step": 2532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856581230813289e-05,
+      "loss": 1.0379,
+      "step": 2533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8564677599860394e-05,
+      "loss": 0.9124,
+      "step": 2534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856354245388436e-05,
+      "loss": 1.0452,
+      "step": 2535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856240687022218e-05,
+      "loss": 0.8378,
+      "step": 2536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8561270848891245e-05,
+      "loss": 0.9712,
+      "step": 2537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.856013438990898e-05,
+      "loss": 0.8118,
+      "step": 2538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855899749329279e-05,
+      "loss": 0.9695,
+      "step": 2539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855786015906009e-05,
+      "loss": 0.8991,
+      "step": 2540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855672238722833e-05,
+      "loss": 0.9438,
+      "step": 2541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855558417781492e-05,
+      "loss": 0.8778,
+      "step": 2542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.85544455308373e-05,
+      "loss": 0.7706,
+      "step": 2543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855330644631294e-05,
+      "loss": 0.8995,
+      "step": 2544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855216692425928e-05,
+      "loss": 0.7804,
+      "step": 2545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.855102696469378e-05,
+      "loss": 0.8088,
+      "step": 2546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8549886567633914e-05,
+      "loss": 0.8271,
+      "step": 2547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854874573309715e-05,
+      "loss": 0.8524,
+      "step": 2548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8547604461100965e-05,
+      "loss": 0.9939,
+      "step": 2549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854646275166287e-05,
+      "loss": 0.8088,
+      "step": 2550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854532060480034e-05,
+      "loss": 0.7712,
+      "step": 2551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854417802053088e-05,
+      "loss": 0.7362,
+      "step": 2552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8543034998872e-05,
+      "loss": 1.1896,
+      "step": 2553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8541891539841214e-05,
+      "loss": 0.8333,
+      "step": 2554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.854074764345604e-05,
+      "loss": 0.9296,
+      "step": 2555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8539603309734015e-05,
+      "loss": 0.841,
+      "step": 2556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853845853869267e-05,
+      "loss": 0.6995,
+      "step": 2557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853731333034953e-05,
+      "loss": 0.9407,
+      "step": 2558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8536167684722177e-05,
+      "loss": 0.8335,
+      "step": 2559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8535021601828136e-05,
+      "loss": 0.7811,
+      "step": 2560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853387508168498e-05,
+      "loss": 0.6827,
+      "step": 2561
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.853272812431029e-05,
+      "loss": 0.7576,
+      "step": 2562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8531580729721614e-05,
+      "loss": 0.8141,
+      "step": 2563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8530432897936556e-05,
+      "loss": 0.8002,
+      "step": 2564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8529284628972696e-05,
+      "loss": 0.9692,
+      "step": 2565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852813592284763e-05,
+      "loss": 1.1197,
+      "step": 2566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852698677957896e-05,
+      "loss": 1.0411,
+      "step": 2567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8525837199184286e-05,
+      "loss": 0.9226,
+      "step": 2568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8524687181681245e-05,
+      "loss": 0.808,
+      "step": 2569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852353672708744e-05,
+      "loss": 0.7351,
+      "step": 2570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.852238583542051e-05,
+      "loss": 0.7913,
+      "step": 2571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.8521234506698075e-05,
+      "loss": 0.9852,
+      "step": 2572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.85200827409378e-05,
+      "loss": 0.9676,
+      "step": 2573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.851893053815732e-05,
+      "loss": 0.8136,
+      "step": 2574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8517777898374285e-05,
+      "loss": 0.9954,
+      "step": 2575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8516624821606374e-05,
+      "loss": 0.8145,
+      "step": 2576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8515471307871234e-05,
+      "loss": 0.9711,
+      "step": 2577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8514317357186567e-05,
+      "loss": 0.8164,
+      "step": 2578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8513162969570036e-05,
+      "loss": 0.9764,
+      "step": 2579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8512008145039335e-05,
+      "loss": 0.8766,
+      "step": 2580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.851085288361216e-05,
+      "loss": 0.858,
+      "step": 2581
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.850969718530621e-05,
+      "loss": 0.6965,
+      "step": 2582
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.85085410501392e-05,
+      "loss": 0.9253,
+      "step": 2583
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8507384478128846e-05,
+      "loss": 0.9537,
+      "step": 2584
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8506227469292864e-05,
+      "loss": 0.9269,
+      "step": 2585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8505070023648985e-05,
+      "loss": 0.6588,
+      "step": 2586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8503912141214955e-05,
+      "loss": 0.8167,
+      "step": 2587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.85027538220085e-05,
+      "loss": 0.8099,
+      "step": 2588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.850159506604738e-05,
+      "loss": 0.8317,
+      "step": 2589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.850043587334935e-05,
+      "loss": 0.9807,
+      "step": 2590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849927624393217e-05,
+      "loss": 0.9469,
+      "step": 2591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84981161778136e-05,
+      "loss": 0.9626,
+      "step": 2592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849695567501144e-05,
+      "loss": 0.701,
+      "step": 2593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8495794735543455e-05,
+      "loss": 0.7764,
+      "step": 2594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849463335942743e-05,
+      "loss": 0.947,
+      "step": 2595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8493471546681175e-05,
+      "loss": 0.9453,
+      "step": 2596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.849230929732249e-05,
+      "loss": 0.9124,
+      "step": 2597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8491146611369176e-05,
+      "loss": 1.1347,
+      "step": 2598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8489983488839056e-05,
+      "loss": 0.8718,
+      "step": 2599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8488819929749957e-05,
+      "loss": 0.8153,
+      "step": 2600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848765593411969e-05,
+      "loss": 0.7117,
+      "step": 2601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848649150196611e-05,
+      "loss": 0.8868,
+      "step": 2602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848532663330706e-05,
+      "loss": 0.8298,
+      "step": 2603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8484161328160376e-05,
+      "loss": 0.878,
+      "step": 2604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8482995586543924e-05,
+      "loss": 0.9174,
+      "step": 2605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848182940847556e-05,
+      "loss": 0.8791,
+      "step": 2606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.848066279397317e-05,
+      "loss": 1.0338,
+      "step": 2607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8479495743054606e-05,
+      "loss": 1.1381,
+      "step": 2608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847832825573776e-05,
+      "loss": 0.9975,
+      "step": 2609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847716033204054e-05,
+      "loss": 0.688,
+      "step": 2610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847599197198082e-05,
+      "loss": 0.7088,
+      "step": 2611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8474823175576514e-05,
+      "loss": 0.8872,
+      "step": 2612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847365394284552e-05,
+      "loss": 0.9356,
+      "step": 2613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8472484273805774e-05,
+      "loss": 0.9295,
+      "step": 2614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.847131416847518e-05,
+      "loss": 0.7403,
+      "step": 2615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8470143626871675e-05,
+      "loss": 0.7433,
+      "step": 2616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8468972649013194e-05,
+      "loss": 0.8849,
+      "step": 2617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846780123491769e-05,
+      "loss": 0.9199,
+      "step": 2618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.84666293846031e-05,
+      "loss": 0.7646,
+      "step": 2619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846545709808739e-05,
+      "loss": 0.8796,
+      "step": 2620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8464284375388514e-05,
+      "loss": 0.829,
+      "step": 2621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846311121652445e-05,
+      "loss": 0.8582,
+      "step": 2622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846193762151318e-05,
+      "loss": 0.8295,
+      "step": 2623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.846076359037266e-05,
+      "loss": 0.8084,
+      "step": 2624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8459589123120916e-05,
+      "loss": 0.9663,
+      "step": 2625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845841421977592e-05,
+      "loss": 0.8334,
+      "step": 2626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845723888035569e-05,
+      "loss": 0.9585,
+      "step": 2627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845606310487821e-05,
+      "loss": 0.8276,
+      "step": 2628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845488689336154e-05,
+      "loss": 0.8376,
+      "step": 2629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.8453710245823666e-05,
+      "loss": 0.7261,
+      "step": 2630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845253316228263e-05,
+      "loss": 0.9728,
+      "step": 2631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845135564275647e-05,
+      "loss": 1.0412,
+      "step": 2632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.845017768726323e-05,
+      "loss": 1.1526,
+      "step": 2633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.844899929582096e-05,
+      "loss": 1.0792,
+      "step": 2634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.844782046844772e-05,
+      "loss": 1.0213,
+      "step": 2635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.844664120516156e-05,
+      "loss": 0.9465,
+      "step": 2636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.844546150598057e-05,
+      "loss": 0.8128,
+      "step": 2637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.84442813709228e-05,
+      "loss": 0.9347,
+      "step": 2638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.844310080000636e-05,
+      "loss": 0.9015,
+      "step": 2639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.844191979324933e-05,
+      "loss": 0.7106,
+      "step": 2640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.844073835066982e-05,
+      "loss": 0.8049,
+      "step": 2641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8439556472285895e-05,
+      "loss": 0.9685,
+      "step": 2642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.843837415811571e-05,
+      "loss": 0.7792,
+      "step": 2643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8437191408177356e-05,
+      "loss": 1.0453,
+      "step": 2644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.843600822248896e-05,
+      "loss": 1.0426,
+      "step": 2645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8434824601068664e-05,
+      "loss": 0.7891,
+      "step": 2646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.843364054393459e-05,
+      "loss": 0.6665,
+      "step": 2647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8432456051104885e-05,
+      "loss": 0.6935,
+      "step": 2648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.843127112259771e-05,
+      "loss": 0.8823,
+      "step": 2649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.843008575843121e-05,
+      "loss": 0.8358,
+      "step": 2650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842889995862355e-05,
+      "loss": 0.9765,
+      "step": 2651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842771372319291e-05,
+      "loss": 0.8353,
+      "step": 2652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842652705215745e-05,
+      "loss": 0.6889,
+      "step": 2653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842533994553537e-05,
+      "loss": 1.0,
+      "step": 2654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8424152403344856e-05,
+      "loss": 0.7488,
+      "step": 2655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.84229644256041e-05,
+      "loss": 0.832,
+      "step": 2656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842177601233131e-05,
+      "loss": 0.8274,
+      "step": 2657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.842058716354469e-05,
+      "loss": 1.0078,
+      "step": 2658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8419397879262466e-05,
+      "loss": 0.9177,
+      "step": 2659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841820815950286e-05,
+      "loss": 0.7091,
+      "step": 2660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.84170180042841e-05,
+      "loss": 0.8225,
+      "step": 2661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841582741362442e-05,
+      "loss": 0.9407,
+      "step": 2662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841463638754207e-05,
+      "loss": 0.9825,
+      "step": 2663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8413444926055286e-05,
+      "loss": 0.9835,
+      "step": 2664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841225302918235e-05,
+      "loss": 1.0968,
+      "step": 2665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.841106069694151e-05,
+      "loss": 0.9814,
+      "step": 2666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840986792935103e-05,
+      "loss": 0.9686,
+      "step": 2667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.84086747264292e-05,
+      "loss": 0.9755,
+      "step": 2668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8407481088194305e-05,
+      "loss": 0.8977,
+      "step": 2669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840628701466463e-05,
+      "loss": 1.0561,
+      "step": 2670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840509250585846e-05,
+      "loss": 0.7794,
+      "step": 2671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840389756179412e-05,
+      "loss": 0.833,
+      "step": 2672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.84027021824899e-05,
+      "loss": 1.2115,
+      "step": 2673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8401506367964146e-05,
+      "loss": 0.9222,
+      "step": 2674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.840031011823515e-05,
+      "loss": 0.6824,
+      "step": 2675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8399113433321256e-05,
+      "loss": 0.7521,
+      "step": 2676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.83979163132408e-05,
+      "loss": 0.8835,
+      "step": 2677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839671875801214e-05,
+      "loss": 0.8967,
+      "step": 2678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.83955207676536e-05,
+      "loss": 0.9244,
+      "step": 2679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839432234218356e-05,
+      "loss": 0.9237,
+      "step": 2680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.839312348162037e-05,
+      "loss": 0.9029,
+      "step": 2681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8391924185982406e-05,
+      "loss": 0.7644,
+      "step": 2682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8390724455288045e-05,
+      "loss": 1.0343,
+      "step": 2683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8389524289555665e-05,
+      "loss": 0.8741,
+      "step": 2684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8388323688803656e-05,
+      "loss": 0.9404,
+      "step": 2685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838712265305044e-05,
+      "loss": 0.726,
+      "step": 2686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838592118231438e-05,
+      "loss": 0.8075,
+      "step": 2687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838471927661392e-05,
+      "loss": 0.8919,
+      "step": 2688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.838351693596746e-05,
+      "loss": 0.9568,
+      "step": 2689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8382314160393424e-05,
+      "loss": 1.1428,
+      "step": 2690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8381110949910245e-05,
+      "loss": 0.8234,
+      "step": 2691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.837990730453637e-05,
+      "loss": 0.9621,
+      "step": 2692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.837870322429023e-05,
+      "loss": 0.7265,
+      "step": 2693
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8377498709190284e-05,
+      "loss": 0.8357,
+      "step": 2694
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8376293759254984e-05,
+      "loss": 0.9935,
+      "step": 2695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8375088374502794e-05,
+      "loss": 0.8898,
+      "step": 2696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8373882554952186e-05,
+      "loss": 0.8592,
+      "step": 2697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.8372676300621634e-05,
+      "loss": 0.8023,
+      "step": 2698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.837146961152962e-05,
+      "loss": 0.99,
+      "step": 2699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8370262487694644e-05,
+      "loss": 1.056,
+      "step": 2700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.83690549291352e-05,
+      "loss": 1.0143,
+      "step": 2701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.836784693586978e-05,
+      "loss": 1.0216,
+      "step": 2702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8366638507916905e-05,
+      "loss": 0.9825,
+      "step": 2703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.83654296452951e-05,
+      "loss": 0.7674,
+      "step": 2704
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.836422034802287e-05,
+      "loss": 0.7506,
+      "step": 2705
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.836301061611876e-05,
+      "loss": 0.8232,
+      "step": 2706
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8361800449601295e-05,
+      "loss": 0.9575,
+      "step": 2707
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.836058984848903e-05,
+      "loss": 0.952,
+      "step": 2708
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.835937881280051e-05,
+      "loss": 0.7523,
+      "step": 2709
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.835816734255429e-05,
+      "loss": 0.8493,
+      "step": 2710
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8356955437768935e-05,
+      "loss": 0.9073,
+      "step": 2711
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8355743098463025e-05,
+      "loss": 0.9371,
+      "step": 2712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.835453032465512e-05,
+      "loss": 0.8424,
+      "step": 2713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8353317116363817e-05,
+      "loss": 0.771,
+      "step": 2714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8352103473607706e-05,
+      "loss": 0.8239,
+      "step": 2715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8350889396405375e-05,
+      "loss": 0.7397,
