diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18604 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999192571659266,
+  "eval_steps": 500,
+  "global_step": 3096,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1505376344086024e-07,
+      "loss": 11.2523,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.301075268817205e-07,
+      "loss": 11.1603,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.451612903225807e-07,
+      "loss": 11.2967,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.60215053763441e-07,
+      "loss": 11.5578,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0752688172043011e-06,
+      "loss": 11.3434,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2903225806451614e-06,
+      "loss": 10.852,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5053763440860217e-06,
+      "loss": 11.1243,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.720430107526882e-06,
+      "loss": 10.3581,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.935483870967742e-06,
+      "loss": 10.4285,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1505376344086023e-06,
+      "loss": 9.8411,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3655913978494625e-06,
+      "loss": 9.0887,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.580645161290323e-06,
+      "loss": 8.9608,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7956989247311827e-06,
+      "loss": 8.4732,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0107526881720433e-06,
+      "loss": 7.9622,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 7.8852,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.440860215053764e-06,
+      "loss": 7.9042,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.655913978494624e-06,
+      "loss": 7.452,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.870967741935484e-06,
+      "loss": 7.3886,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.086021505376344e-06,
+      "loss": 7.131,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3010752688172045e-06,
+      "loss": 7.1192,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.516129032258065e-06,
+      "loss": 6.7896,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.731182795698925e-06,
+      "loss": 6.7783,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.946236559139785e-06,
+      "loss": 6.219,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.161290322580646e-06,
+      "loss": 7.0534,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.376344086021506e-06,
+      "loss": 7.0028,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.591397849462365e-06,
+      "loss": 6.4491,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.806451612903226e-06,
+      "loss": 6.3982,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.021505376344087e-06,
+      "loss": 6.395,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.236559139784947e-06,
+      "loss": 6.1269,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 6.0543,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 5.8655,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.881720430107528e-06,
+      "loss": 5.8824,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.096774193548388e-06,
+      "loss": 6.0531,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.311827956989248e-06,
+      "loss": 6.4295,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.526881720430108e-06,
+      "loss": 5.9465,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.741935483870968e-06,
+      "loss": 6.3217,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.956989247311828e-06,
+      "loss": 5.8956,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.172043010752689e-06,
+      "loss": 6.0283,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.387096774193549e-06,
+      "loss": 5.9659,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.602150537634409e-06,
+      "loss": 6.27,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.81720430107527e-06,
+      "loss": 6.4405,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.03225806451613e-06,
+      "loss": 6.3341,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.24731182795699e-06,
+      "loss": 6.6001,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.46236559139785e-06,
+      "loss": 5.9685,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 6.7039,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.89247311827957e-06,
+      "loss": 6.2619,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0107526881720431e-05,
+      "loss": 6.2191,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0322580645161291e-05,
+      "loss": 6.0644,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0537634408602151e-05,
+      "loss": 6.445,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 6.3045,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.096774193548387e-05,
+      "loss": 6.2348,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.118279569892473e-05,
+      "loss": 6.4836,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1397849462365593e-05,
+      "loss": 5.8163,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1612903225806453e-05,
+      "loss": 6.0588,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1827956989247313e-05,
+      "loss": 6.0818,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2043010752688173e-05,
+      "loss": 6.5693,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2258064516129034e-05,
+      "loss": 5.8655,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2473118279569894e-05,
+      "loss": 6.4145,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2688172043010754e-05,
+      "loss": 5.6631,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 6.2009,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3118279569892473e-05,
+      "loss": 6.2652,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 5.4556,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3548387096774194e-05,
+      "loss": 5.9838,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3763440860215056e-05,
+      "loss": 6.2262,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3978494623655916e-05,
+      "loss": 5.7353,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4193548387096776e-05,
+      "loss": 6.0541,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4408602150537636e-05,
+      "loss": 6.4737,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4623655913978497e-05,
+      "loss": 5.8877,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4838709677419357e-05,
+      "loss": 5.3909,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5053763440860215e-05,
+      "loss": 5.9032,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5268817204301076e-05,
+      "loss": 5.7652,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5483870967741936e-05,
+      "loss": 5.6813,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5698924731182796e-05,
+      "loss": 5.0512,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5913978494623657e-05,
+      "loss": 5.3157,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 6.0397,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6344086021505377e-05,
+      "loss": 5.2705,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6559139784946237e-05,
+      "loss": 6.0639,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6774193548387098e-05,
+      "loss": 6.012,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6989247311827958e-05,
+      "loss": 6.0576,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7204301075268818e-05,
+      "loss": 6.0368,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.741935483870968e-05,
+      "loss": 5.5931,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.763440860215054e-05,
+      "loss": 5.7729,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.78494623655914e-05,
+      "loss": 5.9606,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.806451612903226e-05,
+      "loss": 5.4084,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.827956989247312e-05,
+      "loss": 5.898,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.849462365591398e-05,
+      "loss": 6.2293,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.870967741935484e-05,
+      "loss": 5.6815,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.89247311827957e-05,
+      "loss": 6.4054,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.913978494623656e-05,
+      "loss": 5.6638,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 5.8248,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.956989247311828e-05,
+      "loss": 6.0076,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.978494623655914e-05,
+      "loss": 6.1857,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 5.5555,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994527836745e-05,
+      "loss": 5.5281,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997811135296e-05,
+      "loss": 5.6892,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999950750566616e-05,
+      "loss": 5.9195,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999912445507662e-05,
+      "loss": 6.1625,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999986319621801e-05,
+      "loss": 5.7769,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999980300275157e-05,
+      "loss": 5.6907,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999731865174213e-05,
+      "loss": 5.4982,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999649783563798e-05,
+      "loss": 6.0839,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999556758010158e-05,
+      "loss": 5.5607,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99994527886151e-05,
+      "loss": 5.5967,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999337875492412e-05,
+      "loss": 5.8768,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999921201876786e-05,
+      "loss": 5.9076,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999075218579184e-05,
+      "loss": 6.0399,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998927475076107e-05,
+      "loss": 6.3103,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998768788420322e-05,
+      "loss": 6.083,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99985991587855e-05,
+      "loss": 5.6962,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998418586357288e-05,
+      "loss": 6.0606,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999822707133331e-05,
+      "loss": 6.1549,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999802461392317e-05,
+      "loss": 6.308,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997811214348447e-05,
+      "loss": 6.2313,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997586872842683e-05,
+      "loss": 5.6793,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999735158965141e-05,
+      "loss": 6.0456,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999710536503213e-05,
+      "loss": 5.4685,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996848199254315e-05,
+      "loss": 5.9847,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996580092599424e-05,
+      "loss": 5.6301,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996301045360874e-05,
+      "loss": 6.1908,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996011057844063e-05,
+      "loss": 5.4774,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999571013036637e-05,
+      "loss": 6.0477,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999539826325713e-05,
+      "loss": 5.8704,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995075456857672e-05,
+      "loss": 6.4527,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994741711521278e-05,
+      "loss": 5.6719,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999439702761321e-05,
+      "loss": 6.0086,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994041405510705e-05,
+      "loss": 5.7683,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993674845602966e-05,
+      "loss": 6.2303,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993297348291168e-05,
+      "loss": 5.4419,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992908913988454e-05,
+      "loss": 5.5274,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992509543119945e-05,
+      "loss": 5.8157,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992099236122718e-05,
+      "loss": 5.4889,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991677993445832e-05,
+      "loss": 6.1498,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991245815550304e-05,
+      "loss": 5.5528,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999080270290913e-05,
+      "loss": 5.7821,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999034865600726e-05,
+      "loss": 5.497,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989883675341625e-05,
+      "loss": 6.2006,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998940776142111e-05,
+      "loss": 5.8677,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998892091476657e-05,
+      "loss": 5.8513,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988423135910827e-05,
+      "loss": 5.6117,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987914425398673e-05,
+      "loss": 5.679,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987394783786845e-05,
+      "loss": 5.5721,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998686421164407e-05,
+      "loss": 6.0701,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986322709551008e-05,
+      "loss": 6.5103,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998577027810031e-05,
+      "loss": 5.6403,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985206917896563e-05,
+      "loss": 5.633,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998463262955634e-05,
+      "loss": 5.674,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984047413708153e-05,
+      "loss": 5.8374,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998345127099248e-05,
+      "loss": 5.5087,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998284420206176e-05,
+      "loss": 5.6504,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982226207580397e-05,
+      "loss": 6.2382,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981597288224733e-05,
+      "loss": 5.972,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980957444683085e-05,
+      "loss": 5.7115,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980306677655715e-05,
+      "loss": 5.5847,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979644987854846e-05,
+      "loss": 5.8033,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978972376004654e-05,
+      "loss": 5.8481,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997828884284126e-05,
+      "loss": 6.0275,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977594389112753e-05,
+      "loss": 5.6279,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976889015579167e-05,
+      "loss": 5.7205,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997617272301248e-05,
+      "loss": 5.539,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975445512196624e-05,
+      "loss": 5.5349,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997470738392749e-05,
+      "loss": 5.7883,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973958339012907e-05,
+      "loss": 5.8294,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997319837827265e-05,
+      "loss": 5.5742,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972427502538453e-05,
+      "loss": 5.4151,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997164571265398e-05,
+      "loss": 6.1941,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997085300947485e-05,
+      "loss": 5.7661,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970049393868624e-05,
+      "loss": 5.7286,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969234866714803e-05,
+      "loss": 5.3034,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968409428904834e-05,
+      "loss": 5.5373,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967573081342103e-05,
+      "loss": 5.7259,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966725824941933e-05,
+      "loss": 5.4789,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965867660631595e-05,
+      "loss": 6.1486,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964998589350292e-05,
+      "loss": 5.3499,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964118612049155e-05,
+      "loss": 5.211,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963227729691267e-05,
+      "loss": 5.3859,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996232594325164e-05,
+      "loss": 5.7187,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961413253717214e-05,
+      "loss": 5.6355,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960489662086867e-05,
+      "loss": 6.1098,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959555169371412e-05,
+      "loss": 5.9268,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958609776593586e-05,
+      "loss": 5.4417,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957653484788054e-05,
+      "loss": 5.6361,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956686295001416e-05,
+      "loss": 5.9034,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955708208292196e-05,
+      "loss": 5.6908,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954719225730847e-05,
+      "loss": 5.6977,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995371934839974e-05,
+      "loss": 5.478,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952708577393176e-05,
+      "loss": 5.2431,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995168691381737e-05,
+      "loss": 6.1743,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995065435879047e-05,
+      "loss": 5.9625,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994961091344254e-05,
+      "loss": 4.9168,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994855657891555e-05,
+      "loss": 5.978,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994749135636341e-05,
+      "loss": 5.7182,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946415246951928e-05,
+      "loss": 5.7296,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945328251858837e-05,
+      "loss": 5.6075,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944230372273772e-05,
+      "loss": 5.4285,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943121609398297e-05,
+      "loss": 5.6908,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942001964445875e-05,
+      "loss": 5.1918,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994087143864188e-05,
+      "loss": 4.999,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99397300332236e-05,
+      "loss": 5.4761,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938577749440226e-05,
+      "loss": 5.9446,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937414588552852e-05,
+      "loss": 5.6314,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936240551834484e-05,
+      "loss": 4.7865,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935055640570017e-05,
+      "loss": 5.6759,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933859856056264e-05,
+      "loss": 5.7187,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932653199601933e-05,
+      "loss": 5.6459,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931435672527625e-05,
+      "loss": 5.6789,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993020727616584e-05,
+      "loss": 5.8367,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928968011860973e-05,
+      "loss": 6.2878,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992771788096932e-05,
+      "loss": 5.5897,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926456884859063e-05,
+      "loss": 5.6318,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992518502491028e-05,
+      "loss": 6.1508,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992390230251493e-05,
+      "loss": 5.6096,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922608719076874e-05,
+      "loss": 5.5019,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921304276011845e-05,
+      "loss": 5.4271,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919988974747473e-05,
+      "loss": 5.8629,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991866281672326e-05,
+      "loss": 5.7247,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991732580339061e-05,
+      "loss": 5.4441,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915977936212778e-05,
+      "loss": 5.1901,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914619216664922e-05,
+      "loss": 5.5426,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913249646234072e-05,
+      "loss": 5.5594,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911869226419124e-05,
+      "loss": 5.6131,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991047795873086e-05,
+      "loss": 5.7572,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990907584469192e-05,
+      "loss": 5.9789,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907662885836836e-05,
+      "loss": 6.2657,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906239083711988e-05,
+      "loss": 5.7575,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904804439875635e-05,
+      "loss": 5.9298,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903358955897892e-05,
+      "loss": 5.899,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901902633360754e-05,
+      "loss": 6.0377,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900435473858057e-05,
+      "loss": 5.464,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898957478995515e-05,
+      "loss": 6.0851,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897468650390694e-05,
+      "loss": 5.2966,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895968989673014e-05,
+      "loss": 5.9391,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894458498483754e-05,
+      "loss": 5.8891,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892937178476044e-05,
+      "loss": 5.7219,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891405031314867e-05,
+      "loss": 5.5409,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889862058677056e-05,
+      "loss": 5.6944,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 5.9884,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988674364373809e-05,
+      "loss": 5.9536,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988516820484983e-05,
+      "loss": 5.8912,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988358194731072e-05,
+      "loss": 5.8294,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988198487285682e-05,
+      "loss": 5.591,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988037698323601e-05,
+      "loss": 5.3667,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878758280208018e-05,
+      "loss": 5.7412,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987712876554441e-05,
+      "loss": 5.5993,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987548844102858e-05,
+      "loss": 5.4538,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987383730845575e-05,
+      "loss": 5.4855,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872175369632975e-05,
+      "loss": 5.9244,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870502626379127e-05,
+      "loss": 5.6939,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986881908052493e-05,
+      "loss": 5.8225,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986712473391289e-05,
+      "loss": 5.6172,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986541958839737e-05,
+      "loss": 5.9472,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986370364584453e-05,
+      "loss": 5.7957,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861976908132356e-05,
+      "loss": 5.7991,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860239377150646e-05,
+      "loss": 4.9695,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858491054801007e-05,
+      "loss": 5.7282,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856731942996868e-05,
+      "loss": 5.9198,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854962043663452e-05,
+      "loss": 5.6327,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853181358737796e-05,
+      "loss": 5.2667,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851389890168738e-05,
+      "loss": 5.6768,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849587639916924e-05,
+      "loss": 5.3621,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847774609954794e-05,
+      "loss": 5.3538,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845950802266584e-05,
+      "loss": 5.1158,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844116218848335e-05,
+      "loss": 5.4266,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842270861707867e-05,
+      "loss": 5.5394,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840414732864804e-05,
+      "loss": 5.2664,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838547834350555e-05,
+      "loss": 5.56,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836670168208313e-05,
+      "loss": 5.4799,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834781736493057e-05,
+      "loss": 5.6981,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832882541271545e-05,
+      "loss": 5.5166,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830972584622327e-05,
+      "loss": 5.5523,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982905186863571e-05,
+      "loss": 5.9833,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98271203954138e-05,
+      "loss": 5.9419,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825178167070457e-05,
+      "loss": 5.7361,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823225185731325e-05,
+      "loss": 5.4065,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821261453533803e-05,
+      "loss": 5.5331,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819286972627066e-05,
+      "loss": 5.0584,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981730174517206e-05,
+      "loss": 5.0447,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815305773341467e-05,
+      "loss": 6.046,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813299059319753e-05,
+      "loss": 5.2723,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981128160530313e-05,
+      "loss": 5.9329,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809253413499565e-05,
+      "loss": 5.7179,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807214486128777e-05,
+      "loss": 5.6529,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980516482542224e-05,
+      "loss": 5.4845,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980310443362316e-05,
+      "loss": 5.5007,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801033312986503e-05,
+      "loss": 5.4484,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979895146577897e-05,
+      "loss": 5.453,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796858894279004e-05,
+      "loss": 5.685,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979475560077678e-05,
+      "loss": 5.3919,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792641587574212e-05,
+      "loss": 5.8338,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979051685698495e-05,
+      "loss": 6.2729,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788381411334364e-05,
+      "loss": 5.6173,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786235252959555e-05,
+      "loss": 5.673,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978407838420935e-05,
+      "loss": 6.093,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97819108074443e-05,
+      "loss": 5.6624,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779732525036663e-05,
+      "loss": 5.7209,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977754353937043e-05,
+      "loss": 5.3588,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775343852841297e-05,
+      "loss": 5.6003,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773133467856672e-05,
+      "loss": 5.7414,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977091238683567e-05,
+      "loss": 5.4276,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976868061220912e-05,
+      "loss": 5.03,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766438146419547e-05,
+      "loss": 5.7274,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764184991921178e-05,
+      "loss": 5.2163,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761921151179937e-05,
+      "loss": 6.0195,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759646626673445e-05,
+      "loss": 5.8819,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757361420891022e-05,
+      "loss": 5.502,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755065536333664e-05,
+      "loss": 5.8538,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975275897551407e-05,
+      "loss": 5.6033,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750441740956608e-05,
+      "loss": 6.1968,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974811383519734e-05,
+      "loss": 5.4511,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745775260783997e-05,
+      "loss": 5.7708,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743426020275996e-05,
+      "loss": 5.6127,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741066116244423e-05,
+      "loss": 5.9458,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973869555127203e-05,
+      "loss": 5.7762,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736314327953242e-05,
+      "loss": 5.4952,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973392244889415e-05,
+      "loss": 5.4392,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731519916712504e-05,
+      "loss": 5.9811,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729106734037712e-05,
+      "loss": 5.2171,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972668290351084e-05,
+      "loss": 5.2674,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972424842778461e-05,
+      "loss": 5.773,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972180330952339e-05,
+      "loss": 5.2495,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719347551403193e-05,
+      "loss": 5.304,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716881156111685e-05,
+      "loss": 5.2438,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971440412634817e-05,
+      "loss": 5.1921,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971191646482359e-05,
+      "loss": 5.8696,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 5.6236,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9706909257393178e-05,
+      "loss": 5.7124,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704389716967396e-05,
+      "loss": 5.2619,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701859555740647e-05,
+      "loss": 5.5584,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699318776482017e-05,
+      "loss": 5.941,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696767381972225e-05,
+      "loss": 5.7166,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694205375003596e-05,
+      "loss": 5.4249,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691632758380074e-05,
+      "loss": 5.6803,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968904953491721e-05,
+      "loss": 5.6692,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686455707442177e-05,
+      "loss": 5.3837,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968385127879374e-05,
+      "loss": 5.4445,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681236251822274e-05,
+      "loss": 5.5735,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678610629389748e-05,
+      "loss": 5.1102,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675974414369726e-05,
+      "loss": 5.5418,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673327609647367e-05,
+      "loss": 5.2347,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670670218119426e-05,
+      "loss": 5.7103,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966800224269424e-05,
+      "loss": 6.3578,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665323686291717e-05,
+      "loss": 5.4782,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966263455184337e-05,
+      "loss": 5.2717,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965993484229227e-05,
+      "loss": 5.9077,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965722456059307e-05,
+      "loss": 6.0143,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654503709711984e-05,
+      "loss": 5.2354,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651772292626804e-05,
+      "loss": 5.41,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649030312326885e-05,
+      "loss": 5.6595,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646277771813134e-05,
+      "loss": 5.5018,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643514674098026e-05,
+      "loss": 5.4212,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640741022205584e-05,
+      "loss": 5.3674,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963795681917138e-05,
+      "loss": 5.6751,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635162068042547e-05,
+      "loss": 5.7527,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632356771877735e-05,
+      "loss": 5.7428,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629540933747168e-05,
+      "loss": 4.6071,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626714556732582e-05,
+      "loss": 5.8738,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9623877643927258e-05,
+      "loss": 5.7195,
+      "step": 356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621030198436007e-05,
+      "loss": 5.7148,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618172223375166e-05,
+      "loss": 5.4018,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615303721872598e-05,
+      "loss": 5.4383,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961242469706768e-05,
+      "loss": 5.3423,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609535152111313e-05,
+      "loss": 5.2229,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606635090165914e-05,
+      "loss": 5.498,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603724514405398e-05,
+      "loss": 5.8307,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96008034280152e-05,
+      "loss": 5.8714,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597871834192248e-05,
+      "loss": 5.2466,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 5.1189,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959197713709331e-05,
+      "loss": 5.1863,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589014040268678e-05,
+      "loss": 5.4508,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586040448913982e-05,
+      "loss": 5.7852,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583056366283618e-05,
+      "loss": 5.4393,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958006179564347e-05,
+      "loss": 6.1222,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577056740270887e-05,
+      "loss": 5.4123,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574041203454707e-05,
+      "loss": 4.9911,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571015188495223e-05,
+      "loss": 5.102,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956797869870421e-05,
+      "loss": 5.7436,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564931737404896e-05,
+      "loss": 5.8948,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561874307931986e-05,
+      "loss": 5.2555,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558806413631622e-05,
+      "loss": 5.5007,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 4.9589,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95526392439904e-05,
+      "loss": 4.9332,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549539975399096e-05,
+      "loss": 5.4381,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546430255479437e-05,
+      "loss": 5.907,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543310087634798e-05,
+      "loss": 5.3406,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954017947528e-05,
+      "loss": 5.2588,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953703842184128e-05,
+      "loss": 5.2448,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533886930756318e-05,
+      "loss": 5.6152,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9530725005474195e-05,
+      "loss": 5.3286,
+      "step": 387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952755264945544e-05,
+      "loss": 5.4163,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524369866171976e-05,
+      "loss": 5.8276,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952117665910714e-05,
+      "loss": 5.5625,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517973031755693e-05,
+      "loss": 5.2268,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514758987623784e-05,
+      "loss": 5.4272,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511534530228968e-05,
+      "loss": 5.6319,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508299663100195e-05,
+      "loss": 6.0987,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505054389777815e-05,
+      "loss": 5.1525,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950179871381355e-05,
+      "loss": 5.356,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498532638770528e-05,
+      "loss": 5.6566,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495256168223246e-05,
+      "loss": 5.7933,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949196930575758e-05,
+      "loss": 5.2578,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488672054970777e-05,
+      "loss": 5.4439,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485364419471454e-05,
+      "loss": 5.2028,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482046402879607e-05,
+      "loss": 5.4753,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947871800882657e-05,
+      "loss": 5.1778,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475379240955048e-05,
+      "loss": 5.3738,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472030102919102e-05,
+      "loss": 4.8285,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468670598384135e-05,
+      "loss": 5.5062,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94653007310269e-05,
+      "loss": 5.2454,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946192050453549e-05,
+      "loss": 5.1467,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945852992260933e-05,
+      "loss": 5.7146,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945512898895919e-05,
+      "loss": 5.3914,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945171770730716e-05,
+      "loss": 5.8769,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448296081386656e-05,
+      "loss": 5.7974,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444864114942424e-05,
+      "loss": 5.8411,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441421811730515e-05,
+      "loss": 6.1292,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437969175518296e-05,
+      "loss": 5.5491,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943450621008445e-05,
+      "loss": 5.6618,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431032919218957e-05,
+      "loss": 5.5782,
+      "step": 417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94275493067231e-05,
+      "loss": 5.8389,
+      "step": 418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424055376409457e-05,
+      "loss": 5.6693,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420551132101904e-05,
+      "loss": 5.4265,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941703657763559e-05,
+      "loss": 5.6138,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413511716856973e-05,
+      "loss": 5.3944,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409976553623767e-05,
+      "loss": 5.4957,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940643109180497e-05,
+      "loss": 5.4157,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940287533528085e-05,
+      "loss": 5.1284,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399309287942945e-05,
+      "loss": 5.4513,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939573295369406e-05,
+      "loss": 5.5275,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392146336448235e-05,
+      "loss": 5.4687,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93885494401308e-05,
+      "loss": 5.4805,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384942268678305e-05,
+      "loss": 5.4937,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938132482603856e-05,
+      "loss": 5.9957,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377697116170614e-05,
+      "loss": 5.0352,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374059143044745e-05,
+      "loss": 5.0728,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370410910642473e-05,
+      "loss": 5.8465,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366752422956542e-05,
+      "loss": 5.6267,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936308368399092e-05,
+      "loss": 5.8,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93594046977608e-05,
+      "loss": 5.5685,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355715468292576e-05,
+      "loss": 6.0706,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352015999623867e-05,