+      "step": 2716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834967488477543e-05,
+      "loss": 0.9055,
+      "step": 2717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834845993873648e-05,
+      "loss": 0.7127,
+      "step": 2718
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834724455830714e-05,
+      "loss": 0.6087,
+      "step": 2719
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834602874350605e-05,
+      "loss": 0.8261,
+      "step": 2720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834481249435183e-05,
+      "loss": 0.8744,
+      "step": 2721
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834359581086311e-05,
+      "loss": 0.7797,
+      "step": 2722
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8342378693058545e-05,
+      "loss": 0.8412,
+      "step": 2723
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.834116114095677e-05,
+      "loss": 0.8774,
+      "step": 2724
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833994315457646e-05,
+      "loss": 0.8627,
+      "step": 2725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8338724733936264e-05,
+      "loss": 0.8983,
+      "step": 2726
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833750587905487e-05,
+      "loss": 0.7768,
+      "step": 2727
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833628658995093e-05,
+      "loss": 0.9208,
+      "step": 2728
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833506686664314e-05,
+      "loss": 0.8801,
+      "step": 2729
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.83338467091502e-05,
+      "loss": 0.8473,
+      "step": 2730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8332626117490784e-05,
+      "loss": 0.8373,
+      "step": 2731
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8331405091683614e-05,
+      "loss": 0.9429,
+      "step": 2732
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.833018363174741e-05,
+      "loss": 0.8188,
+      "step": 2733
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832896173770085e-05,
+      "loss": 0.8972,
+      "step": 2734
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832773940956269e-05,
+      "loss": 0.7996,
+      "step": 2735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832651664735166e-05,
+      "loss": 0.8705,
+      "step": 2736
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832529345108648e-05,
+      "loss": 0.9317,
+      "step": 2737
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.83240698207859e-05,
+      "loss": 0.866,
+      "step": 2738
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832284575646867e-05,
+      "loss": 0.8441,
+      "step": 2739
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.832162125815355e-05,
+      "loss": 1.0172,
+      "step": 2740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8320396325859295e-05,
+      "loss": 1.1381,
+      "step": 2741
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.831917095960468e-05,
+      "loss": 0.9441,
+      "step": 2742
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.831794515940849e-05,
+      "loss": 0.7327,
+      "step": 2743
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.83167189252895e-05,
+      "loss": 0.8327,
+      "step": 2744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8315492257266504e-05,
+      "loss": 0.9002,
+      "step": 2745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8314265155358286e-05,
+      "loss": 0.9393,
+      "step": 2746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.831303761958366e-05,
+      "loss": 0.8373,
+      "step": 2747
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8311809649961446e-05,
+      "loss": 0.7778,
+      "step": 2748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.831058124651043e-05,
+      "loss": 0.8918,
+      "step": 2749
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830935240924946e-05,
+      "loss": 0.851,
+      "step": 2750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830812313819736e-05,
+      "loss": 0.7348,
+      "step": 2751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830689343337297e-05,
+      "loss": 0.9065,
+      "step": 2752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8305663294795125e-05,
+      "loss": 0.9098,
+      "step": 2753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830443272248268e-05,
+      "loss": 0.8485,
+      "step": 2754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830320171645449e-05,
+      "loss": 0.9896,
+      "step": 2755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8301970276729414e-05,
+      "loss": 1.0409,
+      "step": 2756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.830073840332633e-05,
+      "loss": 0.9402,
+      "step": 2757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8299506096264105e-05,
+      "loss": 0.9495,
+      "step": 2758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.829827335556163e-05,
+      "loss": 0.8584,
+      "step": 2759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.82970401812378e-05,
+      "loss": 0.7561,
+      "step": 2760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8295806573311494e-05,
+      "loss": 0.7702,
+      "step": 2761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.8294572531801626e-05,
+      "loss": 0.8691,
+      "step": 2762
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.82933380567271e-05,
+      "loss": 0.9514,
+      "step": 2763
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.829210314810683e-05,
+      "loss": 0.9492,
+      "step": 2764
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.829086780595975e-05,
+      "loss": 0.8482,
+      "step": 2765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8289632030304785e-05,
+      "loss": 1.38,
+      "step": 2766
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.828839582116086e-05,
+      "loss": 0.759,
+      "step": 2767
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8287159178546943e-05,
+      "loss": 0.7925,
+      "step": 2768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.828592210248196e-05,
+      "loss": 1.0741,
+      "step": 2769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.828468459298488e-05,
+      "loss": 1.1599,
+      "step": 2770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.828344665007465e-05,
+      "loss": 1.0457,
+      "step": 2771
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.828220827377026e-05,
+      "loss": 0.7207,
+      "step": 2772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.828096946409066e-05,
+      "loss": 0.8282,
+      "step": 2773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.827973022105486e-05,
+      "loss": 1.1739,
+      "step": 2774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.827849054468184e-05,
+      "loss": 0.9244,
+      "step": 2775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.827725043499059e-05,
+      "loss": 0.7593,
+      "step": 2776
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.827600989200012e-05,
+      "loss": 0.9636,
+      "step": 2777
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8274768915729426e-05,
+      "loss": 0.9343,
+      "step": 2778
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.827352750619753e-05,
+      "loss": 0.8717,
+      "step": 2779
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8272285663423465e-05,
+      "loss": 1.0705,
+      "step": 2780
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.827104338742625e-05,
+      "loss": 0.7806,
+      "step": 2781
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826980067822493e-05,
+      "loss": 0.7877,
+      "step": 2782
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8268557535838527e-05,
+      "loss": 0.985,
+      "step": 2783
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826731396028612e-05,
+      "loss": 0.7551,
+      "step": 2784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8266069951586726e-05,
+      "loss": 0.806,
+      "step": 2785
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826482550975945e-05,
+      "loss": 0.7527,
+      "step": 2786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826358063482333e-05,
+      "loss": 0.8643,
+      "step": 2787
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.826233532679745e-05,
+      "loss": 0.7149,
+      "step": 2788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8261089585700894e-05,
+      "loss": 0.9954,
+      "step": 2789
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.825984341155275e-05,
+      "loss": 0.8063,
+      "step": 2790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.825859680437211e-05,
+      "loss": 0.7053,
+      "step": 2791
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.825734976417809e-05,
+      "loss": 0.8194,
+      "step": 2792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.825610229098979e-05,
+      "loss": 0.9131,
+      "step": 2793
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8254854384826324e-05,
+      "loss": 1.0719,
+      "step": 2794
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.82536060457068e-05,
+      "loss": 0.9988,
+      "step": 2795
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8252357273650366e-05,
+      "loss": 0.9416,
+      "step": 2796
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8251108068676155e-05,
+      "loss": 0.9572,
+      "step": 2797
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8249858430803306e-05,
+      "loss": 0.5827,
+      "step": 2798
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824860836005097e-05,
+      "loss": 0.7759,
+      "step": 2799
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8247357856438296e-05,
+      "loss": 0.7607,
+      "step": 2800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824610691998445e-05,
+      "loss": 0.8712,
+      "step": 2801
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824485555070859e-05,
+      "loss": 0.95,
+      "step": 2802
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824360374862991e-05,
+      "loss": 0.814,
+      "step": 2803
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824235151376758e-05,
+      "loss": 0.755,
+      "step": 2804
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.824109884614079e-05,
+      "loss": 0.5987,
+      "step": 2805
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823984574576874e-05,
+      "loss": 0.8547,
+      "step": 2806
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823859221267063e-05,
+      "loss": 0.9679,
+      "step": 2807
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8237338246865655e-05,
+      "loss": 0.9331,
+      "step": 2808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823608384837304e-05,
+      "loss": 0.779,
+      "step": 2809
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823482901721201e-05,
+      "loss": 0.7932,
+      "step": 2810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8233573753401794e-05,
+      "loss": 0.7173,
+      "step": 2811
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8232318056961615e-05,
+      "loss": 0.7261,
+      "step": 2812
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.823106192791072e-05,
+      "loss": 0.9513,
+      "step": 2813
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822980536626837e-05,
+      "loss": 0.7787,
+      "step": 2814
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8228548372053806e-05,
+      "loss": 0.7926,
+      "step": 2815
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822729094528628e-05,
+      "loss": 0.9069,
+      "step": 2816
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822603308598508e-05,
+      "loss": 1.0402,
+      "step": 2817
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8224774794169466e-05,
+      "loss": 1.0377,
+      "step": 2818
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822351606985873e-05,
+      "loss": 0.9752,
+      "step": 2819
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.822225691307215e-05,
+      "loss": 0.9062,
+      "step": 2820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.8220997323829026e-05,
+      "loss": 0.7865,
+      "step": 2821
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821973730214866e-05,
+      "loss": 0.9337,
+      "step": 2822
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821847684805036e-05,
+      "loss": 1.2445,
+      "step": 2823
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821721596155343e-05,
+      "loss": 0.9232,
+      "step": 2824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821595464267721e-05,
+      "loss": 0.9734,
+      "step": 2825
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.821469289144101e-05,
+      "loss": 0.951,
+      "step": 2826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.821343070786416e-05,
+      "loss": 0.9823,
+      "step": 2827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8212168091966023e-05,
+      "loss": 0.9633,
+      "step": 2828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8210905043765944e-05,
+      "loss": 0.8794,
+      "step": 2829
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.820964156328325e-05,
+      "loss": 1.2386,
+      "step": 2830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.820837765053732e-05,
+      "loss": 0.7865,
+      "step": 2831
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.820711330554753e-05,
+      "loss": 0.8183,
+      "step": 2832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8205848528333246e-05,
+      "loss": 0.9486,
+      "step": 2833
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8204583318913836e-05,
+      "loss": 1.0446,
+      "step": 2834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8203317677308706e-05,
+      "loss": 1.0434,
+      "step": 2835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.820205160353723e-05,
+      "loss": 0.9014,
+      "step": 2836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.820078509761883e-05,
+      "loss": 0.6215,
+      "step": 2837
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8199518159572896e-05,
+      "loss": 0.9617,
+      "step": 2838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.819825078941886e-05,
+      "loss": 0.904,
+      "step": 2839
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8196982987176116e-05,
+      "loss": 0.8824,
+      "step": 2840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8195714752864114e-05,
+      "loss": 0.8858,
+      "step": 2841
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8194446086502276e-05,
+      "loss": 0.9166,
+      "step": 2842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.819317698811005e-05,
+      "loss": 0.7808,
+      "step": 2843
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.819190745770687e-05,
+      "loss": 0.7849,
+      "step": 2844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8190637495312205e-05,
+      "loss": 0.7316,
+      "step": 2845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.81893671009455e-05,
+      "loss": 0.8739,
+      "step": 2846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818809627462623e-05,
+      "loss": 0.8246,
+      "step": 2847
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818682501637387e-05,
+      "loss": 0.754,
+      "step": 2848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.81855533262079e-05,
+      "loss": 1.032,
+      "step": 2849
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8184281204147794e-05,
+      "loss": 0.8286,
+      "step": 2850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818300865021306e-05,
+      "loss": 0.7428,
+      "step": 2851
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818173566442319e-05,
+      "loss": 0.94,
+      "step": 2852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.818046224679769e-05,
+      "loss": 0.8893,
+      "step": 2853
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817918839735608e-05,
+      "loss": 0.878,
+      "step": 2854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817791411611787e-05,
+      "loss": 1.0722,
+      "step": 2855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817663940310259e-05,
+      "loss": 0.7854,
+      "step": 2856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8175364258329775e-05,
+      "loss": 0.8122,
+      "step": 2857
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817408868181896e-05,
+      "loss": 0.8998,
+      "step": 2858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8172812673589696e-05,
+      "loss": 0.9026,
+      "step": 2859
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8171536233661535e-05,
+      "loss": 0.8689,
+      "step": 2860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817025936205404e-05,
+      "loss": 0.6992,
+      "step": 2861
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816898205878675e-05,
+      "loss": 0.8932,
+      "step": 2862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8167704323879276e-05,
+      "loss": 0.7985,
+      "step": 2863
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816642615735118e-05,
+      "loss": 0.7668,
+      "step": 2864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816514755922204e-05,
+      "loss": 0.8344,
+      "step": 2865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816386852951145e-05,
+      "loss": 0.7565,
+      "step": 2866
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816258906823903e-05,
+      "loss": 0.7241,
+      "step": 2867
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8161309175424365e-05,
+      "loss": 0.8619,
+      "step": 2868
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.816002885108707e-05,
+      "loss": 0.8691,
+      "step": 2869
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815874809524677e-05,
+      "loss": 0.9688,
+      "step": 2870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815746690792308e-05,
+      "loss": 0.7769,
+      "step": 2871
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8156185289135635e-05,
+      "loss": 0.9315,
+      "step": 2872
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8154903238904073e-05,
+      "loss": 0.7914,
+      "step": 2873
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815362075724806e-05,
+      "loss": 0.7869,
+      "step": 2874
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.815233784418722e-05,
+      "loss": 0.8652,
+      "step": 2875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8151054499741227e-05,
+      "loss": 0.8921,
+      "step": 2876
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814977072392973e-05,
+      "loss": 1.0745,
+      "step": 2877
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814848651677242e-05,
+      "loss": 0.93,
+      "step": 2878
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8147201878288965e-05,
+      "loss": 0.7591,
+      "step": 2879
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814591680849905e-05,
+      "loss": 0.8078,
+      "step": 2880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814463130742237e-05,
+      "loss": 0.7923,
+      "step": 2881
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814334537507863e-05,
+      "loss": 1.1142,
+      "step": 2882
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8142059011487516e-05,
+      "loss": 0.8914,
+      "step": 2883
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814077221666875e-05,
+      "loss": 0.8941,
+      "step": 2884
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813948499064205e-05,
+      "loss": 0.9648,
+      "step": 2885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813819733342713e-05,
+      "loss": 0.7897,
+      "step": 2886
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8136909245043745e-05,
+      "loss": 0.6895,
+      "step": 2887
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8135620725511605e-05,
+      "loss": 0.8899,
+      "step": 2888
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.813433177485048e-05,
+      "loss": 0.7159,
+      "step": 2889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.81330423930801e-05,
+      "loss": 1.1911,
+      "step": 2890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.813175258022024e-05,
+      "loss": 0.9766,
+      "step": 2891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8130462336290645e-05,
+      "loss": 0.9178,
+      "step": 2892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.81291716613111e-05,
+      "loss": 0.9796,
+      "step": 2893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812788055530137e-05,
+      "loss": 0.8741,
+      "step": 2894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812658901828126e-05,
+      "loss": 0.8536,
+      "step": 2895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812529705027054e-05,
+      "loss": 0.5934,
+      "step": 2896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812400465128901e-05,
+      "loss": 0.8574,
+      "step": 2897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8122711821356494e-05,
+      "loss": 0.7867,
+      "step": 2898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812141856049277e-05,
+      "loss": 0.6735,
+      "step": 2899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812012486871768e-05,
+      "loss": 0.8695,
+      "step": 2900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811883074605104e-05,
+      "loss": 0.9385,
+      "step": 2901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811753619251268e-05,
+      "loss": 0.7954,
+      "step": 2902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811624120812243e-05,
+      "loss": 0.7688,
+      "step": 2903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811494579290014e-05,
+      "loss": 0.8031,
+      "step": 2904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8113649946865666e-05,
+      "loss": 0.9752,
+      "step": 2905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811235367003886e-05,
+      "loss": 1.0325,
+      "step": 2906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811105696243957e-05,
+      "loss": 0.8664,
+      "step": 2907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810975982408769e-05,
+      "loss": 0.9182,
+      "step": 2908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810846225500309e-05,
+      "loss": 1.0024,
+      "step": 2909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810716425520563e-05,
+      "loss": 0.9013,
+      "step": 2910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8105865824715235e-05,
+      "loss": 1.0323,
+      "step": 2911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810456696355177e-05,
+      "loss": 0.9838,
+      "step": 2912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810326767173517e-05,
+      "loss": 0.8806,
+      "step": 2913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810196794928532e-05,
+      "loss": 0.8062,
+      "step": 2914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810066779622213e-05,
+      "loss": 0.7166,
+      "step": 2915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809936721256554e-05,
+      "loss": 1.1414,
+      "step": 2916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809806619833548e-05,
+      "loss": 0.8332,
+      "step": 2917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809676475355187e-05,
+      "loss": 0.9086,
+      "step": 2918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809546287823467e-05,
+      "loss": 0.7823,
+      "step": 2919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809416057240382e-05,
+      "loss": 1.015,
+      "step": 2920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8092857836079276e-05,
+      "loss": 0.72,
+      "step": 2921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8091554669280995e-05,
+      "loss": 0.8656,
+      "step": 2922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809025107202896e-05,
+      "loss": 1.0351,
+      "step": 2923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808894704434313e-05,
+      "loss": 0.7946,
+      "step": 2924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.80876425862435e-05,
+      "loss": 0.8206,
+      "step": 2925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808633769775006e-05,
+      "loss": 0.6566,
+      "step": 2926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808503237888278e-05,
+      "loss": 0.9285,
+      "step": 2927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.80837266296617e-05,
+      "loss": 0.8645,
+      "step": 2928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8082420450106796e-05,
+      "loss": 0.9212,
+      "step": 2929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8081113840238094e-05,
+      "loss": 0.8663,
+      "step": 2930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807980680007562e-05,
+      "loss": 0.8541,
+      "step": 2931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8078499329639406e-05,
+      "loss": 0.8476,
+      "step": 2932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8077191428949474e-05,
+      "loss": 0.9193,
+      "step": 2933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807588309802587e-05,
+      "loss": 0.9634,
+      "step": 2934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8074574336888635e-05,
+      "loss": 0.9221,
+      "step": 2935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8073265145557844e-05,
+      "loss": 0.7708,
+      "step": 2936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8071955524053544e-05,
+      "loss": 0.7453,
+      "step": 2937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807064547239579e-05,
+      "loss": 0.7953,
+      "step": 2938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806933499060468e-05,
+      "loss": 0.8423,
+      "step": 2939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806802407870029e-05,
+      "loss": 0.8177,
+      "step": 2940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.80667127367027e-05,
+      "loss": 0.8794,
+      "step": 2941
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8065400964632e-05,
+      "loss": 0.9048,
+      "step": 2942
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806408876250831e-05,
+      "loss": 0.986,
+      "step": 2943
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8062776130351706e-05,
+      "loss": 0.9206,
+      "step": 2944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806146306818233e-05,
+      "loss": 0.7681,
+      "step": 2945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8060149576020294e-05,
+      "loss": 0.73,
+      "step": 2946
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8058835653885715e-05,
+      "loss": 0.8575,
+      "step": 2947
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.805752130179874e-05,
+      "loss": 0.7599,
+      "step": 2948
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.805620651977951e-05,
+      "loss": 0.8516,
+      "step": 2949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8054891307848156e-05,
+      "loss": 0.7793,
+      "step": 2950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8053575666024847e-05,
+      "loss": 0.9028,
+      "step": 2951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8052259594329735e-05,
+      "loss": 0.9197,
+      "step": 2952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8050943092782994e-05,
+      "loss": 1.1583,
+      "step": 2953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8049626161404776e-05,
+      "loss": 0.818,
+      "step": 2954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8048308800215295e-05,
+      "loss": 1.0195,
+      "step": 2955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.80469910092347e-05,
+      "loss": 0.843,
+      "step": 2956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.804567278848321e-05,
+      "loss": 0.8036,
+      "step": 2957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.804435413798103e-05,
+      "loss": 0.8469,
+      "step": 2958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8043035057748344e-05,
+      "loss": 0.9036,
+      "step": 2959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.804171554780536e-05,
+      "loss": 0.7874,
+      "step": 2960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8040395608172326e-05,
+      "loss": 0.8331,
+      "step": 2961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.803907523886945e-05,
+      "loss": 0.7866,
+      "step": 2962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8037754439916965e-05,
+      "loss": 0.8261,
+      "step": 2963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.80364332113351e-05,
+      "loss": 0.923,
+      "step": 2964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8035111553144125e-05,
+      "loss": 1.0465,
+      "step": 2965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.803378946536428e-05,
+      "loss": 0.7553,
+      "step": 2966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.803246694801582e-05,
+      "loss": 0.9441,
+      "step": 2967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.803114400111901e-05,
+      "loss": 0.7897,
+      "step": 2968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802982062469413e-05,
+      "loss": 1.6712,
+      "step": 2969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8028496818761445e-05,
+      "loss": 0.927,
+      "step": 2970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802717258334126e-05,
+      "loss": 0.8839,
+      "step": 2971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.802584791845384e-05,
+      "loss": 0.9011,
+      "step": 2972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8024522824119513e-05,
+      "loss": 0.8628,
+      "step": 2973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8023197300358556e-05,
+      "loss": 0.9759,
+      "step": 2974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.80218713471913e-05,
+      "loss": 0.7584,
+      "step": 2975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8020544964638056e-05,
+      "loss": 0.7978,
+      "step": 2976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8019218152719144e-05,
+      "loss": 0.8548,
+      "step": 2977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8017890911454904e-05,
+      "loss": 0.7963,
+      "step": 2978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801656324086567e-05,
+      "loss": 0.7695,
+      "step": 2979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801523514097179e-05,
+      "loss": 0.9493,
+      "step": 2980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.80139066117936e-05,
+      "loss": 0.8616,
+      "step": 2981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801257765335147e-05,
+      "loss": 0.6793,
+      "step": 2982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801124826566576e-05,
+      "loss": 0.8978,
+      "step": 2983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800991844875685e-05,
+      "loss": 0.8008,
+      "step": 2984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.80085882026451e-05,
+      "loss": 0.7666,
+      "step": 2985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800725752735091e-05,
+      "loss": 0.9161,
+      "step": 2986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8005926422894654e-05,
+      "loss": 0.7801,
+      "step": 2987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8004594889296754e-05,
+      "loss": 0.9383,
+      "step": 2988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800326292657758e-05,
+      "loss": 0.742,
+      "step": 2989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800193053475757e-05,
+      "loss": 0.8787,
+      "step": 2990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800059771385712e-05,
+      "loss": 0.8189,
+      "step": 2991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799926446389667e-05,
+      "loss": 0.7069,
+      "step": 2992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799793078489664e-05,
+      "loss": 0.8128,
+      "step": 2993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7996596676877474e-05,
+      "loss": 0.9796,
+      "step": 2994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7995262139859606e-05,
+      "loss": 0.8158,
+      "step": 2995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7993927173863495e-05,
+      "loss": 0.8616,
+      "step": 2996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7992591778909584e-05,
+      "loss": 0.941,
+      "step": 2997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7991255955018344e-05,
+      "loss": 0.7846,
+      "step": 2998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7989919702210246e-05,
+      "loss": 0.8396,
+      "step": 2999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798858302050576e-05,
+      "loss": 0.8338,
+      "step": 3000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7987245909925377e-05,
+      "loss": 0.9602,
+      "step": 3001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798590837048958e-05,
+      "loss": 0.8334,
+      "step": 3002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798457040221885e-05,
+      "loss": 0.8598,
+      "step": 3003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.798323200513372e-05,
+      "loss": 0.7687,
+      "step": 3004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7981893179254685e-05,
+      "loss": 1.0628,
+      "step": 3005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7980553924602236e-05,
+      "loss": 0.831,
+      "step": 3006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7979214241196935e-05,
+      "loss": 0.7613,
+      "step": 3007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797787412905929e-05,
+      "loss": 0.8108,
+      "step": 3008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7976533588209826e-05,
+      "loss": 0.8182,
+      "step": 3009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79751926186691e-05,
+      "loss": 0.8358,
+      "step": 3010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797385122045766e-05,
+      "loss": 0.8087,
+      "step": 3011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797250939359605e-05,
+      "loss": 1.2682,
+      "step": 3012
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797116713810484e-05,
+      "loss": 0.8924,
+      "step": 3013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796982445400459e-05,
+      "loss": 0.9331,
+      "step": 3014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7968481341315884e-05,
+      "loss": 1.0659,
+      "step": 3015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79671378000593e-05,
+      "loss": 0.859,
+      "step": 3016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.796579383025543e-05,
+      "loss": 0.6982,
+      "step": 3017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7964449431924845e-05,
+      "loss": 0.8348,
+      "step": 3018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.796310460508817e-05,
+      "loss": 0.8624,
+      "step": 3019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7961759349766005e-05,
+      "loss": 0.7914,
+      "step": 3020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.796041366597896e-05,
+      "loss": 0.9678,
+      "step": 3021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.795906755374767e-05,
+      "loss": 0.8882,
+      "step": 3022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.795772101309274e-05,
+      "loss": 0.9225,
+      "step": 3023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.795637404403481e-05,
+      "loss": 1.0672,
+      "step": 3024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.795502664659453e-05,
+      "loss": 0.8103,
+      "step": 3025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.795367882079253e-05,
+      "loss": 1.016,
+      "step": 3026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.795233056664948e-05,
+      "loss": 0.959,
+      "step": 3027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.795098188418604e-05,
+      "loss": 0.9914,
+      "step": 3028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7949632773422856e-05,
+      "loss": 0.7706,
+      "step": 3029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794828323438062e-05,
+      "loss": 0.7726,
+      "step": 3030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794693326708e-05,
+      "loss": 0.9432,
+      "step": 3031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794558287154169e-05,
+      "loss": 0.795,
+      "step": 3032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7944232047786374e-05,
+      "loss": 0.7557,
+      "step": 3033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794288079583476e-05,
+      "loss": 0.7635,
+      "step": 3034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794152911570755e-05,
+      "loss": 0.8654,
+      "step": 3035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794017700742545e-05,
+      "loss": 0.9234,
+      "step": 3036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7938824471009185e-05,
+      "loss": 0.9012,
+      "step": 3037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793747150647949e-05,
+      "loss": 0.907,
+      "step": 3038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793611811385708e-05,
+      "loss": 0.9778,
+      "step": 3039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793476429316269e-05,
+      "loss": 0.8535,
+      "step": 3040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.79334100444171e-05,
+      "loss": 1.1206,
+      "step": 3041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7932055367641014e-05,
+      "loss": 0.8869,
+      "step": 3042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.793070026285522e-05,
+      "loss": 0.8001,
+      "step": 3043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7929344730080474e-05,
+      "loss": 0.8532,
+      "step": 3044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792798876933755e-05,
+      "loss": 0.9672,
+      "step": 3045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792663238064721e-05,
+      "loss": 0.9373,
+      "step": 3046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792527556403027e-05,
+      "loss": 0.7849,
+      "step": 3047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7923918319507494e-05,
+      "loss": 0.9443,
+      "step": 3048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792256064709969e-05,
+      "loss": 1.0035,
+      "step": 3049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792120254682766e-05,
+      "loss": 0.732,
+      "step": 3050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791984401871221e-05,
+      "loss": 0.6949,
+      "step": 3051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791848506277417e-05,
+      "loss": 1.0125,
+      "step": 3052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7917125679034353e-05,
+      "loss": 0.7484,
+      "step": 3053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791576586751359e-05,
+      "loss": 0.7173,
+      "step": 3054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791440562823271e-05,
+      "loss": 0.8829,
+      "step": 3055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791304496121258e-05,
+      "loss": 0.8886,
+      "step": 3056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791168386647404e-05,
+      "loss": 0.7099,
+      "step": 3057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.791032234403792e-05,
+      "loss": 0.9718,
+      "step": 3058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790896039392513e-05,
+      "loss": 0.8526,
+      "step": 3059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790759801615649e-05,
+      "loss": 0.9942,
+      "step": 3060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790623521075291e-05,
+      "loss": 1.0329,
+      "step": 3061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790487197773526e-05,
+      "loss": 0.893,
+      "step": 3062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790350831712445e-05,
+      "loss": 0.7748,
+      "step": 3063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790214422894133e-05,
+      "loss": 1.0146,
+      "step": 3064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7900779713206846e-05,
+      "loss": 1.0748,
+      "step": 3065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7899414769941885e-05,
+      "loss": 0.8665,
+      "step": 3066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7898049399167374e-05,
+      "loss": 0.8738,
+      "step": 3067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7896683600904225e-05,
+      "loss": 0.7436,
+      "step": 3068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789531737517337e-05,
+      "loss": 0.9537,
+      "step": 3069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789395072199574e-05,
+      "loss": 1.0001,
+      "step": 3070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789258364139229e-05,
+      "loss": 0.7709,
+      "step": 3071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789121613338396e-05,
+      "loss": 0.9203,
+      "step": 3072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7889848197991696e-05,
+      "loss": 0.7958,
+      "step": 3073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788847983523647e-05,
+      "loss": 0.7457,
+      "step": 3074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788711104513925e-05,
+      "loss": 0.8653,
+      "step": 3075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788574182772101e-05,
+      "loss": 0.8802,
+      "step": 3076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7884372183002716e-05,
+      "loss": 0.944,
+      "step": 3077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788300211100537e-05,
+      "loss": 0.7941,
+      "step": 3078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788163161174997e-05,
+      "loss": 0.9194,
+      "step": 3079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788026068525751e-05,
+      "loss": 0.9148,
+      "step": 3080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7878889331548986e-05,
+      "loss": 0.8089,
+      "step": 3081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7877517550645435e-05,
+      "loss": 1.0798,
+      "step": 3082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7876145342567855e-05,
+      "loss": 0.8604,
+      "step": 3083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7874772707337286e-05,