+      "loss": 5.0806,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934830629580349e-05,
+      "loss": 5.6826,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934458636089147e-05,
+      "loss": 5.4597,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934085619895902e-05,
+      "loss": 5.1316,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337115814088555e-05,
+      "loss": 5.5749,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333365210373668e-05,
+      "loss": 4.9859,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932960439191915e-05,
+      "loss": 5.336,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932583336284096e-05,
+      "loss": 5.5704,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322052127266234e-05,
+      "loss": 5.5642,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931826068933328e-05,
+      "loss": 4.9958,
+      "step": 448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314459053191576e-05,
+      "loss": 5.4844,
+      "step": 449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310647223001752e-05,
+      "loss": 5.3584,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93068252029356e-05,
+      "loss": 5.504,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302992997176066e-05,
+      "loss": 5.2547,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299150609917237e-05,
+      "loss": 5.6756,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929529804536435e-05,
+      "loss": 5.6821,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291435307733777e-05,
+      "loss": 5.6029,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287562401253023e-05,
+      "loss": 5.4263,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283679330160726e-05,
+      "loss": 4.8481,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279786098706646e-05,
+      "loss": 5.0904,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275882711151658e-05,
+      "loss": 5.8117,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927196917176776e-05,
+      "loss": 5.0862,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926804548483806e-05,
+      "loss": 5.7194,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264111654656763e-05,
+      "loss": 6.022,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260167685529184e-05,
+      "loss": 5.3137,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925621358177173e-05,
+      "loss": 5.5181,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252249347711906e-05,
+      "loss": 5.3401,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248274987688296e-05,
+      "loss": 5.1992,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924429050605057e-05,
+      "loss": 5.5666,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240295907159475e-05,
+      "loss": 5.2022,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236291195386828e-05,
+      "loss": 5.7954,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232276375115517e-05,
+      "loss": 5.1637,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228251450739495e-05,
+      "loss": 5.6544,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922421642666377e-05,
+      "loss": 5.4725,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220171307304404e-05,
+      "loss": 5.2713,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216116097088503e-05,
+      "loss": 5.1519,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212050800454228e-05,
+      "loss": 5.0418,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207975421850768e-05,
+      "loss": 5.2584,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203889965738354e-05,
+      "loss": 5.6763,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199794436588244e-05,
+      "loss": 6.2026,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195688838882712e-05,
+      "loss": 5.8034,
+      "step": 479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191573177115064e-05,
+      "loss": 6.0453,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187447455789612e-05,
+      "loss": 5.7255,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183311679421678e-05,
+      "loss": 5.573,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179165852537596e-05,
+      "loss": 5.0769,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917500997967469e-05,
+      "loss": 5.3072,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170844065381287e-05,
+      "loss": 4.8685,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166668114216694e-05,
+      "loss": 5.6319,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162482130751213e-05,
+      "loss": 5.2339,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915828611956612e-05,
+      "loss": 5.5129,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154080085253665e-05,
+      "loss": 5.7043,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914986403241707e-05,
+      "loss": 4.9913,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145637965670523e-05,
+      "loss": 5.4478,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141401889639167e-05,
+      "loss": 5.6672,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137155808959105e-05,
+      "loss": 5.078,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132899728277378e-05,
+      "loss": 5.424,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128633652251994e-05,
+      "loss": 5.1709,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124357585551872e-05,
+      "loss": 5.6506,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120071532856885e-05,
+      "loss": 6.0587,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115775498857827e-05,
+      "loss": 5.1837,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911146948825642e-05,
+      "loss": 5.8565,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107153505765308e-05,
+      "loss": 5.3903,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910282755610803e-05,
+      "loss": 5.4179,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909849164401906e-05,
+      "loss": 5.6284,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909414577424375e-05,
+      "loss": 5.2914,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089789951538373e-05,
+      "loss": 5.3583,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085424180670072e-05,
+      "loss": 5.5266,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081048466416897e-05,
+      "loss": 4.9609,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076662813567772e-05,
+      "loss": 5.4645,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072267226922495e-05,
+      "loss": 5.2537,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067861711291744e-05,
+      "loss": 5.6039,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906344627149706e-05,
+      "loss": 5.0773,
+      "step": 510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059020912370836e-05,
+      "loss": 5.1426,
+      "step": 511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905458563875634e-05,
+      "loss": 5.2761,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905014045550767e-05,
+      "loss": 5.8241,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045685367489786e-05,
+      "loss": 5.0804,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904122037957848e-05,
+      "loss": 6.0953,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036745496660388e-05,
+      "loss": 4.6393,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032260723632957e-05,
+      "loss": 5.7261,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902776606540447e-05,
+      "loss": 6.3172,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023261526894037e-05,
+      "loss": 6.135,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018747113031563e-05,
+      "loss": 5.4533,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9014222828757773e-05,
+      "loss": 5.6885,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 5.698,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900514466879314e-05,
+      "loss": 5.3883,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000590803037732e-05,
+      "loss": 5.2276,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899602708674187e-05,
+      "loss": 5.4838,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991453524900226e-05,
+      "loss": 4.6407,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898687012251826e-05,
+      "loss": 5.9317,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89822768846122e-05,
+      "loss": 5.3226,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897767381620903e-05,
+      "loss": 5.441,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973060922346506e-05,
+      "loss": 5.1898,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968438208073122e-05,
+      "loss": 4.9791,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896380567844813e-05,
+      "loss": 6.0431,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895916333854152e-05,
+      "loss": 5.4643,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954511193434024e-05,
+      "loss": 6.027,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89498492482171e-05,
+      "loss": 4.844,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945177507992933e-05,
+      "loss": 5.5617,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894049597787443e-05,
+      "loss": 5.5111,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935804662985204e-05,
+      "loss": 5.3795,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931103568459593e-05,
+      "loss": 4.9429,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926392699442617e-05,
+      "loss": 5.8612,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8921672061090014e-05,
+      "loss": 5.5807,
+      "step": 541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916941658568202e-05,
+      "loss": 5.6913,
+      "step": 542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912201497054286e-05,
+      "loss": 5.9102,
+      "step": 543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907451581736052e-05,
+      "loss": 5.2359,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902691917811972e-05,
+      "loss": 5.6562,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889792251049117e-05,
+      "loss": 5.1911,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893143364993437e-05,
+      "loss": 5.0219,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888354486549238e-05,
+      "loss": 5.4315,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883555880399666e-05,
+      "loss": 5.5372,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887874755179648e-05,
+      "loss": 5.329,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873929506002065e-05,
+      "loss": 5.5016,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869101748289456e-05,
+      "loss": 6.0721,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864264283942303e-05,
+      "loss": 5.1018,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859417118254884e-05,
+      "loss": 4.7081,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 5.8683,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849693704089457e-05,
+      "loss": 4.6347,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844817466253066e-05,
+      "loss": 5.1226,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883993154835965e-05,
+      "loss": 4.8749,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835035955756505e-05,
+      "loss": 5.5796,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830130693801533e-05,
+      "loss": 5.6056,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825215767863215e-05,
+      "loss": 5.5347,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820291183320602e-05,
+      "loss": 4.8679,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815356945563325e-05,
+      "loss": 5.2475,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881041305999157e-05,
+      "loss": 5.5695,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880545953201609e-05,
+      "loss": 5.5255,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800496367058187e-05,
+      "loss": 5.222,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879552357054971e-05,
+      "loss": 5.8173,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790541147933052e-05,
+      "loss": 5.3241,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785549104661135e-05,
+      "loss": 5.5303,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878054744619742e-05,
+      "loss": 5.5601,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877553617801588e-05,
+      "loss": 5.3023,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770515305601014e-05,
+      "loss": 5.506,
+      "step": 572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765484834447828e-05,
+      "loss": 5.0845,
+      "step": 573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876044477006183e-05,
+      "loss": 5.686,
+      "step": 574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755395117959034e-05,
+      "loss": 5.3264,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750335883665948e-05,
+      "loss": 5.4219,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745267072719555e-05,
+      "loss": 5.5023,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740188690667332e-05,
+      "loss": 5.2031,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873510074306723e-05,
+      "loss": 5.4548,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730003235487658e-05,
+      "loss": 5.3124,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87248961735075e-05,
+      "loss": 5.3489,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719779562716085e-05,
+      "loss": 5.5058,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714653408713204e-05,
+      "loss": 5.5099,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870951771710909e-05,
+      "loss": 5.1902,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87043724935244e-05,
+      "loss": 5.6677,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869921774359025e-05,
+      "loss": 5.4026,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694053472948154e-05,
+      "loss": 5.3938,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688879687250067e-05,
+      "loss": 5.0971,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683696392158343e-05,
+      "loss": 5.3647,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678503593345755e-05,
+      "loss": 5.3657,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8673301296495466e-05,
+      "loss": 5.237,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866808950730104e-05,
+      "loss": 5.2231,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662868231466437e-05,
+      "loss": 5.2224,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657637474705982e-05,
+      "loss": 5.2894,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652397242744392e-05,
+      "loss": 4.9046,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647147541316742e-05,
+      "loss": 5.0644,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641888376168483e-05,
+      "loss": 5.0303,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8636619753055417e-05,
+      "loss": 5.4363,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863134167774369e-05,
+      "loss": 5.8973,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626054156009807e-05,
+      "loss": 5.4797,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620757193640604e-05,
+      "loss": 5.6301,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615450796433247e-05,
+      "loss": 5.8046,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861013497019523e-05,
+      "loss": 5.515,
+      "step": 603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860480972074437e-05,
+      "loss": 5.5799,
+      "step": 604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859947505390879e-05,
+      "loss": 5.5117,
+      "step": 605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594130975526925e-05,
+      "loss": 5.5277,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858877749144751e-05,
+      "loss": 5.5402,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583414607529574e-05,
+      "loss": 5.2643,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578042329642427e-05,
+      "loss": 5.8678,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572660663665672e-05,
+      "loss": 5.2962,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567269615489172e-05,
+      "loss": 5.8732,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561869191013073e-05,
+      "loss": 5.5408,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556459396147777e-05,
+      "loss": 5.3976,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551040236813934e-05,
+      "loss": 5.6873,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854561171894245e-05,
+      "loss": 5.1869,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540173848474477e-05,
+      "loss": 5.8374,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853472663136139e-05,
+      "loss": 5.7683,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852927007356481e-05,
+      "loss": 5.7251,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852380418105657e-05,
+      "loss": 5.3651,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518328959818717e-05,
+      "loss": 4.8609,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512844415843514e-05,
+      "loss": 5.586,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507350555133425e-05,
+      "loss": 5.3998,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501847383701112e-05,
+      "loss": 5.3386,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496334907569424e-05,
+      "loss": 5.2288,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490813132771396e-05,
+      "loss": 5.9018,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485282065350237e-05,
+      "loss": 5.568,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847974171135933e-05,
+      "loss": 5.5551,
+      "step": 627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474192076862217e-05,
+      "loss": 5.8372,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468633167932598e-05,
+      "loss": 5.5272,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846306499065433e-05,
+      "loss": 4.6007,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84574875511214e-05,
+      "loss": 5.5733,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845190085543795e-05,
+      "loss": 5.687,
+      "step": 632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446304909718235e-05,
+      "loss": 5.4838,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844069972008664e-05,
+      "loss": 4.8584,
+      "step": 634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435085292677674e-05,
+      "loss": 5.0836,
+      "step": 635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429461633635943e-05,
+      "loss": 5.9206,
+      "step": 636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423828749116163e-05,
+      "loss": 6.1072,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418186645283154e-05,
+      "loss": 5.4725,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 5.3176,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406874804387126e-05,
+      "loss": 5.4354,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840120507970416e-05,
+      "loss": 5.0889,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839552616046804e-05,
+      "loss": 5.5976,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838983805289396e-05,
+      "loss": 5.6075,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838414076320718e-05,
+      "loss": 5.8046,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837843429764299e-05,
+      "loss": 5.3243,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837271866244674e-05,
+      "loss": 5.3155,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83669938638738e-05,
+      "loss": 5.353,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361259908189584e-05,
+      "loss": 4.944,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355516801669516e-05,
+      "loss": 5.4575,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834976455059904e-05,
+      "loss": 5.0001,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834400316127361e-05,
+      "loss": 5.5756,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338232639998672e-05,
+      "loss": 5.0029,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833245299308968e-05,
+      "loss": 4.895,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8326664226872063e-05,
+      "loss": 5.3797,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320866347681244e-05,
+      "loss": 5.2599,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83150593618626e-05,
+      "loss": 5.1337,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830924327577149e-05,
+      "loss": 5.2403,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303418095773234e-05,
+      "loss": 5.0438,
+      "step": 658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297583828243096e-05,
+      "loss": 5.3398,
+      "step": 659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291740479566286e-05,
+      "loss": 5.2488,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285888056137957e-05,
+      "loss": 4.9762,
+      "step": 661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828002656436319e-05,
+      "loss": 5.1471,
+      "step": 662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8274156010657e-05,
+      "loss": 5.87,
+      "step": 663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268276401444306e-05,
+      "loss": 5.2467,
+      "step": 664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 5.1873,
+      "step": 665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256490042248668e-05,
+      "loss": 5.6211,
+      "step": 666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250583305165098e-05,
+      "loss": 5.2212,
+      "step": 667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824466753837376e-05,
+      "loss": 5.469,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823874274834907e-05,
+      "loss": 5.1473,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823280894157531e-05,
+      "loss": 4.928,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822686612454663e-05,
+      "loss": 4.9096,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220914303767047e-05,
+      "loss": 5.2418,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821495348575042e-05,
+      "loss": 5.426,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820898367702047e-05,
+      "loss": 5.5417,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820300488411075e-05,
+      "loss": 5.0751,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819701711356464e-05,
+      "loss": 5.7463,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191020371935363e-05,
+      "loss": 5.2119,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185014665785936e-05,
+      "loss": 5.5146,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179000001689207e-05,
+      "loss": 5.7423,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817297638622782e-05,
+      "loss": 5.5143,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816694382599422e-05,
+      "loss": 5.2286,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816090232759063e-05,
+      "loss": 5.4669,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154851897629065e-05,
+      "loss": 5.4148,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148792542731322e-05,
+      "loss": 5.4881,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8142724269528944e-05,
+      "loss": 4.7664,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136647084663258e-05,
+      "loss": 5.5262,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130560994785325e-05,
+      "loss": 4.8627,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124466006555967e-05,
+      "loss": 5.0584,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118362126645736e-05,
+      "loss": 4.6303,
+      "step": 689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112249361734914e-05,
+      "loss": 5.991,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810612771851352e-05,
+      "loss": 5.2386,
+      "step": 691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099997203681264e-05,
+      "loss": 5.664,
+      "step": 692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093857823947598e-05,
+      "loss": 5.2677,
+      "step": 693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808770958603165e-05,
+      "loss": 5.1878,
+      "step": 694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081552496662256e-05,
+      "loss": 5.5768,
+      "step": 695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807538656257794e-05,
+      "loss": 4.9638,
+      "step": 696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806921179052689e-05,
+      "loss": 5.2604,
+      "step": 697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063028187266987e-05,
+      "loss": 4.8481,
+      "step": 698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805683575956577e-05,
+      "loss": 5.1362,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050634514200423e-05,
+      "loss": 5.1505,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044424457957804e-05,
+      "loss": 5.5548,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038205597634392e-05,
+      "loss": 5.3597,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031977940036318e-05,
+      "loss": 5.2963,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025741491979326e-05,
+      "loss": 4.97,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019496260288795e-05,
+      "loss": 5.4965,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801324225179971e-05,
+      "loss": 5.4298,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800697947335666e-05,
+      "loss": 5.4997,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000707931813835e-05,
+      "loss": 5.0289,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7994427634035016e-05,
+      "loss": 5.1102,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7988138586893556e-05,
+      "loss": 5.6927,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798184079727241e-05,
+      "loss": 5.8337,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975534272064076e-05,
+      "loss": 5.511,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969219018170618e-05,
+      "loss": 4.9349,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962895042503663e-05,
+      "loss": 5.4191,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795656235198437e-05,
+      "loss": 5.1395,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795022095354345e-05,
+      "loss": 5.2878,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943870854121126e-05,
+      "loss": 5.2701,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937512060667165e-05,
+      "loss": 5.5402,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793114458014083e-05,
+      "loss": 5.5687,
+      "step": 719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924768419510906e-05,
+      "loss": 5.415,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918383585755665e-05,
+      "loss": 5.4488,
+      "step": 721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7911990085862886e-05,
+      "loss": 5.4041,
+      "step": 722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905587926829815e-05,
+      "loss": 5.4082,
+      "step": 723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7899177115663186e-05,
+      "loss": 5.1113,
+      "step": 724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892757659379203e-05,
+      "loss": 5.4496,
+      "step": 725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788632956500353e-05,
+      "loss": 5.2661,
+      "step": 726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879892839571278e-05,
+      "loss": 4.8748,
+      "step": 727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873447490127015e-05,
+      "loss": 5.1638,
+      "step": 728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866993523724735e-05,
+      "loss": 5.5465,
+      "step": 729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860530947427878e-05,
+      "loss": 5.6778,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854059768309292e-05,
+      "loss": 5.4998,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847579993451246e-05,
+      "loss": 5.5828,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841091629945423e-05,
+      "loss": 5.3102,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834594684892898e-05,
+      "loss": 4.8154,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828089165404138e-05,
+      "loss": 5.6881,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821575078598996e-05,
+      "loss": 5.0055,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815052431606702e-05,
+      "loss": 5.4128,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7808521231565853e-05,
+      "loss": 5.2742,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780198148562441e-05,
+      "loss": 4.7828,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795433200939684e-05,
+      "loss": 5.1243,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788876384678325e-05,
+      "loss": 5.2997,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778231104401634e-05,
+      "loss": 4.9705,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777573718613904e-05,
+      "loss": 5.2981,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776915481824107e-05,
+      "loss": 4.6106,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77625639475264e-05,
+      "loss": 5.6805,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775596458120828e-05,
+      "loss": 5.7755,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7749356726509286e-05,
+      "loss": 5.447,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7742740390661256e-05,
+      "loss": 4.7257,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736115580905335e-05,
+      "loss": 5.0536,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7729482304491927e-05,
+      "loss": 5.1614,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722840568680704e-05,
+      "loss": 5.5444,
+      "step": 751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77161903807406e-05,
+      "loss": 5.1274,
+      "step": 752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7709531747949796e-05,
+      "loss": 5.3046,
+      "step": 753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702864677595723e-05,
+      "loss": 5.1192,
+      "step": 754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696189176975037e-05,
+      "loss": 5.6285,
+      "step": 755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768950525339362e-05,
+      "loss": 5.4723,
+      "step": 756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682812914166586e-05,
+      "loss": 5.3738,
+      "step": 757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676112166618236e-05,
+      "loss": 5.542,
+      "step": 758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669403018082102e-05,
+      "loss": 5.2168,
+      "step": 759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662685475900882e-05,
+      "loss": 5.3207,
+      "step": 760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7655959547426484e-05,
+      "loss": 4.7926,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649225240019974e-05,
+      "loss": 5.3133,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642482561051605e-05,
+      "loss": 5.1031,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635731517900782e-05,
+      "loss": 5.1474,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762897211795607e-05,
+      "loss": 5.2085,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622204368615176e-05,
+      "loss": 4.9216,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615428277284937e-05,
+      "loss": 5.4598,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608643851381344e-05,
+      "loss": 5.7588,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7601851098329484e-05,
+      "loss": 5.4233,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759505002556357e-05,
+      "loss": 5.0015,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758824064052692e-05,
+      "loss": 5.4576,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581422950671942e-05,
+      "loss": 5.0575,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574596963460145e-05,
+      "loss": 5.1211,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756776268636211e-05,
+      "loss": 4.5395,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756092012685749e-05,
+      "loss": 5.7545,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7554069292435012e-05,
+      "loss": 5.0344,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547210190592446e-05,
+      "loss": 5.2662,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754034282883662e-05,
+      "loss": 5.243,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533467214683394e-05,
+      "loss": 5.7569,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526583355657675e-05,
+      "loss": 5.015,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519691259293377e-05,
+      "loss": 4.8673,
+      "step": 781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7512790933133435e-05,
+      "loss": 5.3919,
+      "step": 782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505882384729793e-05,
+      "loss": 4.9753,
+      "step": 783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749896562164339e-05,
+      "loss": 4.9479,
+      "step": 784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749204065144416e-05,
+      "loss": 5.7124,
+      "step": 785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 4.7567,
+      "step": 786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7478166120031838e-05,
+      "loss": 5.2022,
+      "step": 787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7471216574003488e-05,
+      "loss": 5.668,
+      "step": 788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464258851231774e-05,
+      "loss": 5.1876,
+      "step": 789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7457292959331455e-05,
+      "loss": 4.9557,
+      "step": 790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450318905926227e-05,
+      "loss": 5.022,
+      "step": 791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7443336698648728e-05,
+      "loss": 5.0482,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743634634514051e-05,
+      "loss": 5.2632,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429347853052036e-05,
+      "loss": 5.3865,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74223412300427e-05,
+      "loss": 5.4536,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741532648378077e-05,
+      "loss": 4.6899,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408303621943417e-05,
+      "loss": 5.3472,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401272652216686e-05,
+      "loss": 5.6949,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73942335822955e-05,
+      "loss": 5.5201,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738718641988365e-05,
+      "loss": 5.4359,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738013117269378e-05,
+      "loss": 5.3639,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737306784844738e-05,
+      "loss": 5.2194,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736599645487479e-05,
+      "loss": 5.2366,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358916999715164e-05,
+      "loss": 5.0215,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7351829490716498e-05,
+      "loss": 5.3718,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7344733935635586e-05,
+      "loss": 5.5891,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733763034223804e-05,
+      "loss": 5.3203,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 5.3335,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323399071599453e-05,
+      "loss": 5.449,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731627140993357e-05,
+      "loss": 4.9075,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7309135741101376e-05,
+      "loss": 5.1489,
+      "step": 811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7301992072912372e-05,
+      "loss": 5.8455,
+      "step": 812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294840413184818e-05,
+      "loss": 5.6709,
+      "step": 813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728768076974573e-05,
+      "loss": 5.6517,
+      "step": 814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280513150430856e-05,
+      "loss": 5.3582,
+      "step": 815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727333756308467e-05,
+      "loss": 5.3793,
+      "step": 816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266154015560373e-05,
+      "loss": 5.2881,
+      "step": 817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7258962515719867e-05,
+      "loss": 5.1702,
+      "step": 818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251763071433767e-05,
+      "loss": 4.9943,
+      "step": 819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244555690581382e-05,
+      "loss": 5.3293,
+      "step": 820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72373403810507e-05,
+      "loss": 5.6325,
+      "step": 821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230117150738403e-05,
+      "loss": 5.4688,
+      "step": 822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7222886007549817e-05,
+      "loss": 4.968,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721564695939895e-05,
+      "loss": 5.504,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208400014208445e-05,
+      "loss": 4.8441,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72011451799096e-05,
+      "loss": 4.8649,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193882464442343e-05,
+      "loss": 5.0812,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186611875755227e-05,
+      "loss": 5.3846,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717933342180542e-05,
+      "loss": 5.5867,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71720471105587e-05,
+      "loss": 5.284,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7164752949989444e-05,
+      "loss": 5.2402,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715745094808062e-05,
+      "loss": 5.1778,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150141112823778e-05,
+      "loss": 4.5919,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142823452219036e-05,
+      "loss": 5.4338,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135497974275088e-05,
+      "loss": 5.2938,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712816468700917e-05,
+      "loss": 5.1408,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120823598447077e-05,
+      "loss": 5.1271,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7113474716623128e-05,
+      "loss": 4.9855,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7106118049580187e-05,
+      "loss": 5.9474,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7098753605369624e-05,
+      "loss": 5.1687,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091381392051333e-05,
+      "loss": 5.6371,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084001417693702e-05,
+      "loss": 5.0552,
+      "step": 842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7076613690373615e-05,
+      "loss": 5.1417,
+      "step": 843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7069218218176444e-05,
+      "loss": 5.5534,
+      "step": 844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061815009196038e-05,
+      "loss": 4.9545,
+      "step": 845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7054404071534702e-05,
+      "loss": 5.3138,
+      "step": 846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046985413303215e-05,
+      "loss": 5.1759,
+      "step": 847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70395590426208e-05,
+      "loss": 6.0155,
+      "step": 848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7032124967615112e-05,
+      "loss": 5.2462,
+      "step": 849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024683196422255e-05,
+      "loss": 4.9173,
+      "step": 850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7017233737186737e-05,
+      "loss": 5.1785,
+      "step": 851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009776598061496e-05,
+      "loss": 5.6221,
+      "step": 852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700231178720787e-05,
+      "loss": 5.4612,
+      "step": 853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994839312795578e-05,
+      "loss": 5.0423,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698735918300276e-05,
+      "loss": 5.3052,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69798714060159e-05,
+      "loss": 5.5842,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697237599002987e-05,
+      "loss": 5.2446,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964872943247902e-05,
+      "loss": 5.3414,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957362273881572e-05,
+      "loss": 5.4239,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949843990150798e-05,
+      "loss": 4.8978,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6942318100283837e-05,
+      "loss": 5.3718,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693478461251727e-05,
+      "loss": 5.0032,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927243535095995e-05,
+      "loss": 5.7803,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691969487627321e-05,
+      "loss": 5.7153,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912138644310415e-05,
+      "loss": 4.4574,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6904574847477392e-05,
+      "loss": 5.4424,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897003494052217e-05,
+      "loss": 5.4463,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6889424592321216e-05,
+      "loss": 5.5298,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6881838150578994e-05,
+      "loss": 5.6594,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874244177128395e-05,
+      "loss": 5.8581,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866642680280516e-05,
+      "loss": 5.4232,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859033668354684e-05,
+      "loss": 4.6907,
+      "step": 872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851417149678442e-05,
+      "loss": 5.733,
+      "step": 873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684379313258757e-05,
+      "loss": 5.4404,
+      "step": 874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836161625426028e-05,
+      "loss": 5.2226,
+      "step": 875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6828522636545992e-05,
+      "loss": 5.8068,
+      "step": 876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682087617430782e-05,
+      "loss": 5.3048,
+      "step": 877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681322224708006e-05,
+      "loss": 5.7264,
+      "step": 878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805560863239403e-05,
+      "loss": 5.3599,
+      "step": 879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679789203117073e-05,
+      "loss": 5.3591,
+      "step": 880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679021575926705e-05,
+      "loss": 4.7874,
+      "step": 881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678253205592954e-05,
+      "loss": 5.3179,
+      "step": 882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6774840929567485e-05,
+      "loss": 4.9787,
+      "step": 883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676714238859831e-05,
+      "loss": 5.2381,
+      "step": 884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6759436441447544e-05,
+      "loss": 5.0632,
+      "step": 885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6751723096548834e-05,
+      "loss": 4.693,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744002362343913e-05,
+      "loss": 4.7287,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6736274247282605e-05,
+      "loss": 5.3859,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672853875982281e-05,
+      "loss": 5.7829,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6720795908430497e-05,
+      "loss": 4.7737,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713045701579705e-05,
+      "loss": 4.6038,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6705288147752502e-05,
+      "loss": 5.7575,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669752325543901e-05,
+      "loss": 4.7589,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668975103313739e-05,
+      "loss": 5.1326,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6681971489353805e-05,
+      "loss": 5.5436,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674184632602447e-05,
+      "loss": 5.4478,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6666390471405504e-05,
+      "loss": 5.0867,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665858901429316e-05,
+      "loss": 5.1199,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6650780269803587e-05,
+      "loss": 5.371,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664296424648293e-05,
+      "loss": 5.7879,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635140952885298e-05,
+      "loss": 5.1653,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627310397572762e-05,
+      "loss": 5.2919,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619472589115334e-05,
+      "loss": 5.1469,
+      "step": 903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611627536090967e-05,
+      "loss": 5.5859,
+      "step": 904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660377524708555e-05,
+      "loss": 5.1974,
+      "step": 905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6595915730692876e-05,
+      "loss": 5.3049,
+      "step": 906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6588048995514657e-05,
+      "loss": 5.843,
+      "step": 907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658017505016051e-05,
+      "loss": 5.0835,
+      "step": 908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657229390324794e-05,
+      "loss": 5.2163,
+      "step": 909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564405563402322e-05,
+      "loss": 5.1366,
+      "step": 910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655651003925692e-05,
+      "loss": 5.7192,
+      "step": 911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 4.7482,
+      "step": 912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654069747263909e-05,
+      "loss": 5.5648,
+      "step": 913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653278044747245e-05,
+      "loss": 5.6641,
+      "step": 914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524856272617582e-05,
+      "loss": 5.3765,
+      "step": 915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6516924956746963e-05,
+      "loss": 5.5352,
+      "step": 916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6508986508540884e-05,
+      "loss": 5.7084,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6501040936687444e-05,
+      "loss": 5.0796,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6493088249882533e-05,
+      "loss": 5.4057,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485128456829835e-05,
+      "loss": 5.2352,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64771615662408e-05,
+      "loss": 5.074,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646918758683466e-05,
+      "loss": 5.4777,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461206527338397e-05,
+      "loss": 5.4299,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453218396486744e-05,
+      "loss": 4.6229,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644522320302217e-05,
+      "loss": 5.4563,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643722095569487e-05,
+      "loss": 5.8901,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642921166326278e-05,
+      "loss": 6.0232,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421195334491516e-05,
+      "loss": 4.7125,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641317197815442e-05,
+      "loss": 4.7826,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405141603032512e-05,
+      "loss": 5.3212,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397104217914496e-05,
+      "loss": 5.6565,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638905983159675e-05,
+      "loss": 4.6067,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638100845288331e-05,
+      "loss": 5.6763,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372950090585868e-05,
+      "loss": 5.2652,
+      "step": 934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364884753523765e-05,
+      "loss": 4.9159,
+      "step": 935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6356812450523962e-05,
+      "loss": 4.9952,
+      "step": 936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348733190421055e-05,
+      "loss": 4.8927,
+      "step": 937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634064698205725e-05,
+      "loss": 5.3544,
+      "step": 938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6332553834282352e-05,
+      "loss": 5.4248,
+      "step": 939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324453755953772e-05,
+      "loss": 5.329,
+      "step": 940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63163467559365e-05,
+      "loss": 5.0095,
+      "step": 941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6308232843103102e-05,
+      "loss": 4.6932,
+      "step": 942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63001120263337e-05,
+      "loss": 5.2144,
+      "step": 943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291984314516e-05,
+      "loss": 4.8207,
+      "step": 944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283849716545217e-05,
+      "loss": 4.9841,
+      "step": 945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275708241324128e-05,
+      "loss": 4.8411,
+      "step": 946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626755989776303e-05,
+      "loss": 5.1419,
+      "step": 947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625940469477973e-05,
+      "loss": 4.8427,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251242641299557e-05,
+      "loss": 4.7908,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243073746255325e-05,
+      "loss": 5.1104,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 4.8471,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622671546724338e-05,
+      "loss": 4.6093,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218526101178707e-05,
+      "loss": 4.8199,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210329929356017e-05,
+      "loss": 5.3694,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202126960745484e-05,
+      "loss": 4.6345,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193917204324697e-05,
+      "loss": 5.56,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6185700669078674e-05,
+      "loss": 4.9259,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617747736399987e-05,
+      "loss": 5.0963,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169247298088136e-05,
+      "loss": 5.3194,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161010480350725e-05,
+      "loss": 5.2244,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6152766919802275e-05,
+      "loss": 6.0543,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6144516625464813e-05,
+      "loss": 5.1469,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6136259606367727e-05,
+      "loss": 5.5881,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612799587154777e-05,
+      "loss": 5.2376,
+      "step": 964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6119725430049045e-05,
+      "loss": 5.0323,
+      "step": 965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611144829092299e-05,
+      "loss": 5.3893,
+      "step": 966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6103164463228377e-05,
+      "loss": 5.1956,
+      "step": 967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60948739560313e-05,
+      "loss": 5.227,
+      "step": 968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086576778405156e-05,
+      "loss": 4.9657,
+      "step": 969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607827293943065e-05,
+      "loss": 5.6051,
+      "step": 970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069962448195778e-05,
+      "loss": 5.0406,
+      "step": 971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6061645313795805e-05,
+      "loss": 5.5661,
+      "step": 972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605332154533328e-05,
+      "loss": 5.5688,
+      "step": 973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604499115191801e-05,
+      "loss": 4.6433,
+      "step": 974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6036654142667043e-05,
+      "loss": 5.0352,
+      "step": 975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028310526704678e-05,
+      "loss": 4.9061,
+      "step": 976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6019960313162436e-05,
+      "loss": 5.5943,
+      "step": 977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6011603511179067e-05,
+      "loss": 5.163,
+      "step": 978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003240129900528e-05,
+      "loss": 5.4297,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599487017847998e-05,
+      "loss": 5.1729,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986493666077763e-05,
+      "loss": 5.4179,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5978110601861408e-05,
+      "loss": 5.8035,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596972099500562e-05,
+      "loss": 4.9084,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5961324854692254e-05,
+      "loss": 5.6465,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5952922190110322e-05,
+      "loss": 4.8062,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5944513010455973e-05,
+      "loss": 5.4547,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936097324932487e-05,
+      "loss": 5.1981,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927675142750264e-05,
+      "loss": 4.9893,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5919246473126825e-05,
+      "loss": 5.3153,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910811325286768e-05,
+      "loss": 5.0187,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59023697084618e-05,
+      "loss": 5.2819,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58939216318907e-05,
+      "loss": 5.2459,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885467104819326e-05,
+      "loss": 4.8135,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587700613650058e-05,
+      "loss": 4.9549,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868538736194427e-05,
+      "loss": 4.7911,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5860064913167863e-05,
+      "loss": 4.8176,
+      "step": 996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585158467669492e-05,
+      "loss": 4.9403,
+      "step": 997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584309803605664e-05,
+      "loss": 5.1862,
+      "step": 998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834605000541088e-05,
+      "loss": 6.0152,
+      "step": 999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5826105579443314e-05,
+      "loss": 5.1954,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5817599782065364e-05,
+      "loss": 5.5539,
+      "step": 1001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5809087617716255e-05,
+      "loss": 5.9387,
+      "step": 1002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800569095711983e-05,
+      "loss": 5.5961,
+      "step": 1003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57920442253755e-05,
+      "loss": 5.3307,
+      "step": 1004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5783513016036693e-05,
+      "loss": 5.5618,
+      "step": 1005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57749754770324e-05,
+      "loss": 5.2745,
+      "step": 1006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766431617706387e-05,
+      "loss": 5.4927,
+      "step": 1007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575788144740933e-05,
+      "loss": 5.1563,
+      "step": 1008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574932497549881e-05,
+      "loss": 5.676,
+      "step": 1009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5740762211339314e-05,
+      "loss": 4.8081,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732193164302212e-05,
+      "loss": 5.2442,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572361784376574e-05,
+      "loss": 5.2197,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715036259115022e-05,
+      "loss": 5.3626,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5706448419742014e-05,
+      "loss": 5.2727,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697854335045534e-05,
+      "loss": 5.2437,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689254014431225e-05,
+      "loss": 4.9385,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 5.788,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672034703105822e-05,
+      "loss": 5.1206,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566341573124011e-05,
+      "loss": 5.295,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56547905611473e-05,
+      "loss": 5.0552,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5646159202267064e-05,
+      "loss": 5.4433,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563752166404584e-05,
+      "loss": 5.3993,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5628877955936833e-05,
+      "loss": 5.3923,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620228087400004e-05,
+      "loss": 4.5872,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611572067902042e-05,
+      "loss": 5.1961,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602909906916385e-05,
+      "loss": 5.3667,
+      "step": 1026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559424161392318e-05,
+      "loss": 5.4144,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55855671984093e-05,
+      "loss": 4.6756,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576886669868297e-05,
+      "loss": 4.8076,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556820003780043e-05,
+      "loss": 4.8578,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5559507311712626e-05,
+      "loss": 5.3028,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5550808501118497e-05,
+      "loss": 5.1985,
+      "step": 1032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542103615538304e-05,
+      "loss": 5.0974,
+      "step": 1033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533392664498954e-05,
+      "loss": 5.2687,
+      "step": 1034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5524675657534e-05,
+      "loss": 5.5768,
+      "step": 1035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5515952604183618e-05,
+      "loss": 5.3478,
+      "step": 1036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55072235139946e-05,
+      "loss": 5.1798,
+      "step": 1037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549848839652035e-05,
+      "loss": 5.2626,
+      "step": 1038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5489747261320866e-05,
+      "loss": 5.0025,
+      "step": 1039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548100011796273e-05,
+      "loss": 5.2976,
+      "step": 1040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472246976019104e-05,
+      "loss": 5.1774,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463487845069708e-05,
+      "loss": 5.4544,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454722734700823e-05,
+      "loss": 4.9301,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445951654505276e-05,
+      "loss": 4.8776,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437174614082418e-05,
+      "loss": 5.2563,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542839162303813e-05,
+      "loss": 5.1019,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419602690984805e-05,
+      "loss": 5.2514,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5410807827541338e-05,
+      "loss": 5.3718,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5402007042333113e-05,
+      "loss": 5.0391,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5393200344991993e-05,
+      "loss": 5.0215,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5384387745156323e-05,
+      "loss": 5.6943,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375569252470897e-05,
+      "loss": 5.16,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536674487658696e-05,
+      "loss": 5.5103,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53579146271622e-05,
+      "loss": 4.7369,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349078513860728e-05,
+      "loss": 5.5024,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5340236546353067e-05,
+      "loss": 4.9048,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331388734316173e-05,
+      "loss": 5.2931,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5322535087433367e-05,
+      "loss": 5.4055,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313675615394373e-05,
+      "loss": 5.0189,
+      "step": 1059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304810327895287e-05,
+      "loss": 5.0885,
+      "step": 1060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295939234638566e-05,
+      "loss": 5.3797,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528706234533303e-05,
+      "loss": 5.1285,
+      "step": 1062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278179669693828e-05,
+      "loss": 5.0539,
+      "step": 1063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526929121744246e-05,
+      "loss": 5.41,
+      "step": 1064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526039699830673e-05,
+      "loss": 4.9828,
+      "step": 1065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5251497022020764e-05,
+      "loss": 4.9217,
+      "step": 1066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524259129832499e-05,
+      "loss": 5.2512,
+      "step": 1067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5233679836966122e-05,
+      "loss": 5.0339,
+      "step": 1068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5224762647697153e-05,
+      "loss": 5.1233,
+      "step": 1069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215839740277342e-05,
+      "loss": 5.0003,
+      "step": 1070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206911124472217e-05,
+      "loss": 5.5443,
+      "step": 1071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197976810053546e-05,
+      "loss": 5.6363,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189036806799333e-05,
+      "loss": 5.6182,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180091124493805e-05,
+      "loss": 5.3924,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171139772927415e-05,
+      "loss": 4.7445,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5162182761896808e-05,
+      "loss": 4.7621,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153220101204839e-05,
+      "loss": 5.4392,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514425180066053e-05,
+      "loss": 5.7834,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135277870079082e-05,
+      "loss": 5.2813,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126298319281859e-05,
+      "loss": 4.8336,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117313158096371e-05,
+      "loss": 5.4568,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.510832239635628e-05,
+      "loss": 5.3143,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5099326043901361e-05,
+      "loss": 5.0802,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509032411057752e-05,
+      "loss": 5.2395,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5081316606236764e-05,
+      "loss": 5.5049,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50723035407372e-05,
+      "loss": 4.8854,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506328492394303e-05,
+      "loss": 5.1871,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054260765724511e-05,
+      "loss": 5.4402,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504523107595798e-05,
+      "loss": 4.8296,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036195864525827e-05,
+      "loss": 5.1363,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5027155141316479e-05,
+      "loss": 5.2168,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5018108916224404e-05,
+      "loss": 4.9145,
+      "step": 1092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500905719915008e-05,
+      "loss": 4.8859,
+      "step": 1093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 5.2689,
+      "step": 1094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4990937328686665e-05,
+      "loss": 5.2569,
+      "step": 1095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981869195128557e-05,
+      "loss": 4.8897,
+      "step": 1096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972795609250141e-05,
+      "loss": 5.57,
+      "step": 1097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4963716580981836e-05,
+      "loss": 5.6917,
+      "step": 1098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954632120260034e-05,
+      "loss": 5.1033,
+      "step": 1099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494554223702706e-05,
+      "loss": 4.7082,
+      "step": 1100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936446941231186e-05,
+      "loss": 5.2319,
+      "step": 1101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927346242826597e-05,
+      "loss": 4.7165,
+      "step": 1102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918240151773393e-05,
+      "loss": 4.9807,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4909128678037581e-05,
+      "loss": 5.6086,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.490001183159105e-05,
+      "loss": 5.7596,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890889622411582e-05,
+      "loss": 5.0485,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4881762060482814e-05,
+      "loss": 5.3573,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872629155794251e-05,
+      "loss": 4.8693,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863490918341241e-05,
+      "loss": 5.2144,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854347358124969e-05,
+      "loss": 5.1952,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.484519848515245e-05,
+      "loss": 4.561,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836044309436504e-05,
+      "loss": 5.0016,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482688484099576e-05,
+      "loss": 4.8499,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817720089854644e-05,
+      "loss": 5.3619,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4808550066043353e-05,
+      "loss": 5.4856,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799374779597866e-05,
+      "loss": 5.5355,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790194240559911e-05,
+      "loss": 4.9194,
+      "step": 1117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4781008458976972e-05,
+      "loss": 5.2108,
+      "step": 1118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771817444902273e-05,
+      "loss": 4.9892,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476262120839475e-05,
+      "loss": 5.1553,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753419759519074e-05,
+      "loss": 5.6352,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744213108345605e-05,
+      "loss": 5.5793,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4735001264950406e-05,
+      "loss": 5.3461,
+      "step": 1123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4725784239415214e-05,
+      "loss": 5.8831,
+      "step": 1124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4716562041827451e-05,
+      "loss": 5.4436,
+      "step": 1125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707334682280186e-05,
+      "loss": 5.0783,
+      "step": 1126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698102170872142e-05,
+      "loss": 5.4177,
+      "step": 1127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688864517707682e-05,
+      "loss": 5.1168,
+      "step": 1128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4679621732896796e-05,
+      "loss": 5.2188,
+      "step": 1129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467037382655509e-05,
+      "loss": 5.6484,
+      "step": 1130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661120808803771e-05,
+      "loss": 5.063,
+      "step": 1131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651862689769646e-05,
+      "loss": 5.215,
+      "step": 1132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642599479585106e-05,
+      "loss": 5.1374,
+      "step": 1133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633331188388107e-05,
+      "loss": 4.7817,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624057826322169e-05,
+      "loss": 5.2558,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614779403536366e-05,
+      "loss": 4.9512,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4605495930185303e-05,
+      "loss": 5.3208,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4596207416429119e-05,
+      "loss": 4.8476,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4586913872433461e-05,
+      "loss": 5.087,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4577615308369495e-05,
+      "loss": 5.1913,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456831173441387e-05,
+      "loss": 5.4371,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4559003160748719e-05,
+      "loss": 5.2031,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549689597561652e-05,
+      "loss": 4.9489,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4540371055045735e-05,
+      "loss": 5.2792,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531047543399485e-05,
+      "loss": 4.4508,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4521719072826858e-05,
+      "loss": 5.6152,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4512385653537238e-05,
+      "loss": 4.6536,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503047295745423e-05,
+      "loss": 5.4616,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493704009671614e-05,
+      "loss": 5.2848,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4484355805541413e-05,
+      "loss": 5.0954,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475002693585798e-05,
+      "loss": 5.4551,
+      "step": 1151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465644684041116e-05,
+      "loss": 4.9894,
+      "step": 1152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456281787149084e-05,
+      "loss": 4.507,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446914013156762e-05,
+      "loss": 5.1341,
+      "step": 1154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4437541372316543e-05,
+      "loss": 5.1967,
+      "step": 1155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428163874886154e-05,
+      "loss": 5.7001,
+      "step": 1156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418781531128636e-05,
+      "loss": 5.4433,
+      "step": 1157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409394351312329e-05,
+      "loss": 5.0393,
+      "step": 1158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400002345710871e-05,
+      "loss": 4.8125,
+      "step": 1159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.439060552460318e-05,
+      "loss": 5.0623,
+      "step": 1160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4381203898273442e-05,
+      "loss": 5.2254,
+      "step": 1161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4371797477011104e-05,
+      "loss": 5.788,
+      "step": 1162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362386271110864e-05,
+      "loss": 5.2643,
+      "step": 1163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435297029087265e-05,
+      "loss": 5.1987,
+      "step": 1164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343549546601618e-05,
+      "loss": 5.8559,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433412404860814e-05,
+      "loss": 5.0148,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324693807207785e-05,
+      "loss": 4.9229,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4315258832721321e-05,
+      "loss": 4.7785,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305819135474692e-05,
+      "loss": 4.9925,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429637472579901e-05,
+      "loss": 5.2576,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286925614030542e-05,
+      "loss": 5.1927,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427747181051071e-05,
+      "loss": 5.2959,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4268013325586063e-05,
+      "loss": 4.9987,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258550169608276e-05,
+      "loss": 4.7928,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249082352934134e-05,
+      "loss": 5.7524,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239609885925525e-05,
+      "loss": 5.0545,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423013277894943e-05,
+      "loss": 5.2359,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42206510423779e-05,
+      "loss": 5.516,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421116468658806e-05,
+      "loss": 4.9595,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201673721962084e-05,
+      "loss": 5.5467,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4192178158887198e-05,
+      "loss": 5.0348,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4182678007755653e-05,
+      "loss": 5.2111,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173173278964726e-05,
+      "loss": 5.521,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4163663982916702e-05,
+      "loss": 4.991,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 5.1392,
+      "step": 1185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4144631730683487e-05,
+      "loss": 5.3329,
+      "step": 1186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135108795327815e-05,
+      "loss": 5.3169,
+      "step": 1187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412558133437406e-05,
+      "loss": 5.0944,
+      "step": 1188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116049358249386e-05,
+      "loss": 5.017,
+      "step": 1189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4106512877385898e-05,
+      "loss": 5.539,
+      "step": 1190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096971902220635e-05,
+      "loss": 5.3821,
+      "step": 1191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4087426443195549e-05,
+      "loss": 4.7846,
+      "step": 1192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077876510757502e-05,
+      "loss": 5.084,
+      "step": 1193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068322115358255e-05,
+      "loss": 5.5383,
+      "step": 1194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058763267454447e-05,
+      "loss": 5.1855,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049199977507595e-05,
+      "loss": 4.8645,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039632255984078e-05,
+      "loss": 4.9856,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4030060113355118e-05,
+      "loss": 5.5759,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402048356009678e-05,
+      "loss": 4.7834,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401090260668996e-05,
+      "loss": 4.6414,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4001317263620366e-05,
+      "loss": 5.5301,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399172754137851e-05,
+      "loss": 5.6045,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982133450459699e-05,
+      "loss": 5.0482,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972535001364014e-05,
+      "loss": 5.5105,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962932204596316e-05,
+      "loss": 5.2909,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953325070666215e-05,
+      "loss": 5.7185,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394371361008808e-05,
+      "loss": 5.3463,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3934097833381e-05,
+      "loss": 5.0331,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3924477751068792e-05,
+      "loss": 5.3888,
+      "step": 1209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3914853373679997e-05,
+      "loss": 5.0406,
+      "step": 1210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905224711747844e-05,
+      "loss": 5.7974,
+      "step": 1211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3895591775810255e-05,
+      "loss": 5.7021,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388595457640983e-05,
+      "loss": 5.1425,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876313124093837e-05,
+      "loss": 5.606,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3866667429414188e-05,
+      "loss": 5.1917,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3857017502927457e-05,
+      "loss": 4.6696,
+      "step": 1216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3847363355194831e-05,
+      "loss": 5.1887,
+      "step": 1217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3837704996782127e-05,
+      "loss": 4.9903,
+      "step": 1218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828042438259768e-05,
+      "loss": 5.3093,
+      "step": 1219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3818375690202774e-05,
+      "loss": 5.1576,
+      "step": 1220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808704763190748e-05,
+      "loss": 4.5165,
+      "step": 1221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799029667807871e-05,
+      "loss": 5.1624,
+      "step": 1222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3789350414642879e-05,
+      "loss": 5.6895,
+      "step": 1223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779667014289067e-05,
+      "loss": 5.3834,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769979477344262e-05,
+      "loss": 5.1839,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3760287814410822e-05,
+      "loss": 4.8007,
+      "step": 1226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375059203609562e-05,
+      "loss": 5.3282,
+      "step": 1227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374089215301003e-05,
+      "loss": 5.5233,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3731188175769923e-05,
+      "loss": 5.4167,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721480114995647e-05,
+      "loss": 5.5174,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3711767981312021e-05,
+      "loss": 4.9862,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702051785348323e-05,
+      "loss": 5.2894,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692331537738271e-05,
+      "loss": 5.6331,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682607249120024e-05,
+      "loss": 5.2932,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672878930136162e-05,
+      "loss": 5.1124,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663146591433676e-05,
+      "loss": 5.33,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 5.1815,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3643669897482768e-05,
+      "loss": 5.4397,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633925563550276e-05,
+      "loss": 4.7793,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624177252530994e-05,
+      "loss": 5.0236,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614424975093794e-05,
+      "loss": 4.7638,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604668741911884e-05,
+      "loss": 5.2216,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594908563662802e-05,
+      "loss": 5.1676,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3585144451028414e-05,
+      "loss": 5.6134,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575376414694876e-05,
+      "loss": 4.7659,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3565604465352651e-05,
+      "loss": 5.2929,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555828613696473e-05,
+      "loss": 5.0092,
+      "step": 1247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 5.3301,
+      "step": 1248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536265246242575e-05,
+      "loss": 5.4262,
+      "step": 1249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526477751855645e-05,
+      "loss": 4.8284,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3516686397976317e-05,
+      "loss": 5.5501,
+      "step": 1251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506891195320571e-05,
+      "loss": 5.2313,
+      "step": 1252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3497092154608592e-05,
+      "loss": 5.1631,
+      "step": 1253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3487289286564775e-05,
+      "loss": 5.0007,
+      "step": 1254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34774826019177e-05,
+      "loss": 5.3228,
+      "step": 1255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467672111400121e-05,
+      "loss": 5.658,
+      "step": 1256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3457857825748959e-05,
+      "loss": 4.8321,
+      "step": 1257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448039755705283e-05,
+      "loss": 5.2913,
+      "step": 1258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438217912014318e-05,
+      "loss": 5.6164,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3428392305425408e-05,
+      "loss": 5.7248,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418562946692018e-05,
+      "loss": 5.537,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408729846571716e-05,
+      "loss": 4.8351,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398893015826166e-05,
+      "loss": 4.9485,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3389052465221126e-05,
+      "loss": 5.5652,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379208205526411e-05,