+      "loss": 0.7382,
+      "step": 3084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.787339964497475e-05,
+      "loss": 0.8876,
+      "step": 3085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7872026155501306e-05,
+      "loss": 0.7772,
+      "step": 3086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.787065223893799e-05,
+      "loss": 0.8813,
+      "step": 3087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.786927789530585e-05,
+      "loss": 0.7416,
+      "step": 3088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.786790312462594e-05,
+      "loss": 0.8042,
+      "step": 3089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.786652792691934e-05,
+      "loss": 0.9892,
+      "step": 3090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7865152302207124e-05,
+      "loss": 1.0201,
+      "step": 3091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7863776250510354e-05,
+      "loss": 0.7742,
+      "step": 3092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7862399771850134e-05,
+      "loss": 0.8317,
+      "step": 3093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.786102286624754e-05,
+      "loss": 0.9102,
+      "step": 3094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785964553372369e-05,
+      "loss": 1.1938,
+      "step": 3095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785826777429966e-05,
+      "loss": 0.9047,
+      "step": 3096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.78568895879966e-05,
+      "loss": 0.7838,
+      "step": 3097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785551097483559e-05,
+      "loss": 0.935,
+      "step": 3098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785413193483779e-05,
+      "loss": 1.0018,
+      "step": 3099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.78527524680243e-05,
+      "loss": 0.7317,
+      "step": 3100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.785137257441628e-05,
+      "loss": 0.8079,
+      "step": 3101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7849992254034866e-05,
+      "loss": 0.7856,
+      "step": 3102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784861150690119e-05,
+      "loss": 0.8693,
+      "step": 3103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7847230333036454e-05,
+      "loss": 0.825,
+      "step": 3104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7845848732461786e-05,
+      "loss": 0.938,
+      "step": 3105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7844466705198377e-05,
+      "loss": 0.9043,
+      "step": 3106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784308425126738e-05,
+      "loss": 0.7469,
+      "step": 3107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784170137069e-05,
+      "loss": 0.8424,
+      "step": 3108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.784031806348743e-05,
+      "loss": 0.8965,
+      "step": 3109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783893432968084e-05,
+      "loss": 0.7011,
+      "step": 3110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783755016929147e-05,
+      "loss": 0.8209,
+      "step": 3111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7836165582340496e-05,
+      "loss": 0.9023,
+      "step": 3112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783478056884914e-05,
+      "loss": 0.8944,
+      "step": 3113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783339512883866e-05,
+      "loss": 0.8123,
+      "step": 3114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783200926233024e-05,
+      "loss": 0.7892,
+      "step": 3115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.783062296934514e-05,
+      "loss": 0.9534,
+      "step": 3116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782923624990459e-05,
+      "loss": 0.78,
+      "step": 3117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782784910402986e-05,
+      "loss": 1.0248,
+      "step": 3118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782646153174219e-05,
+      "loss": 0.9757,
+      "step": 3119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782507353306285e-05,
+      "loss": 0.7871,
+      "step": 3120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782368510801309e-05,
+      "loss": 0.966,
+      "step": 3121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.78222962566142e-05,
+      "loss": 0.8724,
+      "step": 3122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.782090697888746e-05,
+      "loss": 0.7816,
+      "step": 3123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.781951727485416e-05,
+      "loss": 0.6571,
+      "step": 3124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7818127144535597e-05,
+      "loss": 0.8442,
+      "step": 3125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7816736587953063e-05,
+      "loss": 0.9274,
+      "step": 3126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.781534560512787e-05,
+      "loss": 0.735,
+      "step": 3127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7813954196081344e-05,
+      "loss": 0.9335,
+      "step": 3128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.781256236083478e-05,
+      "loss": 0.8102,
+      "step": 3129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7811170099409526e-05,
+      "loss": 0.6926,
+      "step": 3130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7809777411826925e-05,
+      "loss": 0.7972,
+      "step": 3131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780838429810829e-05,
+      "loss": 0.9333,
+      "step": 3132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780699075827498e-05,
+      "loss": 0.9069,
+      "step": 3133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780559679234835e-05,
+      "loss": 0.8229,
+      "step": 3134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780420240034977e-05,
+      "loss": 0.7892,
+      "step": 3135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7802807582300586e-05,
+      "loss": 0.9862,
+      "step": 3136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780141233822218e-05,
+      "loss": 0.9937,
+      "step": 3137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.780001666813594e-05,
+      "loss": 0.7911,
+      "step": 3138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7798620572063244e-05,
+      "loss": 0.7155,
+      "step": 3139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779722405002548e-05,
+      "loss": 0.8475,
+      "step": 3140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7795827102044065e-05,
+      "loss": 0.7171,
+      "step": 3141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7794429728140383e-05,
+      "loss": 0.775,
+      "step": 3142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.779303192833586e-05,
+      "loss": 1.1115,
+      "step": 3143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7791633702651913e-05,
+      "loss": 0.6798,
+      "step": 3144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7790235051109965e-05,
+      "loss": 0.7635,
+      "step": 3145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7788835973731445e-05,
+      "loss": 0.8583,
+      "step": 3146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.778743647053781e-05,
+      "loss": 0.9198,
+      "step": 3147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.778603654155047e-05,
+      "loss": 0.9261,
+      "step": 3148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7784636186790905e-05,
+      "loss": 0.7098,
+      "step": 3149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.778323540628056e-05,
+      "loss": 0.7735,
+      "step": 3150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.778183420004091e-05,
+      "loss": 0.9318,
+      "step": 3151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.778043256809342e-05,
+      "loss": 0.9259,
+      "step": 3152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.777903051045956e-05,
+      "loss": 0.7572,
+      "step": 3153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.777762802716083e-05,
+      "loss": 0.7302,
+      "step": 3154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.777622511821871e-05,
+      "loss": 0.7444,
+      "step": 3155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7774821783654695e-05,
+      "loss": 0.8031,
+      "step": 3156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.77734180234903e-05,
+      "loss": 1.0208,
+      "step": 3157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7772013837747024e-05,
+      "loss": 0.8885,
+      "step": 3158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.777060922644639e-05,
+      "loss": 0.8959,
+      "step": 3159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776920418960992e-05,
+      "loss": 0.9826,
+      "step": 3160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7767798727259135e-05,
+      "loss": 0.9568,
+      "step": 3161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776639283941558e-05,
+      "loss": 0.8754,
+      "step": 3162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.77649865261008e-05,
+      "loss": 0.682,
+      "step": 3163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7763579787336335e-05,
+      "loss": 0.8901,
+      "step": 3164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776217262314375e-05,
+      "loss": 0.6951,
+      "step": 3165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.776076503354461e-05,
+      "loss": 0.9004,
+      "step": 3166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7759357018560474e-05,
+      "loss": 0.886,
+      "step": 3167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.775794857821292e-05,
+      "loss": 0.9503,
+      "step": 3168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7756539712523534e-05,
+      "loss": 0.8349,
+      "step": 3169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.77551304215139e-05,
+      "loss": 0.84,
+      "step": 3170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.775372070520561e-05,
+      "loss": 0.7684,
+      "step": 3171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7752310563620284e-05,
+      "loss": 0.7384,
+      "step": 3172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7750899996779496e-05,
+      "loss": 0.9848,
+      "step": 3173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774948900470489e-05,
+      "loss": 0.7753,
+      "step": 3174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774807758741809e-05,
+      "loss": 0.7594,
+      "step": 3175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7746665744940694e-05,
+      "loss": 0.8416,
+      "step": 3176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774525347729436e-05,
+      "loss": 0.8191,
+      "step": 3177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774384078450071e-05,
+      "loss": 1.0057,
+      "step": 3178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.774242766658141e-05,
+      "loss": 0.7683,
+      "step": 3179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7741014123558105e-05,
+      "loss": 0.723,
+      "step": 3180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773960015545246e-05,
+      "loss": 0.803,
+      "step": 3181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7738185762286137e-05,
+      "loss": 0.7827,
+      "step": 3182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.77367709440808e-05,
+      "loss": 0.7553,
+      "step": 3183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773535570085815e-05,
+      "loss": 0.9225,
+      "step": 3184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7733940032639855e-05,
+      "loss": 0.8649,
+      "step": 3185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773252393944761e-05,
+      "loss": 0.821,
+      "step": 3186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.773110742130313e-05,
+      "loss": 0.9535,
+      "step": 3187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7729690478228104e-05,
+      "loss": 0.9309,
+      "step": 3188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7728273110244245e-05,
+      "loss": 0.8173,
+      "step": 3189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772685531737328e-05,
+      "loss": 0.7762,
+      "step": 3190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772543709963693e-05,
+      "loss": 0.8151,
+      "step": 3191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772401845705692e-05,
+      "loss": 0.7848,
+      "step": 3192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7722599389655e-05,
+      "loss": 0.9179,
+      "step": 3193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.772117989745291e-05,
+      "loss": 0.7909,
+      "step": 3194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.77197599804724e-05,
+      "loss": 0.8343,
+      "step": 3195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7718339638735226e-05,
+      "loss": 0.8821,
+      "step": 3196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.771691887226316e-05,
+      "loss": 0.6802,
+      "step": 3197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7715497681077954e-05,
+      "loss": 0.8979,
+      "step": 3198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.771407606520141e-05,
+      "loss": 0.8977,
+      "step": 3199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.77126540246553e-05,
+      "loss": 0.6405,
+      "step": 3200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7711231559461415e-05,
+      "loss": 0.8373,
+      "step": 3201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.7709808669641544e-05,
+      "loss": 0.9387,
+      "step": 3202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.77083853552175e-05,
+      "loss": 0.9061,
+      "step": 3203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.77069616162111e-05,
+      "loss": 1.0295,
+      "step": 3204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770553745264414e-05,
+      "loss": 0.7449,
+      "step": 3205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770411286453846e-05,
+      "loss": 0.9151,
+      "step": 3206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.770268785191588e-05,
+      "loss": 0.7929,
+      "step": 3207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.770126241479823e-05,
+      "loss": 0.9207,
+      "step": 3208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.769983655320738e-05,
+      "loss": 0.878,
+      "step": 3209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.769841026716515e-05,
+      "loss": 0.9182,
+      "step": 3210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76969835566934e-05,
+      "loss": 0.7948,
+      "step": 3211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.769555642181401e-05,
+      "loss": 0.7555,
+      "step": 3212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.769412886254882e-05,
+      "loss": 1.0779,
+      "step": 3213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.769270087891973e-05,
+      "loss": 1.0188,
+      "step": 3214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7691272470948616e-05,
+      "loss": 0.8532,
+      "step": 3215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.768984363865735e-05,
+      "loss": 0.8383,
+      "step": 3216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7688414382067844e-05,
+      "loss": 0.9099,
+      "step": 3217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.768698470120199e-05,
+      "loss": 1.1705,
+      "step": 3218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.768555459608169e-05,
+      "loss": 0.7161,
+      "step": 3219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.768412406672888e-05,
+      "loss": 0.728,
+      "step": 3220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.768269311316547e-05,
+      "loss": 0.841,
+      "step": 3221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7681261735413364e-05,
+      "loss": 0.9946,
+      "step": 3222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767982993349453e-05,
+      "loss": 1.0392,
+      "step": 3223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7678397707430884e-05,
+      "loss": 0.8078,
+      "step": 3224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767696505724439e-05,
+      "loss": 0.9884,
+      "step": 3225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767553198295698e-05,
+      "loss": 0.8376,
+      "step": 3226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767409848459063e-05,
+      "loss": 1.0456,
+      "step": 3227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76726645621673e-05,
+      "loss": 1.067,
+      "step": 3228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.767123021570896e-05,
+      "loss": 0.9058,
+      "step": 3229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76697954452376e-05,
+      "loss": 0.8194,
+      "step": 3230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.766836025077519e-05,
+      "loss": 1.0702,
+      "step": 3231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.766692463234373e-05,
+      "loss": 0.6222,
+      "step": 3232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7665488589965216e-05,
+      "loss": 0.7506,
+      "step": 3233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.766405212366166e-05,
+      "loss": 1.1405,
+      "step": 3234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.766261523345507e-05,
+      "loss": 0.8803,
+      "step": 3235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7661177919367453e-05,
+      "loss": 0.9205,
+      "step": 3236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765974018142083e-05,
+      "loss": 0.9248,
+      "step": 3237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765830201963727e-05,
+      "loss": 0.9296,
+      "step": 3238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765686343403877e-05,
+      "loss": 1.1017,
+      "step": 3239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765542442464739e-05,
+      "loss": 0.7672,
+      "step": 3240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765398499148518e-05,
+      "loss": 0.756,
+      "step": 3241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765254513457418e-05,
+      "loss": 0.877,
+      "step": 3242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.765110485393648e-05,
+      "loss": 0.96,
+      "step": 3243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7649664149594136e-05,
+      "loss": 0.8893,
+      "step": 3244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7648223021569224e-05,
+      "loss": 0.7901,
+      "step": 3245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7646781469883836e-05,
+      "loss": 0.8101,
+      "step": 3246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764533949456004e-05,
+      "loss": 0.85,
+      "step": 3247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764389709561996e-05,
+      "loss": 0.812,
+      "step": 3248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7642454273085675e-05,
+      "loss": 0.9814,
+      "step": 3249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.764101102697931e-05,
+      "loss": 0.8539,
+      "step": 3250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763956735732296e-05,
+      "loss": 0.8511,
+      "step": 3251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763812326413878e-05,
+      "loss": 0.9723,
+      "step": 3252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763667874744886e-05,
+      "loss": 1.0472,
+      "step": 3253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763523380727536e-05,
+      "loss": 0.7903,
+      "step": 3254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763378844364041e-05,
+      "loss": 1.0481,
+      "step": 3255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7632342656566166e-05,
+      "loss": 0.9124,
+      "step": 3256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.763089644607478e-05,
+      "loss": 0.9304,
+      "step": 3257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.762944981218839e-05,
+      "loss": 0.9843,
+      "step": 3258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7628002754929204e-05,
+      "loss": 0.8906,
+      "step": 3259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.762655527431937e-05,
+      "loss": 1.1359,
+      "step": 3260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.762510737038107e-05,
+      "loss": 1.1298,
+      "step": 3261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.762365904313649e-05,
+      "loss": 0.8493,
+      "step": 3262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.762221029260783e-05,
+      "loss": 0.8508,
+      "step": 3263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.762076111881729e-05,
+      "loss": 0.8173,
+      "step": 3264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.761931152178707e-05,
+      "loss": 0.8222,
+      "step": 3265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7617861501539385e-05,
+      "loss": 0.9608,
+      "step": 3266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7616411058096455e-05,
+      "loss": 0.7273,
+      "step": 3267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.76149601914805e-05,