+      "loss": 4.9113,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369360247515899e-05,
+      "loss": 5.24,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359508601967517e-05,
+      "loss": 5.6333,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349653279663223e-05,
+      "loss": 5.678,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339794291389015e-05,
+      "loss": 4.6867,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329931647934884e-05,
+      "loss": 5.7029,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332006536009483e-05,
+      "loss": 5.1677,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310195438666842e-05,
+      "loss": 4.9195,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300321894452877e-05,
+      "loss": 5.1314,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3290444738258876e-05,
+      "loss": 5.2474,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280563980894716e-05,
+      "loss": 5.3227,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3270679633174219e-05,
+      "loss": 4.842,
+      "step": 1276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3260791705915138e-05,
+      "loss": 4.9719,
+      "step": 1277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3250900209939144e-05,
+      "loss": 5.0415,
+      "step": 1278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241005156071814e-05,
+      "loss": 5.2236,
+      "step": 1279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323110655514262e-05,
+      "loss": 5.2174,
+      "step": 1280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3221204417984907e-05,
+      "loss": 5.2954,
+      "step": 1281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3211298755435903e-05,
+      "loss": 4.9414,
+      "step": 1282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320138957833669e-05,
+      "loss": 5.5273,
+      "step": 1283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191476897532191e-05,
+      "loss": 5.4813,
+      "step": 1284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.318156072387117e-05,
+      "loss": 4.9352,
+      "step": 1285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3171641068206208e-05,
+      "loss": 5.5353,
+      "step": 1286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3161717941393703e-05,
+      "loss": 4.9012,
+      "step": 1287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151791354293848e-05,
+      "loss": 5.2476,
+      "step": 1288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141861317770628e-05,
+      "loss": 5.1678,
+      "step": 1289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131927842691793e-05,
+      "loss": 5.1259,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121990939928862e-05,
+      "loss": 5.3073,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311205062035711e-05,
+      "loss": 5.0676,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3102106894855546e-05,
+      "loss": 5.6219,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309215977430691e-05,
+      "loss": 5.7764,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082209269597652e-05,
+      "loss": 5.5695,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307225539161793e-05,
+      "loss": 4.8968,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3062298151261592e-05,
+      "loss": 5.5122,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052337559426171e-05,
+      "loss": 5.5213,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042373627012865e-05,
+      "loss": 5.2258,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3032406364926519e-05,
+      "loss": 4.8909,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022435784075638e-05,
+      "loss": 4.7584,
+      "step": 1301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3012461895372343e-05,
+      "loss": 4.4644,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300248470973239e-05,
+      "loss": 5.1438,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299250423807514e-05,
+      "loss": 5.1262,
+      "step": 1304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982520491323537e-05,
+      "loss": 4.8788,
+      "step": 1305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972533480404126e-05,
+      "loss": 5.7905,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962543216247014e-05,
+      "loss": 5.2525,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295254970978588e-05,
+      "loss": 5.6479,
+      "step": 1308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2942552971957936e-05,
+      "loss": 4.7224,
+      "step": 1309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293255301370394e-05,
+      "loss": 4.6543,
+      "step": 1310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922549845968174e-05,
+      "loss": 5.43,
+      "step": 1311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2912543479698429e-05,
+      "loss": 5.5025,
+      "step": 1312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2902533925846006e-05,
+      "loss": 4.9862,
+      "step": 1313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2892521195365679e-05,
+      "loss": 5.1762,
+      "step": 1314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882505299215711e-05,
+      "loss": 5.0528,
+      "step": 1315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872486248357825e-05,
+      "loss": 5.8582,
+      "step": 1316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862464053757196e-05,
+      "loss": 4.8227,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852438726382448e-05,
+      "loss": 5.5658,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284241027720562e-05,
+      "loss": 4.9496,
+      "step": 1319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2832378717202174e-05,
+      "loss": 5.5484,
+      "step": 1320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282234405735098e-05,
+      "loss": 4.8963,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281230630863429e-05,
+      "loss": 4.7517,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2802265482037758e-05,
+      "loss": 5.8092,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792221588550383e-05,
+      "loss": 4.8576,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782174639164528e-05,
+      "loss": 5.2403,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772124644875907e-05,
+      "loss": 5.21,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276207161668356e-05,
+      "loss": 5.3323,
+      "step": 1327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2752015565589852e-05,
+      "loss": 5.0548,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741956502600452e-05,
+      "loss": 4.7638,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2731894438724325e-05,
+      "loss": 5.0042,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2721829384973723e-05,
+      "loss": 5.32,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711761352364172e-05,
+      "loss": 4.6454,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701690351914453e-05,
+      "loss": 5.0403,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26916163946466e-05,
+      "loss": 5.4895,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681539491585876e-05,
+      "loss": 4.9983,
+      "step": 1335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2671459653760781e-05,
+      "loss": 5.2099,
+      "step": 1336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266137689220301e-05,
+      "loss": 5.3804,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2651291217947472e-05,
+      "loss": 5.3292,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641202642032257e-05,
+      "loss": 4.7578,
+      "step": 1339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631111175498632e-05,
+      "loss": 5.5785,
+      "step": 1340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2621016829391022e-05,
+      "loss": 4.5863,
+      "step": 1341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610919614757018e-05,
+      "loss": 4.9592,
+      "step": 1342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600819542647338e-05,
+      "loss": 5.187,
+      "step": 1343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590716624115829e-05,
+      "loss": 4.4405,
+      "step": 1344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580610870219455e-05,
+      "loss": 5.8358,
+      "step": 1345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2570502292018281e-05,
+      "loss": 4.9229,
+      "step": 1346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2560390900575472e-05,
+      "loss": 4.781,
+      "step": 1347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255027670695726e-05,
+      "loss": 5.0497,
+      "step": 1348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254015972223295e-05,
+      "loss": 5.6266,
+      "step": 1349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.25300399574749e-05,
+      "loss": 5.0979,
+      "step": 1350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251991742375851e-05,
+      "loss": 5.0601,
+      "step": 1351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2509792132162213e-05,
+      "loss": 5.2294,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499664093767458e-05,
+      "loss": 5.2359,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489533319658703e-05,
+      "loss": 4.8565,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2479399820923392e-05,
+      "loss": 5.0199,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2469263608651963e-05,
+      "loss": 5.2576,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459124693937817e-05,
+      "loss": 5.3721,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244898308787731e-05,
+      "loss": 5.3065,
+      "step": 1358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2438838801569749e-05,
+      "loss": 5.4408,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2428691846117371e-05,
+      "loss": 4.7533,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418542232625337e-05,
+      "loss": 4.7377,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408389972201716e-05,
+      "loss": 4.7416,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239823507595747e-05,
+      "loss": 4.6607,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388077555006453e-05,
+      "loss": 5.3663,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377917420465382e-05,
+      "loss": 5.1557,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367754683453846e-05,
+      "loss": 5.3696,
+      "step": 1366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2357589355094275e-05,
+      "loss": 5.0532,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2347421446511931e-05,
+      "loss": 5.1965,
+      "step": 1368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337250968834912e-05,
+      "loss": 5.1502,
+      "step": 1369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327077933194117e-05,
+      "loss": 5.1554,
+      "step": 1370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316902350723253e-05,
+      "loss": 5.2746,
+      "step": 1371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2306724232558802e-05,
+      "loss": 4.9455,
+      "step": 1372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2296543589840033e-05,
+      "loss": 5.3616,
+      "step": 1373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286360433708976e-05,
+      "loss": 5.4109,
+      "step": 1374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2276174775310405e-05,
+      "loss": 4.8969,
+      "step": 1375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265986625791842e-05,
+      "loss": 4.5085,
+      "step": 1376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2255795996303526e-05,
+      "loss": 5.3357,
+      "step": 1377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245602897998417e-05,
+      "loss": 5.2152,
+      "step": 1378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2235407342032172e-05,
+      "loss": 4.7289,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 5.4509,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2215008901752361e-05,
+      "loss": 4.9955,
+      "step": 1381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220480603976351e-05,
+      "loss": 4.8379,
+      "step": 1382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194600764762941e-05,
+      "loss": 4.997,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184393087919635e-05,
+      "loss": 4.5783,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2174183020405213e-05,
+      "loss": 4.8103,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163970573393899e-05,
+      "loss": 5.3459,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2153755758062538e-05,
+      "loss": 5.5204,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143538585590548e-05,
+      "loss": 4.9278,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2133319067159941e-05,
+      "loss": 5.3312,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123097213955292e-05,
+      "loss": 5.2826,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2112873037163728e-05,
+      "loss": 5.0091,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102646547974925e-05,
+      "loss": 5.1318,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2092417757581085e-05,
+      "loss": 4.8779,
+      "step": 1393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082186677176933e-05,
+      "loss": 5.7135,
+      "step": 1394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071953317959692e-05,
+      "loss": 5.3381,
+      "step": 1395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206171769112909e-05,
+      "loss": 5.2513,
+      "step": 1396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2051479807887325e-05,
+      "loss": 5.545,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2041239679439077e-05,
+      "loss": 5.2373,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2030997316991473e-05,
+      "loss": 4.6366,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020752731754091e-05,
+      "loss": 5.1838,
+      "step": 1400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010505934938938e-05,
+      "loss": 5.6558,
+      "step": 1401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000256937760446e-05,
+      "loss": 4.5717,
+      "step": 1402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1990005751435448e-05,
+      "loss": 4.8856,
+      "step": 1403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979752387183179e-05,
+      "loss": 5.2536,
+      "step": 1404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969496856225256e-05,
+      "loss": 5.2425,
+      "step": 1405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959239169785668e-05,
+      "loss": 5.5192,
+      "step": 1406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948979339090758e-05,
+      "loss": 5.7989,
+      "step": 1407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1938717375369225e-05,
+      "loss": 5.305,
+      "step": 1408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192845328985209e-05,
+      "loss": 5.3712,
+      "step": 1409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918187093772714e-05,
+      "loss": 5.2837,
+      "step": 1410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907918798366746e-05,
+      "loss": 5.6754,
+      "step": 1411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897648414872149e-05,
+      "loss": 5.0749,
+      "step": 1412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887375954529167e-05,
+      "loss": 5.1941,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877101428580316e-05,
+      "loss": 4.7915,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1866824848270371e-05,
+      "loss": 5.3078,
+      "step": 1415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856546224846354e-05,
+      "loss": 4.8239,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1846265569557534e-05,
+      "loss": 5.6101,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183598289365539e-05,
+      "loss": 5.333,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182569820839362e-05,
+      "loss": 5.1734,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815411525028116e-05,
+      "loss": 4.982,
+      "step": 1420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1805122854816963e-05,
+      "loss": 5.1124,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794832209020422e-05,
+      "loss": 4.9197,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1784539598900905e-05,
+      "loss": 5.1912,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1774245035722984e-05,
+      "loss": 4.9932,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763948530753364e-05,
+      "loss": 4.7128,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1753650095260875e-05,
+      "loss": 5.3433,
+      "step": 1426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743349740516466e-05,
+      "loss": 4.9009,
+      "step": 1427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733047477793174e-05,
+      "loss": 5.7388,
+      "step": 1428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.172274331836614e-05,
+      "loss": 5.9899,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1712437273512562e-05,
+      "loss": 5.4967,
+      "step": 1430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1702129354511724e-05,
+      "loss": 4.6708,
+      "step": 1431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1691819572644941e-05,
+      "loss": 4.4706,
+      "step": 1432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1681507939195577e-05,
+      "loss": 5.1386,
+      "step": 1433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1671194465449022e-05,
+      "loss": 5.0881,
+      "step": 1434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660879162692676e-05,
+      "loss": 5.016,
+      "step": 1435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1650562042215943e-05,
+      "loss": 5.43,
+      "step": 1436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640243115310219e-05,
+      "loss": 5.5026,
+      "step": 1437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1629922393268873e-05,
+      "loss": 5.863,
+      "step": 1438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1619599887387242e-05,
+      "loss": 5.2473,
+      "step": 1439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609275608962612e-05,
+      "loss": 5.2661,
+      "step": 1440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598949569294207e-05,
+      "loss": 4.875,
+      "step": 1441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588621779683186e-05,
+      "loss": 4.936,
+      "step": 1442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1578292251432622e-05,
+      "loss": 5.1614,
+      "step": 1443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567960995847485e-05,
+      "loss": 4.6709,
+      "step": 1444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155762802423463e-05,
+      "loss": 4.7922,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1547293347902813e-05,
+      "loss": 5.4235,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1536956978162634e-05,
+      "loss": 5.0176,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1526618926326556e-05,
+      "loss": 5.0563,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1516279203708875e-05,
+      "loss": 5.1407,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1505937821625724e-05,
+      "loss": 4.8101,
+      "step": 1450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149559479139505e-05,
+      "loss": 5.3288,
+      "step": 1451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1485250124336607e-05,
+      "loss": 5.1357,
+      "step": 1452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147490383177193e-05,
+      "loss": 5.7774,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1464555925024342e-05,
+      "loss": 4.8587,
+      "step": 1454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454206415418928e-05,
+      "loss": 5.3073,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144385531428253e-05,
+      "loss": 5.1754,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433502632943735e-05,
+      "loss": 5.3893,
+      "step": 1457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 5.0747,
+      "step": 1458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412792574981913e-05,
+      "loss": 5.0746,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402435221024645e-05,
+      "loss": 5.3059,
+      "step": 1460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1392076332196483e-05,
+      "loss": 4.8791,
+      "step": 1461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1381715919834529e-05,
+      "loss": 5.1558,
+      "step": 1462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371353995277561e-05,
+      "loss": 4.9763,
+      "step": 1463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1360990569865999e-05,
+      "loss": 4.9389,
+      "step": 1464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1350625654941919e-05,
+      "loss": 4.6413,
+      "step": 1465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1340259261849023e-05,
+      "loss": 4.9395,
+      "step": 1466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329891401932631e-05,
+      "loss": 5.0794,
+      "step": 1467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1319522086539666e-05,
+      "loss": 5.3237,
+      "step": 1468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1309151327018641e-05,
+      "loss": 4.9743,
+      "step": 1469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298779134719662e-05,
+      "loss": 5.486,
+      "step": 1470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128840552099439e-05,
+      "loss": 4.9151,
+      "step": 1471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127803049719605e-05,
+      "loss": 5.2551,
+      "step": 1472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12676540746794e-05,
+      "loss": 5.2075,
+      "step": 1473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125727626480074e-05,
+      "loss": 4.986,
+      "step": 1474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1246897078917884e-05,
+      "loss": 5.1318,
+      "step": 1475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236516528390154e-05,
+      "loss": 4.7924,
+      "step": 1476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226134624578362e-05,
+      "loss": 4.9045,
+      "step": 1477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12157513788448e-05,
+      "loss": 5.3899,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 5.3347,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1194980907068878e-05,
+      "loss": 4.9772,
+      "step": 1480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184593703758404e-05,
+      "loss": 4.9746,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11742052039899e-05,
+      "loss": 4.9794,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163815419132884e-05,
+      "loss": 5.3759,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1153424360558268e-05,
+      "loss": 4.8732,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143032039638373e-05,
+      "loss": 5.0355,
+      "step": 1485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132638467746892e-05,
+      "loss": 5.0984,
+      "step": 1486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1122243656258894e-05,
+      "loss": 6.3004,
+      "step": 1487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111184761655079e-05,
+      "loss": 4.7857,
+      "step": 1488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1101450360000351e-05,
+      "loss": 5.0079,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091051897986677e-05,
+      "loss": 5.5767,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108065224189018e-05,
+      "loss": 4.8648,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1070251403092587e-05,
+      "loss": 5.0258,
+      "step": 1492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059849392976912e-05,
+      "loss": 4.6293,
+      "step": 1493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1049446222927451e-05,
+      "loss": 4.9393,
+      "step": 1494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1039041904329782e-05,
+      "loss": 4.9289,
+      "step": 1495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1028636448570729e-05,
+      "loss": 4.883,
+      "step": 1496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1018229867038358e-05,
+      "loss": 4.8865,
+      "step": 1497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007822171121972e-05,
+      "loss": 4.4123,
+      "step": 1498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997413372212094e-05,
+      "loss": 5.193,
+      "step": 1499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987003481700456e-05,
+      "loss": 5.7485,
+      "step": 1500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0976592510979982e-05,
+      "loss": 5.5676,
+      "step": 1501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0966180471444776e-05,
+      "loss": 5.1571,
+      "step": 1502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955767374490116e-05,
+      "loss": 5.2443,
+      "step": 1503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094535323151243e-05,
+      "loss": 5.2498,
+      "step": 1504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934938053909303e-05,
+      "loss": 5.027,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0924521853079444e-05,
+      "loss": 5.2154,
+      "step": 1506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091410464042268e-05,
+      "loss": 5.437,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903686427339952e-05,
+      "loss": 5.5828,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0893267225233289e-05,
+      "loss": 5.2729,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0882847045505809e-05,
+      "loss": 5.3805,
+      "step": 1510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872425899561698e-05,
+      "loss": 5.3566,
+      "step": 1511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862003798806195e-05,
+      "loss": 4.7676,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085158075464559e-05,
+      "loss": 5.0111,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841156778487199e-05,
+      "loss": 4.8287,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083073188173937e-05,
+      "loss": 5.1255,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082030607581144e-05,
+      "loss": 5.1387,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809879372113758e-05,
+      "loss": 5.0201,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0799451782057648e-05,
+      "loss": 5.1596,
+      "step": 1518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0789023317055402e-05,
+      "loss": 5.3236,
+      "step": 1519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0778593988520278e-05,
+      "loss": 5.2206,
+      "step": 1520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076816380786647e-05,
+      "loss": 5.4391,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075773278650911e-05,
+      "loss": 4.825,
+      "step": 1522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 5.1352,
+      "step": 1523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0736868267348834e-05,
+      "loss": 4.978,
+      "step": 1524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726434792380736e-05,
+      "loss": 4.9565,
+      "step": 1525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071600052237868e-05,
+      "loss": 5.3019,
+      "step": 1526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0705565468762274e-05,
+      "loss": 5.0677,
+      "step": 1527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.069512964295198e-05,
+      "loss": 5.032,
+      "step": 1528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0684693056369106e-05,
+      "loss": 5.0894,
+      "step": 1529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06742557204358e-05,
+      "loss": 5.1077,
+      "step": 1530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663817646575014e-05,
+      "loss": 5.5783,
+      "step": 1531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0653378846210522e-05,
+      "loss": 4.9575,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642939330766885e-05,
+      "loss": 4.6717,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632499111669453e-05,
+      "loss": 4.4598,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622058200344345e-05,
+      "loss": 5.3604,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0611616608218429e-05,
+      "loss": 4.9338,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601174346719327e-05,
+      "loss": 5.3768,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0590731427275387e-05,
+      "loss": 4.8145,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058028786131569e-05,
+      "loss": 4.7843,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569843660270007e-05,
+      "loss": 5.1213,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0559398835568818e-05,
+      "loss": 5.6926,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548953398643276e-05,
+      "loss": 5.5413,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053850736092521e-05,
+      "loss": 5.0182,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0528060733847108e-05,
+      "loss": 4.6931,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0517613528842096e-05,
+      "loss": 4.4346,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507165757343936e-05,
+      "loss": 4.5961,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0496717430787014e-05,
+      "loss": 5.3929,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486268560606312e-05,
+      "loss": 5.4168,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0475819158237426e-05,
+      "loss": 5.0141,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0465369235116514e-05,
+      "loss": 4.7931,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0454918802680316e-05,
+      "loss": 5.4649,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444467872366127e-05,
+      "loss": 4.8626,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0434016455611781e-05,
+      "loss": 5.0403,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423564563855663e-05,
+      "loss": 4.8304,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413112208536654e-05,
+      "loss": 4.4961,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402659401094154e-05,
+      "loss": 4.8371,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0392206152968058e-05,
+      "loss": 4.7079,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038175247559874e-05,
+      "loss": 4.6514,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037129838042705e-05,
+      "loss": 4.6424,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0360843878894292e-05,
+      "loss": 5.4549,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350388982442205e-05,
+      "loss": 4.9904,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0339933702512978e-05,
+      "loss": 5.0736,
+      "step": 1562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0329478050549208e-05,
+      "loss": 4.9361,
+      "step": 1563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0319022037993903e-05,
+      "loss": 5.3813,
+      "step": 1564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030856567629046e-05,
+      "loss": 4.8939,
+      "step": 1565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0298108976882665e-05,
+      "loss": 4.8331,
+      "step": 1566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0287651951214675e-05,
+      "loss": 4.8615,
+      "step": 1567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0277194610730996e-05,
+      "loss": 5.8272,
+      "step": 1568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0266736966876484e-05,
+      "loss": 4.912,
+      "step": 1569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256279031096327e-05,
+      "loss": 5.0969,
+      "step": 1570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024582081483603e-05,
+      "loss": 5.1326,
+      "step": 1571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0235362329541407e-05,
+      "loss": 4.8906,
+      "step": 1572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224903586658566e-05,
+      "loss": 4.8685,
+      "step": 1573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0214444597633895e-05,
+      "loss": 4.4575,
+      "step": 1574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203985373914056e-05,
+      "loss": 5.317,
+      "step": 1575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0193525926945962e-05,
+      "loss": 4.9219,
+      "step": 1576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0183066268176775e-05,
+      "loss": 5.6178,
+      "step": 1577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172606409053887e-05,
+      "loss": 4.4386,
+      "step": 1578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016214636102491e-05,
+      "loss": 5.0682,
+      "step": 1579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015168613553766e-05,
+      "loss": 5.3407,
+      "step": 1580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0141225744040148e-05,
+      "loss": 5.9358,
+      "step": 1581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130765197980574e-05,
+      "loss": 5.0937,
+      "step": 1582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0120304508807297e-05,
+      "loss": 4.5694,
+      "step": 1583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109843687968837e-05,
+      "loss": 4.9348,
+      "step": 1584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0099382746913859e-05,
+      "loss": 5.4197,
+      "step": 1585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0088921697091156e-05,
+      "loss": 5.492,
+      "step": 1586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0078460549949647e-05,
+      "loss": 5.3402,
+      "step": 1587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067999316938348e-05,
+      "loss": 4.653,
+      "step": 1588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057538009506378e-05,
+      "loss": 5.1526,
+      "step": 1589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004707663910293e-05,
+      "loss": 5.2502,
+      "step": 1590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036615217177273e-05,
+      "loss": 5.2078,
+      "step": 1591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026153755178725e-05,
+      "loss": 5.7775,
+      "step": 1592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0015692264556653e-05,
+      "loss": 5.1879,
+      "step": 1593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0005230756760456e-05,
+      "loss": 5.3672,
+      "step": 1594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.994769243239544e-06,
+      "loss": 5.6319,
+      "step": 1595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984307735443347e-06,
+      "loss": 4.729,
+      "step": 1596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973846244821279e-06,
+      "loss": 4.9599,
+      "step": 1597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963384782822732e-06,
+      "loss": 4.9639,
+      "step": 1598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.952923360897073e-06,
+      "loss": 4.9581,
+      "step": 1599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.942461990493627e-06,
+      "loss": 5.1175,
+      "step": 1600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932000683061654e-06,
+      "loss": 5.2154,
+      "step": 1601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.921539450050356e-06,
+      "loss": 5.1612,
+      "step": 1602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.911078302908849e-06,
+      "loss": 4.8348,
+      "step": 1603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900617253086146e-06,
+      "loss": 4.9023,
+      "step": 1604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890156312031165e-06,
+      "loss": 4.8777,
+      "step": 1605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879695491192706e-06,
+      "loss": 5.3361,
+      "step": 1606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869234802019428e-06,
+      "loss": 4.627,
+      "step": 1607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.858774255959855e-06,
+      "loss": 4.4992,
+      "step": 1608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848313864462345e-06,
+      "loss": 5.3327,
+      "step": 1609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837853638975095e-06,
+      "loss": 5.2832,
+      "step": 1610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827393590946116e-06,
+      "loss": 5.0458,
+      "step": 1611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81693373182323e-06,
+      "loss": 5.131,
+      "step": 1612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.806474073054043e-06,
+      "loss": 4.9969,
+      "step": 1613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79601462608595e-06,
+      "loss": 4.9842,
+      "step": 1614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785555402366105e-06,
+      "loss": 5.2133,
+      "step": 1615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775096413341438e-06,
+      "loss": 5.1059,
+      "step": 1616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.764637670458595e-06,
+      "loss": 5.1413,
+      "step": 1617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754179185163972e-06,
+      "loss": 5.6031,
+      "step": 1618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743720968903676e-06,
+      "loss": 5.1056,
+      "step": 1619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.733263033123517e-06,
+      "loss": 5.1456,
+      "step": 1620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.722805389269007e-06,
+      "loss": 5.3463,
+      "step": 1621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71234804878533e-06,
+      "loss": 5.0504,
+      "step": 1622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.701891023117338e-06,
+      "loss": 4.4573,
+      "step": 1623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691434323709546e-06,
+      "loss": 4.7796,
+      "step": 1624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.6809779620061e-06,
+      "loss": 4.8802,
+      "step": 1625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670521949450793e-06,
+      "loss": 5.1207,
+      "step": 1626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660066297487024e-06,
+      "loss": 5.2394,
+      "step": 1627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649611017557797e-06,
+      "loss": 4.8261,
+      "step": 1628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.639156121105713e-06,
+      "loss": 5.1212,
+      "step": 1629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628701619572952e-06,
+      "loss": 5.6109,
+      "step": 1630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618247524401261e-06,
+      "loss": 5.115,
+      "step": 1631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607793847031948e-06,
+      "loss": 4.9965,
+      "step": 1632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597340598905851e-06,
+      "loss": 4.5535,
+      "step": 1633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586887791463348e-06,
+      "loss": 5.4033,
+      "step": 1634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576435436144338e-06,
+      "loss": 4.9846,
+      "step": 1635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.565983544388217e-06,
+      "loss": 5.0161,
+      "step": 1636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555532127633877e-06,
+      "loss": 4.9386,
+      "step": 1637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545081197319687e-06,
+      "loss": 5.3462,
+      "step": 1638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53463076488349e-06,
+      "loss": 5.1464,
+      "step": 1639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.524180841762577e-06,
+      "loss": 5.1949,
+      "step": 1640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513731439393691e-06,
+      "loss": 5.4737,
+      "step": 1641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.503282569212993e-06,
+      "loss": 4.9294,
+      "step": 1642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492834242656069e-06,
+      "loss": 5.4278,
+      "step": 1643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482386471157905e-06,
+      "loss": 5.332,
+      "step": 1644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.471939266152894e-06,
+      "loss": 5.3843,
+      "step": 1645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46149263907479e-06,
+      "loss": 4.9178,
+      "step": 1646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451046601356725e-06,
+      "loss": 5.359,
+      "step": 1647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.440601164431186e-06,
+      "loss": 5.2771,
+      "step": 1648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430156339729995e-06,
+      "loss": 4.8366,
+      "step": 1649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419712138684313e-06,
+      "loss": 5.3991,
+      "step": 1650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.409268572724616e-06,
+      "loss": 5.0262,
+      "step": 1651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398825653280678e-06,
+      "loss": 5.1075,
+      "step": 1652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.388383391781576e-06,
+      "loss": 4.8154,
+      "step": 1653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377941799655656e-06,
+      "loss": 5.257,
+      "step": 1654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367500888330546e-06,
+      "loss": 5.333,
+      "step": 1655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357060669233115e-06,
+      "loss": 5.3043,
+      "step": 1656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.346621153789481e-06,
+      "loss": 5.2189,
+      "step": 1657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336182353424988e-06,
+      "loss": 4.6755,
+      "step": 1658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325744279564203e-06,
+      "loss": 5.4333,
+      "step": 1659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.315306943630897e-06,
+      "loss": 4.8102,
+      "step": 1660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.304870357048025e-06,
+      "loss": 5.1959,
+      "step": 1661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.294434531237733e-06,
+      "loss": 5.4958,
+      "step": 1662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.283999477621327e-06,
+      "loss": 5.2643,
+      "step": 1663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273565207619268e-06,
+      "loss": 4.921,
+      "step": 1664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263131732651167e-06,
+      "loss": 5.1483,
+      "step": 1665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 5.3825,
+      "step": 1666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242267213490895e-06,
+      "loss": 4.7724,
+      "step": 1667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231836192133532e-06,
+      "loss": 4.9969,
+      "step": 1668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.221406011479725e-06,
+      "loss": 5.8218,
+      "step": 1669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.210976682944601e-06,
+      "loss": 4.7477,
+      "step": 1670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.200548217942357e-06,
+      "loss": 5.1129,
+      "step": 1671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190120627886247e-06,
+      "loss": 5.5323,
+      "step": 1672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179693924188564e-06,
+      "loss": 5.5013,
+      "step": 1673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.169268118260634e-06,
+      "loss": 5.3579,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1588432215128e-06,
+      "loss": 5.1692,
+      "step": 1675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148419245354412e-06,
+      "loss": 5.1909,
+      "step": 1676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137996201193807e-06,
+      "loss": 5.0067,
+      "step": 1677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127574100438303e-06,
+      "loss": 4.8343,
+      "step": 1678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.117152954494195e-06,
+      "loss": 5.2398,
+      "step": 1679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106732774766715e-06,
+      "loss": 4.5781,
+      "step": 1680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096313572660053e-06,
+      "loss": 4.7616,
+      "step": 1681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085895359577324e-06,
+      "loss": 5.826,
+      "step": 1682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.075478146920561e-06,
+      "loss": 4.5681,
+      "step": 1683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.065061946090698e-06,
+      "loss": 4.782,
+      "step": 1684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054646768487572e-06,
+      "loss": 5.1408,
+      "step": 1685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.044232625509887e-06,
+      "loss": 5.3124,
+      "step": 1686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033819528555227e-06,
+      "loss": 5.3182,
+      "step": 1687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.02340748902002e-06,
+      "loss": 5.084,
+      "step": 1688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012996518299547e-06,
+      "loss": 5.0874,
+      "step": 1689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00258662778791e-06,
+      "loss": 4.4176,
+      "step": 1690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992177828878033e-06,
+      "loss": 4.4811,
+      "step": 1691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.981770132961649e-06,
+      "loss": 5.3105,
+      "step": 1692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971363551429274e-06,
+      "loss": 4.9046,
+      "step": 1693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96095809567022e-06,
+      "loss": 5.1857,
+      "step": 1694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.950553777072549e-06,
+      "loss": 5.2556,
+      "step": 1695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940150607023091e-06,
+      "loss": 5.6373,
+      "step": 1696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.929748596907416e-06,
+      "loss": 5.1288,
+      "step": 1697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.919347758109824e-06,
+      "loss": 4.9654,
+      "step": 1698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908948102013326e-06,
+      "loss": 5.1238,
+      "step": 1699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89854963999965e-06,
+      "loss": 4.7403,
+      "step": 1700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888152383449214e-06,
+      "loss": 5.1092,
+      "step": 1701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877756343741111e-06,
+      "loss": 5.1689,
+      "step": 1702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867361532253108e-06,
+      "loss": 5.2545,
+      "step": 1703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856967960361627e-06,
+      "loss": 5.2309,
+      "step": 1704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.846575639441732e-06,
+      "loss": 4.8149,
+      "step": 1705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83618458086712e-06,
+      "loss": 4.9727,
+      "step": 1706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825794796010101e-06,
+      "loss": 4.5819,
+      "step": 1707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8154062962416e-06,
+      "loss": 4.7236,
+      "step": 1708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805019092931124e-06,
+      "loss": 5.6412,
+      "step": 1709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 5.0681,
+      "step": 1710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784248621155204e-06,
+      "loss": 4.9009,
+      "step": 1711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773865375421643e-06,
+      "loss": 4.8711,
+      "step": 1712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763483471609848e-06,
+      "loss": 5.0116,
+      "step": 1713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753102921082116e-06,
+      "loss": 5.282,
+      "step": 1714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742723735199262e-06,
+      "loss": 5.1583,
+      "step": 1715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.732345925320602e-06,
+      "loss": 5.1573,
+      "step": 1716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.721969502803954e-06,
+      "loss": 4.6374,
+      "step": 1717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.711594479005614e-06,
+      "loss": 5.5667,
+      "step": 1718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701220865280341e-06,
+      "loss": 5.0683,
+      "step": 1719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69084867298136e-06,
+      "loss": 5.1426,
+      "step": 1720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680477913460339e-06,
+      "loss": 5.2418,
+      "step": 1721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670108598067372e-06,
+      "loss": 5.5929,
+      "step": 1722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659740738150977e-06,
+      "loss": 5.173,
+      "step": 1723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649374345058083e-06,
+      "loss": 4.8633,
+      "step": 1724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.639009430134003e-06,
+      "loss": 5.1888,
+      "step": 1725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628646004722442e-06,
+      "loss": 5.2444,
+      "step": 1726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618284080165473e-06,
+      "loss": 4.9859,
+      "step": 1727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607923667803519e-06,
+      "loss": 4.843,
+      "step": 1728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597564778975356e-06,
+      "loss": 5.1768,
+      "step": 1729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58720742501809e-06,
+      "loss": 5.0317,
+      "step": 1730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 4.9745,
+      "step": 1731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566497367056268e-06,
+      "loss": 5.1942,
+      "step": 1732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556144685717469e-06,
+      "loss": 5.5576,
+      "step": 1733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545793584581074e-06,
+      "loss": 4.954,
+      "step": 1734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53544407497566e-06,
+      "loss": 5.8361,
+      "step": 1735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525096168228073e-06,
+      "loss": 4.6843,
+      "step": 1736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514749875663397e-06,
+      "loss": 4.8864,
+      "step": 1737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504405208604951e-06,
+      "loss": 4.952,
+      "step": 1738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49406217837428e-06,
+      "loss": 5.319,
+      "step": 1739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48372079629113e-06,
+      "loss": 4.9072,
+      "step": 1740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47338107367345e-06,
+      "loss": 5.4414,
+      "step": 1741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.46304302183737e-06,
+      "loss": 4.7452,
+      "step": 1742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.452706652097187e-06,
+      "loss": 5.079,
+      "step": 1743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.442371975765368e-06,
+      "loss": 5.4446,
+      "step": 1744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43203900415252e-06,
+      "loss": 4.6905,
+      "step": 1745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.421707748567382e-06,
+      "loss": 5.4145,
+      "step": 1746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411378220316816e-06,
+      "loss": 5.4359,
+      "step": 1747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401050430705794e-06,
+      "loss": 4.6547,
+      "step": 1748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390724391037393e-06,
+      "loss": 5.2869,
+      "step": 1749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380400112612761e-06,
+      "loss": 4.8692,
+      "step": 1750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370077606731128e-06,
+      "loss": 4.8819,
+      "step": 1751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359756884689785e-06,
+      "loss": 4.7486,
+      "step": 1752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349437957784056e-06,
+      "loss": 5.1121,
+      "step": 1753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339120837307324e-06,
+      "loss": 5.1565,
+      "step": 1754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32880553455098e-06,
+      "loss": 4.8604,
+      "step": 1755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318492060804426e-06,
+      "loss": 5.2516,
+      "step": 1756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.308180427355062e-06,
+      "loss": 4.7461,
+      "step": 1757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29787064548828e-06,
+      "loss": 5.3507,
+      "step": 1758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28756272648744e-06,
+      "loss": 4.4691,
+      "step": 1759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277256681633865e-06,
+      "loss": 5.2005,
+      "step": 1760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266952522206828e-06,
+      "loss": 4.7273,
+      "step": 1761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256650259483536e-06,
+      "loss": 4.612,
+      "step": 1762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246349904739125e-06,
+      "loss": 5.8041,
+      "step": 1763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236051469246637e-06,
+      "loss": 4.901,
+      "step": 1764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.225754964277018e-06,
+      "loss": 4.9684,
+      "step": 1765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215460401099098e-06,
+      "loss": 5.4027,
+      "step": 1766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20516779097958e-06,
+      "loss": 5.5841,
+      "step": 1767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.194877145183038e-06,
+      "loss": 5.1882,
+      "step": 1768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184588474971885e-06,
+      "loss": 4.6662,
+      "step": 1769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174301791606384e-06,
+      "loss": 4.5927,
+      "step": 1770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.164017106344614e-06,
+      "loss": 4.5368,
+      "step": 1771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.153734430442466e-06,
+      "loss": 4.7197,
+      "step": 1772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143453775153646e-06,
+      "loss": 5.2034,
+      "step": 1773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13317515172963e-06,
+      "loss": 5.3274,
+      "step": 1774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122898571419688e-06,
+      "loss": 4.3052,
+      "step": 1775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112624045470834e-06,
+      "loss": 5.2325,
+      "step": 1776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102351585127853e-06,
+      "loss": 5.1161,
+      "step": 1777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092081201633257e-06,
+      "loss": 5.3941,
+      "step": 1778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.081812906227291e-06,
+      "loss": 4.6555,
+      "step": 1779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071546710147911e-06,
+      "loss": 4.9592,
+      "step": 1780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06128262463078e-06,
+      "loss": 5.1949,
+      "step": 1781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051020660909242e-06,
+      "loss": 5.206,
+      "step": 1782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040760830214334e-06,
+      "loss": 4.9952,
+      "step": 1783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030503143774744e-06,
+      "loss": 5.3774,
+      "step": 1784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020247612816823e-06,
+      "loss": 5.6332,
+      "step": 1785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.009994248564554e-06,
+      "loss": 4.8213,
+      "step": 1786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999743062239557e-06,
+      "loss": 4.6384,
+      "step": 1787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989494065061064e-06,
+      "loss": 4.8133,
+      "step": 1788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97924726824591e-06,
+      "loss": 5.2983,
+      "step": 1789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.96900268300853e-06,
+      "loss": 4.6624,
+      "step": 1790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.958760320560928e-06,
+      "loss": 5.352,
+      "step": 1791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948520192112675e-06,
+      "loss": 5.0028,
+      "step": 1792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.938282308870912e-06,
+      "loss": 5.1274,
+      "step": 1793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.928046682040311e-06,
+      "loss": 5.3006,
+      "step": 1794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.917813322823072e-06,
+      "loss": 5.0771,
+      "step": 1795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907582242418916e-06,
+      "loss": 5.3273,
+      "step": 1796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897353452025077e-06,
+      "loss": 4.6367,
+      "step": 1797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.887126962836274e-06,
+      "loss": 5.2203,
+      "step": 1798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876902786044711e-06,
+      "loss": 5.014,
+      "step": 1799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866680932840064e-06,
+      "loss": 5.2738,
+      "step": 1800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856461414409458e-06,
+      "loss": 5.1192,
+      "step": 1801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846244241937463e-06,
+      "loss": 5.1593,
+      "step": 1802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8360294266061e-06,
+      "loss": 4.6498,
+      "step": 1803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825816979594792e-06,
+      "loss": 4.899,
+      "step": 1804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815606912080368e-06,
+      "loss": 5.0179,
+      "step": 1805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805399235237064e-06,
+      "loss": 4.8886,
+      "step": 1806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795193960236492e-06,
+      "loss": 4.7126,
+      "step": 1807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.784991098247642e-06,
+      "loss": 5.2822,
+      "step": 1808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 5.2653,
+      "step": 1809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764592657967833e-06,
+      "loss": 4.8438,
+      "step": 1810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75439710200159e-06,
+      "loss": 5.7323,
+      "step": 1811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.744204003696476e-06,
+      "loss": 5.2025,
+      "step": 1812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734013374208163e-06,
+      "loss": 5.114,
+      "step": 1813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.723825224689598e-06,
+      "loss": 4.6795,
+      "step": 1814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713639566291028e-06,
+      "loss": 5.1609,
+      "step": 1815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70345641015997e-06,
+      "loss": 4.5701,
+      "step": 1816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.693275767441201e-06,
+      "loss": 5.7481,
+      "step": 1817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683097649276752e-06,
+      "loss": 4.9849,
+      "step": 1818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672922066805888e-06,
+      "loss": 5.3652,
+      "step": 1819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662749031165093e-06,
+      "loss": 4.8812,
+      "step": 1820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.652578553488069e-06,
+      "loss": 4.5092,
+      "step": 1821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.642410644905726e-06,
+      "loss": 5.314,
+      "step": 1822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632245316546156e-06,
+      "loss": 4.9964,
+      "step": 1823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.62208257953462e-06,
+      "loss": 5.6026,
+      "step": 1824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6119224449935515e-06,
+      "loss": 4.9541,
+      "step": 1825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601764924042533e-06,
+      "loss": 4.7081,
+      "step": 1826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591610027798287e-06,
+      "loss": 5.3911,
+      "step": 1827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5814577673746655e-06,
+      "loss": 5.0922,
+      "step": 1828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.571308153882634e-06,
+      "loss": 4.8694,
+      "step": 1829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.561161198430257e-06,
+      "loss": 4.9366,
+      "step": 1830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551016912122692e-06,
+      "loss": 5.2836,
+      "step": 1831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5408753060621875e-06,
+      "loss": 5.2909,
+      "step": 1832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530736391348039e-06,
+      "loss": 5.3022,
+      "step": 1833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.52060017907661e-06,
+      "loss": 5.0472,
+      "step": 1834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5104666803413015e-06,
+      "loss": 5.2119,
+      "step": 1835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500335906232544e-06,
+      "loss": 4.7481,
+      "step": 1836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.49020786783779e-06,
+      "loss": 5.2613,
+      "step": 1837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.480082576241492e-06,
+      "loss": 4.7173,
+      "step": 1838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469960042525105e-06,
+      "loss": 4.7816,
+      "step": 1839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.459840277767054e-06,
+      "loss": 5.3118,
+      "step": 1840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44972329304274e-06,
+      "loss": 5.2133,
+      "step": 1841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4396090994245295e-06,
+      "loss": 5.1075,
+      "step": 1842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.42949770798172e-06,
+      "loss": 4.9571,
+      "step": 1843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419389129780549e-06,
+      "loss": 5.8429,
+      "step": 1844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409283375884175e-06,
+      "loss": 5.3518,
+      "step": 1845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399180457352666e-06,
+      "loss": 4.9286,
+      "step": 1846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389080385242984e-06,
+      "loss": 5.2723,
+      "step": 1847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.378983170608982e-06,
+      "loss": 4.9394,
+      "step": 1848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.368888824501375e-06,
+      "loss": 4.8893,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358797357967749e-06,
+      "loss": 5.2987,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.348708782052528e-06,
+      "loss": 4.9046,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3386231077969915e-06,
+      "loss": 5.2586,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328540346239222e-06,
+      "loss": 4.9943,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318460508414124e-06,
+      "loss": 4.864,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308383605353404e-06,
+      "loss": 5.2319,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29830964808555e-06,
+      "loss": 4.9874,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2882386476358304e-06,
+      "loss": 4.8368,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278170615026281e-06,
+      "loss": 4.8509,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2681055612756805e-06,
+      "loss": 4.8947,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.258043497399554e-06,
+      "loss": 5.0632,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24798443441015e-06,
+      "loss": 4.9426,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.237928383316441e-06,
+      "loss": 4.8914,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227875355124094e-06,
+      "loss": 4.9179,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217825360835475e-06,
+      "loss": 5.4482,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207778411449621e-06,
+      "loss": 4.5401,
+      "step": 1865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.197734517962245e-06,
+      "loss": 5.1092,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1876936913657115e-06,
+      "loss": 5.1282,
+      "step": 1867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.177655942649027e-06,
+      "loss": 5.003,
+      "step": 1868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.167621282797832e-06,
+      "loss": 5.1361,
+      "step": 1869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.157589722794386e-06,
+      "loss": 4.6505,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147561273617555e-06,
+      "loss": 4.8419,
+      "step": 1871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.137535946242804e-06,
+      "loss": 5.2277,
+      "step": 1872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1275137516421776e-06,
+      "loss": 5.3085,
+      "step": 1873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117494700784292e-06,
+      "loss": 4.8935,
+      "step": 1874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107478804634324e-06,
+      "loss": 5.5909,
+      "step": 1875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.097466074153998e-06,
+      "loss": 4.4948,
+      "step": 1876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087456520301574e-06,
+      "loss": 4.9133,
+      "step": 1877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0774501540318305e-06,
+      "loss": 5.2184,
+      "step": 1878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0674469862960645e-06,
+      "loss": 4.8392,
+      "step": 1879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.057447028042069e-06,
+      "loss": 5.4022,
+      "step": 1880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047450290214123e-06,
+      "loss": 4.3086,
+      "step": 1881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037456783752985e-06,
+      "loss": 5.3022,
+      "step": 1882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027466519595876e-06,
+      "loss": 4.7866,
+      "step": 1883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017479508676466e-06,
+      "loss": 5.2542,
+      "step": 1884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007495761924863e-06,
+      "loss": 5.4878,
+      "step": 1885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.997515290267611e-06,
+      "loss": 5.017,
+      "step": 1886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9875381046276605e-06,
+      "loss": 5.0288,
+      "step": 1887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977564215924368e-06,
+      "loss": 4.5947,
+      "step": 1888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967593635073484e-06,
+      "loss": 4.6664,
+      "step": 1889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957626372987139e-06,
+      "loss": 4.7067,
+      "step": 1890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947662440573828e-06,
+      "loss": 4.9267,
+      "step": 1891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.937701848738407e-06,
+      "loss": 5.0054,
+      "step": 1892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.927744608382072e-06,
+      "loss": 4.785,
+      "step": 1893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.917790730402351e-06,
+      "loss": 5.0426,
+      "step": 1894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.907840225693092e-06,
+      "loss": 4.903,
+      "step": 1895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.897893105144458e-06,
+      "loss": 4.9103,
+      "step": 1896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.887949379642893e-06,
+      "loss": 5.0072,
+      "step": 1897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878009060071143e-06,
+      "loss": 5.5567,
+      "step": 1898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868072157308213e-06,
+      "loss": 5.285,
+      "step": 1899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8581386822293765e-06,
+      "loss": 4.7752,
+      "step": 1900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848208645706152e-06,
+      "loss": 5.1672,
+      "step": 1901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.838282058606298e-06,
+      "loss": 5.0616,
+      "step": 1902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828358931793793e-06,
+      "loss": 4.4036,
+      "step": 1903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.818439276128832e-06,
+      "loss": 4.3973,
+      "step": 1904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808523102467812e-06,
+      "loss": 4.8813,
+      "step": 1905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.798610421663313e-06,
+      "loss": 4.6785,
+      "step": 1906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.788701244564099e-06,
+      "loss": 4.3545,
+      "step": 1907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.778795582015096e-06,
+      "loss": 5.2176,
+      "step": 1908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7688934448573854e-06,
+      "loss": 4.8374,
+      "step": 1909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758994843928187e-06,
+      "loss": 5.0234,
+      "step": 1910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.749099790060858e-06,
+      "loss": 5.0483,
+      "step": 1911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739208294084865e-06,
+      "loss": 4.6987,
+      "step": 1912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.729320366825785e-06,
+      "loss": 5.3323,
+      "step": 1913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719436019105287e-06,
+      "loss": 4.9036,
+      "step": 1914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.709555261741127e-06,
+      "loss": 4.9043,
+      "step": 1915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699678105547125e-06,
+      "loss": 4.7282,
+      "step": 1916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.689804561333164e-06,
+      "loss": 4.4873,
+      "step": 1917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679934639905174e-06,
+      "loss": 5.2638,
+      "step": 1918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67006835206512e-06,
+      "loss": 4.9211,
+      "step": 1919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660205708610987e-06,
+      "loss": 5.4178,
+      "step": 1920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650346720336777e-06,
+      "loss": 4.6077,
+      "step": 1921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640491398032487e-06,
+      "loss": 4.4994,
+      "step": 1922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630639752484105e-06,
+      "loss": 5.1185,
+      "step": 1923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620791794473591e-06,
+      "loss": 5.3843,
+      "step": 1924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610947534778877e-06,
+      "loss": 4.9296,
+      "step": 1925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601106984173835e-06,
+      "loss": 5.2757,
+      "step": 1926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591270153428288e-06,
+      "loss": 5.1111,
+      "step": 1927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581437053307986e-06,
+      "loss": 5.0841,
+      "step": 1928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.571607694574596e-06,
+      "loss": 4.6124,
+      "step": 1929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561782087985682e-06,
+      "loss": 4.6248,
+      "step": 1930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551960244294717e-06,
+      "loss": 5.3775,
+      "step": 1931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.542142174251045e-06,
+      "loss": 4.8835,
+      "step": 1932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532327888599881e-06,
+      "loss": 5.3231,
+      "step": 1933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522517398082303e-06,
+      "loss": 4.8339,
+      "step": 1934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.512710713435226e-06,
+      "loss": 4.606,
+      "step": 1935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.502907845391411e-06,
+      "loss": 5.0772,
+      "step": 1936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4931088046794355e-06,
+      "loss": 4.7496,
+      "step": 1937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4833136020236875e-06,
+      "loss": 5.1408,
+      "step": 1938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473522248144359e-06,
+      "loss": 5.0475,
+      "step": 1939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.463734753757426e-06,
+      "loss": 5.0729,
+      "step": 1940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 4.8658,
+      "step": 1941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444171386303529e-06,
+      "loss": 4.7884,
+      "step": 1942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434395534647353e-06,
+      "loss": 5.1142,
+      "step": 1943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424623585305127e-06,
+      "loss": 5.3875,
+      "step": 1944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41485554897159e-06,
+      "loss": 5.0583,
+      "step": 1945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4050914363371985e-06,
+      "loss": 5.1744,
+      "step": 1946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39533125808812e-06,
+      "loss": 5.2892,
+      "step": 1947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3855750249062096e-06,
+      "loss": 4.3343,
+      "step": 1948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375822747469007e-06,
+      "loss": 5.3719,
+      "step": 1949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366074436449725e-06,
+      "loss": 4.7026,
+      "step": 1950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3563301025172345e-06,
+      "loss": 5.1167,
+      "step": 1951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.34658975633605e-06,
+      "loss": 5.4952,
+      "step": 1952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336853408566328e-06,
+      "loss": 4.8779,
+      "step": 1953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.32712106986384e-06,
+      "loss": 5.024,
+      "step": 1954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317392750879978e-06,
+      "loss": 5.1431,
+      "step": 1955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.307668462261732e-06,
+      "loss": 5.2556,
+      "step": 1956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297948214651681e-06,
+      "loss": 4.5772,
+      "step": 1957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.288232018687983e-06,
+      "loss": 5.4631,
+      "step": 1958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.278519885004355e-06,
+      "loss": 4.7473,
+      "step": 1959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.268811824230078e-06,
+      "loss": 4.9313,
+      "step": 1960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259107846989971e-06,
+      "loss": 5.2045,
+      "step": 1961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.249407963904381e-06,
+      "loss": 4.3187,
+      "step": 1962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.239712185589181e-06,
+      "loss": 5.0633,
+      "step": 1963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230020522655741e-06,
+      "loss": 5.2094,
+      "step": 1964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.220332985710936e-06,
+      "loss": 5.6512,
+      "step": 1965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2106495853571236e-06,
+      "loss": 5.1533,
+      "step": 1966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200970332192133e-06,
+      "loss": 4.843,
+      "step": 1967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191295236809255e-06,
+      "loss": 4.8119,
+      "step": 1968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181624309797228e-06,
+      "loss": 5.1993,
+      "step": 1969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.171957561740233e-06,
+      "loss": 5.07,
+      "step": 1970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.162295003217873e-06,
+      "loss": 4.9458,
+      "step": 1971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.152636644805172e-06,
+      "loss": 4.9231,
+      "step": 1972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142982497072546e-06,
+      "loss": 4.9664,
+      "step": 1973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133332570585813e-06,
+      "loss": 4.6146,
+      "step": 1974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123686875906167e-06,
+      "loss": 5.605,
+      "step": 1975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1140454235901714e-06,
+      "loss": 5.0907,
+      "step": 1976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104408224189746e-06,
+      "loss": 5.0932,
+      "step": 1977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094775288252157e-06,
+      "loss": 4.8014,
+      "step": 1978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.085146626320003e-06,
+      "loss": 5.2007,
+      "step": 1979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0755222489312085e-06,
+      "loss": 5.024,
+      "step": 1980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.065902166619004e-06,
+      "loss": 5.2362,
+      "step": 1981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056286389911925e-06,
+      "loss": 5.2703,
+      "step": 1982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046674929333787e-06,
+      "loss": 5.5876,
+      "step": 1983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.037067795403689e-06,
+      "loss": 5.1654,
+      "step": 1984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02746499863599e-06,
+      "loss": 5.3817,
+      "step": 1985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0178665495403055e-06,
+      "loss": 5.0092,
+      "step": 1986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008272458621492e-06,
+      "loss": 5.3997,
+      "step": 1987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.998682736379638e-06,
+      "loss": 4.8982,
+      "step": 1988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9890973933100405e-06,
+      "loss": 4.3196,
+      "step": 1989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.979516439903221e-06,
+      "loss": 4.6368,
+      "step": 1990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9699398866448846e-06,
+      "loss": 5.3231,
+      "step": 1991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960367744015927e-06,
+      "loss": 4.7225,
+      "step": 1992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.950800022492407e-06,
+      "loss": 5.2379,
+      "step": 1993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.941236732545554e-06,
+      "loss": 4.8781,
+      "step": 1994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931677884641747e-06,
+      "loss": 5.4668,
+      "step": 1995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922123489242499e-06,
+      "loss": 4.7585,
+      "step": 1996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.912573556804453e-06,
+      "loss": 4.9749,
+      "step": 1997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.90302809777937e-06,
+      "loss": 5.3935,
+      "step": 1998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.893487122614102e-06,
+      "loss": 5.0976,
+      "step": 1999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883950641750615e-06,
+      "loss": 5.0468,
+      "step": 2000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.874418665625942e-06,
+      "loss": 5.5313,
+      "step": 2001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864891204672187e-06,
+      "loss": 5.349,
+      "step": 2002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.855368269316515e-06,
+      "loss": 4.7699,
+      "step": 2003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 5.2818,
+      "step": 2004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836336017083303e-06,
+      "loss": 4.7392,
+      "step": 2005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8268267210352795e-06,
+      "loss": 4.7679,
+      "step": 2006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8173219922443516e-06,
+      "loss": 5.2697,
+      "step": 2007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807821841112807e-06,
+      "loss": 4.8352,
+      "step": 2008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798326278037916e-06,
+      "loss": 5.1412,
+      "step": 2009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788835313411941e-06,
+      "loss": 5.3274,
+      "step": 2010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7793489576220994e-06,
+      "loss": 5.6741,
+      "step": 2011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.769867221050571e-06,
+      "loss": 5.4612,
+      "step": 2012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760390114074475e-06,
+      "loss": 4.6335,
+      "step": 2013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750917647065871e-06,
+      "loss": 5.0604,
+      "step": 2014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.74144983039173e-06,
+      "loss": 5.2675,
+      "step": 2015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731986674413941e-06,
+      "loss": 5.3016,