+      "loss": 0.9783,
+      "step": 3268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.761350890171376e-05,
+      "loss": 1.0992,
+      "step": 3269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.761205718881846e-05,
+      "loss": 0.9846,
+      "step": 3270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.761060505281687e-05,
+      "loss": 0.8753,
+      "step": 3271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7609152493731215e-05,
+      "loss": 0.7244,
+      "step": 3272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.760769951158377e-05,
+      "loss": 0.7854,
+      "step": 3273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.760624610639678e-05,
+      "loss": 1.1023,
+      "step": 3274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.760479227819254e-05,
+      "loss": 0.7876,
+      "step": 3275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.760333802699332e-05,
+      "loss": 1.0016,
+      "step": 3276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.76018833528214e-05,
+      "loss": 0.8806,
+      "step": 3277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.760042825569906e-05,
+      "loss": 0.8346,
+      "step": 3278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7598972735648615e-05,
+      "loss": 0.8503,
+      "step": 3279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7597516792692364e-05,
+      "loss": 0.8378,
+      "step": 3280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.759606042685261e-05,
+      "loss": 0.8547,
+      "step": 3281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.759460363815167e-05,
+      "loss": 0.9074,
+      "step": 3282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.759314642661187e-05,
+      "loss": 0.9278,
+      "step": 3283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7591688792255543e-05,
+      "loss": 1.0365,
+      "step": 3284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.759023073510502e-05,
+      "loss": 1.0217,
+      "step": 3285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7588772255182645e-05,
+      "loss": 0.9282,
+      "step": 3286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758731335251076e-05,
+      "loss": 0.8024,
+      "step": 3287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758585402711173e-05,
+      "loss": 0.8699,
+      "step": 3288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758439427900791e-05,
+      "loss": 1.0281,
+      "step": 3289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7582934108221666e-05,
+      "loss": 0.9186,
+      "step": 3290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758147351477539e-05,
+      "loss": 0.9257,
+      "step": 3291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.758001249869143e-05,
+      "loss": 0.886,
+      "step": 3292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7578551059992203e-05,
+      "loss": 0.982,
+      "step": 3293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7577089198700104e-05,
+      "loss": 0.6894,
+      "step": 3294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7575626914837504e-05,
+      "loss": 0.8782,
+      "step": 3295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.757416420842684e-05,
+      "loss": 0.8551,
+      "step": 3296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7572701079490506e-05,
+      "loss": 1.0872,
+      "step": 3297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.757123752805094e-05,
+      "loss": 0.7165,
+      "step": 3298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7569773554130545e-05,
+      "loss": 1.0773,
+      "step": 3299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7568309157751776e-05,
+      "loss": 0.798,
+      "step": 3300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.756684433893706e-05,
+      "loss": 0.8414,
+      "step": 3301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7565379097708854e-05,
+      "loss": 0.9254,
+      "step": 3302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7563913434089595e-05,
+      "loss": 0.724,
+      "step": 3303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7562447348101754e-05,
+      "loss": 0.8654,
+      "step": 3304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.756098083976778e-05,
+      "loss": 0.8773,
+      "step": 3305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755951390911017e-05,
+      "loss": 0.8331,
+      "step": 3306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755804655615138e-05,
+      "loss": 1.0145,
+      "step": 3307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7556578780913905e-05,
+      "loss": 0.7573,
+      "step": 3308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755511058342023e-05,
+      "loss": 0.9433,
+      "step": 3309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.755364196369287e-05,
+      "loss": 0.9131,
+      "step": 3310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.75521729217543e-05,
+      "loss": 0.7741,
+      "step": 3311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7550703457627045e-05,
+      "loss": 1.0939,
+      "step": 3312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7549233571333635e-05,
+      "loss": 0.8469,
+      "step": 3313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.754776326289658e-05,
+      "loss": 0.7552,
+      "step": 3314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.75462925323384e-05,
+      "loss": 0.8368,
+      "step": 3315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.754482137968165e-05,
+      "loss": 0.6884,
+      "step": 3316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7543349804948856e-05,
+      "loss": 0.7987,
+      "step": 3317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.754187780816259e-05,
+      "loss": 0.8281,
+      "step": 3318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.754040538934539e-05,
+      "loss": 0.8777,
+      "step": 3319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.753893254851983e-05,
+      "loss": 0.8009,
+      "step": 3320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.753745928570845e-05,
+      "loss": 0.8184,
+      "step": 3321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7535985600933865e-05,
+      "loss": 0.8217,
+      "step": 3322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7534511494218625e-05,
+      "loss": 0.7694,
+      "step": 3323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.753303696558534e-05,
+      "loss": 0.6984,
+      "step": 3324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.75315620150566e-05,
+      "loss": 0.8589,
+      "step": 3325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.753008664265499e-05,
+      "loss": 1.0246,
+      "step": 3326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752861084840313e-05,
+      "loss": 0.7429,
+      "step": 3327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752713463232365e-05,
+      "loss": 0.7504,
+      "step": 3328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7525657994439136e-05,
+      "loss": 0.906,
+      "step": 3329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752418093477224e-05,
+      "loss": 0.9113,
+      "step": 3330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752270345334559e-05,
+      "loss": 0.8933,
+      "step": 3331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752122555018181e-05,
+      "loss": 0.7479,
+      "step": 3332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751974722530358e-05,
+      "loss": 0.9481,
+      "step": 3333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.751826847873352e-05,
+      "loss": 0.8658,
+      "step": 3334
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.751678931049431e-05,
+      "loss": 0.8725,
+      "step": 3335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7515309720608604e-05,
+      "loss": 0.9344,
+      "step": 3336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.751382970909907e-05,
+      "loss": 1.043,
+      "step": 3337
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.75123492759884e-05,
+      "loss": 0.7964,
+      "step": 3338
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.751086842129927e-05,
+      "loss": 1.0589,
+      "step": 3339
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.750938714505438e-05,
+      "loss": 0.7644,
+      "step": 3340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7507905447276425e-05,
+      "loss": 0.8111,
+      "step": 3341
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.75064233279881e-05,
+      "loss": 0.7552,
+      "step": 3342
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.750494078721213e-05,
+      "loss": 0.8253,
+      "step": 3343
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.750345782497122e-05,
+      "loss": 1.1123,
+      "step": 3344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.75019744412881e-05,
+      "loss": 1.1677,
+      "step": 3345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.75004906361855e-05,
+      "loss": 0.9168,
+      "step": 3346
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7499006409686156e-05,
+      "loss": 0.824,
+      "step": 3347
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7497521761812814e-05,
+      "loss": 0.8629,
+      "step": 3348
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.749603669258822e-05,
+      "loss": 0.9071,
+      "step": 3349
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.749455120203513e-05,
+      "loss": 0.7635,
+      "step": 3350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.74930652901763e-05,
+      "loss": 0.9986,
+      "step": 3351
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.749157895703452e-05,
+      "loss": 0.8913,
+      "step": 3352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.749009220263254e-05,
+      "loss": 0.9807,
+      "step": 3353
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7488605026993155e-05,
+      "loss": 0.9133,
+      "step": 3354
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748711743013917e-05,
+      "loss": 0.9263,
+      "step": 3355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748562941209334e-05,
+      "loss": 0.7619,
+      "step": 3356
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7484140972878506e-05,
+      "loss": 0.7327,
+      "step": 3357
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748265211251744e-05,
+      "loss": 0.8056,
+      "step": 3358
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.748116283103298e-05,
+      "loss": 0.7632,
+      "step": 3359
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.747967312844795e-05,
+      "loss": 0.9315,
+      "step": 3360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.747818300478517e-05,
+      "loss": 0.9811,
+      "step": 3361
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.747669246006745e-05,
+      "loss": 0.816,
+      "step": 3362
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7475201494317665e-05,
+      "loss": 1.0016,
+      "step": 3363
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7473710107558655e-05,
+      "loss": 0.8566,
+      "step": 3364
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.747221829981325e-05,
+      "loss": 0.792,
+      "step": 3365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.747072607110433e-05,
+      "loss": 0.8402,
+      "step": 3366
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.746923342145477e-05,
+      "loss": 0.7695,
+      "step": 3367
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.746774035088742e-05,
+      "loss": 0.665,
+      "step": 3368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7466246859425156e-05,
+      "loss": 0.9334,
+      "step": 3369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7464752947090884e-05,
+      "loss": 0.7347,
+      "step": 3370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7463258613907485e-05,
+      "loss": 0.7571,
+      "step": 3371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.746176385989785e-05,
+      "loss": 0.8816,
+      "step": 3372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7460268685084905e-05,
+      "loss": 0.892,
+      "step": 3373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745877308949153e-05,
+      "loss": 0.9219,
+      "step": 3374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745727707314067e-05,
+      "loss": 0.738,
+      "step": 3375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745578063605523e-05,
+      "loss": 0.7781,
+      "step": 3376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745428377825816e-05,
+      "loss": 0.8767,
+      "step": 3377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745278649977238e-05,
+      "loss": 0.7422,
+      "step": 3378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.745128880062084e-05,
+      "loss": 0.9563,
+      "step": 3379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.744979068082649e-05,
+      "loss": 0.7316,
+      "step": 3380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.744829214041228e-05,
+      "loss": 1.0626,
+      "step": 3381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.744679317940117e-05,
+      "loss": 0.7742,
+      "step": 3382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7445293797816136e-05,
+      "loss": 0.7436,
+      "step": 3383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.744379399568016e-05,
+      "loss": 0.9278,
+      "step": 3384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7442293773016225e-05,
+      "loss": 0.7741,
+      "step": 3385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.744079312984729e-05,
+      "loss": 1.0803,
+      "step": 3386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7439292066196386e-05,
+      "loss": 0.9049,
+      "step": 3387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7437790582086496e-05,
+      "loss": 0.653,
+      "step": 3388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743628867754063e-05,
+      "loss": 0.7259,
+      "step": 3389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7434786352581804e-05,
+      "loss": 0.8873,
+      "step": 3390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743328360723303e-05,
+      "loss": 0.8102,
+      "step": 3391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743178044151734e-05,
+      "loss": 0.7935,
+      "step": 3392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.743027685545778e-05,
+      "loss": 0.8646,
+      "step": 3393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.742877284907737e-05,
+      "loss": 1.0234,
+      "step": 3394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.7427268422399173e-05,
+      "loss": 0.8277,
+      "step": 3395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.742576357544623e-05,
+      "loss": 0.7306,
+      "step": 3396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.74242583082416e-05,
+      "loss": 0.8032,
+      "step": 3397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.742275262080836e-05,
+      "loss": 0.7809,
+      "step": 3398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.742124651316957e-05,
+      "loss": 0.724,
+      "step": 3399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.741973998534831e-05,
+      "loss": 0.8029,
+      "step": 3400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.741823303736769e-05,
+      "loss": 0.7207,
+      "step": 3401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.741672566925076e-05,
+      "loss": 0.7918,
+      "step": 3402
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.741521788102064e-05,
+      "loss": 0.9699,
+      "step": 3403
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7413709672700434e-05,
+      "loss": 0.7723,
+      "step": 3404
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.741220104431324e-05,
+      "loss": 0.8198,
+      "step": 3405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.741069199588219e-05,
+      "loss": 0.7474,
+      "step": 3406
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.740918252743041e-05,
+      "loss": 0.9119,
+      "step": 3407
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.740767263898102e-05,
+      "loss": 0.9542,
+      "step": 3408
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.740616233055715e-05,
+      "loss": 0.8444,
+      "step": 3409
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.740465160218197e-05,
+      "loss": 0.8428,
+      "step": 3410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7403140453878595e-05,
+      "loss": 0.8522,
+      "step": 3411
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.740162888567019e-05,
+      "loss": 0.906,
+      "step": 3412
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.740011689757995e-05,
+      "loss": 1.0557,
+      "step": 3413
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7398604489631e-05,
+      "loss": 1.0203,
+      "step": 3414
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739709166184653e-05,
+      "loss": 0.9766,
+      "step": 3415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739557841424973e-05,
+      "loss": 0.7574,
+      "step": 3416
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.739406474686379e-05,
+      "loss": 0.643,
+      "step": 3417
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7392550659711875e-05,
+      "loss": 0.8301,
+      "step": 3418
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7391036152817224e-05,
+      "loss": 0.7831,
+      "step": 3419
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738952122620302e-05,
+      "loss": 0.7274,
+      "step": 3420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7388005879892486e-05,
+      "loss": 0.9218,
+      "step": 3421
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738649011390884e-05,
+      "loss": 0.9007,
+      "step": 3422
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7384973928275306e-05,
+      "loss": 0.6938,
+      "step": 3423
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738345732301512e-05,
+      "loss": 0.9108,
+      "step": 3424
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.738194029815152e-05,
+      "loss": 0.8406,
+      "step": 3425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7380422853707754e-05,
+      "loss": 0.7734,
+      "step": 3426
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7378904989707067e-05,
+      "loss": 0.7478,
+      "step": 3427
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.737738670617273e-05,
+      "loss": 0.7898,
+      "step": 3428
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.737586800312799e-05,
+      "loss": 0.8973,
+      "step": 3429
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.737434888059614e-05,
+      "loss": 0.9726,
+      "step": 3430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7372829338600445e-05,
+      "loss": 0.7603,
+      "step": 3431
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.737130937716418e-05,
+      "loss": 0.9601,
+      "step": 3432
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736978899631066e-05,
+      "loss": 1.0828,
+      "step": 3433
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736826819606317e-05,
+      "loss": 0.855,
+      "step": 3434
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7366746976445006e-05,
+      "loss": 0.6756,
+      "step": 3435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736522533747949e-05,
+      "loss": 0.9506,
+      "step": 3436
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736370327918993e-05,
+      "loss": 1.0216,
+      "step": 3437
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736218080159965e-05,
+      "loss": 0.748,
+      "step": 3438
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.736065790473199e-05,
+      "loss": 0.8162,
+      "step": 3439
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.735913458861027e-05,
+      "loss": 0.9372,
+      "step": 3440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.735761085325784e-05,
+      "loss": 0.7687,
+      "step": 3441
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.735608669869805e-05,
+      "loss": 0.8981,
+      "step": 3442
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.735456212495425e-05,
+      "loss": 0.7922,
+      "step": 3443
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.735303713204981e-05,
+      "loss": 0.9608,
+      "step": 3444
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.73515117200081e-05,
+      "loss": 0.8875,
+      "step": 3445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7349985888852465e-05,
+      "loss": 0.6312,
+      "step": 3446
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.734845963860632e-05,
+      "loss": 0.9071,
+      "step": 3447
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.734693296929304e-05,
+      "loss": 0.7376,
+      "step": 3448
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7345405880936005e-05,
+      "loss": 1.0976,
+      "step": 3449
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7343878373558645e-05,
+      "loss": 0.8503,
+      "step": 3450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.734235044718434e-05,
+      "loss": 0.9299,