+      "step": 2016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722528189489294e-06,
+      "loss": 4.8605,
+      "step": 2017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713074385969457e-06,
+      "loss": 5.5039,
+      "step": 2018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.703625274200992e-06,
+      "loss": 4.8306,
+      "step": 2019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694180864525308e-06,
+      "loss": 4.4065,
+      "step": 2020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.684741167278679e-06,
+      "loss": 5.4708,
+      "step": 2021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675306192792216e-06,
+      "loss": 5.4329,
+      "step": 2022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665875951391866e-06,
+      "loss": 4.9221,
+      "step": 2023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656450453398386e-06,
+      "loss": 4.8176,
+      "step": 2024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.647029709127355e-06,
+      "loss": 5.0494,
+      "step": 2025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63761372888914e-06,
+      "loss": 5.4005,
+      "step": 2026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.628202522988899e-06,
+      "loss": 5.1268,
+      "step": 2027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.618796101726558e-06,
+      "loss": 4.957,
+      "step": 2028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.609394475396821e-06,
+      "loss": 4.8936,
+      "step": 2029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.599997654289129e-06,
+      "loss": 5.5233,
+      "step": 2030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590605648687673e-06,
+      "loss": 5.1851,
+      "step": 2031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.581218468871365e-06,
+      "loss": 5.2942,
+      "step": 2032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57183612511385e-06,
+      "loss": 5.1119,
+      "step": 2033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562458627683462e-06,
+      "loss": 4.9895,
+      "step": 2034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.553085986843243e-06,
+      "loss": 5.0446,
+      "step": 2035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5437182128509195e-06,
+      "loss": 5.1654,
+      "step": 2036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5343553159588884e-06,
+      "loss": 4.938,
+      "step": 2037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524997306414204e-06,
+      "loss": 5.1817,
+      "step": 2038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.515644194458588e-06,
+      "loss": 4.9664,
+      "step": 2039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5062959903283855e-06,
+      "loss": 5.3054,
+      "step": 2040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496952704254579e-06,
+      "loss": 5.6855,
+      "step": 2041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.487614346462766e-06,
+      "loss": 5.0024,
+      "step": 2042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.478280927173145e-06,
+      "loss": 4.9738,
+      "step": 2043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468952456600519e-06,
+      "loss": 4.7157,
+      "step": 2044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459628944954269e-06,
+      "loss": 5.5214,
+      "step": 2045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450310402438353e-06,
+      "loss": 4.8575,
+      "step": 2046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.440996839251284e-06,
+      "loss": 4.6823,
+      "step": 2047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43168826558613e-06,
+      "loss": 5.0541,
+      "step": 2048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422384691630505e-06,
+      "loss": 4.6301,
+      "step": 2049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413086127566538e-06,
+      "loss": 4.8295,
+      "step": 2050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403792583570884e-06,
+      "loss": 4.8202,
+      "step": 2051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3945040698147026e-06,
+      "loss": 4.9071,
+      "step": 2052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.385220596463638e-06,
+      "loss": 5.4797,
+      "step": 2053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.375942173677835e-06,
+      "loss": 5.4089,
+      "step": 2054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.366668811611897e-06,
+      "loss": 4.4526,
+      "step": 2055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.357400520414898e-06,
+      "loss": 4.8095,
+      "step": 2056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348137310230357e-06,
+      "loss": 4.6416,
+      "step": 2057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.338879191196229e-06,
+      "loss": 4.6671,
+      "step": 2058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.329626173444913e-06,
+      "loss": 4.733,
+      "step": 2059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.320378267103206e-06,
+      "loss": 5.1388,
+      "step": 2060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311135482292319e-06,
+      "loss": 4.8114,
+      "step": 2061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3018978291278635e-06,
+      "loss": 4.4139,
+      "step": 2062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292665317719819e-06,
+      "loss": 4.8877,
+      "step": 2063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283437958172553e-06,
+      "loss": 4.7254,
+      "step": 2064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274215760584789e-06,
+      "loss": 5.236,
+      "step": 2065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2649987350496e-06,
+      "loss": 5.3157,
+      "step": 2066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2557868916543996e-06,
+      "loss": 5.0922,
+      "step": 2067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246580240480928e-06,
+      "loss": 4.565,
+      "step": 2068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237378791605249e-06,
+      "loss": 5.0883,
+      "step": 2069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228182555097729e-06,
+      "loss": 5.2033,
+      "step": 2070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218991541023031e-06,
+      "loss": 5.4431,
+      "step": 2071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209805759440093e-06,
+      "loss": 4.7419,
+      "step": 2072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.200625220402139e-06,
+      "loss": 5.6618,
+      "step": 2073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1914499339566514e-06,
+      "loss": 4.8643,
+      "step": 2074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1822799101453615e-06,
+      "loss": 5.0976,
+      "step": 2075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173115159004244e-06,
+      "loss": 4.8559,
+      "step": 2076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163955690563498e-06,
+      "loss": 5.147,
+      "step": 2077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154801514847552e-06,
+      "loss": 4.9246,
+      "step": 2078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145652641875029e-06,
+      "loss": 4.5951,
+      "step": 2079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136509081658758e-06,
+      "loss": 4.4565,
+      "step": 2080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127370844205753e-06,
+      "loss": 4.3819,
+      "step": 2081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.11823793951719e-06,
+      "loss": 4.7872,
+      "step": 2082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109110377588422e-06,
+      "loss": 5.192,
+      "step": 2083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099988168408952e-06,
+      "loss": 4.9528,
+      "step": 2084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090871321962424e-06,
+      "loss": 5.7689,
+      "step": 2085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08175984822661e-06,
+      "loss": 4.7245,
+      "step": 2086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072653757173403e-06,
+      "loss": 5.213,
+      "step": 2087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.063553058768814e-06,
+      "loss": 5.2496,
+      "step": 2088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05445776297294e-06,
+      "loss": 4.8386,
+      "step": 2089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045367879739971e-06,
+      "loss": 5.0027,
+      "step": 2090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0362834190181685e-06,
+      "loss": 4.9478,
+      "step": 2091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027204390749865e-06,
+      "loss": 4.9713,
+      "step": 2092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018130804871444e-06,
+      "loss": 5.1083,
+      "step": 2093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.009062671313336e-06,
+      "loss": 4.5411,
+      "step": 2094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 5.171,
+      "step": 2095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.990942800849926e-06,
+      "loss": 4.9168,
+      "step": 2096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.981891083775597e-06,
+      "loss": 5.1068,
+      "step": 2097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.97284485868352e-06,
+      "loss": 4.8378,
+      "step": 2098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9638041354741725e-06,
+      "loss": 4.9284,
+      "step": 2099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954768924042024e-06,
+      "loss": 5.53,
+      "step": 2100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.945739234275494e-06,
+      "loss": 5.1416,
+      "step": 2101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9367150760569746e-06,
+      "loss": 5.0625,
+      "step": 2102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.927696459262801e-06,
+      "loss": 5.6099,
+      "step": 2103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.91868339376324e-06,
+      "loss": 5.7142,
+      "step": 2104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.909675889422485e-06,
+      "loss": 5.584,
+      "step": 2105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900673956098644e-06,
+      "loss": 5.2364,
+      "step": 2106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891677603643721e-06,
+      "loss": 4.6975,
+      "step": 2107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882686841903627e-06,
+      "loss": 5.1836,
+      "step": 2108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.873701680718146e-06,
+      "loss": 4.8778,
+      "step": 2109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8647221299209225e-06,
+      "loss": 5.4437,
+      "step": 2110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855748199339474e-06,
+      "loss": 4.761,
+      "step": 2111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8467798987951645e-06,
+      "loss": 4.8644,
+      "step": 2112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837817238103193e-06,
+      "loss": 5.3385,
+      "step": 2113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.828860227072589e-06,
+      "loss": 5.1151,
+      "step": 2114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8199088755062e-06,
+      "loss": 5.1411,
+      "step": 2115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.810963193200673e-06,
+      "loss": 5.4431,
+      "step": 2116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.802023189946454e-06,
+      "loss": 4.6416,
+      "step": 2117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7930888755277815e-06,
+      "loss": 4.623,
+      "step": 2118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78416025972266e-06,
+      "loss": 5.002,
+      "step": 2119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.775237352302852e-06,
+      "loss": 5.1893,
+      "step": 2120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.766320163033882e-06,
+      "loss": 5.0284,
+      "step": 2121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757408701675013e-06,
+      "loss": 4.4095,
+      "step": 2122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.748502977979239e-06,
+      "loss": 5.1867,
+      "step": 2123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739603001693275e-06,
+      "loss": 5.217,
+      "step": 2124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.730708782557546e-06,
+      "loss": 5.0611,
+      "step": 2125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7218203303061746e-06,
+      "loss": 5.4167,
+      "step": 2126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.712937654666971e-06,
+      "loss": 4.6786,
+      "step": 2127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7040607653614335e-06,
+      "loss": 5.2581,
+      "step": 2128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695189672104717e-06,
+      "loss": 5.3403,
+      "step": 2129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686324384605629e-06,
+      "loss": 5.4073,
+      "step": 2130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.677464912566636e-06,
+      "loss": 4.6305,
+      "step": 2131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66861126568383e-06,
+      "loss": 5.2972,
+      "step": 2132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6597634536469326e-06,
+      "loss": 5.0182,
+      "step": 2133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6509214861392785e-06,
+      "loss": 4.9776,
+      "step": 2134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642085372837804e-06,
+      "loss": 5.1141,
+      "step": 2135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.633255123413043e-06,
+      "loss": 5.1117,
+      "step": 2136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6244307475291025e-06,
+      "loss": 5.0514,
+      "step": 2137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.61561225484368e-06,
+      "loss": 5.0094,
+      "step": 2138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606799655008009e-06,
+      "loss": 4.6737,
+      "step": 2139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597992957666893e-06,
+      "loss": 5.1473,
+      "step": 2140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589192172458668e-06,
+      "loss": 5.2802,
+      "step": 2141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5803973090152e-06,
+      "loss": 4.9902,
+      "step": 2142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571608376961875e-06,
+      "loss": 4.9052,
+      "step": 2143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.562825385917586e-06,
+      "loss": 5.4661,
+      "step": 2144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5540483454947285e-06,
+      "loss": 5.1214,
+      "step": 2145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545277265299175e-06,
+      "loss": 4.8249,
+      "step": 2146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5365121549302916e-06,
+      "loss": 4.5764,
+      "step": 2147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5277530239809e-06,
+      "loss": 5.0413,
+      "step": 2148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5189998820372725e-06,
+      "loss": 4.767,
+      "step": 2149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.510252738679136e-06,
+      "loss": 5.4236,
+      "step": 2150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501511603479653e-06,
+      "loss": 5.0046,
+      "step": 2151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492776486005402e-06,
+      "loss": 4.8704,
+      "step": 2152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4840473958163854e-06,
+      "loss": 4.9487,
+      "step": 2153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475324342466003e-06,
+      "loss": 4.195,
+      "step": 2154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4666073355010485e-06,
+      "loss": 5.4903,
+      "step": 2155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457896384461696e-06,
+      "loss": 4.7603,
+      "step": 2156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4491914988815055e-06,
+      "loss": 5.5847,
+      "step": 2157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.440492688287377e-06,
+      "loss": 5.2795,
+      "step": 2158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.431799962199576e-06,
+      "loss": 4.6786,
+      "step": 2159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423113330131708e-06,
+      "loss": 4.9986,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.414432801590703e-06,
+      "loss": 4.9194,
+      "step": 2161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40575838607682e-06,
+      "loss": 4.8368,
+      "step": 2162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397090093083617e-06,
+      "loss": 4.8397,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38842793209796e-06,
+      "loss": 5.1216,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379771912600001e-06,
+      "loss": 4.8685,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371122044063164e-06,
+      "loss": 5.5846,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.362478335954162e-06,
+      "loss": 4.7654,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353840797732939e-06,
+      "loss": 5.2476,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345209438852702e-06,
+      "loss": 5.296,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336584268759892e-06,
+      "loss": 5.3866,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327965296894179e-06,
+      "loss": 5.16,
+      "step": 2171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 5.2224,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.310745985568779e-06,
+      "loss": 4.6929,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302145664954469e-06,
+      "loss": 5.0222,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.293551580257988e-06,
+      "loss": 5.1668,
+      "step": 2175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284963740884981e-06,
+      "loss": 5.2567,
+      "step": 2176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276382156234261e-06,
+      "loss": 4.7709,
+      "step": 2177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267806835697793e-06,
+      "loss": 5.1349,
+      "step": 2178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.259237788660689e-06,
+      "loss": 5.7443,
+      "step": 2179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250675024501193e-06,
+      "loss": 5.1151,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242118552590675e-06,
+      "loss": 5.0435,
+      "step": 2181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233568382293616e-06,
+      "loss": 5.3005,
+      "step": 2182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.225024522967602e-06,
+      "loss": 4.9374,
+      "step": 2183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216486983963311e-06,
+      "loss": 4.6949,
+      "step": 2184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.207955774624506e-06,
+      "loss": 4.7518,
+      "step": 2185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19943090428802e-06,
+      "loss": 5.4177,
+      "step": 2186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190912382283749e-06,
+      "loss": 4.9447,
+      "step": 2187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182400217934643e-06,
+      "loss": 5.02,
+      "step": 2188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.17389442055669e-06,
+      "loss": 4.7113,
+      "step": 2189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165394999458915e-06,
+      "loss": 5.194,
+      "step": 2190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156901963943361e-06,
+      "loss": 4.8004,
+      "step": 2191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148415323305084e-06,
+      "loss": 5.1893,
+      "step": 2192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13993508683214e-06,
+      "loss": 5.1163,
+      "step": 2193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1314612638055766e-06,
+      "loss": 5.2813,
+      "step": 2194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122993863499423e-06,
+      "loss": 5.2048,
+      "step": 2195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114532895180677e-06,
+      "loss": 5.2234,
+      "step": 2196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.106078368109301e-06,
+      "loss": 4.8755,
+      "step": 2197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.097630291538204e-06,
+      "loss": 5.0577,
+      "step": 2198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0891886747132356e-06,
+      "loss": 4.9527,
+      "step": 2199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08075352687318e-06,
+      "loss": 5.1385,
+      "step": 2200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0723248572497365e-06,
+      "loss": 5.0598,
+      "step": 2201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063902675067516e-06,
+      "loss": 5.0314,
+      "step": 2202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.055486989544031e-06,
+      "loss": 4.9068,
+      "step": 2203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047077809889681e-06,
+      "loss": 4.8398,
+      "step": 2204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038675145307747e-06,
+      "loss": 4.982,
+      "step": 2205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.030279004994382e-06,
+      "loss": 5.479,
+      "step": 2206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0218893981385935e-06,
+      "loss": 4.8428,
+      "step": 2207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013506333922242e-06,
+      "loss": 4.6023,
+      "step": 2208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005129821520025e-06,
+      "loss": 5.1211,
+      "step": 2209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996759870099474e-06,
+      "loss": 4.8126,
+      "step": 2210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988396488820936e-06,
+      "loss": 5.0912,
+      "step": 2211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9800396868375675e-06,
+      "loss": 4.6438,
+      "step": 2212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971689473295327e-06,
+      "loss": 4.8596,
+      "step": 2213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.96334585733296e-06,
+      "loss": 5.2758,
+      "step": 2214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.955008848081993e-06,
+      "loss": 5.3967,
+      "step": 2215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946678454666719e-06,
+      "loss": 5.4729,
+      "step": 2216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9383546862041955e-06,
+      "loss": 4.8453,
+      "step": 2217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930037551804226e-06,
+      "loss": 5.0931,
+      "step": 2218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921727060569351e-06,
+      "loss": 5.2697,
+      "step": 2219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.913423221594847e-06,
+      "loss": 5.3073,
+      "step": 2220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905126043968705e-06,
+      "loss": 4.8709,
+      "step": 2221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896835536771626e-06,
+      "loss": 5.0632,
+      "step": 2222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8885517090770135e-06,
+      "loss": 5.277,
+      "step": 2223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.880274569950958e-06,
+      "loss": 4.6664,
+      "step": 2224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872004128452231e-06,
+      "loss": 4.7663,
+      "step": 2225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863740393632276e-06,
+      "loss": 5.236,
+      "step": 2226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855483374535191e-06,
+      "loss": 4.7781,
+      "step": 2227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847233080197728e-06,
+      "loss": 5.1872,
+      "step": 2228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8389895196492785e-06,
+      "loss": 5.1456,
+      "step": 2229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830752701911867e-06,
+      "loss": 5.3849,
+      "step": 2230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822522636000131e-06,
+      "loss": 5.0291,
+      "step": 2231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8142993309213283e-06,
+      "loss": 5.1655,
+      "step": 2232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8060827956753087e-06,
+      "loss": 4.864,
+      "step": 2233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7978730392545184e-06,
+      "loss": 4.9345,
+      "step": 2234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7896700706439826e-06,
+      "loss": 5.31,
+      "step": 2235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7814738988212972e-06,
+      "loss": 5.1756,
+      "step": 2236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7732845327566225e-06,
+      "loss": 4.5538,
+      "step": 2237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 5.0414,
+      "step": 2238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7569262537446794e-06,
+      "loss": 4.775,
+      "step": 2239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748757358700447e-06,
+      "loss": 5.1046,
+      "step": 2240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7405953052202727e-06,
+      "loss": 4.8692,
+      "step": 2241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.732440102236975e-06,
+      "loss": 5.4876,
+      "step": 2242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.724291758675875e-06,
+      "loss": 5.0728,
+      "step": 2243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.716150283454786e-06,
+      "loss": 4.5064,
+      "step": 2244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7080156854840035e-06,
+      "loss": 5.132,
+      "step": 2245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6998879736662975e-06,
+      "loss": 5.33,
+      "step": 2246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6917671568969006e-06,
+      "loss": 4.8326,
+      "step": 2247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.683653244063501e-06,
+      "loss": 4.6378,
+      "step": 2248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6755462440462288e-06,
+      "loss": 5.1157,
+      "step": 2249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.66744616571765e-06,
+      "loss": 5.0878,
+      "step": 2250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.659353017942754e-06,
+      "loss": 5.5365,
+      "step": 2251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6512668095789473e-06,
+      "loss": 5.3966,
+      "step": 2252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6431875494760403e-06,
+      "loss": 5.0751,
+      "step": 2253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6351152464762373e-06,
+      "loss": 4.9067,
+      "step": 2254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.627049909414132e-06,
+      "loss": 5.2524,
+      "step": 2255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6189915471166927e-06,
+      "loss": 5.5216,
+      "step": 2256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.610940168403253e-06,
+      "loss": 4.6528,
+      "step": 2257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.602895782085506e-06,
+      "loss": 4.6817,
+      "step": 2258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.594858396967489e-06,
+      "loss": 4.996,
+      "step": 2259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58682802184558e-06,
+      "loss": 4.8859,
+      "step": 2260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578804665508484e-06,
+      "loss": 5.5829,
+      "step": 2261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.570788336737223e-06,
+      "loss": 4.5921,
+      "step": 2262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5627790443051303e-06,
+      "loss": 5.3123,
+      "step": 2263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5547767969778355e-06,
+      "loss": 5.0221,
+      "step": 2264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5467816035132596e-06,
+      "loss": 5.1458,
+      "step": 2265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.538793472661606e-06,
+      "loss": 4.7244,
+      "step": 2266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5308124131653423e-06,
+      "loss": 4.7928,
+      "step": 2267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5228384337592023e-06,
+      "loss": 5.0745,
+      "step": 2268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5148715431701696e-06,
+      "loss": 5.0509,
+      "step": 2269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.506911750117469e-06,
+      "loss": 5.8305,
+      "step": 2270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4989590633125583e-06,
+      "loss": 5.2939,
+      "step": 2271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4910134914591177e-06,
+      "loss": 4.9,
+      "step": 2272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483075043253039e-06,
+      "loss": 4.8939,
+      "step": 2273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.475143727382421e-06,
+      "loss": 5.7129,
+      "step": 2274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467219552527553e-06,
+      "loss": 4.7538,
+      "step": 2275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4593025273609127e-06,
+      "loss": 4.7391,
+      "step": 2276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 4.8892,
+      "step": 2277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4434899607430806e-06,
+      "loss": 5.141,
+      "step": 2278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.435594436597679e-06,
+      "loss": 4.9304,
+      "step": 2279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4277060967520627e-06,
+      "loss": 4.9538,
+      "step": 2280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4198249498394897e-06,
+      "loss": 5.1489,
+      "step": 2281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4119510044853434e-06,
+      "loss": 4.9298,
+      "step": 2282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4040842693071275e-06,
+      "loss": 5.385,
+      "step": 2283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3962247529144544e-06,
+      "loss": 4.7856,
+      "step": 2284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3883724639090343e-06,
+      "loss": 4.8533,
+      "step": 2285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3805274108846685e-06,
+      "loss": 5.361,
+      "step": 2286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3726896024272406e-06,
+      "loss": 5.0915,
+      "step": 2287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3648590471147035e-06,
+      "loss": 5.2812,
+      "step": 2288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3570357535170715e-06,
+      "loss": 4.3488,
+      "step": 2289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3492197301964145e-06,
+      "loss": 5.5152,
+      "step": 2290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.341410985706842e-06,
+      "loss": 4.8099,
+      "step": 2291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.333609528594499e-06,
+      "loss": 4.9824,
+      "step": 2292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.325815367397557e-06,
+      "loss": 5.2483,
+      "step": 2293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3180285106461973e-06,
+      "loss": 5.3092,
+      "step": 2294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3102489668626125e-06,
+      "loss": 4.9927,
+      "step": 2295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3024767445609896e-06,
+      "loss": 4.7236,
+      "step": 2296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2947118522475e-06,
+      "loss": 5.127,
+      "step": 2297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2869542984202974e-06,
+      "loss": 5.2606,
+      "step": 2298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.279204091569502e-06,
+      "loss": 5.0081,
+      "step": 2299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2714612401771916e-06,
+      "loss": 4.7011,
+      "step": 2300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2637257527173984e-06,
+      "loss": 5.2286,
+      "step": 2301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2559976376560897e-06,
+      "loss": 5.651,
+      "step": 2302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.248276903451171e-06,
+      "loss": 5.3053,
+      "step": 2303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2405635585524566e-06,
+      "loss": 5.0114,
+      "step": 2304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232857611401693e-06,
+      "loss": 5.3149,
+      "step": 2305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2251590704325162e-06,
+      "loss": 5.3178,
+      "step": 2306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2174679440704616e-06,
+      "loss": 4.762,
+      "step": 2307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2097842407329492e-06,
+      "loss": 5.3755,
+      "step": 2308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2021079688292735e-06,
+      "loss": 5.4986,
+      "step": 2309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1944391367605987e-06,
+      "loss": 5.2452,
+      "step": 2310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1867777529199437e-06,
+      "loss": 5.3684,
+      "step": 2311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1791238256921785e-06,
+      "loss": 5.3193,
+      "step": 2312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.171477363454012e-06,
+      "loss": 4.8897,
+      "step": 2313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1638383745739755e-06,
+      "loss": 5.1463,
+      "step": 2314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1562068674124347e-06,
+      "loss": 4.9683,
+      "step": 2315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1485828503215588e-06,
+      "loss": 4.7816,
+      "step": 2316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1409663316453197e-06,
+      "loss": 4.9111,
+      "step": 2317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1333573197194857e-06,
+      "loss": 4.822,
+      "step": 2318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.125755822871607e-06,
+      "loss": 4.4861,
+      "step": 2319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1181618494210097e-06,
+      "loss": 5.079,
+      "step": 2320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1105754076787865e-06,
+      "loss": 4.8788,
+      "step": 2321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1029965059477897e-06,
+      "loss": 4.7235,
+      "step": 2322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.095425152522612e-06,
+      "loss": 5.1758,
+      "step": 2323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087861355689589e-06,
+      "loss": 5.3083,
+      "step": 2324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0803051237267924e-06,
+      "loss": 4.6789,
+      "step": 2325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0727564649040066e-06,
+      "loss": 4.6664,
+      "step": 2326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0652153874827308e-06,
+      "loss": 4.475,
+      "step": 2327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.057681899716165e-06,
+      "loss": 5.2598,
+      "step": 2328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0501560098492056e-06,
+      "loss": 5.0088,
+      "step": 2329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0426377261184327e-06,
+      "loss": 4.8588,
+      "step": 2330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0351270567520995e-06,
+      "loss": 5.2674,
+      "step": 2331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0276240099701324e-06,
+      "loss": 4.7772,
+      "step": 2332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0201285939841006e-06,
+      "loss": 5.3374,
+      "step": 2333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.012640816997242e-06,
+      "loss": 5.0173,
+      "step": 2334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0051606872044202e-06,
+      "loss": 4.888,
+      "step": 2335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997688212792136e-06,
+      "loss": 4.9968,
+      "step": 2336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9902234019385056e-06,
+      "loss": 5.3148,
+      "step": 2337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9827662628132646e-06,
+      "loss": 4.991,
+      "step": 2338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.975316803577749e-06,
+      "loss": 4.9897,
+      "step": 2339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9678750323848893e-06,
+      "loss": 4.5098,
+      "step": 2340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.960440957379204e-06,
+      "loss": 5.0038,
+      "step": 2341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9530145866967897e-06,
+      "loss": 4.8655,
+      "step": 2342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9455959284653e-06,
+      "loss": 4.3695,
+      "step": 2343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.938184990803966e-06,
+      "loss": 5.4478,
+      "step": 2344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9307817818235573e-06,
+      "loss": 4.8994,
+      "step": 2345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.923386309626386e-06,
+      "loss": 5.0184,
+      "step": 2346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9159985823062997e-06,
+      "loss": 4.9461,
+      "step": 2347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.908618607948669e-06,
+      "loss": 4.5539,
+      "step": 2348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.901246394630377e-06,
+      "loss": 5.1423,
+      "step": 2349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8938819504198157e-06,
+      "loss": 4.5916,
+      "step": 2350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8865252833768764e-06,
+      "loss": 4.2338,
+      "step": 2351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.879176401552929e-06,
+      "loss": 5.0636,
+      "step": 2352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8718353129908307e-06,
+      "loss": 5.3664,
+      "step": 2353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8645020257249136e-06,
+      "loss": 5.0269,
+      "step": 2354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8571765477809645e-06,
+      "loss": 4.8935,
+      "step": 2355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8498588871762244e-06,
+      "loss": 4.7899,
+      "step": 2356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8425490519193823e-06,
+      "loss": 5.3168,
+      "step": 2357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8352470500105576e-06,
+      "loss": 5.1396,
+      "step": 2358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.827952889441302e-06,
+      "loss": 5.2845,
+      "step": 2359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8206665781945817e-06,
+      "loss": 5.3653,
+      "step": 2360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.813388124244778e-06,
+      "loss": 4.4348,
+      "step": 2361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806117535557661e-06,
+      "loss": 4.7069,
+      "step": 2362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.798854820090401e-06,
+      "loss": 5.1714,
+      "step": 2363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7915999857915577e-06,
+      "loss": 5.3089,
+      "step": 2364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.784353040601053e-06,
+      "loss": 4.9871,
+      "step": 2365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7771139924501845e-06,
+      "loss": 5.1453,
+      "step": 2366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7698828492615992e-06,
+      "loss": 4.8582,
+      "step": 2367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7626596189492983e-06,
+      "loss": 4.6739,
+      "step": 2368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7554443094186202e-06,
+      "loss": 5.2113,
+      "step": 2369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.748236928566238e-06,
+      "loss": 4.7391,
+      "step": 2370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.741037484280139e-06,
+      "loss": 4.9206,
+      "step": 2371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7338459844396337e-06,
+      "loss": 5.1011,
+      "step": 2372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7266624369153307e-06,
+      "loss": 4.788,
+      "step": 2373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719486849569144e-06,
+      "loss": 4.9496,
+      "step": 2374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.71231923025427e-06,
+      "loss": 5.7367,
+      "step": 2375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7051595868151825e-06,
+      "loss": 4.9868,
+      "step": 2376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6980079270876314e-06,
+      "loss": 5.1055,
+      "step": 2377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6908642588986254e-06,
+      "loss": 4.891,
+      "step": 2378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6837285900664302e-06,
+      "loss": 5.133,
+      "step": 2379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6766009284005546e-06,