+      "step": 3451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.734082210183651e-05,
+      "loss": 0.8395,
+      "step": 3452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733929333753858e-05,
+      "loss": 0.7955,
+      "step": 3453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7337764154313973e-05,
+      "loss": 0.9611,
+      "step": 3454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733623455218611e-05,
+      "loss": 1.0845,
+      "step": 3455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733470453117845e-05,
+      "loss": 0.8908,
+      "step": 3456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7333174091314425e-05,
+      "loss": 1.0487,
+      "step": 3457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.733164323261749e-05,
+      "loss": 0.9639,
+      "step": 3458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7330111955111104e-05,
+      "loss": 0.8574,
+      "step": 3459
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.732858025881872e-05,
+      "loss": 0.762,
+      "step": 3460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.7327048143763826e-05,
+      "loss": 0.777,
+      "step": 3461
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7325515609969885e-05,
+      "loss": 0.6247,
+      "step": 3462
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7323982657460387e-05,
+      "loss": 0.8719,
+      "step": 3463
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7322449286258816e-05,
+      "loss": 0.8745,
+      "step": 3464
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.732091549638867e-05,
+      "loss": 0.9881,
+      "step": 3465
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.731938128787346e-05,
+      "loss": 0.9245,
+      "step": 3466
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7317846660736685e-05,
+      "loss": 0.838,
+      "step": 3467
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.731631161500187e-05,
+      "loss": 0.8639,
+      "step": 3468
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.731477615069252e-05,
+      "loss": 0.9705,
+      "step": 3469
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7313240267832185e-05,
+      "loss": 0.7457,
+      "step": 3470
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.731170396644439e-05,
+      "loss": 0.8175,
+      "step": 3471
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7310167246552665e-05,
+      "loss": 0.9398,
+      "step": 3472
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7308630108180586e-05,
+      "loss": 0.944,
+      "step": 3473
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.730709255135167e-05,
+      "loss": 0.9212,
+      "step": 3474
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.73055545760895e-05,
+      "loss": 0.8672,
+      "step": 3475
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.730401618241765e-05,
+      "loss": 0.8692,
+      "step": 3476
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7302477370359674e-05,
+      "loss": 0.8906,
+      "step": 3477
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.730093813993916e-05,
+      "loss": 0.8189,
+      "step": 3478
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729939849117969e-05,
+      "loss": 0.89,
+      "step": 3479
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7297858424104874e-05,
+      "loss": 0.7572,
+      "step": 3480
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729631793873829e-05,
+      "loss": 0.9342,
+      "step": 3481
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729477703510355e-05,
+      "loss": 1.1163,
+      "step": 3482
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729323571322426e-05,
+      "loss": 0.7844,
+      "step": 3483
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729169397312405e-05,
+      "loss": 0.7418,
+      "step": 3484
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.729015181482653e-05,
+      "loss": 0.889,
+      "step": 3485
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.728860923835535e-05,
+      "loss": 0.8103,
+      "step": 3486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.728706624373413e-05,
+      "loss": 1.2371,
+      "step": 3487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.728552283098653e-05,
+      "loss": 1.1201,
+      "step": 3488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.728397900013619e-05,
+      "loss": 0.8554,
+      "step": 3489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7282434751206747e-05,
+      "loss": 0.9282,
+      "step": 3490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7280890084221904e-05,
+      "loss": 0.7663,
+      "step": 3491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7279344999205295e-05,
+      "loss": 0.7506,
+      "step": 3492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7277799496180624e-05,
+      "loss": 0.8441,
+      "step": 3493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7276253575171554e-05,
+      "loss": 0.6451,
+      "step": 3494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7274707236201786e-05,
+      "loss": 0.7823,
+      "step": 3495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7273160479295e-05,
+      "loss": 0.916,
+      "step": 3496
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727161330447491e-05,
+      "loss": 0.7879,
+      "step": 3497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.727006571176522e-05,
+      "loss": 0.7913,
+      "step": 3498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.726851770118964e-05,
+      "loss": 0.8879,
+      "step": 3499
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.726696927277189e-05,
+      "loss": 0.7664,
+      "step": 3500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.726542042653571e-05,
+      "loss": 0.9078,
+      "step": 3501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.726387116250482e-05,
+      "loss": 0.7161,
+      "step": 3502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.726232148070297e-05,
+      "loss": 0.9574,
+      "step": 3503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7260771381153896e-05,
+      "loss": 0.9959,
+      "step": 3504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.725922086388135e-05,
+      "loss": 1.1417,
+      "step": 3505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7257669928909104e-05,
+      "loss": 0.8102,
+      "step": 3506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7256118576260913e-05,
+      "loss": 0.9233,
+      "step": 3507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.725456680596055e-05,
+      "loss": 0.8761,
+      "step": 3508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.725301461803179e-05,
+      "loss": 0.8901,
+      "step": 3509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.725146201249843e-05,
+      "loss": 0.7932,
+      "step": 3510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7249908989384245e-05,
+      "loss": 0.779,
+      "step": 3511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724835554871304e-05,
+      "loss": 0.7341,
+      "step": 3512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724680169050863e-05,
+      "loss": 0.9169,
+      "step": 3513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7245247414794806e-05,
+      "loss": 0.9437,
+      "step": 3514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724369272159539e-05,
+      "loss": 1.0744,
+      "step": 3515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724213761093421e-05,
+      "loss": 1.1385,
+      "step": 3516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.724058208283509e-05,
+      "loss": 0.899,
+      "step": 3517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723902613732187e-05,
+      "loss": 0.7967,
+      "step": 3518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7237469774418396e-05,
+      "loss": 0.89,
+      "step": 3519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.72359129941485e-05,
+      "loss": 0.8412,
+      "step": 3520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723435579653606e-05,
+      "loss": 0.9009,
+      "step": 3521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7232798181604914e-05,
+      "loss": 0.8379,
+      "step": 3522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.723124014937895e-05,
+      "loss": 0.7938,
+      "step": 3523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.722968169988203e-05,
+      "loss": 0.8175,
+      "step": 3524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.7228122833138045e-05,
+      "loss": 0.7844,
+      "step": 3525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.722656354917086e-05,
+      "loss": 0.9855,
+      "step": 3526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.722500384800439e-05,
+      "loss": 0.7358,
+      "step": 3527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.722344372966253e-05,
+      "loss": 0.7555,
+      "step": 3528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.722188319416918e-05,
+      "loss": 1.0051,
+      "step": 3529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.722032224154826e-05,
+      "loss": 0.8693,
+      "step": 3530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.721876087182368e-05,
+      "loss": 0.6871,
+      "step": 3531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.721719908501937e-05,
+      "loss": 0.8222,
+      "step": 3532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7215636881159275e-05,
+      "loss": 0.833,
+      "step": 3533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7214074260267306e-05,
+      "loss": 0.7451,
+      "step": 3534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7212511222367424e-05,
+      "loss": 0.8314,
+      "step": 3535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.721094776748358e-05,
+      "loss": 0.7031,
+      "step": 3536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7209383895639725e-05,
+      "loss": 1.1049,
+      "step": 3537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.720781960685983e-05,
+      "loss": 0.9183,
+      "step": 3538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7206254901167855e-05,
+      "loss": 0.7733,
+      "step": 3539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.720468977858779e-05,
+      "loss": 0.8088,
+      "step": 3540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.72031242391436e-05,
+      "loss": 0.8849,
+      "step": 3541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.720155828285928e-05,
+      "loss": 1.062,
+      "step": 3542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7199991909758844e-05,
+      "loss": 0.7298,
+      "step": 3543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.719842511986628e-05,
+      "loss": 0.8731,
+      "step": 3544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7196857913205584e-05,
+      "loss": 0.8323,
+      "step": 3545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7195290289800793e-05,
+      "loss": 0.8088,
+      "step": 3546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.719372224967592e-05,
+      "loss": 0.9214,
+      "step": 3547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.719215379285498e-05,
+      "loss": 0.9159,
+      "step": 3548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.719058491936202e-05,
+      "loss": 0.7797,
+      "step": 3549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.718901562922108e-05,
+      "loss": 0.9196,
+      "step": 3550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71874459224562e-05,
+      "loss": 0.9414,
+      "step": 3551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.718587579909144e-05,
+      "loss": 0.92,
+      "step": 3552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7184305259150856e-05,
+      "loss": 0.7306,
+      "step": 3553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.718273430265851e-05,
+      "loss": 0.9383,
+      "step": 3554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7181162929638484e-05,
+      "loss": 0.6836,
+      "step": 3555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.717959114011485e-05,
+      "loss": 0.8175,
+      "step": 3556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.717801893411169e-05,
+      "loss": 0.9896,
+      "step": 3557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7176446311653104e-05,
+      "loss": 0.9442,
+      "step": 3558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7174873272763176e-05,
+      "loss": 0.9052,
+      "step": 3559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.717329981746603e-05,
+      "loss": 0.7436,
+      "step": 3560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.717172594578575e-05,
+      "loss": 0.7093,
+      "step": 3561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7170151657746484e-05,
+      "loss": 0.8999,
+      "step": 3562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716857695337233e-05,
+      "loss": 0.9106,
+      "step": 3563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716700183268743e-05,
+      "loss": 0.7956,
+      "step": 3564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7165426295715917e-05,
+      "loss": 0.7175,
+      "step": 3565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716385034248192e-05,
+      "loss": 0.8505,
+      "step": 3566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716227397300962e-05,
+      "loss": 0.8807,
+      "step": 3567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.716069718732315e-05,
+      "loss": 0.7993,
+      "step": 3568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715911998544666e-05,
+      "loss": 0.8267,
+      "step": 3569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715754236740434e-05,
+      "loss": 0.8602,
+      "step": 3570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715596433322036e-05,
+      "loss": 0.7566,
+      "step": 3571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7154385882918896e-05,
+      "loss": 0.9616,
+      "step": 3572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7152807016524134e-05,
+      "loss": 0.8889,
+      "step": 3573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.715122773406026e-05,
+      "loss": 0.8673,
+      "step": 3574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.71496480355515e-05,
+      "loss": 0.9267,
+      "step": 3575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7148067921022026e-05,
+      "loss": 0.937,
+      "step": 3576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7146487390496084e-05,
+      "loss": 0.8741,
+      "step": 3577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.714490644399787e-05,
+      "loss": 0.847,
+      "step": 3578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7143325081551614e-05,
+      "loss": 1.0489,
+      "step": 3579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.714174330318155e-05,
+      "loss": 0.8454,
+      "step": 3580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7140161108911915e-05,
+      "loss": 0.7527,
+      "step": 3581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.713857849876695e-05,
+      "loss": 0.8807,
+      "step": 3582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7136995472770916e-05,
+      "loss": 0.8266,
+      "step": 3583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7135412030948064e-05,
+      "loss": 0.8681,
+      "step": 3584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.7133828173322655e-05,
+      "loss": 0.8284,
+      "step": 3585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.713224389991895e-05,
+      "loss": 0.9658,
+      "step": 3586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.713065921076126e-05,
+      "loss": 0.9168,
+      "step": 3587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.712907410587382e-05,
+      "loss": 0.9173,
+      "step": 3588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7127488585280954e-05,
+      "loss": 0.8712,
+      "step": 3589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7125902649006944e-05,
+      "loss": 0.9122,
+      "step": 3590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7124316297076086e-05,
+      "loss": 0.939,
+      "step": 3591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.712272952951271e-05,
+      "loss": 0.838,
+      "step": 3592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7121142346341106e-05,
+      "loss": 0.6489,
+      "step": 3593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.711955474758561e-05,
+      "loss": 1.0083,
+      "step": 3594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.711796673327053e-05,
+      "loss": 0.7653,
+      "step": 3595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7116378303420226e-05,
+      "loss": 0.8627,
+      "step": 3596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7114789458059034e-05,
+      "loss": 0.8693,
+      "step": 3597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.711320019721127e-05,
+      "loss": 1.0112,
+      "step": 3598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.711161052090132e-05,
+      "loss": 0.7851,
+      "step": 3599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7110020429153534e-05,
+      "loss": 0.942,
+      "step": 3600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7108429921992276e-05,
+      "loss": 0.6923,
+      "step": 3601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.710683899944191e-05,
+      "loss": 0.9518,
+      "step": 3602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7105247661526815e-05,
+      "loss": 0.826,
+      "step": 3603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7103655908271393e-05,
+      "loss": 0.8471,
+      "step": 3604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.710206373970001e-05,
+      "loss": 0.9205,
+      "step": 3605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7100471155837095e-05,
+      "loss": 0.8239,
+      "step": 3606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.709887815670702e-05,
+      "loss": 0.8911,
+      "step": 3607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.70972847423342e-05,
+      "loss": 0.9126,
+      "step": 3608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.709569091274307e-05,
+      "loss": 0.8138,
+      "step": 3609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.709409666795804e-05,
+      "loss": 0.7823,
+      "step": 3610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7092502008003534e-05,
+      "loss": 0.6681,
+      "step": 3611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7090906932904005e-05,
+      "loss": 0.6889,
+      "step": 3612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.708931144268387e-05,
+      "loss": 0.9562,
+      "step": 3613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.70877155373676e-05,
+      "loss": 0.7421,
+      "step": 3614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.708611921697964e-05,
+      "loss": 0.7932,
+      "step": 3615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.708452248154444e-05,
+      "loss": 0.7188,
+      "step": 3616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7082925331086494e-05,
+      "loss": 0.9117,
+      "step": 3617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.708132776563025e-05,
+      "loss": 0.78,
+      "step": 3618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7079729785200194e-05,
+      "loss": 0.8944,
+      "step": 3619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707813138982082e-05,
+      "loss": 1.1354,
+      "step": 3620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7076532579516614e-05,
+      "loss": 0.7223,
+      "step": 3621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707493335431208e-05,
+      "loss": 0.7371,
+      "step": 3622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707333371423172e-05,
+      "loss": 0.8691,
+      "step": 3623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707173365930004e-05,
+      "loss": 0.8643,
+      "step": 3624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.707013318954157e-05,
+      "loss": 0.8276,
+      "step": 3625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706853230498083e-05,
+      "loss": 0.7929,
+      "step": 3626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706693100564235e-05,
+      "loss": 0.8285,
+      "step": 3627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706532929155067e-05,
+      "loss": 0.8917,
+      "step": 3628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7063727162730324e-05,
+      "loss": 0.85,
+      "step": 3629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706212461920587e-05,
+      "loss": 0.8016,
+      "step": 3630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.706052166100186e-05,
+      "loss": 0.753,
+      "step": 3631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7058918288142854e-05,
+      "loss": 0.8771,
+      "step": 3632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7057314500653425e-05,
+      "loss": 0.88,
+      "step": 3633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.705571029855815e-05,
+      "loss": 0.8629,
+      "step": 3634
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7054105681881614e-05,
+      "loss": 0.8681,
+      "step": 3635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7052500650648395e-05,
+      "loss": 0.9756,
+      "step": 3636
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.705089520488309e-05,
+      "loss": 0.7511,
+      "step": 3637
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.704928934461031e-05,
+      "loss": 0.7585,
+      "step": 3638
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.704768306985465e-05,
+      "loss": 0.7951,
+      "step": 3639
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7046076380640727e-05,
+      "loss": 1.0109,
+      "step": 3640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7044469276993165e-05,
+      "loss": 0.912,
+      "step": 3641
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7042861758936576e-05,
+      "loss": 0.7612,
+      "step": 3642
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.704125382649561e-05,
+      "loss": 0.8732,
+      "step": 3643
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.7039645479694895e-05,
+      "loss": 0.7645,
+      "step": 3644
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703803671855908e-05,
+      "loss": 0.8527,
+      "step": 3645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703642754311282e-05,
+      "loss": 0.8125,
+      "step": 3646
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703481795338077e-05,
+      "loss": 0.9808,
+      "step": 3647
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703320794938758e-05,
+      "loss": 1.0778,
+      "step": 3648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.703159753115794e-05,
+      "loss": 0.9586,
+      "step": 3649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.702998669871652e-05,
+      "loss": 0.7123,
+      "step": 3650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.702837545208801e-05,
+      "loss": 1.0529,
+      "step": 3651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.702676379129709e-05,
+      "loss": 0.9432,
+      "step": 3652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.702515171636844e-05,
+      "loss": 0.7419,
+      "step": 3653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.7023539227326805e-05,
+      "loss": 1.0096,
+      "step": 3654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.702192632419686e-05,
+      "loss": 0.8256,
+      "step": 3655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.7020313007003325e-05,
+      "loss": 0.9308,
+      "step": 3656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.701869927577092e-05,
+      "loss": 0.9622,
+      "step": 3657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.7017085130524395e-05,
+      "loss": 1.0758,
+      "step": 3658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.701547057128846e-05,
+      "loss": 0.8885,
+      "step": 3659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.701385559808786e-05,
+      "loss": 0.8894,
+      "step": 3660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.701224021094735e-05,
+      "loss": 0.7462,
+      "step": 3661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.701062440989166e-05,
+      "loss": 0.8707,
+      "step": 3662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.700900819494558e-05,
+      "loss": 0.8916,
+      "step": 3663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.7007391566133865e-05,
+      "loss": 0.9413,
+      "step": 3664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.7005774523481274e-05,
+      "loss": 0.7433,
+      "step": 3665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.700415706701259e-05,
+      "loss": 0.837,
+      "step": 3666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.7002539196752616e-05,
+      "loss": 1.0277,
+      "step": 3667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.700092091272613e-05,
+      "loss": 0.7544,
+      "step": 3668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.699930221495793e-05,
+      "loss": 0.7582,
+      "step": 3669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6997683103472804e-05,
+      "loss": 0.8907,
+      "step": 3670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.699606357829559e-05,
+      "loss": 1.0788,
+      "step": 3671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.699444363945109e-05,
+      "loss": 0.7802,
+      "step": 3672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.699282328696412e-05,
+      "loss": 0.8326,
+      "step": 3673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.699120252085953e-05,
+      "loss": 0.8323,
+      "step": 3674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.698958134116214e-05,
+      "loss": 0.9244,
+      "step": 3675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6987959747896785e-05,
+      "loss": 0.7619,
+      "step": 3676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6986337741088324e-05,
+      "loss": 0.9446,
+      "step": 3677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.698471532076161e-05,
+      "loss": 0.7955,
+      "step": 3678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.698309248694151e-05,
+      "loss": 0.7897,
+      "step": 3679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.698146923965288e-05,
+      "loss": 0.6905,
+      "step": 3680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6979845578920595e-05,
+      "loss": 0.7873,
+      "step": 3681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.697822150476954e-05,
+      "loss": 0.7391,
+      "step": 3682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.69765970172246e-05,
+      "loss": 0.7727,
+      "step": 3683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6974972116310674e-05,
+      "loss": 0.7613,
+      "step": 3684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.697334680205264e-05,
+      "loss": 0.8733,
+      "step": 3685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6971721074475426e-05,
+      "loss": 0.7133,
+      "step": 3686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6970094933603925e-05,
+      "loss": 0.8079,
+      "step": 3687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696846837946307e-05,
+      "loss": 0.7466,
+      "step": 3688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696684141207777e-05,
+      "loss": 0.7914,
+      "step": 3689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6965214031472966e-05,
+      "loss": 0.8169,
+      "step": 3690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.69635862376736e-05,
+      "loss": 0.9603,
+      "step": 3691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6961958030704605e-05,
+      "loss": 0.9453,
+      "step": 3692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.696032941059093e-05,
+      "loss": 0.8067,
+      "step": 3693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6958700377357536e-05,
+      "loss": 0.7865,
+      "step": 3694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695707093102938e-05,
+      "loss": 0.7548,
+      "step": 3695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695544107163143e-05,
+      "loss": 0.8368,
+      "step": 3696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6953810799188675e-05,
+      "loss": 0.8402,
+      "step": 3697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695218011372607e-05,
+      "loss": 0.8165,
+      "step": 3698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.695054901526863e-05,
+      "loss": 0.7267,
+      "step": 3699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.694891750384133e-05,
+      "loss": 0.8898,
+      "step": 3700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.694728557946918e-05,
+      "loss": 0.9361,
+      "step": 3701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.694565324217717e-05,
+      "loss": 0.8752,
+      "step": 3702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.694402049199035e-05,
+      "loss": 0.7538,
+      "step": 3703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6942387328933695e-05,
+      "loss": 0.9575,
+      "step": 3704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6940753753032246e-05,
+      "loss": 0.8471,
+      "step": 3705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6939119764311056e-05,
+      "loss": 0.8788,
+      "step": 3706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6937485362795135e-05,
+      "loss": 0.7983,
+      "step": 3707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6935850548509536e-05,
+      "loss": 0.7583,
+      "step": 3708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6934215321479316e-05,
+      "loss": 0.8503,
+      "step": 3709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.693257968172952e-05,
+      "loss": 0.7157,
+      "step": 3710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6930943629285224e-05,
+      "loss": 0.9318,
+      "step": 3711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6929307164171495e-05,
+      "loss": 0.9299,
+      "step": 3712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.69276702864134e-05,
+      "loss": 0.8713,
+      "step": 3713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.692603299603603e-05,
+      "loss": 0.8826,
+      "step": 3714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.6924395293064475e-05,
+      "loss": 0.9231,
+      "step": 3715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.692275717752382e-05,
+      "loss": 0.7256,
+      "step": 3716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6921118649439184e-05,
+      "loss": 0.8401,
+      "step": 3717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.691947970883565e-05,
+      "loss": 0.8512,
+      "step": 3718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.691784035573834e-05,
+      "loss": 1.0566,
+      "step": 3719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.691620059017239e-05,
+      "loss": 0.7834,
+      "step": 3720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.691456041216291e-05,
+      "loss": 0.8115,
+      "step": 3721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.691291982173504e-05,
+      "loss": 0.7448,
+      "step": 3722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.691127881891392e-05,
+      "loss": 0.9824,
+      "step": 3723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.69096374037247e-05,
+      "loss": 0.826,
+      "step": 3724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6907995576192507e-05,
+      "loss": 1.0273,
+      "step": 3725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6906353336342525e-05,
+      "loss": 0.8798,
+      "step": 3726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6904710684199914e-05,
+      "loss": 0.9045,
+      "step": 3727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6903067619789834e-05,
+      "loss": 0.7972,
+      "step": 3728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.690142414313748e-05,
+      "loss": 0.8377,
+      "step": 3729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6899780254268014e-05,
+      "loss": 0.821,
+      "step": 3730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6898135953206636e-05,
+      "loss": 0.9496,
+      "step": 3731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6896491239978544e-05,
+      "loss": 0.8124,
+      "step": 3732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.689484611460895e-05,
+      "loss": 0.8097,
+      "step": 3733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.689320057712303e-05,
+      "loss": 0.8292,
+      "step": 3734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.689155462754604e-05,
+      "loss": 0.922,
+      "step": 3735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.688990826590317e-05,
+      "loss": 0.9261,
+      "step": 3736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6888261492219664e-05,
+      "loss": 0.9805,
+      "step": 3737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.688661430652075e-05,
+      "loss": 0.966,
+      "step": 3738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.688496670883167e-05,
+      "loss": 0.7135,
+      "step": 3739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.688331869917767e-05,
+      "loss": 0.85,
+      "step": 3740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6881670277584e-05,
+      "loss": 0.7824,
+      "step": 3741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.688002144407593e-05,
+      "loss": 1.0682,
+      "step": 3742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6878372198678704e-05,
+      "loss": 1.1557,
+      "step": 3743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6876722541417616e-05,
+      "loss": 0.7407,
+      "step": 3744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687507247231793e-05,
+      "loss": 0.7667,
+      "step": 3745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687342199140494e-05,
+      "loss": 0.8481,
+      "step": 3746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.687177109870393e-05,
+      "loss": 0.9687,
+      "step": 3747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.68701197942402e-05,
+      "loss": 1.0313,
+      "step": 3748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686846807803905e-05,
+      "loss": 0.8977,
+      "step": 3749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.68668159501258e-05,
+      "loss": 0.8811,
+      "step": 3750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686516341052575e-05,
+      "loss": 0.8521,
+      "step": 3751
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686351045926424e-05,
+      "loss": 0.8066,
+      "step": 3752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686185709636658e-05,
+      "loss": 0.7188,
+      "step": 3753
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.686020332185811e-05,
+      "loss": 1.0794,
+      "step": 3754
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685854913576419e-05,
+      "loss": 0.9563,
+      "step": 3755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685689453811015e-05,
+      "loss": 0.7419,
+      "step": 3756
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685523952892133e-05,
+      "loss": 0.8059,
+      "step": 3757
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685358410822312e-05,
+      "loss": 0.9746,
+      "step": 3758
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685192827604087e-05,
+      "loss": 0.7954,
+      "step": 3759
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.685027203239996e-05,
+      "loss": 1.0279,
+      "step": 3760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6848615377325765e-05,
+      "loss": 0.7868,
+      "step": 3761
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.684695831084367e-05,
+      "loss": 0.8534,
+      "step": 3762
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6845300832979054e-05,
+      "loss": 0.7062,
+      "step": 3763
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.684364294375734e-05,
+      "loss": 0.9941,
+      "step": 3764
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.684198464320392e-05,
+      "loss": 0.8075,
+      "step": 3765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.68403259313442e-05,
+      "loss": 0.7188,
+      "step": 3766
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6838666808203604e-05,
+      "loss": 0.9569,
+      "step": 3767
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.683700727380755e-05,
+      "loss": 0.9268,
+      "step": 3768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.683534732818148e-05,
+      "loss": 0.8328,
+      "step": 3769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.683368697135081e-05,
+      "loss": 0.8659,
+      "step": 3770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.683202620334099e-05,
+      "loss": 0.934,
+      "step": 3771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.683036502417747e-05,
+      "loss": 0.8105,
+      "step": 3772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.682870343388572e-05,
+      "loss": 0.8895,
+      "step": 3773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.682704143249117e-05,
+      "loss": 0.6466,
+      "step": 3774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.68253790200193e-05,
+      "loss": 0.8524,
+      "step": 3775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6823716196495604e-05,
+      "loss": 0.8759,
+      "step": 3776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6822052961945524e-05,
+      "loss": 1.0006,
+      "step": 3777
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.682038931639458e-05,
+      "loss": 0.7358,
+      "step": 3778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.6818725259868245e-05,
+      "loss": 0.7638,
+      "step": 3779
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.681706079239202e-05,
+      "loss": 0.8103,
+      "step": 3780
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.681539591399143e-05,
+      "loss": 0.8156,
+      "step": 3781
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.681373062469196e-05,
+      "loss": 0.9126,
+      "step": 3782
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.681206492451914e-05,
+      "loss": 0.9467,
+      "step": 3783
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.681039881349849e-05,
+      "loss": 0.9848,
+      "step": 3784
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.680873229165554e-05,
+      "loss": 0.9721,
+      "step": 3785
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.680706535901583e-05,
+      "loss": 0.7753,
+      "step": 3786
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.68053980156049e-05,
+      "loss": 0.8094,
+      "step": 3787
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.68037302614483e-05,
+      "loss": 0.8372,
+      "step": 3788
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.680206209657159e-05,
+      "loss": 0.9645,
+      "step": 3789
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.680039352100033e-05,
+      "loss": 1.0164,
+      "step": 3790
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6798724534760076e-05,
+      "loss": 0.8931,
+      "step": 3791
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6797055137876414e-05,
+      "loss": 0.9095,
+      "step": 3792
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.679538533037492e-05,
+      "loss": 0.8259,
+      "step": 3793
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6793715112281196e-05,
+      "loss": 0.9325,
+      "step": 3794
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.679204448362081e-05,
+      "loss": 0.7574,
+      "step": 3795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6790373444419374e-05,
+      "loss": 0.779,
+      "step": 3796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.67887019947025e-05,
+      "loss": 0.9895,
+      "step": 3797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678703013449579e-05,
+      "loss": 0.834,
+      "step": 3798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678535786382487e-05,
+      "loss": 0.8952,
+      "step": 3799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6783685182715356e-05,
+      "loss": 0.8823,
+      "step": 3800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.678201209119288e-05,
+      "loss": 0.868,
+      "step": 3801
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6780338589283084e-05,
+      "loss": 0.8439,
+      "step": 3802
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6778664677011616e-05,
+      "loss": 0.7459,
+      "step": 3803
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.677699035440412e-05,
+      "loss": 0.7496,
+      "step": 3804
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.677531562148625e-05,
+      "loss": 0.9774,
+      "step": 3805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.677364047828366e-05,
+      "loss": 0.8558,
+      "step": 3806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.677196492482204e-05,
+      "loss": 1.0808,
+      "step": 3807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6770288961127053e-05,
+      "loss": 1.0274,
+      "step": 3808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6768612587224385e-05,
+      "loss": 0.7813,
+      "step": 3809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.676693580313971e-05,
+      "loss": 1.0452,
+      "step": 3810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6765258608898737e-05,
+      "loss": 0.9665,
+      "step": 3811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.676358100452716e-05,
+      "loss": 0.897,
+      "step": 3812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.676190299005069e-05,
+      "loss": 0.9007,
+      "step": 3813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6760224565495016e-05,
+      "loss": 1.0588,
+      "step": 3814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.675854573088589e-05,
+      "loss": 0.7318,
+      "step": 3815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6756866486249034e-05,
+      "loss": 0.7913,
+      "step": 3816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.675518683161015e-05,
+      "loss": 0.7153,
+      "step": 3817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6753506766995e-05,
+      "loss": 0.837,