+      "loss": 5.2306,
+      "step": 2380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 4.9912,
+      "step": 2381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.662369657761963e-06,
+      "loss": 5.0936,
+      "step": 2382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6552660643644134e-06,
+      "loss": 5.0294,
+      "step": 2383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.648170509283503e-06,
+      "loss": 5.0627,
+      "step": 2384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6410830002848353e-06,
+      "loss": 5.0527,
+      "step": 2385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6340035451252124e-06,
+      "loss": 4.7245,
+      "step": 2386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.626932151552619e-06,
+      "loss": 5.0854,
+      "step": 2387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6198688273062213e-06,
+      "loss": 4.9699,
+      "step": 2388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6128135801163547e-06,
+      "loss": 4.7238,
+      "step": 2389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.605766417704505e-06,
+      "loss": 4.8358,
+      "step": 2390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598727347783321e-06,
+      "loss": 5.0057,
+      "step": 2391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5916963780565897e-06,
+      "loss": 5.4149,
+      "step": 2392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5846735162192304e-06,
+      "loss": 4.911,
+      "step": 2393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5776587699573007e-06,
+      "loss": 4.9403,
+      "step": 2394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.570652146947963e-06,
+      "loss": 5.4263,
+      "step": 2395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.563653654859495e-06,
+      "loss": 5.3632,
+      "step": 2396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5566633013512753e-06,
+      "loss": 5.5245,
+      "step": 2397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5496810940737738e-06,
+      "loss": 4.8265,
+      "step": 2398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5427070406685508e-06,
+      "loss": 4.7791,
+      "step": 2399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5357411487682304e-06,
+      "loss": 4.8624,
+      "step": 2400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.528783425996516e-06,
+      "loss": 5.28,
+      "step": 2401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.521833879968163e-06,
+      "loss": 4.461,
+      "step": 2402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 5.0097,
+      "step": 2403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507959348555842e-06,
+      "loss": 5.0623,
+      "step": 2404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5010343783566104e-06,
+      "loss": 5.1927,
+      "step": 2405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4941176152702073e-06,
+      "loss": 4.8738,
+      "step": 2406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.487209066866565e-06,
+      "loss": 5.2342,
+      "step": 2407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.480308740706623e-06,
+      "loss": 5.3555,
+      "step": 2408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.473416644342328e-06,
+      "loss": 5.2668,
+      "step": 2409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4665327853166076e-06,
+      "loss": 5.3663,
+      "step": 2410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4596571711633864e-06,
+      "loss": 4.6476,
+      "step": 2411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4527898094075573e-06,
+      "loss": 5.2432,
+      "step": 2412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4459307075649907e-06,
+      "loss": 5.0062,
+      "step": 2413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.43907987314251e-06,
+      "loss": 4.9519,
+      "step": 2414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.432237313637892e-06,
+      "loss": 4.945,
+      "step": 2415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4254030365398553e-06,
+      "loss": 5.2905,
+      "step": 2416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.418577049328058e-06,
+      "loss": 5.0318,
+      "step": 2417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.411759359473085e-06,
+      "loss": 5.2985,
+      "step": 2418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4049499744364336e-06,
+      "loss": 4.9503,
+      "step": 2419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.398148901670521e-06,
+      "loss": 4.9297,
+      "step": 2420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.39135614861866e-06,
+      "loss": 4.5377,
+      "step": 2421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3845717227150624e-06,
+      "loss": 5.1104,
+      "step": 2422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377795631384828e-06,
+      "loss": 5.0545,
+      "step": 2423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3710278820439313e-06,
+      "loss": 4.3877,
+      "step": 2424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364268482099218e-06,
+      "loss": 4.6022,
+      "step": 2425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3575174389483955e-06,
+      "loss": 4.9551,
+      "step": 2426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.350774759980027e-06,
+      "loss": 4.7578,
+      "step": 2427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3440404525735215e-06,
+      "loss": 5.4953,
+      "step": 2428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3373145240991213e-06,
+      "loss": 5.3313,
+      "step": 2429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.330596981917903e-06,
+      "loss": 5.098,
+      "step": 2430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323887833381766e-06,
+      "loss": 5.4429,
+      "step": 2431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3171870858334165e-06,
+      "loss": 4.9611,
+      "step": 2432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3104947466063785e-06,
+      "loss": 5.5449,
+      "step": 2433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3038108230249644e-06,
+      "loss": 5.1868,
+      "step": 2434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2971353224042783e-06,
+      "loss": 4.944,
+      "step": 2435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290468252050204e-06,
+      "loss": 5.475,
+      "step": 2436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.283809619259405e-06,
+      "loss": 5.2827,
+      "step": 2437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2771594313193012e-06,
+      "loss": 4.1904,
+      "step": 2438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2705176955080775e-06,
+      "loss": 5.1179,
+      "step": 2439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2638844190946686e-06,
+      "loss": 5.1005,
+      "step": 2440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2572596093387457e-06,
+      "loss": 4.9446,
+      "step": 2441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2506432734907157e-06,
+      "loss": 4.8384,
+      "step": 2442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.244035418791718e-06,
+      "loss": 5.4238,
+      "step": 2443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.237436052473602e-06,
+      "loss": 4.9931,
+      "step": 2444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.230845181758928e-06,
+      "loss": 4.8401,
+      "step": 2445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.224262813860962e-06,
+      "loss": 5.1413,
+      "step": 2446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2176889559836655e-06,
+      "loss": 4.9884,
+      "step": 2447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2111236153216765e-06,
+      "loss": 5.3141,
+      "step": 2448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2045667990603214e-06,
+      "loss": 5.6207,
+      "step": 2449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.198018514375594e-06,
+      "loss": 5.2207,
+      "step": 2450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.19147876843415e-06,
+      "loss": 5.0873,
+      "step": 2451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1849475683932996e-06,
+      "loss": 4.8072,
+      "step": 2452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1784249214010056e-06,
+      "loss": 5.4399,
+      "step": 2453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171910834595864e-06,
+      "loss": 4.4966,
+      "step": 2454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165405315107103e-06,
+      "loss": 5.0477,
+      "step": 2455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1589083700545797e-06,
+      "loss": 4.9174,
+      "step": 2456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1524200065487565e-06,
+      "loss": 4.8758,
+      "step": 2457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.145940231690713e-06,
+      "loss": 4.9536,
+      "step": 2458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1394690525721275e-06,
+      "loss": 4.7445,
+      "step": 2459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133006476275268e-06,
+      "loss": 5.4213,
+      "step": 2460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.126552509872988e-06,
+      "loss": 5.065,
+      "step": 2461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1201071604287227e-06,
+      "loss": 5.2815,
+      "step": 2462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.113670434996472e-06,
+      "loss": 5.1745,
+      "step": 2463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1072423406207966e-06,
+      "loss": 4.667,
+      "step": 2464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.100822884336815e-06,
+      "loss": 5.4248,
+      "step": 2465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09441207317019e-06,
+      "loss": 4.6545,
+      "step": 2466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.088009914137119e-06,
+      "loss": 4.8417,
+      "step": 2467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0816164142443374e-06,
+      "loss": 5.1816,
+      "step": 2468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.075231580489098e-06,
+      "loss": 4.7795,
+      "step": 2469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0688554198591726e-06,
+      "loss": 4.7742,
+      "step": 2470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.062487939332836e-06,
+      "loss": 5.0506,
+      "step": 2471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0561291458788736e-06,
+      "loss": 5.6578,
+      "step": 2472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0497790464565527e-06,
+      "loss": 4.9851,
+      "step": 2473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0434376480156294e-06,
+      "loss": 5.2079,
+      "step": 2474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0371049574963385e-06,
+      "loss": 5.1386,
+      "step": 2475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0307809818293846e-06,
+      "loss": 4.8698,
+      "step": 2476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.024465727935928e-06,
+      "loss": 4.5043,
+      "step": 2477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0181592027275933e-06,
+      "loss": 4.524,
+      "step": 2478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.011861413106445e-06,
+      "loss": 4.5609,
+      "step": 2479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0055723659649907e-06,
+      "loss": 4.7533,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999292068186166e-06,
+      "loss": 4.8989,
+      "step": 2481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.99302052664334e-06,
+      "loss": 5.0546,
+      "step": 2482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9867577482002896e-06,
+      "loss": 5.235,
+      "step": 2483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9805037397112036e-06,
+      "loss": 4.8783,
+      "step": 2484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9742585080206754e-06,
+      "loss": 5.3274,
+      "step": 2485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.968022059963687e-06,
+      "loss": 5.9337,
+      "step": 2486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961794402365611e-06,
+      "loss": 4.8613,
+      "step": 2487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.955575542042201e-06,
+      "loss": 4.8807,
+      "step": 2488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.949365485799579e-06,
+      "loss": 5.2037,
+      "step": 2489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9431642404342354e-06,
+      "loss": 5.0754,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936971812733012e-06,
+      "loss": 4.7229,
+      "step": 2491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.930788209473109e-06,
+      "loss": 5.3113,
+      "step": 2492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9246134374220614e-06,
+      "loss": 4.8299,
+      "step": 2493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9184475033377414e-06,
+      "loss": 4.5803,
+      "step": 2494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912290413968352e-06,
+      "loss": 4.9414,
+      "step": 2495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9061421760524046e-06,
+      "loss": 5.1134,
+      "step": 2496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9000027963187372e-06,
+      "loss": 5.2394,
+      "step": 2497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8938722814864863e-06,
+      "loss": 5.2815,
+      "step": 2498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8877506382650879e-06,
+      "loss": 5.2007,
+      "step": 2499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8816378733542683e-06,
+      "loss": 4.9677,
+      "step": 2500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8755339934440331e-06,
+      "loss": 4.1506,
+      "step": 2501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8694390052146737e-06,
+      "loss": 5.531,
+      "step": 2502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8633529153367435e-06,
+      "loss": 4.628,
+      "step": 2503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8572757304710575e-06,
+      "loss": 5.5932,
+      "step": 2504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8512074572686822e-06,
+      "loss": 5.3995,
+      "step": 2505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.845148102370935e-06,
+      "loss": 4.9142,
+      "step": 2506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8390976724093735e-06,
+      "loss": 4.5101,
+      "step": 2507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8330561740057839e-06,
+      "loss": 5.1636,
+      "step": 2508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.827023613772181e-06,
+      "loss": 4.3459,
+      "step": 2509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.820999998310795e-06,
+      "loss": 5.0047,
+      "step": 2510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8149853342140644e-06,
+      "loss": 4.2515,
+      "step": 2511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8089796280646398e-06,
+      "loss": 4.9195,
+      "step": 2512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8029828864353583e-06,
+      "loss": 5.2617,
+      "step": 2513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.796995115889254e-06,
+      "loss": 4.9443,
+      "step": 2514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.791016322979533e-06,
+      "loss": 4.43,
+      "step": 2515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7850465142495831e-06,
+      "loss": 4.2079,
+      "step": 2516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7790856962329584e-06,
+      "loss": 4.5018,
+      "step": 2517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7731338754533722e-06,
+      "loss": 5.4157,
+      "step": 2518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7671910584246932e-06,
+      "loss": 4.3732,
+      "step": 2519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761257251650933e-06,
+      "loss": 4.8543,
+      "step": 2520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7553324616262401e-06,
+      "loss": 4.94,
+      "step": 2521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7494166948349057e-06,
+      "loss": 5.2381,
+      "step": 2522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743509957751336e-06,
+      "loss": 5.1557,
+      "step": 2523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7376122568400533e-06,
+      "loss": 4.8259,
+      "step": 2524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7317235985556946e-06,
+      "loss": 5.4847,
+      "step": 2525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7258439893430023e-06,
+      "loss": 5.5029,
+      "step": 2526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7199734356368114e-06,
+      "loss": 4.7395,
+      "step": 2527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.714111943862048e-06,
+      "loss": 5.3021,
+      "step": 2528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7082595204337183e-06,
+      "loss": 4.9042,
+      "step": 2529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7024161717569054e-06,
+      "loss": 4.7235,
+      "step": 2530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6965819042267651e-06,
+      "loss": 4.6828,
+      "step": 2531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6907567242285084e-06,
+      "loss": 5.251,
+      "step": 2532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6849406381374033e-06,
+      "loss": 3.9363,
+      "step": 2533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6791336523187608e-06,
+      "loss": 5.0281,
+      "step": 2534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6733357731279375e-06,
+      "loss": 4.8343,
+      "step": 2535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6675470069103228e-06,
+      "loss": 5.3384,
+      "step": 2536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6617673600013295e-06,
+      "loss": 5.2591,
+      "step": 2537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6559968387263935e-06,
+      "loss": 5.0089,
+      "step": 2538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.650235449400962e-06,
+      "loss": 4.8609,
+      "step": 2539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6444831983304832e-06,
+      "loss": 4.7309,
+      "step": 2540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6387400918104156e-06,
+      "loss": 4.6291,
+      "step": 2541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6330061361261985e-06,
+      "loss": 4.2468,
+      "step": 2542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6272813375532637e-06,
+      "loss": 4.501,
+      "step": 2543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6215657023570118e-06,
+      "loss": 5.0121,
+      "step": 2544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.615859236792824e-06,
+      "loss": 5.0428,
+      "step": 2545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6101619471060415e-06,
+      "loss": 4.5576,
+      "step": 2546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6044738395319648e-06,
+      "loss": 5.4032,
+      "step": 2547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5987949202958454e-06,
+      "loss": 4.2911,
+      "step": 2548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5931251956128756e-06,
+      "loss": 4.7631,
+      "step": 2549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 5.346,
+      "step": 2550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5818133547168457e-06,
+      "loss": 4.7034,
+      "step": 2551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5761712508838377e-06,
+      "loss": 5.1047,
+      "step": 2552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5705383663640616e-06,
+      "loss": 5.1441,
+      "step": 2553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5649147073223303e-06,
+      "loss": 4.6449,
+      "step": 2554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5593002799133627e-06,
+      "loss": 4.961,
+      "step": 2555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5536950902817693e-06,
+      "loss": 4.4027,
+      "step": 2556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5480991445620541e-06,
+      "loss": 5.1859,
+      "step": 2557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5425124488786013e-06,
+      "loss": 5.1332,
+      "step": 2558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5369350093456747e-06,
+      "loss": 4.796,
+      "step": 2559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5313668320674025e-06,
+      "loss": 5.1317,
+      "step": 2560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5258079231377853e-06,
+      "loss": 5.1241,
+      "step": 2561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5202582886406747e-06,
+      "loss": 5.0999,
+      "step": 2562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5147179346497665e-06,
+      "loss": 5.2807,
+      "step": 2563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.509186867228607e-06,
+      "loss": 4.357,
+      "step": 2564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5036650924305785e-06,
+      "loss": 5.0778,
+      "step": 2565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4981526162988902e-06,
+      "loss": 4.9603,
+      "step": 2566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4926494448665773e-06,
+      "loss": 4.8795,
+      "step": 2567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4871555841564889e-06,
+      "loss": 5.0385,
+      "step": 2568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4816710401812861e-06,
+      "loss": 4.8887,
+      "step": 2569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.476195818943431e-06,
+      "loss": 5.3169,
+      "step": 2570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4707299264351914e-06,
+      "loss": 5.2643,
+      "step": 2571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.465273368638611e-06,
+      "loss": 4.7244,
+      "step": 2572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4598261515255275e-06,
+      "loss": 4.3784,
+      "step": 2573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4543882810575527e-06,
+      "loss": 5.096,
+      "step": 2574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4489597631860708e-06,
+      "loss": 4.892,
+      "step": 2575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.443540603852227e-06,
+      "loss": 5.0579,
+      "step": 2576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4381308089869283e-06,
+      "loss": 4.7361,
+      "step": 2577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4327303845108299e-06,
+      "loss": 4.657,
+      "step": 2578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4273393363343324e-06,
+      "loss": 5.3106,
+      "step": 2579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4219576703575733e-06,
+      "loss": 4.7292,
+      "step": 2580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4165853924704299e-06,
+      "loss": 4.9793,
+      "step": 2581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4112225085524921e-06,
+      "loss": 4.6937,
+      "step": 2582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4058690244730777e-06,
+      "loss": 4.7273,
+      "step": 2583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.400524946091213e-06,
+      "loss": 4.2588,
+      "step": 2584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3951902792556337e-06,
+      "loss": 5.0244,
+      "step": 2585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3898650298047723e-06,
+      "loss": 5.0506,
+      "step": 2586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3845492035667562e-06,
+      "loss": 4.2999,
+      "step": 2587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3792428063593987e-06,
+      "loss": 4.9877,
+      "step": 2588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.373945843990192e-06,
+      "loss": 4.5569,
+      "step": 2589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.368658322256311e-06,
+      "loss": 4.8628,
+      "step": 2590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3633802469445866e-06,
+      "loss": 4.7298,
+      "step": 2591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3581116238315194e-06,
+      "loss": 4.8075,
+      "step": 2592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3528524586832603e-06,
+      "loss": 5.5911,
+      "step": 2593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3476027572556127e-06,
+      "loss": 4.9771,
+      "step": 2594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.342362525294021e-06,
+      "loss": 5.1276,
+      "step": 2595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3371317685335671e-06,
+      "loss": 4.3483,
+      "step": 2596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3319104926989613e-06,
+      "loss": 4.8787,
+      "step": 2597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3266987035045375e-06,
+      "loss": 5.0185,
+      "step": 2598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3214964066542458e-06,
+      "loss": 5.2302,
+      "step": 2599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.316303607841658e-06,
+      "loss": 5.283,
+      "step": 2600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.311120312749935e-06,
+      "loss": 4.8743,
+      "step": 2601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3059465270518469e-06,
+      "loss": 4.8811,
+      "step": 2602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3007822564097538e-06,
+      "loss": 4.4642,
+      "step": 2603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2956275064756007e-06,
+      "loss": 5.5316,
+      "step": 2604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2904822828909146e-06,
+      "loss": 5.1186,
+      "step": 2605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2853465912867979e-06,
+      "loss": 4.7505,
+      "step": 2606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2802204372839178e-06,
+      "loss": 4.7379,
+      "step": 2607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2751038264925041e-06,
+      "loss": 4.9497,
+      "step": 2608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2699967645123446e-06,
+      "loss": 5.0588,
+      "step": 2609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2648992569327734e-06,
+      "loss": 4.9834,
+      "step": 2610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.259811309332669e-06,
+      "loss": 4.8881,
+      "step": 2611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2547329272804477e-06,
+      "loss": 4.8735,
+      "step": 2612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2496641163340562e-06,
+      "loss": 4.7725,
+      "step": 2613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2446048820409685e-06,
+      "loss": 4.9049,
+      "step": 2614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2395552299381742e-06,
+      "loss": 4.912,
+      "step": 2615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.234515165552177e-06,
+      "loss": 4.8825,
+      "step": 2616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2294846943989892e-06,
+      "loss": 4.4752,
+      "step": 2617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.224463821984122e-06,
+      "loss": 4.6031,
+      "step": 2618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2194525538025826e-06,
+      "loss": 5.119,
+      "step": 2619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2144508953388655e-06,
+      "loss": 4.6682,
+      "step": 2620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2094588520669504e-06,
+      "loss": 5.3079,
+      "step": 2621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2044764294502908e-06,
+      "loss": 5.046,
+      "step": 2622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1995036329418153e-06,
+      "loss": 4.8926,
+      "step": 2623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1945404679839123e-06,
+      "loss": 4.4488,
+      "step": 2624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1895869400084325e-06,
+      "loss": 4.6801,
+      "step": 2625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1846430544366783e-06,
+      "loss": 5.1002,
+      "step": 2626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1797088166794002e-06,
+      "loss": 5.0444,
+      "step": 2627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1747842321367886e-06,
+      "loss": 4.826,
+      "step": 2628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.169869306198469e-06,
+      "loss": 4.9583,
+      "step": 2629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1649640442434983e-06,
+      "loss": 4.6457,
+      "step": 2630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.160068451640355e-06,
+      "loss": 4.8652,
+      "step": 2631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1551825337469346e-06,
+      "loss": 5.0461,
+      "step": 2632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1503062959105448e-06,
+      "loss": 5.2645,
+      "step": 2633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1454397434679022e-06,
+      "loss": 5.216,
+      "step": 2634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.140582881745117e-06,
+      "loss": 4.7196,
+      "step": 2635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1357357160577009e-06,
+      "loss": 5.0051,
+      "step": 2636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.130898251710547e-06,
+      "loss": 5.1401,
+      "step": 2637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.126070493997936e-06,
+      "loss": 4.867,
+      "step": 2638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1212524482035236e-06,
+      "loss": 4.4171,
+      "step": 2639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1164441196003362e-06,
+      "loss": 4.8663,
+      "step": 2640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1116455134507665e-06,
+      "loss": 5.1012,
+      "step": 2641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1068566350065635e-06,
+      "loss": 4.7681,
+      "step": 2642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1020774895088338e-06,
+      "loss": 5.1703,
+      "step": 2643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.09730808218803e-06,
+      "loss": 5.1144,
+      "step": 2644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0925484182639467e-06,
+      "loss": 5.2591,
+      "step": 2645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0877985029457172e-06,
+      "loss": 4.8074,
+      "step": 2646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0830583414318007e-06,
+      "loss": 5.0936,
+      "step": 2647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0783279389099887e-06,
+      "loss": 5.351,
+      "step": 2648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0736073005573856e-06,
+      "loss": 4.8482,
+      "step": 2649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0688964315404116e-06,
+      "loss": 5.2302,
+      "step": 2650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0641953370147973e-06,
+      "loss": 5.104,
+      "step": 2651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0595040221255736e-06,
+      "loss": 4.8638,
+      "step": 2652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0548224920070692e-06,
+      "loss": 4.7967,
+      "step": 2653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0501507517829012e-06,
+      "loss": 4.9088,
+      "step": 2654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0454888065659775e-06,
+      "loss": 4.9879,
+      "step": 2655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040836661458482e-06,
+      "loss": 5.1175,
+      "step": 2656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0361943215518743e-06,
+      "loss": 5.5324,
+      "step": 2657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0315617919268817e-06,
+      "loss": 5.1927,
+      "step": 2658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0269390776534972e-06,
+      "loss": 4.0718,
+      "step": 2659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0223261837909703e-06,
+      "loss": 4.8746,
+      "step": 2660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.017723115387802e-06,
+      "loss": 4.6692,
+      "step": 2661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.013129877481741e-06,
+      "loss": 4.7868,
+      "step": 2662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0085464750997763e-06,
+      "loss": 4.8694,
+      "step": 2663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0039729132581332e-06,
+      "loss": 5.3435,
+      "step": 2664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.994091969622678e-07,
+      "loss": 4.9615,
+      "step": 2665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.9485533120686e-07,
+      "loss": 4.9738,
+      "step": 2666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 5.1588,
+      "step": 2667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.857771712422282e-07,
+      "loss": 5.3394,
+      "step": 2668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.81252886968439e-07,
+      "loss": 5.1639,
+      "step": 2669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.76738473105966e-07,
+      "loss": 5.0288,
+      "step": 2670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.7223393459553e-07,
+      "loss": 5.5039,
+      "step": 2671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.677392763670457e-07,
+      "loss": 5.6598,
+      "step": 2672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.632545033396147e-07,
+      "loss": 5.2658,
+      "step": 2673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.587796204215183e-07,
+      "loss": 5.2923,
+      "step": 2674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.543146325102148e-07,
+      "loss": 5.569,
+      "step": 2675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.49859544492332e-07,
+      "loss": 4.7845,
+      "step": 2676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.45414361243665e-07,
+      "loss": 5.2123,
+      "step": 2677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.409790876291658e-07,
+      "loss": 5.0153,
+      "step": 2678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.365537285029447e-07,
+      "loss": 5.1241,
+      "step": 2679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.321382887082564e-07,
+      "loss": 4.8362,
+      "step": 2680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.277327730775054e-07,
+      "loss": 5.1563,
+      "step": 2681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.233371864322304e-07,
+      "loss": 5.5246,
+      "step": 2682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.189515335831045e-07,
+      "loss": 4.7493,
+      "step": 2683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.145758193299303e-07,
+      "loss": 4.9171,
+      "step": 2684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.102100484616316e-07,
+      "loss": 4.8045,
+      "step": 2685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.058542257562519e-07,
+      "loss": 4.8492,
+      "step": 2686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.015083559809434e-07,
+      "loss": 5.0665,
+      "step": 2687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.971724438919704e-07,
+      "loss": 4.9166,
+      "step": 2688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.928464942346949e-07,
+      "loss": 4.3634,
+      "step": 2689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.885305117435794e-07,
+      "loss": 5.1596,
+      "step": 2690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.842245011421735e-07,
+      "loss": 4.8467,
+      "step": 2691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.799284671431174e-07,
+      "loss": 5.2685,
+      "step": 2692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.756424144481313e-07,
+      "loss": 5.129,
+      "step": 2693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.713663477480105e-07,
+      "loss": 5.0189,
+      "step": 2694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.671002717226218e-07,
+      "loss": 4.6716,
+      "step": 2695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.628441910408991e-07,
+      "loss": 5.0609,
+      "step": 2696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.585981103608343e-07,
+      "loss": 4.9476,
+      "step": 2697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.543620343294789e-07,
+      "loss": 5.2993,
+      "step": 2698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.501359675829313e-07,
+      "loss": 5.0644,
+      "step": 2699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.459199147463371e-07,
+      "loss": 4.928,
+      "step": 2700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.417138804338832e-07,
+      "loss": 4.8491,
+      "step": 2701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.375178692487895e-07,
+      "loss": 4.8832,
+      "step": 2702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.333318857833084e-07,
+      "loss": 5.1938,
+      "step": 2703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.291559346187161e-07,
+      "loss": 5.0225,
+      "step": 2704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.249900203253114e-07,
+      "loss": 5.258,
+      "step": 2705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.208341474624071e-07,
+      "loss": 5.3305,
+      "step": 2706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.166883205783239e-07,
+      "loss": 5.1763,
+      "step": 2707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.12552544210392e-07,
+      "loss": 5.1231,
+      "step": 2708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.08426822884939e-07,
+      "loss": 5.3985,
+      "step": 2709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.043111611172894e-07,
+      "loss": 5.1196,
+      "step": 2710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.002055634117578e-07,
+      "loss": 4.8082,
+      "step": 2711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.961100342616457e-07,
+      "loss": 4.2563,
+      "step": 2712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.92024578149232e-07,
+      "loss": 4.4488,
+      "step": 2713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.879491995457734e-07,
+      "loss": 4.6146,
+      "step": 2714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.838839029114986e-07,
+      "loss": 4.598,
+      "step": 2715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.798286926956e-07,
+      "loss": 4.706,
+      "step": 2716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.757835733362329e-07,
+      "loss": 4.901,
+      "step": 2717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.71748549260507e-07,
+      "loss": 4.7496,
+      "step": 2718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.677236248844855e-07,
+      "loss": 5.1431,
+      "step": 2719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.637088046131758e-07,
+      "loss": 5.4055,
+      "step": 2720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.597040928405297e-07,
+      "loss": 4.8771,
+      "step": 2721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.557094939494325e-07,
+      "loss": 5.5004,
+      "step": 2722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.517250123117059e-07,
+      "loss": 5.1061,
+      "step": 2723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.477506522880951e-07,
+      "loss": 5.7242,
+      "step": 2724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.437864182282706e-07,
+      "loss": 5.0693,
+      "step": 2725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.398323144708186e-07,
+      "loss": 4.3766,
+      "step": 2726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.358883453432398e-07,
+      "loss": 4.7123,
+      "step": 2727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.319545151619434e-07,
+      "loss": 4.9422,
+      "step": 2728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.280308282322413e-07,
+      "loss": 4.9758,
+      "step": 2729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.241172888483428e-07,
+      "loss": 5.2402,
+      "step": 2730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.202139012933562e-07,
+      "loss": 5.114,
+      "step": 2731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.163206698392744e-07,
+      "loss": 5.3363,
+      "step": 2732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.124375987469767e-07,
+      "loss": 4.5294,
+      "step": 2733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.085646922662248e-07,
+      "loss": 5.256,
+      "step": 2734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.047019546356515e-07,
+      "loss": 4.7441,
+      "step": 2735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.008493900827673e-07,
+      "loss": 4.7951,
+      "step": 2736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.970070028239373e-07,
+      "loss": 5.1318,
+      "step": 2737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.931747970644021e-07,
+      "loss": 4.7214,
+      "step": 2738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8935277699825e-07,
+      "loss": 4.9651,
+      "step": 2739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.855409468084251e-07,
+      "loss": 4.316,
+      "step": 2740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.817393106667202e-07,
+      "loss": 4.5725,
+      "step": 2741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.779478727337674e-07,
+      "loss": 4.6831,
+      "step": 2742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.741666371590428e-07,
+      "loss": 5.0537,
+      "step": 2743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.703956080808515e-07,
+      "loss": 4.7235,
+      "step": 2744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.666347896263326e-07,
+      "loss": 4.8368,
+      "step": 2745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.628841859114499e-07,
+      "loss": 4.687,
+      "step": 2746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.591438010409812e-07,
+      "loss": 4.9104,
+      "step": 2747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.554136391085308e-07,
+      "loss": 5.1536,
+      "step": 2748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.516937041965099e-07,
+      "loss": 4.262,
+      "step": 2749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.479840003761339e-07,
+      "loss": 4.3619,
+      "step": 2750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.442845317074253e-07,
+      "loss": 4.9856,
+      "step": 2751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.405953022392031e-07,
+      "loss": 4.2898,
+      "step": 2752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.369163160090808e-07,
+      "loss": 4.8874,
+      "step": 2753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.332475770434598e-07,
+      "loss": 5.3947,
+      "step": 2754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.295890893575296e-07,
+      "loss": 4.4546,
+      "step": 2755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.259408569552594e-07,
+      "loss": 4.7273,
+      "step": 2756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.223028838293898e-07,
+      "loss": 4.4113,
+      "step": 2757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.186751739614405e-07,
+      "loss": 4.6808,
+      "step": 2758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.150577313216954e-07,
+      "loss": 5.1101,
+      "step": 2759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.114505598692011e-07,
+      "loss": 5.3561,
+      "step": 2760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.07853663551764e-07,
+      "loss": 4.6243,
+      "step": 2761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.042670463059452e-07,
+      "loss": 4.8768,
+      "step": 2762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.006907120570559e-07,
+      "loss": 4.6429,
+      "step": 2763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.97124664719152e-07,
+      "loss": 5.0659,
+      "step": 2764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.935689081950346e-07,
+      "loss": 5.2044,
+      "step": 2765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.900234463762367e-07,
+      "loss": 4.6007,
+      "step": 2766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.864882831430274e-07,
+      "loss": 4.8412,
+      "step": 2767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.829634223644076e-07,
+      "loss": 5.0699,
+      "step": 2768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.794488678980981e-07,
+      "loss": 5.09,
+      "step": 2769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.759446235905431e-07,
+      "loss": 5.1681,
+      "step": 2770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.724506932769014e-07,
+      "loss": 4.6311,
+      "step": 2771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.689670807810443e-07,
+      "loss": 4.7775,
+      "step": 2772
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.654937899155511e-07,
+      "loss": 4.9153,
+      "step": 2773
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.620308244817041e-07,
+      "loss": 4.8203,
+      "step": 2774
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.585781882694885e-07,
+      "loss": 5.1915,
+      "step": 2775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.551358850575794e-07,
+      "loss": 4.5119,
+      "step": 2776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.517039186133433e-07,
+      "loss": 4.6659,
+      "step": 2777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.482822926928422e-07,
+      "loss": 5.1254,
+      "step": 2778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.448710110408128e-07,
+      "loss": 5.1925,
+      "step": 2779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.414700773906722e-07,
+      "loss": 5.3479,
+      "step": 2780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.380794954645141e-07,
+      "loss": 4.8399,
+      "step": 2781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.346992689731023e-07,
+      "loss": 4.6783,
+      "step": 2782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.313294016158666e-07,
+      "loss": 5.0103,
+      "step": 2783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.279698970809011e-07,
+      "loss": 5.5592,
+      "step": 2784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.246207590449537e-07,
+      "loss": 5.0901,
+      "step": 2785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.212819911734323e-07,
+      "loss": 5.316,
+      "step": 2786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.179535971203953e-07,
+      "loss": 4.6317,
+      "step": 2787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.146355805285452e-07,
+      "loss": 4.6429,
+      "step": 2788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.113279450292252e-07,
+      "loss": 5.2957,
+      "step": 2789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.08030694242423e-07,
+      "loss": 5.4132,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.04743831776755e-07,
+      "loss": 4.7259,
+      "step": 2791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.014673612294718e-07,
+      "loss": 4.8449,
+      "step": 2792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.982012861864504e-07,
+      "loss": 5.3344,
+      "step": 2793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.949456102221894e-07,
+      "loss": 4.9454,
+      "step": 2794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.917003368998063e-07,
+      "loss": 5.1913,
+      "step": 2795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.884654697710322e-07,
+      "loss": 4.9545,
+      "step": 2796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.852410123762164e-07,
+      "loss": 5.229,
+      "step": 2797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.820269682443068e-07,
+      "loss": 5.3758,
+      "step": 2798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.788233408928588e-07,
+      "loss": 4.3107,
+      "step": 2799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.756301338280267e-07,
+      "loss": 4.5234,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.724473505445626e-07,
+      "loss": 5.0173,
+      "step": 2801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.6927499452580574e-07,
+      "loss": 4.9526,
+      "step": 2802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.661130692436877e-07,
+      "loss": 5.0731,
+      "step": 2803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6296157815872155e-07,
+      "loss": 5.006,
+      "step": 2804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5982052472000383e-07,
+      "loss": 5.1062,
+      "step": 2805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5668991236520243e-07,
+      "loss": 5.5579,
+      "step": 2806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5356974452056557e-07,
+      "loss": 4.386,
+      "step": 2807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.50460024600905e-07,
+      "loss": 5.1556,
+      "step": 2808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.473607560096005e-07,
+      "loss": 4.8819,
+      "step": 2809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 5.0127,
+      "step": 2810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.411935863683803e-07,
+      "loss": 5.0621,
+      "step": 2811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.381256920680155e-07,
+      "loss": 4.8559,
+      "step": 2812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3506826259510416e-07,
+      "loss": 4.9754,
+      "step": 2813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.320213012957941e-07,
+      "loss": 4.7966,
+      "step": 2814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2898481150478013e-07,
+      "loss": 5.3434,
+      "step": 2815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2595879654529605e-07,
+      "loss": 5.2216,
+      "step": 2816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2294325972911274e-07,
+      "loss": 4.6952,
+      "step": 2817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1993820435653034e-07,
+      "loss": 5.2094,
+      "step": 2818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1694363371638126e-07,
+      "loss": 4.987,
+      "step": 2819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1395955108601947e-07,
+      "loss": 4.6845,
+      "step": 2820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.109859597313237e-07,
+      "loss": 4.9553,
+      "step": 2821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0802286290668956e-07,
+      "loss": 4.7679,
+      "step": 2822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 5.0522,
+      "step": 2823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0212816580775585e-07,
+      "loss": 5.0172,
+      "step": 2824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9919657198480453e-07,
+      "loss": 4.385,
+      "step": 2825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.962754855946038e-07,
+      "loss": 5.019,
+      "step": 2826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.933649098340886e-07,
+      "loss": 4.2975,
+      "step": 2827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9046484788868746e-07,
+      "loss": 4.6248,
+      "step": 2828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.875753029323226e-07,
+      "loss": 5.3027,
+      "step": 2829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.846962781274055e-07,
+      "loss": 5.825,
+      "step": 2830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.818277766248357e-07,
+      "loss": 5.5031,
+      "step": 2831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7896980156399533e-07,
+      "loss": 4.7872,
+      "step": 2832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.761223560727445e-07,
+      "loss": 5.2607,
+      "step": 2833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7328544326742155e-07,
+      "loss": 5.4768,
+      "step": 2834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.704590662528351e-07,
+      "loss": 5.4653,
+      "step": 2835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6764322812226416e-07,
+      "loss": 5.3699,
+      "step": 2836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6483793195745686e-07,
+      "loss": 4.7877,
+      "step": 2837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.620431808286184e-07,
+      "loss": 4.5863,
+      "step": 2838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5925897779441663e-07,
+      "loss": 4.6713,
+      "step": 2839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.564853259019752e-07,
+      "loss": 5.1101,
+      "step": 2840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5372222818686685e-07,
+      "loss": 4.8963,
+      "step": 2841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5096968767311834e-07,
+      "loss": 5.2228,
+      "step": 2842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.482277073731988e-07,
+      "loss": 4.834,
+      "step": 2843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.454962902880199e-07,
+      "loss": 5.6606,
+      "step": 2844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.427754394069338e-07,
+      "loss": 4.8979,
+      "step": 2845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4006515770773076e-07,
+      "loss": 4.7597,
+      "step": 2846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3736544815663017e-07,
+      "loss": 4.6179,
+      "step": 2847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.34676313708282e-07,
+      "loss": 4.7012,
+      "step": 2848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.319977573057642e-07,
+      "loss": 4.9011,
+      "step": 2849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.293297818805752e-07,
+      "loss": 4.6616,
+      "step": 2850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.266723903526359e-07,
+      "loss": 4.8252,
+      "step": 2851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2402558563027875e-07,
+      "loss": 5.4724,
+      "step": 2852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.213893706102567e-07,
+      "loss": 5.0164,
+      "step": 2853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.187637481777284e-07,
+      "loss": 5.4649,
+      "step": 2854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.161487212062597e-07,
+      "loss": 4.7952,
+      "step": 2855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.135442925578236e-07,
+      "loss": 5.3605,
+      "step": 2856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.109504650827899e-07,
+      "loss": 4.6091,
+      "step": 2857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.083672416199301e-07,
+      "loss": 5.6734,
+      "step": 2858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0579462499640724e-07,
+      "loss": 5.1408,
+      "step": 2859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.032326180277767e-07,
+      "loss": 4.6121,
+      "step": 2860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0068122351798347e-07,
+      "loss": 4.7205,
+      "step": 2861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9814044425935605e-07,
+      "loss": 5.0505,
+      "step": 2862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9561028303260574e-07,
+      "loss": 4.68,
+      "step": 2863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9309074260682523e-07,
+      "loss": 4.5284,
+      "step": 2864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.905818257394799e-07,
+      "loss": 5.1082,
+      "step": 2865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8808353517641217e-07,
+      "loss": 5.3991,
+      "step": 2866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8559587365183164e-07,
+      "loss": 4.7999,
+      "step": 2867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.83118843888317e-07,
+      "loss": 4.7739,
+      "step": 2868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8065244859680983e-07,
+      "loss": 4.5256,
+      "step": 2869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7819669047661533e-07,
+      "loss": 4.4446,
+      "step": 2870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.757515722153925e-07,
+      "loss": 4.9036,
+      "step": 2871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.733170964891607e-07,
+      "loss": 5.2935,
+      "step": 2872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7089326596228983e-07,
+      "loss": 4.9784,
+      "step": 2873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6848008328749676e-07,
+      "loss": 4.5492,
+      "step": 2874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6607755110584886e-07,
+      "loss": 4.7249,
+      "step": 2875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.636856720467573e-07,
+      "loss": 5.0152,
+      "step": 2876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6130444872797143e-07,
+      "loss": 4.685,
+      "step": 2877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.589338837555799e-07,
+      "loss": 4.6551,
+      "step": 2878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5657397972400523e-07,
+      "loss": 4.2997,
+      "step": 2879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.542247392160058e-07,
+      "loss": 4.9919,
+      "step": 2880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.518861648026649e-07,
+      "loss": 4.7243,
+      "step": 2881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4955825904339647e-07,
+      "loss": 4.9593,
+      "step": 2882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.472410244859347e-07,
+      "loss": 4.8361,
+      "step": 2883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.449344636663375e-07,
+      "loss": 4.7268,
+      "step": 2884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4263857910898025e-07,
+      "loss": 5.1588,
+      "step": 2885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4035337332655504e-07,
+      "loss": 5.037,
+      "step": 2886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3807884882006583e-07,
+      "loss": 5.2703,
+      "step": 2887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3581500807882462e-07,
+      "loss": 4.757,
+      "step": 2888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3356185358045403e-07,
+      "loss": 5.4255,
+      "step": 2889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3131938779088036e-07,
+      "loss": 4.8326,
+      "step": 2890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2908761316432936e-07,
+      "loss": 4.327,
+      "step": 2891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2686653214333053e-07,
+      "loss": 4.8994,
+      "step": 2892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.24656147158705e-07,
+      "loss": 5.236,
+      "step": 2893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2245646062957205e-07,
+      "loss": 5.0185,
+      "step": 2894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2026747496333822e-07,
+      "loss": 5.188,
+      "step": 2895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1808919255570382e-07,
+      "loss": 4.9719,
+      "step": 2896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1592161579065184e-07,
+      "loss": 4.8434,
+      "step": 2897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1376474704044693e-07,
+      "loss": 4.6038,
+      "step": 2898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1161858866563968e-07,
+      "loss": 4.862,
+      "step": 2899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0948314301505346e-07,
+      "loss": 4.7611,
+      "step": 2900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0735841242578992e-07,
+      "loss": 4.8203,
+      "step": 2901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0524439922322336e-07,
+      "loss": 4.9281,
+      "step": 2902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0314110572100087e-07,
+      "loss": 4.6153,
+      "step": 2903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0104853422103332e-07,
+      "loss": 4.9123,
+      "step": 2904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9896668701349987e-07,
+      "loss": 5.3002,
+      "step": 2905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9689556637684237e-07,
+      "loss": 5.2309,
+      "step": 2906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9483517457776436e-07,
+      "loss": 4.4319,
+      "step": 2907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9278551387122312e-07,
+      "loss": 5.0689,
+      "step": 2908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9074658650043764e-07,
+      "loss": 4.6287,
+      "step": 2909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8871839469687291e-07,
+      "loss": 5.4241,
+      "step": 2910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8670094068025e-07,
+      "loss": 4.9413,
+      "step": 2911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8469422665853609e-07,
+      "loss": 4.7419,
+      "step": 2912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8269825482794546e-07,
+      "loss": 5.0517,
+      "step": 2913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8071302737293294e-07,
+      "loss": 4.8835,
+      "step": 2914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7873854646619838e-07,
+      "loss": 4.9955,
+      "step": 2915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.767748142686776e-07,
+      "loss": 4.9119,
+      "step": 2916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.748218329295426e-07,
+      "loss": 5.1046,
+      "step": 2917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7287960458620134e-07,
+      "loss": 4.4443,
+      "step": 2918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7094813136428913e-07,
+      "loss": 4.4292,
+      "step": 2919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6902741537767607e-07,
+      "loss": 4.4239,
+      "step": 2920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6711745872845518e-07,
+      "loss": 5.1637,
+      "step": 2921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6521826350694548e-07,
+      "loss": 4.4102,
+      "step": 2922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.633298317916887e-07,
+      "loss": 4.6981,
+      "step": 2923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.61452165649445e-07,
+      "loss": 4.866,
+      "step": 2924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5958526713519495e-07,
+      "loss": 4.9145,
+      "step": 2925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.577291382921331e-07,
+      "loss": 4.9101,
+      "step": 2926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.558837811516667e-07,
+      "loss": 4.8436,
+      "step": 2927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5404919773341576e-07,
+      "loss": 4.7388,
+      "step": 2928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5222539004520754e-07,
+      "loss": 4.6109,
+      "step": 2929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5041236008307648e-07,
+      "loss": 5.1561,
+      "step": 2930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4861010983126202e-07,
+      "loss": 4.6975,
+      "step": 2931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4681864126220636e-07,
+      "loss": 5.0093,
+      "step": 2932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4503795633655004e-07,
+      "loss": 4.9559,
+      "step": 2933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4326805700313307e-07,
+      "loss": 5.0615,
+      "step": 2934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4150894519899262e-07,
+      "loss": 4.8253,
+      "step": 2935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3976062284935642e-07,
+      "loss": 4.9637,
+      "step": 2936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3802309186764619e-07,
+      "loss": 4.8426,
+      "step": 2937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3629635415547182e-07,
+      "loss": 4.5402,
+      "step": 2938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.345804116026328e-07,
+      "loss": 4.4998,
+      "step": 2939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3287526608711132e-07,
+      "loss": 4.8859,
+      "step": 2940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3118091947507461e-07,
+      "loss": 5.2184,
+      "step": 2941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2949737362087156e-07,
+      "loss": 4.6338,
+      "step": 2942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.278246303670294e-07,
+      "loss": 4.8201,
+      "step": 2943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2616269154425264e-07,
+      "loss": 4.868,
+      "step": 2944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2451155897142186e-07,
+      "loss": 3.9651,
+      "step": 2945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2287123445559045e-07,
+      "loss": 4.6345,
+      "step": 2946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.212417197919824e-07,
+      "loss": 5.2994,
+      "step": 2947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1962301676399335e-07,
+      "loss": 4.7162,
+      "step": 2948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1801512714318286e-07,
+      "loss": 5.2412,
+      "step": 2949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1641805268927886e-07,
+      "loss": 5.2322,
+      "step": 2950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1483179515017095e-07,
+      "loss": 5.1337,
+      "step": 2951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1325635626191267e-07,
+      "loss": 4.884,
+      "step": 2952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 4.5525,
+      "step": 2953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1013794132294641e-07,
+      "loss": 5.2024,
+      "step": 2954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0859496868513397e-07,
+      "loss": 5.2145,
+      "step": 2955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0706282152395776e-07,
+      "loss": 4.9454,
+      "step": 2956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0554150151624864e-07,
+      "loss": 5.2743,
+      "step": 2957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0403101032698815e-07,
+      "loss": 4.5006,
+      "step": 2958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0253134960930832e-07,
+      "loss": 4.7099,
+      "step": 2959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.010425210044863e-07,
+      "loss": 5.0906,
+      "step": 2960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.956452614194534e-08,
+      "loss": 4.9366,
+      "step": 2961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.80973666392504e-08,
+      "loss": 4.757,
+      "step": 2962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.664104410210928e-08,
+      "loss": 5.6603,
+      "step": 2963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.519556012436815e-08,
+      "loss": 4.9403,
+      "step": 2964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.376091628801264e-08,
+      "loss": 5.3389,
+      "step": 2965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.233711416316571e-08,
+      "loss": 5.0345,
+      "step": 2966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.092415530807975e-08,
+      "loss": 4.9176,
+      "step": 2967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.952204126914332e-08,
+      "loss": 5.0991,
+      "step": 2968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.813077358087785e-08,
+      "loss": 5.1621,
+      "step": 2969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.675035376593088e-08,
+      "loss": 4.4437,
+      "step": 2970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.538078333507838e-08,
+      "loss": 5.3506,
+      "step": 2971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.402206378722467e-08,
+      "loss": 5.5727,
+      "step": 2972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.267419660939358e-08,
+      "loss": 4.336,
+      "step": 2973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.133718327673845e-08,
+      "loss": 5.3901,
+      "step": 2974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.00110252525299e-08,
+      "loss": 5.1694,
+      "step": 2975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.869572398815694e-08,
+      "loss": 5.224,
+      "step": 2976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.73912809231292e-08,
+      "loss": 4.6915,
+      "step": 2977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.609769748507135e-08,
+      "loss": 5.0109,
+      "step": 2978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.481497508972313e-08,
+      "loss": 5.2075,
+      "step": 2979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.354311514093825e-08,
+      "loss": 4.986,
+      "step": 2980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.228211903068105e-08,
+      "loss": 4.9793,
+      "step": 2981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.103198813902867e-08,
+      "loss": 4.6875,
+      "step": 2982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.979272383416335e-08,
+      "loss": 4.7929,
+      "step": 2983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.856432747237684e-08,
+      "loss": 4.8525,
+      "step": 2984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.734680039806707e-08,
+      "loss": 5.1488,
+      "step": 2985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.61401439437348e-08,
+      "loss": 5.2169,
+      "step": 2986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.494435942998367e-08,
+      "loss": 4.9253,
+      "step": 2987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.375944816552016e-08,
+      "loss": 4.7658,
+      "step": 2988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.258541144714914e-08,
+      "loss": 4.7215,
+      "step": 2989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.142225055977614e-08,
+      "loss": 4.5906,
+      "step": 2990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.026996677640062e-08,
+      "loss": 4.978,
+      "step": 2991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.912856135812051e-08,
+      "loss": 4.6466,
+      "step": 2992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.799803555412653e-08,
+      "loss": 5.4678,
+      "step": 2993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6878390601704525e-08,
+      "loss": 5.032,
+      "step": 2994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.576962772622874e-08,
+      "loss": 4.5018,
+      "step": 2995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.467174814116738e-08,
+      "loss": 5.382,
+      "step": 2996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3584753048073756e-08,
+      "loss": 4.8314,
+      "step": 2997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2508643636590694e-08,
+      "loss": 4.9055,
+      "step": 2998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.144342108444944e-08,
+      "loss": 4.7875,
+      "step": 2999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0389086557463e-08,
+      "loss": 4.6773,
+      "step": 3000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.934564120952945e-08,
+      "loss": 5.017,
+      "step": 3001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.831308618263086e-08,
+      "loss": 4.4307,
+      "step": 3002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.729142260682662e-08,
+      "loss": 5.1368,
+      "step": 3003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.628065160026118e-08,
+      "loss": 4.7363,
+      "step": 3004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.528077426915412e-08,
+      "loss": 5.0376,
+      "step": 3005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.42917917078034e-08,
+      "loss": 5.61,
+      "step": 3006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.331370499858545e-08,
+      "loss": 5.0544,
+      "step": 3007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.234651521194844e-08,
+      "loss": 4.5339,
+      "step": 3008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.139022340641785e-08,
+      "loss": 4.3246,
+      "step": 3009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0444830628589835e-08,
+      "loss": 4.9655,
+      "step": 3010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9510337913134524e-08,
+      "loss": 4.5874,
+      "step": 3011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.858674628278825e-08,
+      "loss": 4.697,
+      "step": 3012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7674056748362445e-08,
+      "loss": 5.1684,
+      "step": 3013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.677227030873476e-08,
+      "loss": 5.4584,
+      "step": 3014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.588138795084795e-08,
+      "loss": 4.8617,
+      "step": 3015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.500141064971318e-08,
+      "loss": 4.7881,
+      "step": 3016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4132339368405654e-08,
+      "loss": 5.3129,
+      "step": 3017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.327417505806785e-08,
+      "loss": 4.9873,
+      "step": 3018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.242691865790071e-08,
+      "loss": 4.5247,
+      "step": 3019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.159057109516916e-08,
+      "loss": 4.7222,
+      "step": 3020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.076513328520103e-08,
+      "loss": 4.932,
+      "step": 3021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9950606131380344e-08,
+      "loss": 5.1529,
+      "step": 3022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9146990525152906e-08,
+      "loss": 5.4525,
+      "step": 3023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8354287346022967e-08,
+      "loss": 4.8732,
+      "step": 3024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7572497461549884e-08,
+      "loss": 5.1677,
+      "step": 3025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6801621727350347e-08,
+      "loss": 5.2673,
+      "step": 3026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.604166098709504e-08,
+      "loss": 5.435,
+      "step": 3027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5292616072510878e-08,
+      "loss": 4.8598,
+      "step": 3028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4554487803376547e-08,
+      "loss": 5.0177,
+      "step": 3029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3827276987524738e-08,
+      "loss": 4.0303,
+      "step": 3030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.311098442083659e-08,
+      "loss": 5.0516,
+      "step": 3031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2405610887247243e-08,
+      "loss": 4.6765,
+      "step": 3032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.171115715874139e-08,
+      "loss": 4.5651,
+      "step": 3033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1027623995351077e-08,
+      "loss": 5.12,
+      "step": 3034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0355012145156784e-08,
+      "loss": 5.0383,
+      "step": 3035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9693322344287447e-08,
+      "loss": 4.737,
+      "step": 3036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9042555316917123e-08,
+      "loss": 4.6982,
+      "step": 3037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8402711775268312e-08,
+      "loss": 4.8019,
+      "step": 3038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7773792419606418e-08,
+      "loss": 5.2036,
+      "step": 3039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7155797938239737e-08,
+      "loss": 4.922,
+      "step": 3040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.654872900752169e-08,
+      "loss": 5.1986,
+      "step": 3041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5952586291849703e-08,
+      "loss": 4.9693,
+      "step": 3042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.536737044366188e-08,
+      "loss": 5.4208,
+      "step": 3043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4793082103435885e-08,
+      "loss": 5.2676,
+      "step": 3044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4229721899693405e-08,
+      "loss": 4.8466,
+      "step": 3045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3677290448993463e-08,
+      "loss": 5.2775,
+      "step": 3046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3135788355934652e-08,
+      "loss": 5.3913,
+      "step": 3047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2605216213155135e-08,
+      "loss": 5.2533,
+      "step": 3048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2085574601330419e-08,
+      "loss": 4.6075,
+      "step": 3049
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1576864089172247e-08,
+      "loss": 4.8939,
+      "step": 3050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.107908523343193e-08,
+      "loss": 5.3072,
+      "step": 3051
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0592238578892577e-08,
+      "loss": 4.9453,
+      "step": 3052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0116324658376864e-08,
+      "loss": 4.3396,
+      "step": 3053
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.651343992740369e-09,
+      "loss": 4.7392,
+      "step": 3054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.197297090872692e-09,
+      "loss": 4.7345,
+      "step": 3055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.754184449697444e-09,
+      "loss": 4.4741,
+      "step": 3056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.322006554171147e-09,
+      "loss": 5.3617,
+      "step": 3057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.900763877284335e-09,
+      "loss": 4.7998,
+      "step": 3058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.490456880058229e-09,
+      "loss": 5.0834,
+      "step": 3059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.091086011546955e-09,
+      "loss": 5.2989,
+      "step": 3060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.702651708834218e-09,
+      "loss": 4.6822,
+      "step": 3061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.325154397035515e-09,
+      "loss": 4.6773,
+      "step": 3062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.958594489295921e-09,
+      "loss": 5.0517,
+      "step": 3063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6029723867900845e-09,
+      "loss": 5.1481,
+      "step": 3064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.258288478724449e-09,
+      "loss": 4.6049,
+      "step": 3065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.924543142329486e-09,
+      "loss": 4.2156,
+      "step": 3066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.601736742869678e-09,
+      "loss": 5.3152,
+      "step": 3067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2898696336335364e-09,
+      "loss": 5.1463,
+      "step": 3068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.988942155939146e-09,
+      "loss": 4.783,
+      "step": 3069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.698954639129726e-09,
+      "loss": 5.0774,
+      "step": 3070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.419907400579181e-09,
+      "loss": 5.8045,
+      "step": 3071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1518007456854404e-09,
+      "loss": 5.2759,
+      "step": 3072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.894634967872678e-09,
+      "loss": 5.4237,
+      "step": 3073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.648410348592423e-09,
+      "loss": 4.4464,
+      "step": 3074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4131271573191172e-09,
+      "loss": 5.3268,
+      "step": 3075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.188785651555669e-09,
+      "loss": 4.877,
+      "step": 3076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.975386076829011e-09,
+      "loss": 4.7611,
+      "step": 3077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.772928666690099e-09,
+      "loss": 4.8033,
+      "step": 3078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5814136427139137e-09,
+      "loss": 4.6697,
+      "step": 3079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4008412145027904e-09,
+      "loss": 4.8935,
+      "step": 3080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2312115796797586e-09,
+      "loss": 4.8522,
+      "step": 3081
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0725249238940916e-09,
+      "loss": 5.1664,
+      "step": 3082
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.24781420816867e-10,
+      "loss": 4.5877,
+      "step": 3083
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.879812321420765e-10,
+      "loss": 4.7784,
+      "step": 3084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.621245075910665e-10,
+      "loss": 4.8646,
+      "step": 3085
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.472113849036564e-10,
+      "loss": 5.0488,
+      "step": 3086
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.432419898459106e-10,
+      "loss": 5.0274,
+      "step": 3087
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5021643620347655e-10,
+      "loss": 4.4802,
+      "step": 3088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.681348257882466e-10,
+      "loss": 4.8142,
+      "step": 3089
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.969972484316962e-10,
+      "loss": 5.1006,
+      "step": 3090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3680378198932532e-10,
+      "loss": 4.9078,
+      "step": 3091
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.755449234065794e-11,
+      "loss": 4.8671,
+      "step": 3092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9249433383691256e-11,
+      "loss": 4.8119,
+      "step": 3093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1888647040446685e-11,
+      "loss": 5.2319,
+      "step": 3094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.472163256969865e-12,
+      "loss": 4.5746,
+      "step": 3095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 5.4125,
+      "step": 3096
+    },
+    {
+      "epoch": 1.0,
+      "step": 3096,
+      "total_flos": 1.7535635455433328e+17,
+      "train_loss": 5.217048635470467,
+      "train_runtime": 4635.7484,
+      "train_samples_per_second": 42.745,
+      "train_steps_per_second": 0.668
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3096,
+  "num_train_epochs": 1,
+  "save_steps": 2000,
+  "total_flos": 1.7535635455433328e+17,
+  "trial_name": null,
+  "trial_params": null
+}