+      "step": 3818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.675182629242933e-05,
+      "loss": 0.9317,
+      "step": 3819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6750145407938874e-05,
+      "loss": 0.7959,
+      "step": 3820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6748464113549404e-05,
+      "loss": 1.0037,
+      "step": 3821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6746782409286673e-05,
+      "loss": 0.7158,
+      "step": 3822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6745100295176454e-05,
+      "loss": 1.2259,
+      "step": 3823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.674341777124453e-05,
+      "loss": 0.729,
+      "step": 3824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6741734837516666e-05,
+      "loss": 0.7306,
+      "step": 3825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.674005149401866e-05,
+      "loss": 0.96,
+      "step": 3826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.67383677407763e-05,
+      "loss": 0.933,
+      "step": 3827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.673668357781539e-05,
+      "loss": 0.9596,
+      "step": 3828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.673499900516175e-05,
+      "loss": 0.9308,
+      "step": 3829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.673331402284118e-05,
+      "loss": 0.921,
+      "step": 3830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.67316286308795e-05,
+      "loss": 0.7615,
+      "step": 3831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672994282930253e-05,
+      "loss": 0.9329,
+      "step": 3832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672825661813611e-05,
+      "loss": 0.9479,
+      "step": 3833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672656999740609e-05,
+      "loss": 0.8733,
+      "step": 3834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672488296713829e-05,
+      "loss": 0.9706,
+      "step": 3835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.672319552735857e-05,
+      "loss": 0.6743,
+      "step": 3836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.6721507678092797e-05,
+      "loss": 0.9238,
+      "step": 3837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.671981941936682e-05,
+      "loss": 0.8653,
+      "step": 3838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.671813075120652e-05,
+      "loss": 0.8766,
+      "step": 3839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.671644167363776e-05,
+      "loss": 0.9682,
+      "step": 3840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.671475218668644e-05,
+      "loss": 0.9745,
+      "step": 3841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.671306229037844e-05,
+      "loss": 0.7823,
+      "step": 3842
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.671137198473964e-05,
+      "loss": 0.8178,
+      "step": 3843
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6709681269795964e-05,
+      "loss": 0.959,
+      "step": 3844
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6707990145573304e-05,
+      "loss": 0.8689,
+      "step": 3845
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.670629861209759e-05,
+      "loss": 0.9115,
+      "step": 3846
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.670460666939472e-05,
+      "loss": 0.9183,
+      "step": 3847
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.670291431749063e-05,
+      "loss": 0.9209,
+      "step": 3848
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6701221556411257e-05,
+      "loss": 0.7083,
+      "step": 3849
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.669952838618254e-05,
+      "loss": 0.8998,
+      "step": 3850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.669783480683041e-05,
+      "loss": 0.7354,
+      "step": 3851
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6696140818380835e-05,
+      "loss": 0.7339,
+      "step": 3852
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.669444642085976e-05,
+      "loss": 1.0819,
+      "step": 3853
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6692751614293155e-05,
+      "loss": 0.8432,
+      "step": 3854
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6691056398707e-05,
+      "loss": 0.9424,
+      "step": 3855
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6689360774127246e-05,
+      "loss": 0.7105,
+      "step": 3856
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.66876647405799e-05,
+      "loss": 1.0729,
+      "step": 3857
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.668596829809092e-05,
+      "loss": 0.9232,
+      "step": 3858
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6684271446686344e-05,
+      "loss": 0.8453,
+      "step": 3859
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6682574186392135e-05,
+      "loss": 0.8725,
+      "step": 3860
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.668087651723432e-05,
+      "loss": 0.8435,
+      "step": 3861
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.667917843923892e-05,
+      "loss": 0.8023,
+      "step": 3862
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.667747995243193e-05,
+      "loss": 0.8076,
+      "step": 3863
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.667578105683939e-05,
+      "loss": 0.8262,
+      "step": 3864
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.667408175248735e-05,
+      "loss": 0.836,
+      "step": 3865
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.667238203940182e-05,
+      "loss": 0.8076,
+      "step": 3866
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.667068191760885e-05,
+      "loss": 0.8461,
+      "step": 3867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.66689813871345e-05,
+      "loss": 0.8299,
+      "step": 3868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.666728044800484e-05,
+      "loss": 0.8015,
+      "step": 3869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6665579100245906e-05,
+      "loss": 0.757,
+      "step": 3870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6663877343883784e-05,
+      "loss": 0.9807,
+      "step": 3871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6662175178944546e-05,
+      "loss": 0.9275,
+      "step": 3872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.666047260545428e-05,
+      "loss": 0.872,
+      "step": 3873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665876962343907e-05,
+      "loss": 0.8107,
+      "step": 3874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665706623292501e-05,
+      "loss": 0.8234,
+      "step": 3875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665536243393822e-05,
+      "loss": 0.7559,
+      "step": 3876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665365822650478e-05,
+      "loss": 0.9987,
+      "step": 3877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665195361065082e-05,
+      "loss": 0.7966,
+      "step": 3878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.665024858640245e-05,
+      "loss": 0.7403,
+      "step": 3879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6648543153785814e-05,
+      "loss": 0.6765,
+      "step": 3880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.664683731282702e-05,
+      "loss": 0.7655,
+      "step": 3881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.664513106355223e-05,
+      "loss": 0.8926,
+      "step": 3882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.664342440598758e-05,
+      "loss": 0.789,
+      "step": 3883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.664171734015922e-05,
+      "loss": 1.1578,
+      "step": 3884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.664000986609331e-05,
+      "loss": 0.7981,
+      "step": 3885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663830198381601e-05,
+      "loss": 0.8421,
+      "step": 3886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663659369335349e-05,
+      "loss": 0.8899,
+      "step": 3887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6634884994731944e-05,
+      "loss": 0.9627,
+      "step": 3888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6633175887977534e-05,
+      "loss": 0.9034,
+      "step": 3889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.663146637311645e-05,
+      "loss": 0.7791,
+      "step": 3890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.662975645017491e-05,
+      "loss": 0.9641,
+      "step": 3891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.662804611917908e-05,
+      "loss": 0.8168,
+      "step": 3892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6626335380155196e-05,
+      "loss": 0.8074,
+      "step": 3893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6624624233129466e-05,
+      "loss": 0.5774,
+      "step": 3894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.66229126781281e-05,
+      "loss": 1.036,
+      "step": 3895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.662120071517734e-05,
+      "loss": 0.8014,
+      "step": 3896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6619488344303405e-05,
+      "loss": 0.8906,
+      "step": 3897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6617775565532535e-05,
+      "loss": 0.7149,
+      "step": 3898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.661606237889099e-05,
+      "loss": 0.863,
+      "step": 3899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6614348784405e-05,
+      "loss": 0.852,
+      "step": 3900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.6612634782100845e-05,
+      "loss": 0.8563,
+      "step": 3901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.661092037200477e-05,
+      "loss": 0.9132,
+      "step": 3902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660920555414306e-05,
+      "loss": 0.8073,
+      "step": 3903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660749032854199e-05,
+      "loss": 0.8556,
+      "step": 3904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660577469522783e-05,
+      "loss": 0.7746,
+      "step": 3905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.660405865422688e-05,
+      "loss": 0.7357,
+      "step": 3906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.660234220556544e-05,
+      "loss": 0.9936,
+      "step": 3907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6600625349269784e-05,
+      "loss": 0.9015,
+      "step": 3908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.659890808536625e-05,
+      "loss": 0.7739,
+      "step": 3909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6597190413881155e-05,
+      "loss": 0.747,
+      "step": 3910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.659547233484079e-05,
+      "loss": 0.857,
+      "step": 3911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.659375384827151e-05,
+      "loss": 0.8961,
+      "step": 3912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6592034954199626e-05,
+      "loss": 0.9585,
+      "step": 3913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.659031565265149e-05,
+      "loss": 0.7511,
+      "step": 3914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.658859594365344e-05,
+      "loss": 0.8902,
+      "step": 3915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6586875827231835e-05,
+      "loss": 1.0764,
+      "step": 3916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6585155303413035e-05,
+      "loss": 0.851,
+      "step": 3917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.658343437222339e-05,
+      "loss": 0.7478,
+      "step": 3918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.658171303368928e-05,
+      "loss": 0.8645,
+      "step": 3919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.657999128783709e-05,
+      "loss": 0.6094,
+      "step": 3920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6578269134693175e-05,
+      "loss": 0.7361,
+      "step": 3921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6576546574283956e-05,
+      "loss": 0.8726,
+      "step": 3922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.657482360663581e-05,
+      "loss": 0.9057,
+      "step": 3923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6573100231775145e-05,
+      "loss": 0.7937,
+      "step": 3924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.657137644972836e-05,
+      "loss": 0.7599,
+      "step": 3925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.656965226052188e-05,
+      "loss": 0.9047,
+      "step": 3926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.656792766418212e-05,
+      "loss": 0.6939,
+      "step": 3927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6566202660735504e-05,
+      "loss": 0.7418,
+      "step": 3928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6564477250208465e-05,
+      "loss": 0.9094,
+      "step": 3929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.656275143262745e-05,
+      "loss": 0.9633,
+      "step": 3930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6561025208018896e-05,
+      "loss": 1.0619,
+      "step": 3931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.655929857640925e-05,
+      "loss": 1.0488,
+      "step": 3932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.655757153782499e-05,
+      "loss": 0.8198,
+      "step": 3933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.655584409229256e-05,
+      "loss": 0.8928,
+      "step": 3934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.655411623983843e-05,
+      "loss": 0.8305,
+      "step": 3935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.655238798048909e-05,
+      "loss": 0.7445,
+      "step": 3936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6550659314271004e-05,
+      "loss": 0.9237,
+      "step": 3937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654893024121068e-05,
+      "loss": 0.7493,
+      "step": 3938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654720076133461e-05,
+      "loss": 0.8147,
+      "step": 3939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654547087466929e-05,
+      "loss": 0.8667,
+      "step": 3940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654374058124121e-05,
+      "loss": 0.9909,
+      "step": 3941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654200988107692e-05,
+      "loss": 0.6879,
+      "step": 3942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.654027877420292e-05,
+      "loss": 0.7535,
+      "step": 3943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.653854726064573e-05,
+      "loss": 0.7001,
+      "step": 3944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6536815340431884e-05,
+      "loss": 0.9285,
+      "step": 3945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.653508301358794e-05,
+      "loss": 0.9226,
+      "step": 3946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.653335028014042e-05,
+      "loss": 0.7868,
+      "step": 3947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6531617140115884e-05,
+      "loss": 0.8936,
+      "step": 3948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.65298835935409e-05,
+      "loss": 0.9046,
+      "step": 3949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.652814964044201e-05,
+      "loss": 0.7576,
+      "step": 3950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6526415280845804e-05,
+      "loss": 0.7853,
+      "step": 3951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.652468051477885e-05,
+      "loss": 0.8221,
+      "step": 3952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6522945342267715e-05,
+      "loss": 0.9148,
+      "step": 3953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.652120976333901e-05,
+      "loss": 0.7433,
+      "step": 3954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6519473778019324e-05,
+      "loss": 0.8359,
+      "step": 3955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.651773738633525e-05,
+      "loss": 0.7048,
+      "step": 3956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.651600058831341e-05,
+      "loss": 1.0036,
+      "step": 3957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.65142633839804e-05,
+      "loss": 0.8933,
+      "step": 3958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6512525773362846e-05,
+      "loss": 0.7807,
+      "step": 3959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.651078775648738e-05,
+      "loss": 0.8067,
+      "step": 3960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6509049333380626e-05,
+      "loss": 1.0208,
+      "step": 3961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6507310504069227e-05,
+      "loss": 1.05,
+      "step": 3962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6505571268579826e-05,
+      "loss": 0.8168,
+      "step": 3963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.650383162693908e-05,
+      "loss": 0.8681,
+      "step": 3964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.650209157917363e-05,
+      "loss": 0.7636,
+      "step": 3965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.650035112531016e-05,
+      "loss": 1.0796,
+      "step": 3966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6498610265375316e-05,
+      "loss": 0.7397,
+      "step": 3967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6496868999395794e-05,
+      "loss": 0.8808,
+      "step": 3968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.6495127327398266e-05,
+      "loss": 0.89,
+      "step": 3969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.649338524940942e-05,
+      "loss": 0.936,
+      "step": 3970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6491642765455964e-05,
+      "loss": 0.8636,
+      "step": 3971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6489899875564574e-05,
+      "loss": 0.911,
+      "step": 3972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.648815657976197e-05,
+      "loss": 0.8984,
+      "step": 3973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.648641287807486e-05,
+      "loss": 0.819,
+      "step": 3974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6484668770529986e-05,
+      "loss": 1.06,
+      "step": 3975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.648292425715404e-05,
+      "loss": 0.7284,
+      "step": 3976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6481179337973776e-05,
+      "loss": 0.759,
+      "step": 3977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6479434013015915e-05,
+      "loss": 1.2003,
+      "step": 3978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.647768828230722e-05,
+      "loss": 0.867,
+      "step": 3979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.647594214587443e-05,
+      "loss": 0.963,
+      "step": 3980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.64741956037443e-05,
+      "loss": 0.6974,
+      "step": 3981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.647244865594359e-05,
+      "loss": 0.9413,
+      "step": 3982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.647070130249909e-05,
+      "loss": 0.8501,
+      "step": 3983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6468953543437544e-05,
+      "loss": 0.7603,
+      "step": 3984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.646720537878576e-05,
+      "loss": 0.8468,
+      "step": 3985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.646545680857051e-05,
+      "loss": 0.8454,
+      "step": 3986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6463707832818595e-05,
+      "loss": 0.8378,
+      "step": 3987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.646195845155681e-05,
+      "loss": 0.9404,
+      "step": 3988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6460208664811976e-05,
+      "loss": 0.7896,
+      "step": 3989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.645845847261088e-05,
+      "loss": 1.0088,
+      "step": 3990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6456707874980356e-05,
+      "loss": 0.7777,
+      "step": 3991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.645495687194723e-05,
+      "loss": 1.0173,
+      "step": 3992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.645320546353833e-05,
+      "loss": 0.7423,
+      "step": 3993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6451453649780486e-05,
+      "loss": 0.7378,
+      "step": 3994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.644970143070055e-05,
+      "loss": 0.9344,
+      "step": 3995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6447948806325374e-05,
+      "loss": 1.151,
+      "step": 3996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6446195776681814e-05,
+      "loss": 0.7807,
+      "step": 3997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.644444234179672e-05,
+      "loss": 0.767,
+      "step": 3998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.6442688501696966e-05,
+      "loss": 0.6603,
+      "step": 3999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.644093425640944e-05,
+      "loss": 0.6169,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 25404,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 1000,
+  "total_flos": 4.720012287165858e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}